Productivity Report 2019 FIRM PRODUCTIVITY AND ECONOMIC GROWTH IN TURKEY ECONOMIC INTEGRATION INNOVATION HUMAN CAPITAL COMPETITION & BUSINESS REGULATION ECONOMIC GROWTH & PRODUCTIVITY INDUSTRIAL CAPACITY COUNTRY ECONOMIC MEMORANDUM: FIRM PRODUCTIVITY AND ECONOMIC GROWTH IN TURKEY The Country Economic Memorandum (CEM) on Firm Productivity and Economic Growth in Turkey was prepared under the guidance of Johannes Zutt (Former WB Country Director, Turkey), John Panzer (Acting Senior Director, Macroeconomics Trade and Investment), Lalita M. Moorty (Practice Director, Macroeconomics, Trade and Investment), Auguste Tano Kouame (WB Country Director, Turkey), Sandeep Mahajan (Practice Manager, MTI) by a core team led by Habib Rab (EFI Program Leader, Turkey) and Pinar Yasar (Country Economist, MTI GP) and including Gonzalo J. Varela (Senior Economist, MTI GP), David C. Francis (Senior Economist, WB Development Economics Group), Raian Divanbeigi (Senior Economist, WB DEC), Umut Kilinc (Economist, Finance, Competitiveness and Innovation GP), Mattia Makovec (Economist, Social Protection and Jobs). Excellent research analysis was provided by Erdem Atas (Research Analyst, MTI GP), Beyza Polat (Consultant, MTI GP), Stefania Lovo (Consultant, MTI GP) and Aysenur Acar (Consultant, SPJ GP). The team is thankful to Donato De Rosa (Lead Economist, MTI GP), Jose Ramon Perea (Senior Economist, MTI GP), Siddharta Raja (Senior ICT Policy Specialist, SPL GP), Xavier Cirera (Senior Economist, FCI GP), and Ana Paula Cusolito (Senior Economist, FCI GP) for their advice. The team is grateful for inputs from P. Facundo Cuevas (Senior Economist, POV GP), Leonardo Ramiro Lucchetti (Senior Economist, POV GP), Metin Nebiler (Research Analyst, POV GP). N. Nergiz Dincer (Consultant, MTI GP), Ayca Tekin Koru (Consultant, MTI GP), Thomas E. Haven (Senior Private Sector Specialist, FCI GP), Ximena V. Del Carpio (HD Program Leader, Turkey), Sirma Demir Seker (Economist, Social Protection and Jobs GP), Laurent Bossavie (Economist, SPJ GP). Pinar Baydar (Senior Program Assistant) and Selma Karaman (Program Assistant) provided excellent administrative support. Tunya Celasin Aydinalp (Senior Communications Officer) led all communications support. The team thanks Martin Raiser (WB Country Director, Brazil), Sebastian Eckardt (EFI Program Leader, Vietnam), Rafael Munoz Moreno (EFI Program Leader, Brazil) and Karlis Smits (Senior Country Economist, MTI GP) for their peer review comments and advice. The team also thanks Mariana lootty De Paiva Dias (Senior Economist, MTI GP) and Martha Martinez Licetti (Lead Economist, MTI GP) for their guidance. The team is very grateful to colleagues from the Strategy and Budget Unit in the Presidency (former Ministry of Development), the Ministry of Industry and Technology (former Ministry of Science, Industry and Technology), the Central Bank of the Republic of Turkey, the Ministry of Treasury and Finance (former Prime Ministry Under-Secretariat of Treasury; and former Ministry of Finance) and the Turkish Industry and Business Association (TUSIAD) for discussions on productivity and policy priorities. The team is thankful for guidance and assistance to colleagues in the department for the Entrepreneur Information System at the Ministry of Industry and Technology. The CEM is a product of the staff of the World Bank Group. The findings, interpretations, and conclusions expressed in this report do not necessarily reflect the views of the Executive Directors of the World Bank (or the governments they represent), or the Government of the Republic of Turkey. © 2019 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750- 4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey DRAFT – NOT FOR DISCLOSURE OR CITATION "A;<=;=G! ! Figure 1: Turkey’s economic convergence continued to accelerate after the Global Financial Crisis...................3 Figure 2: This enabled Turkey to catch up with high performers and pull away from trapped MICs ..................3 Figure 3: Per capita growth in Turkey is highly volatile particularly relative to high performers...........................4 Figure 4 This is associated with volatility in consumption and investment...............................................................4 Figure 5: Investment efficiency in Turkey is closer to Trapped MICs than high performers ................................5 Figure 6: Whilst private consumption levels are higher than Trapped MICs and high performers ......................5 Figure 7: Macro stability and reforms accelerated capital inflows to plug a large savings gap ...............................5 Figure 8: But composition of external flows has shifted from FDI to volatile portfolio flows .............................5 Figure 9: Credit to the private sector has increasingly become procyclical after 2009 ............................................6 Figure 10: Driven in big part by a sharp increase in debt creating flows to Turkish banks ....................................6 Figure 11: Procyclical credit means tighter access and higher cost of financing during downturns ......................7 Figure 12: Stable employment and rising value-added of industry suggest rising labor productivity ....................7 Figure 13: Trends in industrial employment and value-added shares during UMIC phase ....................................8 Figure 14: Productivity growth in Turkey after the GFC has declined ......................................................................9 Figure 15: Sources of productivity gains and their macro-micro links .................................................................... 10 Figure 16: TFP growth among high performers during their UMIC stage faster than Turkey’s ........................ 11 Figure 17: Per capita GDP growth since 2008 driven by service employment, LFP and demographic shifts . 13 Figure 18: Potential growth has been on declining trend in Turkey since the GFC ............................................. 14 Figure 19: This is largely due to falling potential TFP growth, which is a drag on supply capacity ................... 14 Figure 20: Services are experiencing declining marginal productivity of labor ...................................................... 15 Figure 21: Which is driven by increased employment in low knowledge-intensive sectors ................................. 15 Figure 22: Turkey scores well on connectedness and capabilities but lags on competitiveness .......................... 19 Figure 23: Turkey lags most on labor markets, innovation, financial sector, human capital ............................... 20 Figure 24: Gaps between Turkey and EU average greatest for human capital and labor market ....................... 20 Figure 25: Allocative inefficiencies perpetuated by the banking sector................................................................... 21 Figure 26: Labor productivity and TFP in manufacturing, construction and services on a declining path ...... 25 Figure 27: Manufacturing productivity remains flat whereas construction and services are low and declining 25 Figure 28: Employment growth exceeded growth in value added in construction and services ........................ 26 Figure 29: TFP levels in tradable sectors are higher than non-tradable sectors..................................................... 27 Figure 30: Manufacturing growth constrained by large, low productivity sectors ................................................. 27 Figure 31: Allocative efficiency in manufacturing positive but declining .............................................................. 28 Figure 32: Most efficient sectors have high TFP but experienced a drop in allocative efficiency ..................... 29 Figure 33: Within firm productivity growth helps partially offset negative reallocation and net entry .............. 30 Figure 34: Negative reallocation and net entry effects are evident within manufacturing ................................... 31 Figure 35: Largest firms are more productive, but their productivity growth is slowing ..................................... 32 Figure 36: This may be due to capital deepening among the largest firms ............................................................. 32 Figure 37: TFP across all construction sub-sectors is low relative to manufacturing ........................................... 35 Figure 38: Construction of buildings is most productive with highest VA and employment shares ................. 35 Figure 39: Allocative efficiency in construction has declined together with TFP ................................................. 36 Figure 40: Allocative inefficiency increased in the largest sub-sector ...................................................................... 36 Figure 41: Net entry mostly contributing negatively to construction sector TFP ................................................. 37 Figure 42: Net entry effect is positive only in civil engineering sub-sector ............................................................ 37 Figure 43: TFP in service is low, including within knowledge-intensive sectors ................................................... 38 Figure 44: Poor performance of relatively high KI sectors affects sectors beyond services ............................... 38 Figure 45: Allocative efficiency in services has declined together with TFP .......................................................... 39 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Figure 46: Allocative inefficiencies in important sectors like transport and ICT .................................................. 39 Figure 47: Within firm productivity contributing negatively to service sector TFP.............................................. 40 Figure 48: Within services, sources of productivity changes varies across sectors................................................ 40 Figure 49: Turkey’s merchandise exports grew relatively fast in the period 2000-2015 ....................................... 50 Figure 50: Export growth experienced fast catch-up in the early period and sluggish growth after the GFC . 50 Figure 51: MENA share of exports has increased whilst EU share has decreased ............................................... 51 Figure 52: Turkish exporters reach almost all countries in the world ..................................................................... 51 Figure 53: Export concentration is low and decreasing over time ........................................................................... 52 Figure 54: Turkish exports are highly diversified in terms of products .................................................................. 52 Figure 55: The number of firms entering the export market has increased over time ......................................... 52 Figure 56: Turkish exports show a gradual substitution of low-tech for mid-tech exports ................................. 53 Figure 57: Turkish imports continue being concentrated in more tech-intensive sectors ................................... 53 Figure 58: Turkey’s participation into GVCs both as a buyer and seller has increased since early 2005 ........... 54 Figure 59: Foreign value added is increasingly more important to produce Turkish export products .............. 54 Figure 60: FDI stocks have increased across all sectors ............................................................................................ 54 Figure 61: Turkish exporters enjoy a productivity premium comparable to advanced economies’ ................... 57 Figure 62: Turkish importers show higher productivity growth rates than non-importers ................................. 58 Figure 63: Turkish exporters show higher productivity growth rates than non-exporters .................................. 58 Figure 64: Exporters are more productive independently of their size ................................................................... 59 Figure 65: Firms’ productivity increases after entering the export market ............................................................. 59 Figure 66: The size of first-time manufacturing exporters decreasing over time .................................................. 60 Figure 67: Export intensity and backward linkages linked to within-firm productivity growth .......................... 60 Figure 68: Import intensity and backward linkages linked to within-firm productivity growth .......................... 60 Figure 69: Less FDI restrictions induced reallocation of inputs to less restricted sectors ................................... 61 Figure 70: Sectors with stronger forward linkages show a greater response of within-firm TFP growth to increased liberalization .................................................................................................................................................... 62 Figure 71: Sectors with stronger backward linkages show a greater response of within-firm TFP growth to increased liberalization from the import side .............................................................................................................. 62 Figure 72: Productivity gains from reduced trade costs vary across sectors .......................................................... 65 Figure 73: Productivity gains from reduced trade costs are higher for larger firms and firms with greater absorptive capacity ........................................................................................................................................................... 66 Figure 74: Manufacturing sector: productivity gains from FDI in upstream sectors are greater for medium and large firms. Firms gain mostly from FDI in upstream service sectors .................................................................... 69 Figure 75: Manufacturing sector: productivity gains from FDI in upstream sectors are greater for exporters 69 Figure 76: Services: productivity gains from FDI in upstream sectors are greater for medium and large firms. Firms gain mostly from FDI in upstream non-service sectors ................................................................................. 70 Figure 77: Service sector: productivity gains from FDI in upstream sectors are greater for exporters. ............ 70 Figure 78: Outline of innovation support and firm performance analysis ............................................................. 74 Figure 79: Turkey’s R&D expenditures are catching up with those of peers ......................................................... 76 Figure 80: The private sector’s share of R&D expenditure is close to the 10th NDP target................................ 76 Figure 81: The boost in R&D spending has increased demand for researchers in Turkey.................................. 76 Figure 82: Patent applications remain strong, though like elsewhere have slowed down post GFC ................. 76 Figure 83: IP use charges for Turkey are low .............................................................................................................. 77 Figure 84: Innovation performance on declining trend ............................................................................................. 77 Figure 85: Largest innovation gaps relative to peers include university-company collaboration in R&D, private R&D spending, and quality of research institutions ................................................................................................... 78 Figure 86: KOSGEB grants in evenly distributed across firms regardless of sub-sector ..................................... 80 Figure 87: TUBITAK grants within manufacturing are more targeted to technology intensive industries ...... 81 Figure 88: Youngest firms exhibit significantly higher innovation performance................................................... 82 Figure 89: Young firms that survive the start-up period tend to be more innovative than other firms ............ 82 Figure 90: Innovative firms in both manufacturing and services display higher employment growth .............. 83 Figure 91: KOSGEB has a relatively large positive employment impact in manufacturing ................................ 88 Figure 92: KOSGEB employment impact is also strong in services ....................................................................... 89 Figure 93: TUBITAK has a large and significant impact on R&D across manufacturing ................................... 90 Figure 94: TUBITAK also impacts on R&D in services, whereas KOSGEB support is less significant.......... 90 Figure 95: TUBITAK has a relatively large and significant impact on innovation in manufacturing ................ 91 Figure 96: Impact of public support on innovation in services is small and not very significant ....................... 92 Figure 97: R&D and innovation have a positive and significant impact on TFP in manufacturing .................. 93 Figure 98: The impact of R&D and innovation on service firms’ TFP is also significant ................................... 93 Figure 99: TUBITAK impacts significantly on manufacturing TFP, unlike KOSGEB....................................... 94 Figure 100: Turkey has a rapidly growing working-age population ......................................................................... 98 Figure 101: And relatively low age-dependency, which will last till 2025 ............................................................... 98 Figure 102: Share of workers with low education attainment remains high ........................................................... 99 Figure 103: Some rebalancing recently through high employment growth among more educated ................... 99 Figure 104: Unemployment across levels of education is relatively broad based .................................................. 99 Figure 105: But unemployment among more educated workers is growing more rapidly................................... 99 Figure 106: A large share of 15-29-year-olds are not in employment, education or training............................. 100 Figure 107: Most common field of study among NEET include more technical subjects ................................ 100 Figure 108: Proficiency gaps between Turkey and OECD are smaller among young adults ............................ 102 Figure 109: Steady improvement in PISA scores, but sharp drop in 2015 ........................................................... 103 Figure 110: Performance on reading scale low but in line with trend when controlling for per capita GDP 103 Figure 111: TIMSS math scores have improved but Turkey’s ranking has fallen ............................................... 104 Figure 112: TIMSS science scores have improved whilst Turkey’s ranking has remained the same................ 104 Figure 113: Non-routine manual physical and routine manual skills are becoming less dominant .................. 105 Figure 114: Among less educated, employment mostly in routine tasks at risk of mechanization ................... 105 Figure 115: The proportion of firms offering training in Turkey is relatively low .............................................. 106 Figure 116: The nature of demand and supply mismatches across sectors highly heterogeneous ................... 106 Figure 117: Real wage growth for low skill workers nearly double that of high-skill workers .......................... 107 Figure 118: Secondary graduates’ earnings increasing over time in the formal sector ........................................ 108 Figure 119: Workers in the informal sector derive little benefit from achieving higher education .................. 108 Figure 120: Wage premium for university-educated workers has fallen most recently compared to primary educated workers............................................................................................................................................................ 109 Figure 121: Firms characterized by low skill labor make up a large share of employment, which has grown since 2006 ........................................................................................................................................................................ 111 Figure 122: Employment in manufacturing predominantly in routine manual (low skill) ................................. 112 Figure 123: Employment in wholesale/retail mostly in non-routine manual (medium skill) ............................ 112 Figure 124: No clear association between productivity and incidence of high skilled workers ........................ 112 Figure 125: Firms with higher incidence of employment in non-routine cognitive face sharpest TFP drop . 112 Figure 126: Small firms exhibit low and falling productivity across all skill intensities ...................................... 113 Figure 127: Large firms exhibit higher productivity particularly those with medium-skill labor ...................... 113 Figure 128: Manufacturing firms’ productivity relatively flat across all skill intensities ...................................... 113 Figure 129: Wholesale/retail firms’ productivity on declining trend across all skill intensities ......................... 113 Figure 130: Local competition in Turkey is high relative to comparator countries ............................................ 117 Figure 131: On competition policy however, Turkey falls behind comparator countries .................................. 117 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Figure 132: Manufacturing concentration declining though remains high overall .............................................. 118 Figure 133: Services concentration is lower and has remained relatively flat ....................................................... 118 Figure 134: Construction has even lower concentration but has also declined over the sample period ......... 118 Figure 135: Lower employment concentration but also lower labor productivity in manufacturing ............... 121 Figure 136: Increased employment concentration in low skilled services drags down overall productivity ... 121 Figure 137: Labor share in value added is rising across all sectors ........................................................................ 122 Figure 138: Labor share in high skill services and construction driven by wage and jobs respectively ........... 123 Figure 139: Labor share in high and low skill manufacturing driven by wage and jobs respectively ............... 123 Figure 140: Sharp drop in markups across all services sub-sectors ....................................................................... 124 Figure 141: Sharpest drop in markups in the high-tech sub-sectors sectors within manufacturing ................. 124 Figure 142: OECD Product Market Regulation ....................................................................................................... 127 Figure 143: Turkey is catching up on regulatory reforms as measured by PMR ................................................. 128 Figure 144: Though remains behind recent HIC graduates and other OECD economies................................ 128 Figure 145: With some indication that progress in most recent years has slowed down ................................... 128 Figure 146: Turkey performs well on business entry regulations ........................................................................... 129 Figure 147: Though not very well on regulations for dealing with insolvency .................................................... 129 Figure 148: The private sector in Turkey is broadly adding firms.......................................................................... 130 Figure 149: The net rate of firm entry is declining ................................................................................................... 131 Figure 150: The effect of net entry on productivity has been negative ................................................................. 132 Figure 151: Production stages of the value chain where services are added ........................................................ 134 Y@G=J@?<@A;G! ! 3Cs Competitiveness, Capability and Connectedness ACF Ackerberg, Caves and Frazer ALMP Active Labor Market Policies ARG Argentina B40 Bottom 40 BRA Brazil C/I Consumption-to-investment ratio CAGR Compound annual growth rate CBRT The Central Bank of the Republic of Turkey CEM Country Economic Memorandum CET Central European Time CHL Chile CR20 Concentration of top twenty CR4 Concentration of top four CU Customs Union CZE Czech Republic DB Doing Business DEC Development Economics Vice Presidency EFI Equitable Growth, Finance and Institutions EFTA European Free Trade Association EIS Entrepreneur Information System ENR Engineering News Record EPO European Patent Office EU European Union FCI Finance, Competitiveness and Innovation FDI Foreign direct investment FTA Free trade agreement FX Foreign exchange GATS General Agreement on Trade in Services GCI Global Competitiveness Index GCR Global Competitiveness Report GDP Gross Domestic Product GFC Global Financial Crisis GP Global Practice GVCs Global Value Chains HD Human Development HHI Herfindahl-Hirshchman Index HIC High Income Countries HP Hodrick-Prescott HUN Hungary ICT Information and Communication Technologies IMF International Monetary Fund IO Input-Output IPR Intellectual Property Rights ISCO International Standard Classification of Occupations ISKUR Turkish Employment Agency IT Information Technology K/L Capital-to-Labor ratio KI Knowledge-intensive KIS Knowledge-intensive services KOR Republic of Korea Administration for supporting and developing small and medium size KOSGEB enterprises in Turkey LevPet Levinsohn and Petrin LFP Labor Force Participation LHS Left Hand Side M&E Machinery and Equipment MAX Maximum MENA Middle-East and North Africa MEX Mexico MFN Most-Favored-Nation MIC Middle Income Countries MIN Minimum MOIT Ministry of Industry and Technology MP Melitz-Polanec MPK Marginal Product of Capital MPL Marginal Product of Labor MTI Macroeconomics, Trade and Investment MYS Malaysia NACE Statistical classification of economic activities in the European Community NCES US National Center for Education Statistics NDP National Development Plan NEET Not in Education, Employment, or Training NEP New Economic Program NGO Non-Governmental Organizations OECD Organization for Economic Co-operation and Development OP Olley and Pakes PIAAC OECD Survey of Adult Skills PISA Program for International Student Assessment PMR Product Market Regulation POL Poland POV Poverty PPP Purchasing power parity PSM-DiD Propensity score matching and difference in differences R&D Research and development R&P Rubber and Plastics RHS Right hand side RICA Regional Investment Climate Survey RUS Russia SME Small and medium size entrepreneurs SPJ Social Protection and Jobs STRI Services Trade Restrictiveness Index TDI Trade defense instruments TEPA Trade and Economic Partnership Agreement TFP Total factor productivity TFPQ Quantity total factor productivity TFPR Revenue total factor productivity TFRS Financial Reporting Standards of Turkey TIMMS Trends in International Mathematics and Science Study TIVA Trade in Value Added TL Turkish Lira TUBITAK The Scientific and Technological Research Council of Turkey TUN Tunisia TUR Turkey TURKPATENT Turkish Patent and Trademark Office TURKSTAT Turkish Statistical Institute TUSIAD Turkish Industry and Business Association UK United Kingdom UMIC Upper-middle income countries UN United Nations UNCTAD United Nationals Conference on Trade and Development US United States of America US$ United States of America Dollar VA Value added WB World Bank WBG World Bank Group WDI World Development Indicators WEF World Economic Forum ZAF South Africa EXECUTIVE SUMMARY COUNTRY ECONOMIC MEMORANDUM: FIRM PRODUCTIVITY AND ECONOMIC GROWTH IN TURKEY Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey "#$%&'(!)"#%#*+"!*)*#',%-$*.! "#$%&'(!)"#%#*+"!*)*#',%-$*.! "#$%&'(!)"#%#*+"!*)*#',%-$*.! /+'*!0'#-$"&+1+&(!,%-!)"#%#*+"!2'#3&4!+%!&$'5)(! /+'*!0'#-$"&+1+&(!,%-!)"#%#*+"!2'#3&4!+%!&$'5)(! /+'*!0'#-$"&+1+&(!,%-!)"#%#*+"!2'#3&4!+%!&$'5)(! )Z=FH<@J=!PHLL?>[! )Z=FH<@J=!PHLL?>[! )Z=FH<@J=!PHLL?>[! !! ! 0>AIHF<@J@<[!?;I!=FA;AL@F!Q>AVD=[! 0>AIHF<@J@<[!?;I!=FA;AL@F!Q>AVD=[! 0>AIHF<@J@<[!?;I!=FA;AL@F!Q>AVD=[! “Productivity “Productivity “Productivity isn't isn't isn't everything, everything, everything, but, but, inin but, inthe the the long longrun, run, long itit run, isit is is almost almost almost everything” everything” everything” Paul PaulPaul Krugman Krugman Krugman (1994) (1994) (1994) Turkey’s Turkey’s Turkey’s pace pace pace of of income of income income convergence convergence convergence has has has out out out of ofagriculture. of agriculture. agriculture. ThisThis This led to led led toaa to sharp a sharp sharp drop drop drop ininin the the the globally globally globally beenbeen been oneone one ofofof the the the mostmost most remarkable remarkable remarkable ofofof share share share of ofof the the the population population population withwith with per per capita per capita capita thethe the past past past fifteen fifteen fifteen years years years .. . Reforms Reforms Reforms starting starting starting in inthe in the the expenditure expenditure expenditure below below below the poverty the the poverty poverty lineline line($5.5($5.5 ($5.5 aa day a in day day inin early early early 2000s 2000s 2000s accelerated accelerated accelerated private private private activity activity activity and andand public public public 20112011 2011 PPP)PPP) PPP) from from from 373737 percent percent percent inin2002 in 2002 2002 tototo 99 9 percent percent percent service service service delivery. delivery. delivery. Per Per Per capita capita capita GDPGDP GDP hashas has trebled trebled trebled ininin 151515 inin2017. in 2017. 2017. Turkey Turkey Turkey experienced experienced experienced strong strong strong economic economic economic years, years, years, reaching reaching reaching US$10,500 US$10,500 US$10,500 inin2017 in 2017 2017 and and and resulting resulting resulting in inin convergence convergence convergence (Figure (Figure (Figure 1), 1), coming coming 1),coming close close close to to to oneone one of ofthe ofthethe biggest biggest biggest jumps jumps jumps in inper in per per capita capita capita income income income transitioning transitioning transitioning out of out out ofupper of upper upper middle-income middle-income middle-income status status status ininin rankings rankings rankings since 2002. since since 2002. 2002. Job Job Job creation, creation, creation, particularly particularly particularly ininin lessless than less than than 2020years 20 years years –– a–a a feat feat feat achieved achieved achieved only only only byby aa a by services, services, services, helped helped helped absorb absorb absorb aalot a lot lotofofof labor labor labor transitioning transitioning transitioning handful handful handful of ofof countries countries countries ininrecent in recent recent years. years. years. /@QH>=!9.!P<>A;Q!=FA;AL@F!FA;J=>Q=;F=! /@QH>=!9.!P<>A;Q!=FA;AL@F!FA;J=>Q=;F=! /@QH>=!9.!P<>A;Q!=FA;AL@F!FA;J=>Q=;F=! /@QH>=!7.!-=FB@;@;Q!FA;<>@XH<@A;!AE!K>AIHF<@J@<[! /@QH>=!7.!-=FB@;@;Q!FA;<>@XH<@A;!AE!K>AIHF<@J@<[! /@QH>=!7.!-=FB@;@;Q!FA;<>@XH<@A;!AE!K>AIHF<@J@<[! Convergence Convergence Convergence (Per (Per (Per capita capita capita GDP/US GDP/US GDP/US per per per capita capita capita GDP GDP GDP growth growth growth and and and contributions contributions contributions (%) (%)(%) GDP, GDP, GDP, PPP PPPPPP constant constant constant 2011 2011 2011 international international international $)$) $) 0,550,55 0,55 88 8 33 3 0,450,45 0,45 -2-2-2 0,350,35 0,35 -7-7-7 0,250,25 0,25 -12 -12 -12 89 89 89 911 913 3 935 95 957 97 979 99 091 01 031 03 035 5 057 07 079 099 1 111 3 3 135 155 2017 7 17 90 90 902 92 924 4 946 966 8 8 080 000 2 022 4 4 046 6 068 08 08 100 102 2012 124 2014 164 2016 16 99 99 9 99 99 99 09 00 00 0 00 00 0 01 1 01 1 01 1 01 19 19 19 19 19 19 19 19 19 19 19 19 29 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 99 99 9 99 9 99 9 09 0 00 0 00 0 00 0 00 01 01 01 01 19 19 19 19 19 19 19 19 19 19 29 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 1 1 1 1 1 1 2 2 2 2 2 2 2 2 TFPTFP TFP (gA) (gA) (gA) Capital Capital Capital stock stock stock *(* (!(! gK) ! * gK) gK) 1 1 1 1 1 2 2 2 2 2 2 2 2 High High High performers performers performers Trapped Trapped Trapped MICs MICs MICs Turkey Turkey Turkey Labor Labor Labor ((1- ((1- ! ((1- )! )!* gL) * * gL) ) gL) Real Real RealGDP GDPGDP(gY)(gY) (gY) gY gY gY (period (period (period average) average) average) Sources: Sources: Sources: WDI, WDI, WDI, WB WBWBStaff Staff Staff estimates. estimates. estimates. Sources: Sources: Sources: TURKSTAT, TURKSTAT, TURKSTAT, WDI, WDI, WDI,WBWB WB Staff Staff Staff estimates. estimates. estimates. Note: Note: Note:Convergence Convergence Convergence estimates estimates estimates for for for high high high performers performers performers andand and trapped trapped trapped MICs MICs MICs based based basedon ontheir on their weighted their weighted weighted averages averages averagesof per of of capita per per capita GDP. capita GDP. GDP. Sustaining Sustaining Sustaining growth growth growth and and and improvements improvements improvements ininliving in living living 2). 2). As 2). As aa As result, a result, result,potential potential potential output output output –– what – what what the the the standards standards standards ininTurkey in Turkey Turkey will will willrequire require require higher higher higher economy economy economy cancan can produce produce produce whenwhen when factor factor factor inputs inputs inputs are are are fully fully fully productivity productivity productivity inin the in the the economy economy economy .. Growth . Growth Growth since since since the the the utilized utilized utilized –– has – has flattened has flattened flattened out. Unless out. out. Unless Unless Turkey Turkey Turkey can can can 1980s 1980s 1980s was driven was was driven driven largely largely largely by by an by anincrease an increase increase ininlabor in labor labor produce produce produce moremore more and and and better better better output output output withwith with its itsavailable its available available and and and physical physical physical capital capital capital inputs inputs thanks inputs thanks thanks to toreforms to reforms reforms thatthat that inputs, inputs, inputs, the the return the return return onon those on inputs, those those inputs, inputs, including including including labor, labor, labor, lifted lifted liftedconstraints constraints constraints onon factor on factor factor mobility mobility mobility and reduced and and reduced reduced will stagnate. will will stagnate. stagnate. This This This can can can be be offset be offset offset byby deepening by deepening deepening market market market distortions. distortions. distortions. TheThe The contribution contribution contribution of of of capacity capacity capacity ininindustry in industry industry and and services and services services that that have that have have strong strong strong productivity productivity productivity onon the on the the other other other handhand hand has been has has been been relatively relatively relatively pro-development pro-development pro-development characteristics characteristics characteristics (i.e. (i.e.potential (i.e. potential potential to toto lessless lessand andon andon aa on declining a declining declining path path in path inrecent in years recent recent years years (Figure (Figure (Figure boost boost boost productivity, productivity, productivity, growth growth growth and and jobs). and jobs). jobs). EXECUTIVE SUMMARY help the economy tide over the difficult period This will require a rebalancing of reforms ahead. away from short-term demand management towards long-term structural reforms. Turkey Policies adopted in mid-2018 helped stabilize has weathered several economic shocks since the financial markets and reduce external Global Financial Crisis in 2008. Strong vulnerabilities. Monetary and fiscal policies were institutions, accumulation of fiscal and financial tightened to contain external and internal sector buffers since 2002, and post-GFC global imbalances, whilst the Central Bank policies monetary expansion, enabled effective policy helped relieve liquidity pressures in the financial stimuli that helped avert a prolonged recession. At system. In addition, the authorities released a the same time, short-term policies without long- medium-term New Economic Program (NEP) in term structural reform diminish the sustainability September 2018, which set out several measures of growth. Demand incentives and supply to address immediate vulnerabilities and longer- subsidies provide temporary relief to consumers term productivity. Though developments in the and producers, but in the absence of productivity early part of 2019 highlight ongoing gains they increase chances of boom-bust cycles vulnerabilities. once incentives and subsidies are withdrawn. Longer-term productivity measures include A necessary precondition for productivity structural reforms to deepen supply side growth is macroeconomic stability. capacity. Relative to high performing Macroeconomic imbalances in Turkey intensified comparators, Turkey’s structural reform gaps are economic stress starting in mid-2018. This led to greatest in labor markets, innovation, financial an economic downturn in 2018-19, which can sector, and human capital. Relative to the EU, dent productivity. Consistent economic policies Turkey also faces reform gaps in business are important to support an orderly adjustment. regulations. Weak and/or poor implementation of This means sustaining tight monetary policy, policies and institutions in these areas impact complemented by financial sector policies that negatively on productivity; they can prevent enable gradual deleveraging and enhance financial resources from shifting to higher productivity risk monitoring and management. Critical to sectors and firms, and/or prevent firms from deleveraging is a strong corporate debt investing in productivity enhancing skills and restructuring framework, the absence of which technology. This study examines the impact of could spell the difference between a soft and a structural reforms on productivity in Turkey and hard landing for the economy. Targeted and finite their implications for policy and institutional countercyclical fiscal policies will be important to reform. ,;?B[\@;Q!L@F>ATL?F>A!K>AIHF<@J@<[!B@;D?Q=G!! “The growth of GDP may be measured up in the macroeconomic treetops, but all the action is in the microeconomic undergrowth, where new limbs sprout, and dead wood is cleared away.” The Growth Report: Strategies for Sustained and Inclusive Growth (2008) The impact of structural reforms on freeing up resources for more productive firms, be productivity in Turkey is analyzed using firm- it new entrants or incumbents; (4) that in turn level data. The objective is to link micro analysis help expand the sector; and (5) support structural of firm and sector productivity to macro analysis transformation as factors of production move into of economic growth drivers and challenges in expanding, higher productivity industries, and out Turkey. Micro-macro productivity dynamics can of low productivity ones; (6) a combination of be summarized as follows (Figure 3): (1) raising which accelerates economy-wide productivity and productivity within firms; (2) can force growth. unproductive firms to exit the market; thereby (3) Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!S.!*?F>ATL@F>A!K>AIHF<@J@<[!B@;D?Q=G! ! Manufacturing, services 6. Economy wide and construction Drivers and productivity growth productivity (chapter 2) challenges of through structural transformation and within economic growth sector productivity (chapter 1) 3. Entry of more productive firms through conducive Competition investment climate 4. Within sector 5. Structural and productivity growth as transformation through regulation resources shift to more factor inputs shifting from (chapter 6) efficient firms whose low to high productivity market shares increase sectors 2. Exit of less productive firms through efficient institutions 1. Within firm Economic integration productivity (chapter 3); public sector growth through incentives (chapter 4); technical capacity, human capital (chapter 5) innovation The first two chapters in this study provide a economic growth in Turkey. This is broken down diagnostic of macro-micro productivity further to analyze how within firm productivity is linkages in Turkey, whilst the subsequent four affected by: economic integration as a source of chapters analyze how those dynamics are learning and technology for firms in chapter 3; the affected by structural reforms. To illustrate quality of business support services to promote (Figure 3), chapter 1 analyzes drivers and firm growth and innovation in chapter 4; and challenges of growth in Turkey to take stock human capital as a source of firm capability in economy wide/macro level productivity trends, chapter 5. Finally, chapter 6 looks at how including from structural transformation. Chapter competition and business regulations impact the 2 disentangles productivity trends within economy’s ability to reallocate resources to more manufacturing, construction and services using productive firms and sectors. firm-level data, to see how these add up to 0>ATI=J=BAKL=;G!?;I!L@F>ATL?F>A!K>AIHF<@J@<[!I[;?L@FG A key question for micro-macro productivity In Turkey, firm level analysis shows that dynamics is how, at micro-level, sectors with resources are shifting to low productivity strong pro-development characteristics are sectors with weak development faring, because these sectors are essential for characteristics, even within manufacturing. sustaining growth at macro level. Other than a few breakout industries (e.g. motor Manufacturing has traditionally had strong pro- vehicles, basic metals, textiles), manufacturing in development characteristics (i.e. value addition, Turkey has stagnated in recent years (Figure 4). productivity, tradability, low skill employment). It Construction and services on the other hand have has been the big driver of growth in countries that expanded rapidly, though suffer from low and have transitioned to high income. There is a falling productivity. Services are dominated by less concern however that developing economies more knowledge intensive, low productivity industries recently are transitioning too quickly out of (e.g. wholesale and retail trade), rather than more manufacturing (“premature deindustrialization”) sophisticated services that raise productivity in and into other low skilled areas that do not have manufacturing and other sectors (e.g. ICT). the same pro-development characteristics. These developments weigh on productivity and growth. EXECUTIVE SUMMARY /@QH>=!R.!*?;HE?F@;Q!Q>AV?@;=I!X[!B?>Q=]!BAV!K>AIHF<@J@<[!G=FG! Manufaturing: Weighted TFP levels vs. value added and employment shares within manufacturing sectors (2015) 16% Motor Vehicles 14% 12% Value Added Share Textiles Basic Metals 10% Food Fabricated Metal… Electrical… Manufacturing 8% Oth. Non-Metalic… averages Rubber and Plastic 6% Machinery &… Wearing Apparel Computer, Electronic and Chemicals 4% Optical Products Basic Pharmaceuticals Furniture Wood Paper 2% Beverages Oth. Transport… Oth. Manufacturing Leather 0% Printing and Media 3 4 5 6 7 8 9 TFP Sources: EIS, WB Staff estimates. What are the implications for Turkey? At These sectors however require strong Turkey’s current level of per capita income, it competitiveness, capabilities and cannot compete with low skill, labor intensive connectedness because they are more manufacturing economies. It can also not keep knowledge intensive, and therefore more raising prices to sustain wage increases unless the vulnerable to disruptive technology (i.e. quality of output also improves. At the same time, innovation or technology that uproots markets labor-saving technologies have raised concerns for and firms). High income countries are well placed emerging markets with abundant labor (like for the growth and development of these sectors. Turkey) no longer having manufacturing as a Turkey performs relatively well on connectedness, ladder of development, which others had to climb though it is only just above average on capabilities the income pole. Turkey needs a new ladder of and is a middling performer on competitiveness development. One that takes advantage of new (Figure 5). Therefore, whilst these types of sectors technologies and changing patterns of are critical to sustained growth in Turkey, they are globalization to deepen the country’s also at high risk of being disrupted unless the manufacturing and services capacities. challenges across the “three Cs” can be addressed. A few pro-development sectors stand out as This means deepening supply side capacity having potential for building such a strong through structural reforms that promote ladder of development for Turkey. Basic allocative and productive efficiency, not pharmaceuticals, chemicals, motor vehicles and “picking winners” through distortive policies. transport equipment tend to have the greatest Allocative efficiency is the extent to which scope for productivity growth, innovation and resources in a sector are being channeled to the tradability. Except for basic pharmaceuticals, these most productive firms. This can be supported sectors generated significant employment in recent through policies that address market failures (e.g. years even though their employment shares are anti-competitive practices, price distortions, low. There are other sectors with good potential information asymmetries, inefficient subsidies). but in Turkey are suffering from allocative Productive efficiency is the extent to which an inefficiencies. They include machinery and individual firm is optimizing output with its given equipment, electrical equipment and computers, inputs. This is influenced by a range of factors electronics and optical equipment. Though these including technology adoption, knowledge sectors perform poorly in terms of productivity, transfer, innovation and human capacity. These they are highly traded in international markets and topics cut across the “three Cs.” are associated with more innovative activities. Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!^.!&H>D=[!GFA>=G!V=BB!A;!FA;;=F<=I;=GG!?;I!F?K?X@B@<@=G!XH?<@A;!?;I!E@>L!K>AIHF<@J@<[! “All things will be produced in superior quantity and quality, and with greater ease, when each man works… without meddling with anything else” Plato Reforms that enhance economic integration competitors. ‘Learning by importing’ has been (i.e. connectedness) can impact positively on identified when access to a wider variety of within firm productivity. International evidence intermediates and capital goods relax technological shows that ‘learning by exporting’ has been linked constraints allowing firms to produce in better to productivity gains associated with exposure to conditions.2 Exposure to multinationals through demanding clients, or to highly productive FDI, has also been linked with learning and with 1 Adapted from Hallward-Driemeier, M, G. Nayyar, “Trouble in the Making: The Future of Manufacturing-Led Development,” World Bank Group (2017). 2 See Pavcnik, 2002; Amiti and Konings, 2007; Brandt, Van Biesebroeck, Wang, Zhang, 2017, Atkins et al., 2017. EXECUTIVE SUMMARY increased incentives to innovate. The effects of continue being concentrated in more tech- increasedon openness productivity incentives are not to innovate automatic. . The effects Gains of intensivebeing continue sectors.concentrated in more tech- increased accrue incentives heterogeneously to innovate to openness on productivity are not automatic. . differentThe effects types ofofGains continue being intensive sectors. concentrated in more tech- openness firms. 3 They on productivity can dependto are on not automatic. market structure, Gains intensive Turkey’s sectors. integration into the world economy accrue heterogeneously different types of the accrue overall heterogeneously 3investment to climate, different or types the rigidities of that happened on the back ofthe global value chains; firms. They can depend on market structure, the Turkey’s integration into world economy firms. may 3 They can depend on market structure, the exist in labor and capital Turkeyintegration Turkey’s integrates into the into GVCs world a asvalueeconomy buyer and as overall investment climate, ormarkets the rigidities (i.e. that happened on the back of global chains; overall investment competitiveness andclimate, or the rigidities that capabilities). happened on the back a seller.integrates Turkey’s of global backward value chains; participation may exist in labor and capital markets (i.e. Turkey into GVCs as a buyerinto and as may exist in labor and capital markets (i.e. Turkey GVCs integrates – Turkey’s measured into as theGVCs as a buyer portion of import and as competitiveness and capabilities). a seller. backward participation into competitiveness Turkish firms have and capabilities). become increasingly a seller. content Turkey’s backward participation into GVCs – in Turkey’s measured asexports the portion - hasof increased import GVCs – measured as the portion of import integrated Turkish firms in thehave global become market place. Turkish increasingly dramatically content early in in Turkey’s the turn exports of the - has century, and increased Turkish firms have become increasingly content in Turkey’s exports - has increased merchandise integrated in exports the globalas a share market of global place. Turkish continued growing dramatically early in during the turn the of following the century, fifteen and integrated in the global market place. Turkish dramatically early in the turn of the century, and merchandise merchandise exports exports increased as merchandise exports as a share of global a share from of 0.5 percent in global years. continued In 1995, growingevery US$ during continued growing during the following fifteen 100 the exported following by Turkish fifteen 2000 to merchandise 1 percent exportsin 2015. increased merchandise exports increased from 0.5 Exporters from reach 0.5percentalmost percent inin firms years. years. had InIn US$ 1995, 1995, of 8.9 US$ every every US$import 100 100 content. exported exported byInby 2016, Turkish Turkish this 2000 all to countries 1 percent in the in 2015. world, Exporters though 2000 to 1 percent in 2015. Exporters reach almost exportsreach almost tend to firms had increased US$ to US$8.9 of 16.5. import Forward content. firms had US$ 8.9 of import content. In 2016, this In participation2016, this – beall countries concentrated in the withinworld,the though EU all countries in the world, though exports tend to and exports the tend Middle to increased measured increased toto asUS$ US$the 16.5. portion 16.5. Forward Forward participation of Turkish content participation – – in East be concentrated beconcentrated and North within within Africa the (MENA) the EUEU andand thethe regions. Middle At the Middle measured measured asas other country’s the the portion exports portion ofof Turkish -Turkish has also content increased. content in inIn East product and North level, Africa export East and North Africa (MENA) regions. At(MENA) concentration regions. is At thethe relatively other 2005, other country’s of every country’s exports US$ exports 100 - exported, - has has also also increased. 13.5 ended increased. In In up in product low product level, export concentration is relatively 5 and level, slightly export decreasingconcentration over time. is relatively The top 2005, 2005, of every US$ 100 exported, 13.5 ended upup the of exports every of US$ other 100 exported, countries, 13.5 while ended in 2015, US$ in in low products low andand slightly exported slightly decreasing explain over decreasing over about time. 15 percent time. TheThe top top 55 of the the14.6exports did.5of exports of other Turkish other countries, firms countries, while participate while ininin 2015, GVCs 2015, US$ US$ by products export exported revenues in explain 2016. The products exported explain about 15 percent of about number 15 percent of of exporters 14.6 buying 14.6 did. did. 5 Turkish firms participate in GVCs by 5 intermediates Turkish firms and producing participate in GVCs final by goods, export has export revenues increased revenues byin in 70 2016. percent 2016. The The number over number the of of periodexporters 2006- exporters buying as well intermediates as by producing and producing intermediates buying intermediates and producing final goods, final goods, themselves hasincreased 2016, has increased 4 while the byby 70 number 70 percent percent over ofover importers, thethe period mostly period 2006- 2006-of asas andwell well asasbyby exporting producing them to producing intermediates other firms themselves intermediates themselves producing final 2016, intermediates, 2016, 4 4 while the number of importers, mostly of while the has increased number by 41.5 percent. of importers, mostly of and and exporting goods. To be exporting them sure, them to towhile other otherimport firms producing content of firms producing final final intermediates, Turkish intermediates, exports has has show increased a gradual increased byby 41.5 41.5 percent. of substitution percent. goods. exports goods. To To bebe has sure, increased, sure, while while the import import domesticcontent content of of added in value Turkish low-tech Turkish exports for mid-tech exports show show a gradual aproducts gradual substitution while substitution importsofof exports exports exports hashas has increased, doubled increased, the the since domestic 2005 (Figure domestic value value added 6). added in in ! low-tech low-tech for mid-tech products for mid-tech products while imports while imports exports has doubled since exports has doubled since 2005 (Figure 6). 2005 (Figure 6). ! !/@QH>=!_.!'@G@;Q!IAL=G<@F!J?BH=!?II=I!@;!=ZKA>=!U.!0>AIHF<@J@<[!@;F>=?G=G!?E<=>!=ZKA>=!_.!'@G@;Q!IAL=G<@F!J?BH=!?II=I!@;!=ZKA>=!_.!'@G@;Q!IAL=G<@F!J?BH=!?II=I!@;!=ZKA>=!U.!0>AIHF<@J@<[!@;F>=?G=G!?E<=>!=ZKA>=!U.!0>AIHF<@J@<[!@;F>=?G=G!?E<=>!=ZKA>L!K>AIHF<@J@<[! “Innovation is the market introduction of a technical or organizational novelty, not just its invention.” Joseph A. Schumpeter Firms’ ability to adapt and innovate is a big public procurement, direct grants) can support determinant of within firm productivity firms to invest in technology, spend on R&D, and growth. Adaptation and innovation entail risks innovate. They can help overcome market failures including large fixed costs, long gestation periods, for productivity enhancing investments; or they and uncertain returns. A conducive investment can also distort markets, leading to allocative climate can mitigate those risks through lower inefficiencies. This study assesses the impact of administrative burden on businesses, an efficient incentive grants delivered by two leading tax system, access to long-term finance, and institutions: KOSGEB and TUBITAK. Together agglomeration economies. In addition, public they are the largest providers of public grants for incentives (e.g. tax breaks, preferential credit, EXECUTIVE SUMMARY business support, including for investments in increased demand for researchers, particularly in R&D and innovation. manufacturing and industry. The boost in R&D spending has been associated with intellectual Turkish firms have accelerated R&D property applications, though also from a low expenditure and intellectual property base. In general though, Turkey and its applications in recent years but faces comparators were not the best innovators in the important gaps in innovation capacity relative past ten years. Turkey’s innovation gaps relative to to comparators. R&D expenditure grew rapidly its comparators are greatest in the areas of from a low base, nearly doubling between 2005 university-company collaboration in R&D, private and 2010 standing at just over 1 percent of GDP. R&D spending, and the quality of research The scale up in R&D expenditure has sharply institutions (Figure 8). /@QH>=!O.!Y?>Q=!@;;AJ?<@A;!Q?KG!>=B?<@J=!G Innovation Pillar - Global Competitiveness Index (Turkey vs. Max and Min of selected comparator economies) 5,5 4,5 3,5 2,5 Innovation Availability of Capacity for Government Intellectual University-industry Company spending Quality of scientific scientists and innovation procurement of property protection collaboration in on R&D research engineers advanced tech R&D institutions products MAX MIN TURKEY Source: World Economic Forum, Global Competitiveness Index (2017-2018). Notes: The figure omits the sub-indicator on “Patent Cooperation Treaty patents applications” because of outliers. Country comparators include the High Performers and Trapped MICs listed above. This study shows that TUBITAK grants are The targeting of TUBITAK and KOSGEB better targeted to firm innovation in sectors grants in turn affects their impact on firm with strong pro-development characteristics level performance, which is generally positive. compared to KOSGEB. KOSGEB targets SMEs For example, KOSGEB has a much bigger not just for R&D and innovation, but for broader positive impact on employment growth than business support. KOSGEB grants are evenly TUBITAK. On the other hand, TUBITAK has a distributed across firms (i.e. little variance around much bigger positive impact on R&D investment average size of grants) regardless of which sub- than KOSGEB. It is therefore not surprising also sectors they are in. TUBITAK grants on the other that the impact of TUBITAK support on hand are more targeted to specific projects or firm innovation is larger and significant in more level upgrades. Plus, more technology intensive industries than the impact of KOSGEB. These industries with pro-development characteristics results are in part reflective of the nature of the (e.g. motor vehicles, machinery and equipment two programs; TUBITAK grants are not and computer, electronic and optical products) necessarily aimed at increasing the firm size but receive a relatively largely share of TUBITAK innovation, while the KOSGEB grants are mainly grants. At the same time, there is scope to increase targeted to accelerate the growth of the SMEs. TUBITAK support for younger firms, who tend to be more innovative. Whilst the impact on employment, R&D and innovation is encouraging, ultimately what matters in terms of sustainability is the impact Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey on firm productivity. In other words, without correcting market failures by protecting firms productivity gains, the impact of grants on from market exit. Similarly, the impact of grants employment, R&D investment and innovation on R&D and innovation could also improve may not be sustainable. The evidence for Turkey through better firm selection, particularly by shows that the impact of R&D investment and taking account of firm size and age. innovation is positive for firm productivity. It follows from this that the impact of KOSGEB Several policy implications emerge from these support on productivity is neither large nor results. It is worth reviewing grant award criteria positive, whilst the impact of TUBITAK on firm to; (i) target productivity enhancing interventions productivity is both positive and strong. – these do not necessarily have to be R&D or innovation related, they could include other This raises questions over the sustainability of business improvements that generate efficiencies; results achieved, particularly under KOSGEB. (ii) explicitly target firms that have strong growth KOSGEB programs have been successful in and innovation potential. On the latter, support to employment creation. But KOSGEB support SMEs should distinguish between young and/or does not always impact productivity positively. high growth potential firms from older/lower Therefore, employment creation may be growth potential firms. The evidence in this temporary i.e. supported firm is likely to shrink chapter shows that young firms generally have back to its pre-support size. Alternatively, higher potential to grow and innovate. This KOSGEB grants may also be sustaining less research, and principles behind the efficacy of productive firms that would otherwise go out of public interventions for innovation and R&D business through “creative destruction.” In other could be expanded to other instruments. words, subsidies may be creating rather than 4HL?;!F?K@L!K>AIHF<@J@<[! “Education is another magic formula that failed us in our quest for growth” William Easterly (2002) “In this globalized world, physical capital and technology are always available. But human capital is still very immobile, so you should have good education and job-training programs to acquaint people with the necessary technology.” Han Duck-soo, The Growth Report (2008) New technologies and business lines from qualifications. Both outcomes are inefficient and economic integration and innovation will not accelerate productivity. discussed above can create new and better jobs that will require enhanced human capital. The supply of more educated workers in New technologies and business lines boost the Turkey has outpaced the labor market’s demand for more capable workers, that in turn ability to absorb those educated workers. This enhance firms’ absorptive and technical capacities. is reflected in rapidly growing unemployment These factors work together to accelerate within among the more educated (Figure 9); a relatively firm productivity growth. Investing only in new large share of unemployed workers with technologies and business lines without human vocational and university education; and an capital would create unutilized, excess capacity; important share of people that are not in whilst investing only in human capital without employment, education or training (NEET) with new technologies and business lines would create university degrees in technical subjects (Figure 10). an excess supply of workers with redundant EXECUTIVE SUMMARY /@QH>=!:.!2>AV@;Q!H;=LKBA[L=;=!98.!%))&!?;I!<=FN;@F?B!`H?B@E@F?<@A;G! Growth in unemployment and contribution by Most common fields of study among NEET educational attainment and gender (> 25 years (15-29 years, 2016) old, 2014-2016) Manufacturing and production 28% Teacher training and Female Humanities 18% Social services Business and administ. Humanities 8% Social and behavioural science Male Teacher training -2% Business and administ. Overall Male Female Primary and less than primary Secondary Engineering (thousands) High school Vocational high school 0 50 100 150 200 250 University Total Sources: LFS 2016, WB Staff estimates. There could be a variety of structural reasons investment in training of employees. If demand that explain why the demand for more for more and better skills does not accelerate, the educated workers has not kept pace with the incentives to acquire more and better skills on the supply of educated workers. The first could be supply side will decrease. an excess supply of more educated workers. This could be linked to the pace of economic An analysis of wage developments and premia integration and innovation discussed in chapters 3 point to the relative importance of demand and 4. For example more integration through FDI side factors in deterring the acquisition of can boost demand for more educated and skilled deeper skills. This is reflected in the high wage workers in a way that domestic investment cannot. growth for low skill workers relative to high skill The second could be due to labor market rigidities workers, and the decline in wage premia for and other market failures discussed in chapter 1. university educated workers relative to primary These can prevent labor from being efficiently educated workers. Whilst the impact of minimum reallocated to more productive areas. Thirdly, it wages certainly plays a role in driving up earnings could be due to the quality of education. Fourthly, of those at the lower end of the wage spectrum, it could reflect a lack of skills that cannot be offset these developments together with the above by higher educational attainment only. Various findings suggest that a demand side boost is skills assessments show that Turkey does indeed needed to deepen the skills of the Turkish labor perform less well compared to other OECD force. economies. The study finds no significant relationship The demand for more cognitive and technical between skill composition and firm skills in the Turkish economy has increased, productivity. Firms characterized mostly by albeit quite slowly. This is in line with earlier medium-skills absorb nearly half of all findings on the rapid increase in low-skilled employment. Low skill firms absorb around 35 employment relative to high-skilled employment. percent of employment, whereas high skill firms Whilst the public sector is working on better account for around 16 percent of employment. matching the supply of skills to the demand for One might expect labor productivity to be higher skills, the role of the private sector in fostering the in firms that have higher incidence of high skill demand for more skills is also limited. This is workers. However, there is little evidence of a evident in the relatively low levels of firm clear association. Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey These counter-intuitive results are likely rising labor supply will not be absorbed in good linked to the prevalence of small-micro firms quality jobs. This in turn could reverse gains in in the economy. These firms absorb the bulk of standards of living over the past few years. employment and suffer from declining Boosting demand for more educated and skilled productivity. Larger firms exhibit constant workers links back to other parts of the report productivity over the last 10 years across all skill including the deepening of economic integration incidences. In other words, small firms will not and more effective support to firm level drive demand for higher skills. This means that innovation, both of which would help to deepen even as the supply of skilled workers increases in production capacity of firms, which in turn would the future, the prevalence of small-micro firms require more qualified workers. It would require limits the labor market’s capacity to absorb fully more targeted education policies that would help the next cohort of skilled graduates strengthen the relationship between skills and firm-level productivity. It would also require a Turkey needs a boost in the demand for more reduction in the mandatory costs of labor through educated and skilled workers. Without this, reform of labor market regulations. "ALK=<@<@A;]!>=QHB?<@A;!?;I!K>AIHF<@J@<[! A famous economist once said that the best of all monopoly profits is a quiet life. You don’t want a quiet life for a firm; you want it forever to improve its productivity” Robert Solow, The Growth Report (2008) The final chapter in the study analyzes the terms of employment shares of large firms, has impact of competition and business been on a downward path. regulations on firm and sector level productivity. Barriers to competition and With reduced concentration, labor share is burdensome business regulations distort markets rising whilst markups are declining. Total and prevent a reallocation of resources to more labor compensation as a share of firm revenues efficient firms. Competition may free up the has consistently risen over the past decade largely efficient re-allocation of market resources from due to a shift towards labor-intensive production. low- to high-productivity firms as well as from This is positive for Turkey given a more global low-efficiency sectors to high-efficiency ones. concern over the secular decline of the once-stable Turkey has over time reformed its business share of revenues going to labor. In more regulatory framework, which has gradually helped concentrated sectors, firms can command higher increase market competition. The big wave of prices and revenues, with a lower share of total reforms in the early 2000s targeted a reduction of firm revenues accruing to labor. Firm-level state intervention in the economy and markups (i.e. price above marginal cost of establishment of independent regulators, and producing one unit) have been declining across all through these aimed to encourage more private sectors. In general, if market power is declining, investment and competition in key sectors of the firms cannot command prices too far above the economy. marginal cost required to produce a good or service (i.e. a markup); this will also result in a Firm level analysis shows that these reforms larger share of the bottom line accruing to labor, increased competition across manufacturing as is the case in Turkey. and services in Turkey over the past ten years. There has been reduced market concentration Despite higher labor share and lower across all major sectors, which bodes well for markups, the observed declines in productivity, though manufacturing concentration concentration have also been associated with remains high. More high-tech manufacturing and declining productivity. This should not be taken skill intensive services tend to be more to say that more competitive markets result in concentrated as these sectors are more difficult to declining productivity. First, manufacturing break into. But even here, the trend, particularly in remains notably concentrated. Second, under a EXECUTIVE SUMMARY model of ‘winner take most’ competition, where a The second aspect are regulations in the few highly productive firms capture increasing services sector, which in Turkey pose market share, such trends are consistent with the obstacles to competition and productivity. lack of highly productive market superstars and Regulatory reform in services has a positive the presence of increasing numbers of less impact on downstream manufacturing and productive (and possibly more labor-intensive) services firms’ productivity. When service inputs firms. That is, the prevailing market conditions are are effectively supplied, this has a knock-on effect such that there has been declining market on industries that use services. A recent World concentration at the same time as a re-allocative Bank study finds that Turkey has much potential shift toward less productive activities. This is to increase the impact of services on consistent with findings across the rest of the manufacturing productivity (Haven and Van Der report. Marel, 2018). There is much scope to reduce restrictions on FDI in services; domestic The study looks at two aspects of the regulatory barriers in services that affect Turkish regulatory framework that affect competition and foreign firms; discriminatory services trade and productivity. The first relates to regulatory barriers that prevent foreign service providers barriers for firm entry and exit, which are implicit from entering and operating in Turkey. constraints on competition. Regulatory barriers to firm entry and exit can dampen within sector These results have several policy implications. productivity gains as more efficient entrants are To increase dynamism in high-value added prevented from displacing less efficient industries with strong pro-development incumbents. While Starting a Business in Turkey is characteristics, there is scope to increase comparatively efficient and less burdensome, an competition (e.g. through anti-trust policies) as opposite conclusion emerges when looking at some of these sectors remain relatively more Resolving Insolvency. In economies where these concentrated. Moreover, aligning capital risk, via procedures are burdensome, less productive firms policies that, for instance, would increase creditor are more likely to remain in the market, recovery rates would help support more activity in dampening overall productivity. To address these these sectors. Secondly, to help release resources challenges, a new ‘concordat’ procedure was for more productive firms by enabling exit of less recently introduced, which enables authorities to productive ones will require review of bankruptcy set timelines for the procedure, and puts a heavy and insolvency frameworks, which is already focus on business continuation rather than its underway (e.g. through the concordat reform). liquidation through new financing, confirmation Thirdly, in the services sector, there is much scope of contracts and sale of essential assets in to remove FDI restrictions and opening up bankruptcy. professional services to greater competition. "A;FBHG@A;!! The analysis of micro-macro productivity This will require reforms that promote within linkages in this study reveal that the recent firm productivity growth. The study looks at years’ focus on managing demand shocks has three inter-related factors (Figure 11). The first diverted resources away from sectors with two – economic integration and innovation – can strong pro-development characteristics. Other be big sources of new technology and business than a few breakout sectors (e.g. motor vehicles, lines. The third is human capital, which is basic metals, textiles), manufacturing in has necessary to increase firms’ ability to absorb new stagnated in recent years (Figure 4). Construction technology and develop business lines. The study and services on the other hand have expanded finds that economic integration and innovation rapidly, though suffer from low and falling have boosted firm-level productivity, though productivity. However, Turkey has scope to reforms could further accelerate these positive expand sectors that have strong potential for impacts. Particularly as productivity gains have not boosting productivity, economic growth and jobs. sufficiently increased the demand for more educated and skilled workers. The growth of more Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey productive of could firms The growth more in be accelerated turn also firms productive could in turn also! be accelerated through reforms that increase through reforms competition that and increase reduce competition regulatory and burden. ! reduce regulatory burden. ! /@QH>=!99.!'=EA>LG!EA>!V@L!?;I!V@!K>AIHF<@J@<[!Q>AV@J=>G!?;I!FN?BB=;Q=G!AE!=FA;AL@F!Q>AVA;Q!=FA;AL@F!FA;J=>Q=;F=!;AL!GHKKB[!>=EA>LG! b?c! PNA>L!KAB@F@=G!N=BK!L?;?Q=!I=L?;ITG@I=!GNAFDG!X=!0=>EA>L=>!?;I!&>?KK=I!*+"!FALK?>?G Several parts of this report use two sets of comparator countries. One set includes “high performer” countries that recently graduated from Upper Middle Income to High Income in less than 20 years: Chile; Czech Republic; Korea, Rep; Poland. The other set, referred to as “trapped MICs,” includes countries that have remained in the Upper Middle Income category for more than 20 years: Argentina; Brazil; Malaysia; Mexico; and South Africa. The time series data on transition across income categories is based on World Bank data on per capita GNI and Felipe et al. (2012). The twenty-year threshold is based on a similar approach used in the WBG (2017) report: Lessons from Poland, Insights for Poland. 6 WBG, “Turkey’s Transitions: Integration, Inclusion, Institutions,” (December 2014). Page | 1 1 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey The measure of comparison is limited to income graduation. It does not include the distribution of income or social outcomes. The report however assesses possible economic and structural characteristics that can enable, or pose challenges to, graduation from UMIC to HIC. Comparators and reform paths are diverse, but there are common traits. For example, most trapped MICs are relatively resource rich. Among them, Brazil and Malaysia are 2 of only 13 economies that since 1950 experienced high sustained growth (average of 7 percent a year or more for 25 years or longer); but Brazil and Malaysia did not cross the HIC threshold.7 Korea, Rep., on the other hand, also among the 13 high growth countries, did graduate to HIC. Successful high growth economies have some common features. The Commission on Growth and Development highlights the following:8 (i) integration in the global economy; (ii) maintenance of macroeconomic stability; (iii) high rates of (domestic) savings and investment; (iv) market structures to allocate resources; (v) committed, credible and capable governments. 4.! Turkey’s convergence is even more remarkable given the shocks it suffered since 2008. Strong institutions, accumulation of fiscal and financial sector buffers since 2002 (World Bank, 2014), and post-GFC global monetary expansion, enabled effective policy stimuli. After the 2016 crisis for example, fiscal incentives boosted demand and sustained supply through subsidies to employers. These short-term measures helped offset labor market rigidities that otherwise would have precipitated retrenchments, particularly among low income earners. Similarly, government guarantees in 2017 for credit to Small and Medium Enterprises (SMEs) helped maintain access to working capital during the downturn. Banks would have otherwise not extended credit due to financial market frictions, increasing the probability of business closures. 5.! Short-term policies since 2008 diverted Turkey from its earlier focus on long-term supply reforms, though also helped avert a recession until recently. Downturns can promote efficient resource allocation; more productive firms drive out less productive ones (creative destruction), and incentives for productive investments rise with falling profits from short-term projects. But structural constraints, which require major reform, pose hurdles to price adjustments and factor mobility needed for efficient reallocation of resources. For example, countercyclical finance for productive investments is limited in Turkey by the depth of the financial sector (capital markets, private equity, insurance) and market imperfections, which are long-term challenges. Therefore, a sharp downturn caused by an exogenous shock like the one in 2016 would have taken a long time to recover from. 7 WB, “The Growth Report: Strategies for Sustained Growth and Inclusive Development,” (2008). 8 Ibid (Part 1). Page | 2 2 CHAPTER 1 / DRIVERS AND CHALLENGES OF ECONOMIC GROWTH /@QH>=!9.!&H>D=[dG!=FA;AL@F!FA;J=>Q=;F=!FA;<@;H=I!=!7.!&N@G!=;?XB=I!&H>D=[!?<=!?E<=>!@G@G! K=>EA>L=>G!?;I!KHBB!?V?[!E>AL!<>?KK=I!*+"G! Convergence (Per capita GDP/US per capita 3,5 Country rank and income growth (GDP per GDP, PPP constant 2011 international $) capita, PPP 2011 international $, 1992-2017) 0,55 Income multiple, 1992 = 1 2,5 0,45 1,5 0,35 0,25 0,5 90 92 94 96 98 00 02 04 06 08 10 12 14 16 75 65 55 45 35 25 19 19 19 19 19 20 20 20 20 20 20 20 20 20 Country rankings 1992-2017 High performers Trapped MICs Turkey High performers Trapped MICs Turkey ! ! Sources: WDI, WB Staff estimates. Sources: WDI, WB Staff estimates. Note: Convergence estimates for high performers and trapped MICs based on their weighted averages of per capita GDP. bXc! -@J=>G@A;!E>AL!BA;QT<=>L!GHKKB[!>=EA>LG!I@L@;@GN!AV=!S.!0=>!F?K@AVD=[!@G!N@QNB[!JAB?<@B=! /@QH>=!R!&N@G!@G!?GGAF@?<=I!V@<@FHB?>B[!>=B?<@J=!EA>L=>G! ?;I!@;J=G=!^.!+;J=GD=[!@G!FBAG=>!=!_.!3N@BG@J?<=!FA;GHLK<@A;!B=J=BG!?>=!N@QN=>! &>?KK=I!*+"G!EA>L=>G! ?KK=I!*+"G!?;I!N@QN!K=>EA>L=>G! Investment/GDP vs. Capital to output ratio Private consumption (% of GDP) 4 66 64 3 62 Capital to output ratio 60 58 2 56 54 1 52 0% 10% 20% 30% 40% 50% 10 11 12 13 t 1 2 3 4 5 6 7 8 9 t+ t+ t+ t+ t+ t+ t+ t+ t+ t+ t+ t+ t+ Investment to GDP High performers Trapped MICs Turkey High performers Trapped UMICs Turkey Sources: WDI, WB Staff estimates. Sources: WDI, WB Staff estimates. Notes: Period for High performers are years they were in UMIC Notes: Periods are the same as Figure 5. group; for Trapped MICs, 2002 onwards, for Turkey 2004 onwards. /@QH>=!U.!*?F>A!G=EA>LG!?FF=B=>?<=I!F?K@=!O.!CH;?B!EBAVG!N?G!GN@E<=I! @;EBAVG!Q=!G?J@;QG!Q?K! E>AL!/-+!=!:.!">=I@@J?<=!G=F!N?G!@;F>=?G@;QB[! /@QH>=!98.!->@J=;!@;!X@Q!K?>K!@;F>=?G=!@;!I=XAF[FB@F?B!?E<=>!788:! F>=?<@;Q!EBAVG!D@GN!X?;DG! GDP growth and private credit growth (%) Turkey capital inflows (% of GDP) 0,04 8% R" = 0,6768 Quarterly private sector growth 0,03 6% 0,02 R" = 0,2463 0,01 4% 0 2% -0,05 0 0,05 0,1 0,15 0,2 -0,01 0% -0,02 -0,03 -2% 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 -0,04 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Quarter GDP growth (Seasonally Adjusted) Portfolio (Equity) Portfolio (Debt) Pre 2009 Post 2009 Other Investment (Banks) Other Investment (Others) ! ! Sources: WDI, WB Staff estimates. Sources: WDI, WB Staff estimates. ! bFc! '@GDG!AE!K>=L?=!I=@;IHG<>@?B@\?<@A;!V@BB!B@L@AV=!99.!0>AF[FB@F?B!F>=I@!?FF=GG!?;I! /@QH>=!97.!P@G@;Q!J?BH=T?II=I!AE! N@QN=>!FAG@;Q!IAV;;G! @;IHG<>[!GHQQ=G@G@;Q!B?XA>!K>AIHF<@J@<[! Quarterly GDP growth and lending rates (%) Industry share of value addition and employment (%) 25 50 R" = 0,7901 Share in value addition 45 20 40 R" = 0,2467 R" = 0,4293 35 15 30 25 R" = 0,245 10 R" = 0,3515 20 15 5 20 25 30 35 40 45 -0,06 -0,04 -0,02 0 0,02 0,04 0,06 Share in employment Consumer lending rate Commercial lending rate Higher performers Trapped MICs Turkey Sources: Haver Analytics, WB Staff estimates. Sources: WDI, WB Staff estimates. Notes: Quarterly growth rates 2008-2017. Notes: Periods are the same as Figure 5. &?XB=!9.!0>ATI=J=BAKL=;?F<=>@G<@FG!?F>AGG!G=FG!@;!&H>D=[9^! ! Exports Employment Low skill R&D Value added per (% of sector (% of total) 2/ employment spending worker output) 1/ (% of sector (% of total) (US$ thousand, employment) 3/ 4/ constant 2010) 5/ Agriculture 11.1 19.5 89.5 7.2 12.7 Industry 60.7 26.8 65.1 61.8 39.2 Services 8.1 53.7 41.7 32.0 39.1 Sources: WDI, Haver Analytics, Turkey Labor Force Survey, MOIT, WB Staff estimates. Notes: 1/ Access to international markets, scope for technology and knowledge spillovers (2017); 2/ Significance as an employer (2016); 3/ Ability to absorb low skilled workers defined as those with less than high school education (2017); 4/ Innovation and technology diffusion (2014); 5/ Labor productivity (2016). 11.! That said, changes in the global nature and distribution of industrial production means that there may be different paths to industrial development and structural transformation. Comparator UMICs seem to have reached their highest shares of industry employment at around US$ 17,000- US$ 18,000 per capita GDP (PPP 2011 $), beyond which the industry employment shares begin to decline (Figure 13). But at this point, industrial labor productivity for high performers is higher than for Turkey and trapped MICs. This could be due to labor saving technology and increased productivity in upstream and downstream services that help deepen industrial production, which in turn requires investment in skills. 15 Adapted from Hallward-Driemeier, M, G. Nayyar, “Trouble in the Making: The Future of Manufacturing-Led Development,” World Bank Group (2017). Page | 7 7 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!9S.!&>=;IG!@;!@;IHG<>@?B!=LKBA[L=;=G!IH>@;Q!$*+"!KN?G=!! High performers Trapped MICs (2002-2017) Turkey (2004-2017) (UMIC years) 1 0,9 0,9 Shares indexed to maximum (max = 1) Shares indexed to maximum (max = 1) Shares indexed to maximum (max = 1) 0,8 0,8 0,95 0,7 0,7 0,6 0,6 0,9 GDP Per capita: $18,130 GDP Per capita: $17,200 GDP Per capita: $17,800 0,5 Employment share: 23% 0,5 Employment share: 32% Employment share: 28% Value added share: 41% Value added share: 49% Value added share: 29% 0,4 0,85 0,4 UMIC years 2002-2017 2004-2017 Poly. (Employment share) Poly. (Employment share) Poly. (Employment share) Poly. (Value added share) Poly. (Value added share) Poly. (Value added share) Sources: WDI, WB Staff estimates. Notes: Periods are the same as for figure 5. GDP Per capita in PPP 2011 $. 12.! Labor-saving technologies have raised concerns about emerging markets like Turkey no longer having the ladder of development that others were able to use to climb up the income pole.16 Some however have noted that countries can prepare and adapt to take advantage of technological and other changes.17 This means better understanding productivity dynamics across and within sectors, particularly as pro-development characteristics (i.e. potential to boost productivity, growth and employment) varies (Chapter 2). Before going into these, the next two sections take stock of aggregate productivity trends and reform gaps in Turkey. ,!K>AIHF<@J@<[!XAAGD=[dG!GAB@I!=FA;AL@F!K=>EA>L?;F=! b?c! 2>AV@J=;!LAG!?FFHLHB?<@A;!?;I!A;B[!K=>@AI@F!XAAGAIHF<@J@<[! 13.! Economic growth in Turkey since the 1980s has been driven largely by factor accumulation, with only periodic boosts in productivity (Figure 14). The liberalization of foreign exchange, price and trade policies in the 1980s, transport and energy investments in the late 1980s, and the commercialization and regulation of public utilities in the late 1990s and early 2000s led to an acceleration in investment and physical capital accumulation.18 Turkey’s relatively young population provided a boost in labor supply, with a near doubling in the labor force between 1988 and 2017. 14.! Productivity boosts were associated with periods of strong reforms (1995-1998, 2002-2007), which also led to growth accelerating to above the period average for more than two years in a row. Periods of economic crisis (1994, 1999, 2001, 2008-09) saw negative contributions of Total Factor Productivity (TFP), which is ascribed to unobservable factors captured in the residual such as a decline in capacity utilization or an accumulation of inventories that are common during economic contractions. 16 Rodrik, D, “No more growth miracles,” Project Syndicate (2012). 17 Hallward-Driemeier, M, G. Nayyar (2017). 18 WBG, “Turkey’s Transitions: Integration, Inclusion, Institutions,” (December 2014). Page | 8 8 CHAPTER 1 / DRIVERS AND CHALLENGES OF ECONOMIC GROWTH /@QH>=!9R.!0>AIHF<@J@<[!Q>AVD=[!?E<=>!=!9R.!0>AIHF<@J@<[!Q>AVD=[!?E<=>!AVAIHF<@J@<[! CAZ!7.!)FA;AL@F!Q>AVAIHF<@J@<[! This first chapter of the report analyses economic growth (Y) in terms of the growth of factor inputs, This first chapter of the report analyses economic growth (Y) in terms of the growth of factor inputs, namely Capital (K) and Labor (L), and Total Factor Productivity (TFP) (Solow growth model): namely Capital (K) and Labor (L), and Total Factor Productivity (TFP) (Solow growth model): !" # $ %" &' "( )*' , ' " )*' !" # $ %" & " (" , where Yt = GDP in year t, At = TFP in year t, Kt = Capital stock in year t, ! = the income share of where Yt = GDP in year t, At = TFP in year t, Kt = Capital stock in year t, ! = the income share of capital, and Lt = Labor force (Population 15 +) in year t. The growth rate is calculated as a geometric capital, and Lt = Labor force (Population 15 +) in year t. The growth rate is calculated as a geometric average growth rate between the beginning and end year using a compound average growth rate. average growth rate between the beginning and end year using a compound average growth rate. The The growth growth rate of rate offactor inputs factor inputs can canaccelerate acceleratethrough throughthe liftingof ofinstitutional thelifting institutionalconstraints constraints(e.g. (e.g. access to finance for investment, elimination of labor market rigidities) or throughexogenous access to finance for investment, elimination of labor market rigidities) or through exogenousfactors factors (e.g. population (e.g. population growth, growth, migration). migration). TFP TFP theefficiency isisthe withwhich efficiencywith whichfactor factorinputs inputsare usedto areused toproduce produce goods goods andand services. services.Long-term Long-term sustainabilityof sustainability growthdepends ofgrowth dependson TFPgrowth. onTFP growth.TFP TFPisisthe theratio ratio of of output output totoinput inputvolumes volumes –– the higher the higher the the ratio,the ratio, lessinput theless inputis neededfor isneeded foraagiven givenoutput. output. The The growth growth TFP inin TFP isis calculated calculated as asaa residual; growthin thegrowth residual;ititisisthe inoutput outputthat isnot thatis notexplained explainedby by the the growth in capital or labor inputs. TFP can be driven by a reallocation of resources or investment in growth in capital or labor inputs. TFP can be driven by a reallocation of resources or investment in skills and skills and technology technology that thatenables enablesthe the economyto economy producemore toproduce morewith less.Labor withless. productivity– Laborproductivity – real real output output ororvalue value added added per per worker worker ––isisalso alsoused usedin inempirical empiricalanalysis analysisas closesubstitute asaaclose substituteto toTFP. TFP. This Thisfirst part first partalso alsouses usesthe the Shapley Shapleymethod methodto todecompose decomposethe growthof thegrowth ofper percapita GDPinto capitaGDP into four four components: components: labor laborproductivity productivitygrowth, growth, employment employmentgrowth, growth,labor forcegrowth, laborforce growth,and andchange changeinin working working age population. age population. This Thiscalculates calculates thecontribution the contributionof differentsectors ofdifferent sectorsto toaggregate aggregateproductivity productivity and employment growth to understand the role of structural transformation in drivingproductivity and employment growth to understand the role of structural transformation in driving productivity vs. vs. the role the roleofofwithin within sector sectorproductivity productivitygains. gains. Page Page 99 || 9 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey The sources of productivity growth, and their micro-macro links can be illustrated by the figure below: (1)! Firm level productivity gains can come from strengthening technical capacity and innovation. Policy areas that impact this include openness to trade and investment, the supply of skills and public sector incentives for growth and innovation. (2)! Firm level productivity gains could lead to less efficient firms exiting the market. Policy areas that impact this include the bankruptcy and insolvency framework, regulations for hiring and firing workers, degree of unionization, and unemployment benefits. (3)! The exit of less productive firms could release resources for more productive incumbents and new entrants. Policy areas that impact this include business entry regulations, competition policy, financial sector depth, and R&D expenditure. /@QH>=!9^.!PAH>F=G!AE!K>AIHF<@J@<[!Q?@;G!?;I!!L?F>ATL@F>A!B@;DG! 6. Economy wide productivity growth through structural transformation and within sector productivity 3. Entry of more productive firms through conducive investment climate 4. Within sector 5. Structural productivity growth as transformation through resources shift to more factor inputs shifting from efficient firms whose low to high productivity market shares increase sectors 2. Exit of less productive firms through efficient institutions 1. Within firm productivity growth through technical capacity, innovation (4)! This process of “creative destruction” – as coined by economist Joseph Schumpeter – that reallocates resources to more efficient firms and sub-sectors can fuel productivity gains within a sector and enable that sector to grow. (5)! At a macro level, this can support a process of structural transformation as factor inputs shift to more productive and rapidly growing sectors. Policy areas that impact this include basic health and education, labor market flexibility, the quality of infrastructure, and the quality of public finances. (6)! A combination of productivity growth within sectors and structural transformation can help accelerate economy wide productivity growth. Based on the above, chapter 1 analyzes productivity and economic growth trends at the macroeconomic level. Chapter 2 digs deeper into productivity dynamics within sectors, reallocation of resources across sub-sectors and firms, and efficiency dynamics within firms and sectors. Chapters 3 to 6 analyze policy and institutional factors that impact on the above productivity dynamics. Source: WB Staff. Page | 10 10 CHAPTER 1 / DRIVERS AND CHALLENGES OF ECONOMIC GROWTH 15.! There is scope for further factor accumulation in Turkey, but TFP growth is needed to sustain improvements in living standards. Relative to high performing comparators, TFP contribution to growth in Turkey over the past 30 years was low, whilst labor input contribution was high (Figure 16). Reforms helped increase labor force participation (LFP), particularly after the GFC, which contributed to growth and poverty reduction.19 Overall LFP however remains low, particularly for women (Table 2).20 This together with Turkey’s young population points to further growth in labor supply in the coming years (Chapter 5). Without TFP growth this labor cannot be absorbed in better jobs with higher earnings. ! /@QH>=!9_.!&/0!Q>AVEA>L=>G!IH>@;Q!!$*+"!G!D=[dG! Contribution to GDP growth 100% 17% 24% 80% 28% 40% 29% 21% 60% 15% 8% 40% 55% 56% 52% 55% 20% 0% -20% 1988-2017 P1 P2 P3 P4 1988-2016 P1 P2 P3 P4 1990-2016 P1 P2 P3 P4 1988-2016 P1 P2 P3 P4 Turkey Korea, Rep. Poland Chile Capital Stock (gK) Labor (gL) Total Factor Productivity (gA) Sources: WDI, WB Staff estimates. Notes: P1 P2 P3 P4 P1 P2 P3 P4 Turkey 1988-2001 2002-2007 2008-2011 2012-2017 Poland 1990-1996 1997-2007 2008-2011 2012-2016 Korea, Rep 1988-1997 1998-2007 2008-2011 2012-2016 Chile 1988-2001 2002-2007 2008-2011 2012-2016 &?XB=!7.!Y?XA>!/A>F=!0?><@F@K?<@A;! Turkey UMICs ECA (exc. HICs) LFP, total (% of total population ages 15+) 52.8 66.6 62.5 LFP, female (% of female population ages 15+) 33.5 NA 53.3 Source: WDI. bXc! +LK>AJ=L=;IG!;A!K>AIHF<@J@<[! 16.! Whilst patterns of structural transformation benefited Turkey till now, more of the same will run into diminishing returns. Increased employment in services played a big role in Turkey’s development, even if, as discussed below and in chapter 2, productivity has been flat compared to industry. But at current levels of income, further growth in low skill, non-tradable services will dent living standards. 19 WBG, “Turkey’s Transitions: Integration, Inclusion, Institutions,” (December 2014) (Chapter 5). 20 WBG, “Turkey’s Future Transitions: Towards Sustainable Poverty Reduction and Shared Prosperity,” (2017). Page | 11 11 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 17.! This can be illustrated by looking at labor productivity gaps across sectors. Large gaps in labor productivity – as measured by the value added per worker in a sector as a share of the value added per worker in agriculture – can signal allocative inefficiency. It suggests that a reallocation of labor from low to high productivity sectors (i.e. structural transformation, Box 2) could accelerate growth, even if productivity within the receiving sector remains weak. 18.! Over time, labor productivity gaps between agriculture and services have gradually narrowed, whereas labor productivity gaps between agriculture and industry have remained stable (Table 3). This is due to a combination of rising labor productivity in agriculture as employment shares have declined, relatively stagnant productivity growth in services as employment shares have increased, and rising labor productivity in industry with slower growth in employment. &?XB=!S.!Y?XA>!K>AIHF<@J@<[!Q?KG!?;I!Q>AVAGG!G=FG! 1988 1990 1995 2000 2005 2010 2015 2016 Labor productivity gaps (Sector VA per worker/Agriculture VA per worker) Agriculture 1 1 1 1 1 1 1 1 Industry 3.0 3.5 3.6 2.8 2.7 2.6 3.0 3.1 Services 4.4 4.9 5.3 4.2 3.6 3.2 3.2 3.1 Marginal Product of Labor (Period change in sector VA/Period change in sector employment) Agriculture -1.6 4.7 -5.7 -4.5 87.6 27.7 10.1 Industry 187.3 21.8 29.5 132.2 43.5 78.6 336.1 Services 50.9 50.3 43.8 62.6 36.6 49.2 23.7 Sources: WDI, WB Staff estimates. 19.! In other words, labor shifted to more productive services, even though productivity growth within services was relatively low, which lowered overall productivity. The expansion of low skill services helped absorb labor from agriculture. The expansion of critical services (transport, logistics, banking and finance) also helped expand industry. However, since services make up around 60 percent of value added and a growing share of employment (from 36 percent in 1988 to 53 percent in 2016), slow productivity growth in services has also contributed to overall stagnation in productivity. 20.! A decomposition of per capita GDP growth by employment dynamics confirms the above patterns of structural transformation and factor accumulation driving growth in Turkey. The decomposition shows how, particularly since 2008, per capita GDP growth was driven by a combination of rapid growth in service sector employment – offsetting contraction in agriculture employment – labor force participation rate, and demographic changes (Figure 17). The contribution of employment growth in industry on the other hand has been more limited. Industry employment overall has grown at a relatively slow pace, and even contracted in the 2012-2016 period, though value addition was high resulting in strong labor productivity growth within industry. Page | 12 12 CHAPTER 1 / DRIVERS AND CHALLENGES OF ECONOMIC GROWTH /@QH>=!9U.!0=>!F?K@AV@J=;!X[!G=>J@F=!=LKBA[L=;<]!Y/0!?;I! I=LAQ>?KN@F!GN@E=!9O.!0A<=;<@?B!Q>AV=;I! /@QH>=!9:.!&N@G!@G!B?>Q=B[!IH=!D=[!G@;F=!AV?Q!A;!GHKKB[!F?K?F@<[! Potential Growth Estimates, % Contributions to Potential Growth, % 100 10 90 80 70 5 60 50 40 0 30 20 10 -5 0 04 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 04 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Upper Bound Lower Bound GDP growth Potential Capital Stock Potential Employment Potential TFP Sources: WDI, TURKSTAT, WB Staff estimates. Notes: Potential growth rates are estimated by HP filter, Multivariate filter, Cobb-Douglas Production and CES production function methodologies. Contributions to potential growth is calculated based on Cobb-Douglas production function estimates. ! 22.! A combination of all the above factors illustrates why it would be difficult for Turkey to sustain improvements in living standards without productivity growth. Labor supply is projected to increase in the coming years. But the Marginal Productivity of Labor (MPL) in services is falling as increased services employment is associated with declining growth in real value added per worker in the sector (Table 3, Figure 20). This is linked to a growing concentration of employment in low skill sectors that suffer from low real value added per worker (e.g. retail and wholesale trade, hospitality services) (Figure 21). Additional labor input in services is therefore associated with diminishing returns. In industry, on the other hand, whilst labor productivity is higher, the pace of job creation is low as noted above and even there it seems that the MPL is declining. This could be a symptom of falling allocative and productive efficiencies within industry and manufacturing, as discussed further in chapter 2.! ! 23.! Therefore, labor incomes will stagnate unless more employment is created in relatively higher productivity industry or more skill intensive services sectors. At Turkey’s current level of per capita income, it cannot compete with low skill, labor intensive manufacturing economies. It can also not keep raising prices to sustain wage increases unless the quality of output also increases. Turkey needs a new ladder of development. One that takes advantage of new technologies and changing patterns of globalization to deepen the country’s industrial and manufacturing base with high quality, complementary services.22 This should help accelerate TFP growth as TFP gains from structural transformation (i.e. labor shifting from agriculture to industry and services) begin to run their course. ! ! 22 Hallward-Driemeier, M, G. Nayyar (2017). Page | 14 14 CHAPTER 1 / DRIVERS AND CHALLENGES OF ECONOMIC GROWTH /@QH>=!78.!P=>J@F=G!?>=!=ZK=>@=;F@;Q!I=FB@;@;Q!L?>Q@;?B! /@QH>=!79.!3N@FN!@G!I>@J=;!X[!@;F>=?G=I!=LKBA[L=;AIHF<@J@<[!AE!B?XA>! BAV!D;AVB=IQ=T@;<=;G@J=!G=FG! Change in sector employment vs. change in real Services labor productivity vs. employment and value added per worker value added shares 4 30% Size of bubble reflects Annual change in real VA per worker 3 25% Trade share in value added Share of services employment 2 20% 1 15% Hospitality 0 10% Transport -1 Real estate Financial -2 5% Arts -3 0% -1000 -500 0 500 (20) 180 380 580 Annual change in employment -5% Agriculture Industry Services Real Value added per worker (TL Thousands) Sources: WDI, TURKSTAT, WB Staff estimates. '=X?B?;F@;Q!>=EA>LG!IG!L?F>A!GHF?B!>=EA>L!Q?KG! b?c! *?F>A=FA;AL@F!G=FA;I@<@A;!?<@;Q!&/0! 24.! How to accelerate TFP growth in Turkey? A near term challenge is to continue restoring macroeconomic stability. Though Turkey has effectively weathered demand shocks in recent years, delays in unwinding policy stimuli in 2017 created macroeconomic imbalances that started unraveling in mid-2018. Market volatility and economic stress rose sharply over the summer of 2018. This led to a 50 percent depreciation in the Lira between January and August 2018, with knock on effects on inflation and external debt burden for corporates. 25.! Market volatility in Turkey temporarily subsided since the turbulence in August, but the economic situation remains fragile.23 The external shock in the summer of 2018 translated into significant real sector impacts, including a sharp acceleration in inflation from already elevated levels. Private consumption and investment contracted in the second half of 2018; this combined with rising production costs has prompted supply side adjustments including rising unemployment. Supply side corrections combined with elevated corporate debt, including FX exposure, have raised corporate solvency and liquidity concerns. Though the financial sector entered the recent period of turbulence with adequate buffers, cracks are beginning to appear because of real sector developments and tighter international finance. 26.! The economic outlook is subject to higher levels of uncertainty than usual given high domestic and external vulnerabilities. Growth is projected to slow to a 10-year low in 2019 followed by a medium-term recovery. Monetary tightening and commitments in the New Economic Program (NEP, September 2019) signal important policy adjustment, which temporarily helped calm the markets, though any 23 For more detailed analysis on this, please see: WBG, “Turkey Economic Monitor: Steadying the Ship,” (Dec 2018). Page | 15 15 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey uncertainty or inaction could tip the economy into further difficulty. The lack of progress on an orderly deleveraging in the private sector could precipitate this tipping point. 27.! A deep recession resulting from a hard landing can be very detrimental to TFP growth.24 Firstly, a contraction in investments, exacerbated by an ongoing credit crunch, will delay much needed productivity enhancing expenditures. And as discussed above, given the depth of the financial market and the pro-cyclical nature of finance in Turkey, the scope for funding long-term projects is very limited. Secondly, deep recessions can also erode human capital through extended unemployment spells. Output contractions can lower potential growth by as much as 1 percentage point on average four to five years after the onset of contraction.25 28.! The authorities’ New Economic Program released in September 2018 provides a good foundation for gradually restoring macro stability. Building on the NEP, a consistent package of economic policies could ensure an orderly adjustment for the Turkish economy. This would include tight monetary policy, complemented by a financial sector response that supports gradual deleveraging and enhances financial risk monitoring and management. Critical to supporting the deleveraging process is a strong corporate debt restructuring framework, the absence of which could spell the difference between an orderly adjustment for the economy and a hard landing. Fiscal adjustment will be necessary to help the economy tide over the difficult period ahead. Some key elements of the economic policy program could include: (i)! Tight monetary policy: Recent monetary tightening through interest rate hikes are helping to restore price stability, maintain exchange rate stability, and rebuild external buffers. Price pressures since the summer have been driven by cost-push rather than demand pull factors. Tight monetary policy should support a gradual deleveraging in the private sector. Macro-policy mix should therefore target a gradual recovery in demand through countercyclical fiscal policy relative to boosting short-term supply through accommodative monetary policy; the latter is likely to be ineffective at a time when the economy is experiencing a negative output gap. Premature loosening of monetary policy could fuel further exchange rate pressures, increased costs, and further supply cuts. In addition, sustaining the monetary policy framework rationalized in May 2018, including adoption of a central policy rate, is important for monetary policy transparency. (ii)! Consistent financial sector policies: Credit to the private sector has started to adjust down very significantly. Evidence from past financial crises that were preceded by credit booms, as in the case of Turkey, suggests that credit plays little role in supporting economic recovery after growth has bottomed out.26 Therefore, efforts to curtail deleveraging (e.g. through credit guarantees, loosening macroprudential regulations) are likely to be counter-productive. The focus should be on analyzing the impact of current conditions (i.e. weak Lira, economic downturn, credit crunch) on banks’ credit risk, liquidity, and capital. This would help target interventions, including potential resolution of problem banks. (iii)! Corporate debt resolution: The above analysis should provide details on the links between the financial system and corporate debt distress. This would provide the basis for a corporate debt resolution framework. There is a Concordat system adopted earlier this year to enable companies to negotiate debt restructuring through the courts with all creditors. The authorities are also 24 WBG, “Global Economic Prospects: Broad-Based Upturn, but for How Long,” (January 2018). 25 Ibid. 26 Takats, E. and Upper, C (July 2013) “Credit and growth after financial crises,” BIS Working Papers (No. 416). Page | 16 16 CHAPTER 1 / DRIVERS AND CHALLENGES OF ECONOMIC GROWTH exploring out of court options like the Istanbul Approach adopted in 2001, and there has been talk of setting up an Asset Management Company to (temporarily) absorb troubled assets. Whatever the mechanism, corporate debt resolution is central to an orderly adjustment; it can help provide much needed breathing space for both corporates and banks, without which there are heightened risks of corporate insolvency, rapid deterioration of banks’ asset quality, debt overhang, and potential government bailout. (iv)! Complementary macro-prudential policies: These processes can help further enhance Turkey’s already extensive macroprudential toolkit,27 which has played an important role in containing risks in the financial sector, including those transmitted through volatile capital inflows. Demand shocks in recent years, however, led to some loosening of macroprudential regulations in 2016.28!Though this contributed to countercyclical finance, the policy mix should now be revisited. Macroprudential instruments are central to the effectiveness of monetary policy targets.29 Macroprudential measures should be focused on financial stability (countercyclical buffers, mitigating systemic risks, liquidity). This means unwinding short-term relaxation of macroprudential policies aimed at accelerating consumption or expanding sector investments. (v)! Fiscal adjustment: Credible tightening of monetary policy, with consistent financial sector and macro-prudential policies, will require careful adjustment to fiscal policy. In the short-term, to ensure that tighter financing does not lead to a sudden stop, supply side subsidies (e.g. minimum wage support, tax relief) need to be withdrawn gradually (which is important too for longer-term productivity as discussed above). There may also be scope to adjust other inefficient expenditure to ease pressures on the supply side of the economy; this requires deeper analysis of public expenditures as proposed in the New Economic Program 2019-2021. But in general, fiscal policy will need to play an important countercyclical role, particularly through public transfers given the projected declined in demand. In addition to short-term fiscal measures, it is also important to maintain momentum medium- term fiscal policy reforms, some of which are highlighted in the New Economic Program (2019- 2021) that are critical to TFP. These are covered in more detail by an earlier World Bank report,30 and include among other things: (i) a rebalancing of tax burden from labor towards capital, including through property tax and rationalization of tax incentives, which can have positive impacts on domestic savings and labor formality; (ii) containing recurrent spending growth, and a slight rebalancing towards good quality public investments. 27 Kara, H. (2016): “A brief assessment of Turkey’s macroprudential policy approach: 2011-2015”, Central Bank Review 16 (2016). 28 Baziki, S.B. (2017): “Impact of macroprudential policies on loan utilization,” CBRT Blog. 29 Chadwick, M.G. (2018): “Effectiveness of monetary and macroprudential shocks on consumer credit growth and volatility in Turkey,” Central Bank Review. 30 WB (May 20, 2014), “Turkey Public Finance Review: Time for a Fiscal Policy Pivot?” Page | 17 17 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 29.! Clear communication of such a package of economic policies is central to avoiding a short- term challenge becoming a longer-term problem. A predictable, credible and transparent policy framework is essential for market stability. This would provide a clearer indication of how the authorities plan to manage a soft landing. This means protecting the integrity of macroeconomic institutions and policy anchors, which Turkey has significantly strengthened over the past decade and a half.31 Key among those institutions and policy anchors are an independent Central Bank; monetary policy framework based on inflation targeting; strong bank supervision; transparency of public finances; a medium-term expenditure framework; and sound public debt management. bXc! 0>@A>@<@=G!@;!&H>D=[dG!G<>HF?B!>=EA>L!Q?KG!! ! 30.! The relative importance of structural reforms could be assessed through competitiveness, capability and connectedness (3Cs) challenges to expanding pro-development sectors. This framework, proposed by Hallward-Driemeier and Nayyar (2017), suggests that success in different manufacturing and service sectors can be a function of performance across the 3Cs. They can determine the extent to which countries are able to take advantage of technological developments and changes to production patterns to deepen supply capacity. 31.! In sum: (i) Competitiveness factors consider the efficiency of the business environment to offset increased labor costs (indicators include ease of doing business, the rule of law, and the use of mobile technologies to complete financial transactions); (ii) Capability factors consider the ability of workers and firms to adopt and use new technologies (indicators include ICT use, tertiary school enrolment rates, and the share of royalty payments and receipts in trade); (iii) Connectedness factors look at not only shifts in international trade, but also the cross sector synergies needed for success in manufacturing (indicators include logistics performance, restrictions on trade in manufactured goods, and the restrictions on trade in professional services). 32.! Turkey performs relatively well on connectedness, though, it is only just above average on capabilities and is a middling performer on competitiveness (Figure 22). High performer comparators have strong scores across all three dimensions, whereas trapped MIC comparators on average do not perform as well as Turkey. These results are broadly consistent with findings from other chapters in the report – chapters 3 on integration; chapter 5 on labor market capabilities; and chapter 6 on competition. The risks and implications in terms of growth of manufacturing sectors is discussed in more detail in chapter 2. In general, though, Turkey needs to address competitiveness challenges most urgently if it is to expand more sophisticated and skill intensive sectors through higher quality financing. 31See IMF, “Structural Reforms and Macroeconomic Performance – Country Cases,” (November 2015); and WBG, “Turkey’s Transitions: Integration, Inclusion, Institutions,” (December 2014). Page | 18 18 CHAPTER 1 / DRIVERS AND CHALLENGES OF ECONOMIC GROWTH /@QH>=!77.!&H>D=[!GFA>=G!V=BB!A;!FA;;=F<=I;=GG!?;I!F?K?X@B@<@=G!XH=!7S.!&H>D=[!B?QG!LAG!L?>D==!7R.!2?KG!X=D=[!?;I!)$!?J=>?Q=! @;;AJ?<@A;]!E@;?;F@?B!G=F]!NHL?;!F?K@=?<=G!NHL?;!F?K@!L?>D==!7^.!,BBAF?<@J=!@;=EE@F@=;F@=G!K=>K=! Sector TFP (weighted average) vs. credit growth and credit outstanding 50% Real Estate Brokerage 40% Hotels and Restaurants Credit growth Construction 30% Manufacturing Transport & Communicatioin Wholesale & 20% Retail 10% 2,5 3,5 4,5 5,5 6,5 TFP (weighted average) Sources: EIS, Haver Analytics, WB Staff estimates. Notes: Credit growth is annual average for the 2010-2017 period; TFP level for 2015; credit outstanding for 2017. "A;FBHG@A;!?;I!KAB@F[!@LKB@F?<@A;G! 37.! The main messages that emerge from the analysis on the drivers and challenges of economic growth are as follows: (i)! Strong economic convergence with limits on sustainability: Turkey’s ability to absorb short- term demand shocks in the past years has helped sustain a strong pace of convergence to high income status. But a lack of focus on long-term supply capacity has dented the sustainability of growth with risks of premature deindustrialization. (ii)! A productivity boost is needed to sustain improvements in living standards: Economic growth in Turkey since the 1980s has been driven largely by factor accumulation, with periodic boosts in productivity. Labor incomes will stagnate unless more employment is created in relatively higher productivity industry or more skill intensive services sectors. (iii)! Restoring macroeconomic stability is a precondition to accelerating TFP: A consistent package of economic policies could be central in ensuring an orderly adjustment – tight monetary policy, corporate deleveraging and debt resolution, consistent macroprudential policies, and fiscal adjustment. Communication of stabilization is key to anchor economic expectations. (iv)! Priorities in Turkey’s structural reform gaps: Address competitiveness challenges most urgently if Turkey is to expand more sophisticated and skill intensive sectors through higher quality financing. Page | 21 21 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 38.! The rest of this report builds on the above to identify challenges and options for accelerating TFP growth through reforms that: (i) deepen Turkey’s international economic integration as a source of learning and technology that enhances firm productivity and growth; (ii) improve the investment climate, including through greater competition, to facilitate the reallocation of resources to more productive firms and sectors; (iii) improve the quality of business support services to promote innovation and firm productivity; and (iv) equip workers with the skills needed to participate in a more modern economy. Issues Policy options Restoring macroeconomic stability Though Turkey has effectively weathered A consistent package of economic policies could be demand shocks in recent years, delays in central in ensuring an orderly adjustment: unwinding policy stimuli in 2017 have led to large macroeconomic imbalances that risk •! Tight monetary policy; operational and policy unraveling with tighter financial conditions at independence of the Central Bank; credible inflation home and abroad. target supported by a transparent and predictable adjustment to policy rates. The risks of a boom-bust scenario have been exacerbated due to heightened external •! Financial sector diagnostic; corporate sector vulnerabilities. deleveraging; corporate debt resolution. •! Unwinding of short-term relaxation of macroprudential policies aimed at accelerating consumption or expanding sector investments. •! Gradually withdraw supply side subsidies provided through the budget. •! Clear communication of stability package. Addressing priority structural reform gaps Turkey performs relatively well on •! Boost demand for more educated and skilled workers connectedness, though, it is only just above in the economy through improved quality of financial average on capabilities and is a middling flows: (i) prioritizing FDI policy including links to performer on competitiveness. local suppliers and the liberalization of foreign Turkey’s biggest competitiveness gaps relative investment in the services sector (Chapters 3, 6); (ii) to high performer comparators are in the areas accelerate development of non-bank financial sector. of labor markets, innovation, financial sector, and human capital. ! Page | 22 22 CHAPTER 1 Chapter 2 Manufacturing, construction and services productivity Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Manufacturing, construction and services 23 productivity Manufacturing: Some breakout sectors but 26 falling allocative efficiency weighs on productivity Construction: Growing but low productivity 33 sector suffering from allocative inefficiency Services: Declining allocative efficiency and 37 within sector productivity growth Boosting pro-development sectors will require 40 competitiveness reforms Conclusion and policy implications 47 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY ++W!*?;HE?F@;Q]!FA;G<>HF<@A;!?;I!G=>J@F=G! K>AIHF<@J@<[! 1.! Turkey has experienced important structural transformation in the past twenty years. A drop in the share of agriculture value added (11 to 7 percent between 1998 and 2016) was offset by industry (26 to 30 percent), whilst services remained relatively flat (63 percent). Agriculture employment share fell sharply (36 to 20 percent between 1998 and 2016), offset by an increase in service’s share (38 percent to 54 percent). Underlying these macroeconomic trends were big shifts within sectors. In industry, the construction sector’s share in value added and employment rose most quickly; within services the value-added share of the financial sector grew quickly from a low base (4 to 8 percent between 1998 and 2016), though it is retail and wholesale, accommodation and food services that absorbed most labor (14 to 19 percent between 1998 and 2016). 2.! This chapter disentangles productivity dynamics within manufacturing, construction and services using firm-level data (Box 3), to see how these add up to explain economic growth in Turkey. Microeconomic analysis of firms and sectors can provide more detail on productivity challenges that constrain growth at a macro level, and from this help target policy more effectively. The chapter: (i) starts with a review of productivity levels across sectors within manufacturing; (ii) it reviews the link between productivity levels and allocative efficiency; (iii) it then analyses drivers of efficiency – namely productivity within firms, resources shifting across firms, and creative destruction. This analysis is repeated for the construction and services sectors. It concludes with a review of risks and implications for long-term growth. CAZ!S.!,KK>A?FN!?;I!L=!E@>LTB=J=B!K>AIHF<@J@<[!?;?B[G@G! Data source and coverage: The firm-level data used in this chapter and the rest of the report comes from the Entrepreneur Information System (EIS) database provided by the Ministry of Industry and Technology. The EIS is compiled from various administrative sources. It contains balance sheet, income statement, and trade flow data for all firms34 in Turkey over the 2006-2016 period. The report however does not analyze all non-agriculture sectors in the economy. Almost all 2-digit manufacturing sectors are covered except for repair and installation of machinery and equipment. In services, sectors excluded from the analysis are finance and insurance, public administration, education, health and other services; these sectors either have very few firms or are state owned. Further detail on the data is provided in Technical Appendix. Productivity measures: This report uses two main measures of productivity. The first is labor productivity which is defined as the real value added per worker. The second is a Total Factor Productivity measure estimated using the Ackerberg, Caves and Frazer (ACF) (2015) methodology. The ACF method is based on a two-step estimation procedure to help overcome the issue of functional dependence when the elasticity of labor is estimated in the first stage as done in Olley and Pakes (1996), and Levinsohn and Petrin (2003). Intermediate inputs are used as a proxy to control for unobserved productivity shocks. While estimating TFP, real value added is used to measure the output of a firm. Since firm level price data is not available, 2-digit sectoral deflators are used to estimate real value added at the firm level. The TFP measure in this report is therefore a revenue-based indicator. 34 The only exception is banking sector which is not covered in the dataset. Page | 23 23 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Challenges with revenue-based TFP (TFPR) measures: Deflating a firm’s nominal output or value added with a sectoral deflator poses several difficulties. Firms within a sector might produce similar goods but may charge different prices for that good depending on a firm’s market power or product quality. Sectoral deflators cannot control for these differences. For firms producing near identical products in a highly competitive market, this is less of an issue. But in other cases, the extent and quality of design, craftsmanship, raw materials, and other inputs might differ, making the final product quite different. TFPR then overestimates the productivity of firms producing high price (quality) products and underestimates the productivity of firms producing low price (quality) product as revenues of two types of firms are deflated by using the same deflator at sector level. A quantity-based TFP (TFPQ) measure: The availability of product-level price data can help address shortcomings of TFPR measures of productivity. Firm-level input and output prices can help disaggregate firm performance by its physical efficiency, market power, and product quality. The residual in the production function (namely TFPQ) therefore gives a more precise estimate of firm productivity relative to market power and product quality. In Turkey, however, the lack of economy- wide information on firm-level input and output prices makes it difficult to use the TFPQ measure. Other ways to estimate TFPQ: To get around the data issues, there are two alternative ways that TFPQ could be estimated: (i) unit value of exports could be used as a proxy for product prices. However, this would mean excluding non-exporting firms, which account for the majority in Turkey; (ii) Hsieh and Klenow (2009) offer an alternative methodology to estimate TFPQ in the absence of firm-level prices. The correlation between the estimated TFPQ and TFPR is around 0.51. But TFPR is still preferred this report as Hsieh and Klenow (2009)’s methodology relies on very strong assumptions and does not control for unobserved productivity shocks unlike the ACF methodology. TFPR vs. TFPQ: Haltiwanger (2016) argues that researchers should not inherently prefer TFPQ measures over TFPR as the latter “have the virtue that they will reflect idiosyncratic profitability factors beyond TFPQ”. In other words, TFPQ is a good measure of technical efficiency. However, if one would like to compare firms in terms of their capacity to earn profits or create value either through producing high quality products or exerting market influence, TFPR seems as a better measure. Additionally, there are studies that measure TFPQ using firm level price and quantity data, which also show that traditional measures of TFPR and TFPQ are highly correlated. Foster, Haltiwanger, and Syverson (2008) consider a traditional revenue productivity measure where output is deflated with industry level price indices and they find a correlation of 0.75 between this traditional measure of TFPR and TFPQ. Using Colombian data that includes data on firm level prices and quantity, Eslava et al. (2004, 2013) find a correlation of 0.69 between TFPR and TFPQ. 3.! Results from aggregating firm level productivity estimates within manufacturing, construction and services are consistent with macroeconomic trends discussed in the previous chapter. The macro analysis looked at total industry trends; the micro analysis on the other hand focuses on manufacturing (63 percent of industry output) and construction (27 percent of industry output). Manufacturing is of big interest given its pro-development characteristics (Chapter 1), and construction is of interest given its rapid expansion in the past ten years. Services account for the largest share in GDP and provide critical inputs that affect the productivity of other sectors in the economy. Firm level analysis of services excludes financial services and public services (e.g. education, health). Results from aggregating firm level TFP and labor productivity in these three sectors show that: Page | 24 24 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY (i)! Aggregate productivity: Has fallen and remained below pre-GFC levels.35 Labor productivity picked up recently but remains low (Figure 26). (ii)! Productivity within manufacturing: Weighted TFP dipped slightly in most recent years, but overall has remained relatively flat (Figure 27). (iii)! Productivity within construction: TFP has remained low and generally on a declining path since 2006 (Figure 27). (iv)! Productivity within services: TFP has remained low and on a declining path (Figure 27), which has driven much of the drop in aggregate TFP fall. /@QH>=!7_.!Y?XA>!K>AIHF<@J@<[!?;I!&/0!@;!L?;HE?F@;Q]! /@QH>=!7U.!*?;HE?F@;Q!K>AIHF<@J@<[!>=L?@;G!EB?HF<@A;!?;I!G=>J@F=G!A;!?!I=FB@;@;Q!K?=?G!FA;G<>HF<@A;!?;I!G=>J@F=G!?>=!BAV!?;I!I=FB@;@;Q! Labor productivity and TFP (manufacturing, TFP (weighted by value added share) construction and service firms, weighted by value added share) 5,7 6,4 5,9 Labor productivity 5,45 5,6 TFP TFP 5,4 5,5 4,9 5,35 5,4 4,4 06 07 08 09 10 11 12 13 14 15 16 06 07 08 09 10 11 12 13 14 15 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Labour Productivity TFP - Secondary Axis Manufacturing Construction Services Sources: EIS, WB Staff estimates. 4.! Firm level data also confirm the importance of labor accumulation as a driver of growth in the past decade relative to productivity. Much of the labor shed from agriculture was absorbed in construction and services rather than manufacturing. Results from aggregating firm level employment and value-added show that (Figure 28): (i)! Manufacturing: Employment growth kept pace with growth in value added, contributing to flat labor productivity over this period. (ii)! Construction: Employment growth far outpaced value-added growth, by 30 percentage points on average per year between 2006 and 2015, contributing to a sharp drop in labor productivity. (iii)! Services: Employment growth also outpaced value-added growth, by 18 percentage points on average per year between 2006 and 2015, contributing to a fall in labor productivity. 35The productivity figures for 2016 are excluded throughout most of the report. In 2016, official employment contracted due the shock of failed coup attempt and the increase in minimum wage by 30 percent. However, this employment contraction is not observed in household labor survey as informal employment increased by around 6 percent in 2016. This distorts the labor productivity estimates. Page | 25 25 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!7O.!)LKBA[L=;AVAVHF<@A;!?;I! G=>J@F=G! Employment and value added growth (2006=100, 2006-2015) 290 Value added growth (2006 = 100) 240 190 140 90 90 140 190 240 290 Employment growth (2006 = 10) Manufacturing Construction Services Sources: EIS, WB Staff estimates. *?;HE?F@;Q.!PAL=!X>=?DAHG!XHAIHF<@J@<[! b?c! 0>AIHF<@J@<[!K?<<=>;G!V@@;Q!V=@QN!A;!AJ=>?BB!L?;HE?F@;Q!&/0! 5.! Within the manufacturing sector, tradable sectors are more productive than non-tradable sectors (Figure 29). Sectors that are more globally integrated and engage in more technologically intensive production have higher than average TFP levels; they include motor vehicles, basic metals, transport equipment, chemicals, pharmaceutical, chemicals, textile and wearing apparel. In contrast, sectors involved in the processing of agricultural products or those that have higher state presence have lower than average productivity levels; they include: food, tobacco, beverages or coke and refined petroleum. 6.! Current productivity patterns within manufacturing have been a drag on the growth of the manufacturing sector (Figure 30). This is driven by low productivity sectors (below average TFP) capturing close to 37 percent of value addition and 44 percent of employment. Relatively large sectors (in terms of value added and employment) that exhibit low productivity include food, electrical equipment, fabricated metals and machinery equipment. These sectors may be experiencing one of or a combination of: low or negative within firm productivity growth; a shift of labor and capital resources less productive firms; limited creative destruction (i.e. net entry of productive firms). This is analyzed in more detail below. 7.! High productivity manufacturing sectors that grew quickly include those that have benefited from foreign investment and exports. Key among those is the motor vehicles sector, which has been an export champion in the last decade. But others that are also relatively more productive, and capture above average share of employment and value addition include: basic metal, rubber plastic, textile and wearing apparel. Other more sophisticated sectors exhibit high productivity, but are relatively small employers and drivers of value addition e.g. basic pharmaceuticals, chemicals. Page | 26 26 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY /@QH>=!7:.!&/0!B=J=BG!@;!<>?I?XB=!G=FG!?>=!N@QN=>!?I?XB=!G=FG! Weighted TFP levels within manufacturing sectors (2015) 8 6 4 2 0 Wood Basic Pharm. Oth. Manuf. Furniture Tobacco Motor Vehicles Leather Comp., Elect.&Optical Basic Metals Paper Chemicals Wearing Apparel Textiles Beverages Food Mach.&Equip. Electr. Equip. Coke&Ref.Petr. Oth. Non-Met. Rubber&Plastic Print.&Rec.Media Fabr.Metal Other Transport TFP Average TFP - manufacturing Sources: EIS, WB Staff estimates. /@QH>=!S8.!*?;HE?F@;Q!Q>AV?@;=I!X[!B?>Q=]!BAV!K>AIHF<@J@<[!G=FG! Manufaturing: Weighted TFP levels vs. value added and employment shares within manufacturing sectors (2015) 16% Motor Vehicles 14% 12% Textiles Basic Metals Value Added Share 10% Food Fabricated Metal… Electrical… Oth. Non-Metalic 8% Manufacturing Mineral averages Rubber and Plastic 6% Machinery & Equipment Wearing Apparel Computer, Electronic and Chemicals 4% Optical Products Basic Pharmaceuticals Wood Paper Furniture Oth. Transport 2% Beverages Oth. Equipment Manufacturing Leather 0% Printing and Media 3 4 5 6 7 8 9 TFP Sources: EIS, WB Staff estimates. ! ! Page | 27 27 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey bXc! ,BBAF?<@J=!=EE@F@=;F[!@;!L?;HE?F@;Q!@G!KAG@<@J=!XH=!S9.!!,BBAF?<@J=!=EE@F@=;F[!@;!L?;HE?F@;Q!KAG@<@J=!XH=!S7.!!*AGG!N?J=!N@QN!&/0!XH@=;F=I!?!I>AK!@;!?BBAF?<@J=!=EE@F@=;F[! Allocative efficiency and TFP within manufacturing sectors (2006 vs. 2015) Motor Basic vehicles metals 0,40 Other transport Chemicals Paper ONM 0,30 R&P Textiles Allocative efficiency Wood Apparel Leather Computers 0,20 Fabr. Metal Basic M&E etc. Prod. Pharma Furniture Tobacco Printing Elec. 0,10 Equip. Coke & Ref Pet. 0,00 Beverages Other manuf. Food -0,10 2,50 3,50 4,50 5,50 6,50 7,50 8,50 TFP Sources: EIS, WB Staff estimates. 11.! Manufacturing TFP in Turkey could grow substantially if efficiency of resource allocation improved. Hsieh and Klenow (2009)38 have developed a methodology to measure how much aggregate manufacturing TFP could increase if capital and labor were reallocated to equalize marginal products across firms within each sector, or if all resource misallocation were removed. In Turkey, if resources were reallocated such that the marginal products of labor and capital were equalized, manufacturing TFP would increase by 63 to 78 percent over the period of analysis. This result is consistent with Nyguen et al.’s (2016) study which find large potential to improve allocative efficiency in the Turkish manufacturing sector. bFc! 3@L!K>AIHF<@J@<[!Q>AV@;Q!AEEG==!SS.!3@L!K>AIHF<@J@<[!Q>AV<@?BB[!AEEG==?BBAF?<@A;!?;I!;=[! Manufacturing TFP growth decomposition (2010-2015) 10% 5% 0% -5% -10% -15% 2007 2008 2009 2010 2011 2012 2013 2014 2015 Within Reallocation Entry Exit Change in Productivity Sources: EIS, WB Staff estimates. 14.! The negative impact of reallocation and net entry on manufacturing productivity is evident across individual manufacturing sectors. Within each sector, the reallocation of resources between firms has made a negative contribution to TFP growth for that sector (Figure 34). Despite some indications of relatively more productive firms entering and relatively less productive ones exiting, the impact of net entry is neither clear nor definitive except for a handful of sectors (e.g. other manufacturing, rubber and plastics, textiles). Even the most productive sector – motor vehicles – was impacted negatively in terms of productivity in the 2010-2015 period by reallocation effects within the sector. Page | 30 30 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY /@QH>=!SR.!%=Q?<@J=!>=?BBAF?<@A;!?;I!;=[!=EE=F=!=J@I=;@;Q! TFPR growth decomposition by manufacturing sectors (2010-2015) 15% TFPR growth and contribution 10% 5% 0% -5% -10% -15% Textiles Basic Metals Rubber&Plastic Furniture Comp., Elect.&Optical Wearing Apparel Wood Food Print.&Rec.Media Mach.&Equip. Electr. Equip. Coke&Ref.Petr. Oth. Manuf. Basic Pharm. Oth. Non-Met. Chemicals Beverages Paper Leather Motor Vehicles Other Transport Fabr.Metal Within Reallocation Entry Exit Change in Productivity Sources: EIS, WB Staff estimates. 15.! Within firm productivity growth helped offset the negative reallocation and net entry impacts on productivity, and may have helped sustain positive, albeit declining, allocative efficiency. Confirming the latter would require more analysis of the interactions between within-firm productivity growth and measures of allocative efficiency.40 But without productivity gains within firms, allocative inefficiencies from negative reallocation and net entry could have dragged manufacturing TFP down. 16.! Much of this within firm productivity growth was driven by larger and more mature firms, as well as firms engaged in external trade (Box 4). The manufacturing sector in Turkey is characterized by a few large firms, together with many small firms. Manufacturing firms with more than 500 employees make up less than 1 percent of all firm in manufacturing. However, these firms account for 28 percent of employment and 53 percent of total value added in manufacturing. Due to their large share in total value added, these very large firms drive the weighted average of productivity in the whole manufacturing industry. 17.! Yet, there is also evidence that the productivity of the largest firms in recent years has slowed down.41 These very large firms are, on average, more productive than other manufacturing firms throughout the period of analysis (Figure 35). However, the productivity of these firms dropped by almost 6 percent on average between 2011 and 2015, whereas the productivity of an average firm increased by 5 percent. Therefore, flat productivity and allocative efficiency trends in manufacturing can also be attributed to very large firm becoming less productive (Figure 31 above). 40 See for example Decker, Ryan A., John Haltiwanger, Ron S. Jarmin, and Javier Miranda (2017). “Declining Dynamism, Allocative Efficiency, and the Productivity Slowdown,” Finance and Economics Discussion Series 2017-019. Washington: Board of Governors of the Federal Reserve System (link). 41 Within the group of large firms, a special group is defined by picking those that have a value-added share of 1 percent or more. This group consists of the largest manufacturing firms in Turkey which are very few in numbers. However, their value-added share in total manufacturing is around 21 percent and their employment share is 5 percent as of 2016. Due to confidentiality rules, the information about the number of firms belonging to this group or NACE code of the sectors they operate in cannot be shared. Page | 31 31 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!S^.!Y?>Q=GLG!?>=!LA>=!K>AIHF<@J=]!XH! /@QH>=!S_.!&N@G!L?[!X=!IH=!AIHF<@J@<[!Q>AVQ=GLG! TFP of very large firms and the rest 2006-2015) Capital to labor ratio of very large firms and the 7,0 4,60 rest (2006-2015) 1900 500 6,9 1700 4,56 450 6,8 1500 6,7 400 4,52 1300 6,6 350 1100 6,5 4,48 900 300 06 07 08 09 10 11 12 13 14 15 06 07 08 09 10 11 12 13 14 15 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 TFP-very large firms (LHS) TFP-Rest (RHS) K/L Ratio- very large firms (LHS) K/L Ratio- Rest (RHS) Sources: EIS, WB Staff estimates. 18.! One of the possible reasons behind this could be capital deepening of very large firms, which may dent productivity in the short-term but boost productivity in the long-term. The evidence does in fact show that the capital to labor ratio (K/L) of the largest firms is much higher over the period of analysis. K/L ratios of the largest and other firms follow similar trends until 2010 (Figure 36). After 2010, the K/L ratios of the largest firms started to rise rapidly whereas the K/L of other firms grew more slowly. This may partly explain the decline in the TFP of the largest firms, especially as this is not matched by a decline in their labor productivity. It is worth exploring whether the increase in the K/L ratio is due to investment in productive capital (such as machinery and equipment or R&D) or due to relatively unproductive capital (such as land). This question deserves more analysis as this will impact on long-term productivity in manufacturing. 19.! The above findings have important policy implications, which are discussed in other parts of the report. Falling allocative efficiency in manufacturing is linked to the shift in policy emphasis towards short-term demand stimulus and supply subsidies (Chapter 1), constraints to competition (Chapter 6), and targeting of public grants for private sector growth and innovation (Chapter 4). Within firm productivity growth is encouraging, but there is more scope for productivity enhancing investments (Chapter 1), technology transfer and learning (Chapter 3), and improving firms’ capacity to absorb technology through human capital (Chapter 5). Page | 32 32 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY CAZ!R.!/@>L!FN?>?F<=>@G<@FG!?;I!K>AIHF<@J@<[! Firm characteristics and productivity: The size, age, ownership and tradability of firms are the most important correlates of firm level productivity.42 These characteristics are also correlated to each other. To understand the relative importance of each characteristic therefore requires controlling for the other characteristics. A TFP regression was run on size, age, exporting status, importing status, technological intensity and year dummies. Manufacturing firm size: Firm size is found to be positively associated with higher TFP. Small firms employing less than 19 employees account for around 70 percent of manufacturing firms in Turkey. Even though large firms constitute a small part of the sector, they produce over two thirds of the value added in the sector, Therefore their productivity performance weighs heavily on aggregate productivity. Manufacturing firm age: More mature firms are found to be more productive than younger firms in the manufacturing sector. This reflects learning effects as firms that defend or extend their market position can learn over time about both what products the firm’s capabilities are best suited for and how to improve those capabilities (World Bank, 2014). At the same time, however, young firms exhibit very strong growth potential in Turkey and tend to be more innovative than older firms (Chapter 4). Manufacturing firms’ external trade: In Turkey, exporter firms are, on average 26 percent more productive than non-exporters. Importers are, on average, 41 percent more productive than non- importers. Competitive pressures from international markets can enhance the efficiency of domestic firms. Exporting can also push firms to learn, access technology and grow.43 Similarly, importing firms can access better quality inputs. Manufacturing firms’ technological intensity: Firms in high technology sectors are, on average, more productive than the rest. But there is no statistically significant difference in average productivity between medium, medium-low and low-tech industries. Most of Turkish manufacturing operates with low and medium-low technology. Since the value-added share of high-tech industries is only around 6 percent (2016), high productivity in those industries does not weigh heavily in aggregate productivity. These are nevertheless important sectors for future growth of manufacturing in Turkey. Service sector firms’ characteristics and productivity Firm size, age, technological intensity, external trade matter for service sector firms’ productivity too. As in manufacturing: (i) productivity increases significantly with age; (ii) exporting and importing firms are significantly more productive than non- exporters and non-importers respectively; (iii) high-technology knowledge intensive services (KIS) sectors are the most productive followed by firms in other KIS, less KI market services and KI market services sectors, respectively. However, the relationship between firm size and productivity in the services sector is very different to that in manufacturing. Firms with 2-9 employees are the most productive group. Foreign ownership: Foreign-owned firms are usually more productive than domestic firms due to their high-quality resources and superior technological, organizational, management and marketing practices. Based on TURKSTAT firm level data,44 domestic firms exhibit very low levels of productivity compared to firms with any type of foreign involvement. This is evident for both manufacturing and service sectors. "A;G<>HF<@A;.!2>AV@;Q!XHAIHF<@J@<[!G=F!GHEE=>@;Q!E>AL!?BBAF?<@J=!@;=EE@F@=;F[! Page | 33 33 services sectors, respectively. However, the relationship between firm size and productivity in the services sector is very different to that in manufacturing. Firms with 2-9 employees are the most productive group. Foreign ownership: Foreign-owned firms are usually more productive than domestic firms due to their high-quality resources and superior Country technological, Economic organizational, Memorandum: management Firm Productivity and marketing and Economic Growth in Turkey practices. Based on TURKSTAT firm level data,44 domestic firms exhibit very low levels of productivity compared to firms with any type of foreign involvement. This is evident for both manufacturing and service sectors. "A;G<>HF<@A;.!2>AV@;Q!XHAIHF<@J@<[!G=F!GHEE=>@;Q!E>AL!?BBAF?<@J=!@;=EE@F@=;F[! b?c! "A;G<>HF<@A;!IAL@;?<=I!X[!B=GG!F?K@AIHF<@J@<[!G=FG 20.! Productivity in the construction is considerably lower than in the manufacturing sector, and has been on a declining trend over the period of analysis. This coincided with a period of rapid | 33 for the construction sector. The construction sector has three sub-sectors namely: expansion Page 45 (i)! Construction of buildings (development of building projects, construction of residential and non-residential buildings): This sub-sector has the highest level of productivity, though that level is low and has been declining over the past decade (Figure 37). It is the largest sub-sector, accounting for around 70 percent of employment and 60 percent of value added in construction sector (Figure 38). It also includes airport and hospital construction projects, which are mostly Public Private Partnership involving a lot of subcontracting enterprises with various specializations. Activity in the construction of buildings sector is highly cyclical, influenced by business and consumer confidence, interest rates and government programs.46 (ii)! Civil engineering construction (constructions of roads, railways, bridges, tunnels and utility projects): Though TFP levels are lower than in the construction of buildings, labor productivity is higher, which is likely due to high capital intensity of civil engineering construction. Civil engineering works are carried out mostly through the public-sector; they play a significant part in the infrastructure. The recent expansion of large PPP infrastructure projects falls under this category. (iii)! Specialized construction activities (demolition, site preparation, electrical, plumping, sewerage activities etc.): Activities are mostly labor-intensive sector and stand out as the worst performer in both labor productivity and TFP. Activities are mostly sub-contracted. has the lowest value-added share but account for a larger share of employment compared to civil engineering. 42 See Diaz and Sanchez (2008), Biesebroeck (2005), Lundvall and Battesse (2000), Bernard and Jensen (1999, 2001), Girma et al. (2004), Halpern et al. (2005), Head and Ries (2003), Castellani (2001) among others. 43 Bernard and Jensen, 1995; 1999; Baldwin, 2000; Giles and Williams, 2000a, 2000b; Yasar et al., 2006; Loecker, 2007. 44 There is no firm ownership information available in EIS dataset. 45 Turkish construction sector expanded significantly in the last decade and Turkish construction companies have increased their global presence. According to rankings by the magazine Engineering News-Record, (ENR the world’s top 250 construction companies by overseas operations), Turkey ranked as a second country in terms of its total number of firms in the list (46 companies) followed by China. 46 http://ec.europa.eu/eurostat/statistics-explained/index.php/Construction_of_buildings_statistics_-_NACE_Rev._2. Page | 34 34 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY /@QH>=!SU.!&/0!?F>AGG!?BB!FA;G<>HF<@A;!GHXTG=FG!@G!BAV! /@QH>=!SO.!"A;G<>HF<@A;!AE!XH@BI@;QG!@G!LAGAIHF<@J=! >=B?<@J=!@;Q! V@=G!! 5,0 70% 4,8 60% Share of value added 50% 4,6 Construction of buildings 40% 4,4 30% Civil 4,2 engineering 20% 4,0 Spec. 10% construction act. 3,8 0% Constr. of Buildings Civil Engineering Speci. Constr. Act. 4,0 4,2 4,4 4,6 4,8 5,0 TFP TFP - Construction average TFP Sources: EIS, WB Staff estimates. 21.! The construction sector is characterized by low skill intensity, with little scope for learning- by-doing and innovation. Hallward- Driemeier and Nayyar (2017) defines construction as a stand-alone sector having low productivity enhancing traits: formal worker training programs, use of foreign technology, exports (direct and indirect), introduction of new products and new processes, and R&D spending. On the other hand, they also argue that there is possibility for technology to enable construction sector to be internationally traded while continuing to generate employment for unskilled labor. ! bXc! ,BBAF?<@J=!@;=EE@F@=;F@=G!@;!FA;G<>HF<@A;!N?J=!@;F>=?G=I!?E<=>!788:! ! 22.! Poor productivity performance in construction is consistent with low allocative efficiency since 2009. Allocative efficiency dropped after the Global Financial Crisis and did not improve much since then (Figure 39). Increased exclusive incentives on construction sector might partially explain the decline in allocative efficiency.47 23.! The construction of buildings experienced both a decline in allocative efficiency and falling productivity in the last decade (Figure 40). This was a big driver of the overall drop in allocative efficiency in the construction sector. This sector has the highest simple average of productivity implying that the average firm in this sector is performing well. However, due to more resources going to the least productive firms, the weighted average of productivity is dragged down. While the relatively small sectors civil engineering and specialized construction activities experienced efficiency improvements in the last decade, they experienced drops in their productivity levels. The improvement in allocative efficiency of these two sub-sectors mitigated the deterioration in the allocative efficiency of construction sector.! ! ! ! 47 See IMF Article IV 2017 Selected Issues for regulatory arrangements on the construction sector. Page | 35 35 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!S:.!,BBAF?<@J=!=EE@F@=;F[!@;!FA;G<>HF<@A;!N?G! /@QH>=!R8.!,BBAF?<@J=!@;=EE@F@=;F[!@;F>=?G=I!@;!Q=G!V@! Construction TFP and allocative efficiency Allocative effiency and TFP within (2006-2015) construction sector (2006 vs. 2015) 4,95 0,11 0,16 Civil 0,10 Engineering 0,14 4,85 Construction 0,09 0,12 of Buildings Allocative efficiency 0,08 4,75 Allocative efficiency 0,07 0,10 0,06 TFP 4,65 0,08 0,05 0,06 4,55 0,04 0,03 0,04 Specialized Construction 4,45 0,02 0,02 Activities 2006-2015 Allocative Efficiency 0,00 TFP - Weighted Average 4,0 4,2 4,4 4,6 4,8 5,0 5,2 Linear (Allocative Efficiency) Linear (TFP - Weighted Average) TFP ! ! Sources: EIS, WB Staff estimates.! ! bFc! '=GAH>F=!L@G?BBAF?<@A;!?;I!V@L!K>AIHF<@J@<[!@LK?FHF<@A;! ! 24.! Poor productivity in the construction sector productivity is due both to allocative inefficiency and within firm productivity (Figure 41). The construction sector suffers from both misallocation of resources and existing firms in construction becoming less productive. As in the services sector, the exit of productive firms weighs heavily on construction productivity. Even though entering firms display a better performance compared to surviving firms in recent period, it did not suffice to offset the large negative effect of productive firms exiting the market. 25.! The negative impact of reallocation and exit on construction productivity is evident across all sub-sectors (Figure 42). In all sub-sectors of construction, the exiting of productive firms is pulling down their productivity and entering firms are impacting positively on productivity. However, the net entry effect is only positive for the civil engineering sector that helped the sector to record productivity growth. The only sector that performed well in the within component following the financial crisis is construction of buildings sub-sector while the other sectors performed very poorly. ! Page | 36 36 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY /@QH>=!R9.!%=[!LAG@XH<@;Q!;=Q?<@J=B[!=!R7.!%=[!=EE=FHF<@A;!G=F!&/0 =;Q@;==>@;Q!GHXTG=F TFP growth and contribution (2007-2015) TFP growth decomposition in construction sector (2010- 2015) 15% 7% 10% 2% 5% -3% 0% -8% -13% -5% -18% -10% -23% 2007 2008 2009 2010 2011 2012 2013 2014 2015 -15% Civil Engineering Speci. Constr. Act. Constr. of Buildings Within Reallocation Entry Exit Within Reallocation Entry Exit Change in Productivity Change in Productivity Sources: EIS, WB Staff estimates. P=>J@F=G.!-=FB@;@;Q!?BBAF?<@J=!=EE@F@=;F[!?;I!V@!K>AIHF<@J@<[!Q>AVJ@F=G!IAL@;?<=I!X[!B=GG!D;AVB=IQ=!@;<=;G@J=]!BAV!K>AIHF<@J@<[!G=FG! 26.! Services in Turkey are dominated by sectors that are relatively less knowledge intensive (KI). Wholesale and retail trade account for nearly half of value addition and employment. The weighted TFP level of the service sector is very close to that of the wholesale and retail trade because of the latter’s dominance. More skill intensive sectors on the other hand – such as ICT; scientific and technical activities – account for only 10-15 percent of value addition and employment. Productivity levels for transportation and storage services are below sector average TFP levels, which significantly impacts services productivity given the size of the transport sector (16 percent of value addition) but also manufacturing (Chapter 3). 27.! Not only is productivity in service sectors low and declining, knowledge intensive (KI) sectors have fared particularly poorly. Firms in relatively less KI sectors have higher than average productivity (e.g. administrative and support, wholesale and retail trade) (Figure 43). Firms in more KI sectors on the other hand (scientific, occupational and technical activities, transportation and storage, ICT) have below average productivity levels (Figure 44). Occupational and technical activities, and ICT, however, can act as productivity enhancers. They have capacity to add value compared to traditional sectors (e.g. retail and wholesale, accommodation and food); they offer more scope for learning-by-doing and innovation. ! Page | 37 37 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!RS.!&/0!@;!G=>J@F=!@G!BAV]!@;FBHI@;Q!V@=!RR.!0AA>!K=>EA>L?;F=!AE!>=B?<@J=B[!N@QN!5+!G=FG! D;AVB=IQ=T@;<=;G@J=!G=FG! ?EE=FG!X=[A;I!G=>J@F=G! Weighted TFP levels within service sectors Services: Weighted TFP levels vs. value added (2015) and employment shares 6 65% Trade 5 55% Weighted TFP 4 45% Share of value added 3 35% 2 Transport Service 1 25% averages Amin & Accommodation 0 and food support 15% in . al e t T e od re ch or tat IC m ltu ICT Fo les Te sp Es Ad Real estate Cu Culture an ho & .& 5% al . Tr m Scientific etc. W up Re co il& cc Ac O ta Re t. ien -5% Sc 2,5 3,5 4,5 5,5 TFP TFP - Service average TFP ! ! Sources: EIS, WB Staff estimates.! ! ! bXc! ,BBAF?<@J=!@;=EE@F@=;F@=G!@;!G=>J@F=G!N?J=!>@G=;!>?K@IB[!G@;F=!788:! ! 28.! Weak productivity outcomes in services is consistent with low and falling allocative efficiency since 2009. Allocative efficiency dropped significantly after the Global Financial Crisis (Figure 45). Fiscal and monetary stimulus, together with a surge in capital inflows from global monetary expansion, fueled a consumption boom and associated growth in inefficient consumer services.! ! 29.! Not only did allocative efficiency decline, but allocative inefficiencies also increased, including in important sectors like ICT and transport (Figure 46). The ICT sector experienced both increased allocative inefficiency and lower productivity in the last decade. This sector has the highest simple average of productivity implying that the average firm in this sector is performing very well. However, due to more resources going to the least productive firms, the weighted average of productivity is dragged down. The transport sector also experienced significant efficiency losses and declining productivity in the last decade. This result partly explains why firms that have post-manufacturing (e.g. transport and distribution) service affiliates are less productive in Turkey (World Bank, 2017). Similar efficiency losses are observed in retail and wholesale sector.! ! Page | 38 38 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY /@QH>=!R^.!,BBAF?<@J=!=EE@F@=;F[!@;!G=>J@F=G!N?G!I=FB@;=I! /@QH>=!R_.!,BBAF?<@J=!@;=EE@F@=;F@=G!@;!@LKA>G! !V@?;GKA>L!K>AIHF<@J@<[!@G!I>?QQ@;Q!IAV;!AJ=>?BB!G=>J@F=G!K>AIHF<@J@<[W! ! 30.! Unlike in the manufacturing sector, within firm productivity in the services sector has contributed negatively to productivity growth (Figure 47). In other words, existing firms in services are getting less productive. The reallocation of resources between those firms therefore also did not help pull productivity up. It is striking that firm exit has weighed so much on productivity. Therefore, exiting firms are relatively more productive than those that are remaining, which is puzzling. Even though new entrants tend to be more productive than incumbents surviving firms in recent years, this was not enough to offset the large negative effect of within firm productivity change and the change due to productive firms exiting the market. 31.! Within the service, the sources of productivity changes vary a lot across sectors (Figure 48). The only common pattern among sub-sectors is that the exiting of productive firms is pulling down their productivity. This signals potential market distortions, including non-competitive practices that allow less productive firms to survive. The ICT and transportation sectors suffer from the misallocation of resources while retail and wholesale, scientific, occupational and technical activities and transportation sectors underperformed in the within component linked to the internal capabilities. Page | 39 39 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!RU.!3@L!K>AIHF<@J@<[!FA;<>@XH<@;Q! /@QH>=!RO.!3@J@F=G]!GAH>F=G!AE!K>AIHF<@J@<[! ;=Q?<@J=B[!J@F=!G=F!&/0! FN?;Q=G!J?>@=G!?F>AGG!G=FG! TFP growth and contribution (2007-2015) TFP growth decomposition in service sectors 13% 7% 8% 3% 2% -2% -7% -3% -12% -17% -8% -22% -27% Scien., Occup.&Tech. Act. Admin Culture Retail&Wholesale Real Estate Accom.&Food ICT Transport. -13% -18% 2007 2008 2009 2010 2011 2012 2013 2014 2015 Within Reallocation Within Reallocation Entry Exit Entry Exit Change in Productivity Change in Productivity ! ! Sources: EIS, WB Staff estimates. CAAG<@;Q!K>ATI=J=BAKL=;G!V@BB!>=`H@>=!FALK=<@<@J=;=GG!>=EA>LG! 32.! Given the above trends in firm employment, growth and productivity across major sectors, how well is Turkey positioned to sustain its strong economic performance? As discussed in the previous chapter, Turkey needs to build its own ladder of development, which enables it to take advantage of new technologies and changing patterns of globalization. This means strengthening the investment climate so that pro-developments sectors can expand and absorb more labor to avoid a stagnation in incomes and living standards (Chapter 1, 3, 4, 6); this includes equipping the workforce with the skills needed to engage in pro- development sectors (Chapter 5). 33.! The analysis above suggests that sectors that could ordinarily be regarded as having strong pro-development characteristics are not faring very well. Except for a few star performers (e.g. motor vehicles), more sophisticated manufacturing sectors suffer from declining allocative efficiency and little job creation (e.g. basic pharmaceuticals), low productivity (e.g. machinery and equipment), or a combination of all three (e.g. computers, electronics and optical products). Complementary high skill services are also suffering from declining allocative efficiency, low productivity, low employment and low value addition. 34.! This section builds on a framework from Hallward- Driemeier and Nayyar (2017) to identify more precisely pro-development manufacturing sectors; it then identifies potential risks and implications for the growth of those pro-development sectors in Turkey drawing on the above analysis. The pro-development characteristics of a sector, as noted in Chapter 1, refer to sectors’ potential to boost productivity, growth and employment. For example, a sector’s exportability shows the potential for spillovers through learning by doing, scale economies, technology diffusion and competition (Hallward- Driemeier and Nayyar 2017). Page | 40 40 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY 35.! Pro-development characteristics in the manufacturing sector in Turkey are defined by 6 criteria:48 (i)! Productivity: Index of average productivity level in 2015 and productivity growth following the 2009 crisis. Indicates the sectors’ potential for growth and convergence. (ii)! Employment: Sector’s share in employment and employment growth. Indicates the sectors’ potential for job creation; (iii)! Value addition: Sector’s share in value added and value-added growth. Indicates the sectors’ significance and growth performance. (iv)! Low skill employees: Share of manual workers in total employment. Indicates the sectors’ potential to employ low-skilled workers. (v)! Tradability: Sector’s export to sales ratio, number of exporter firms. Indicates demand beyond domestic market. (vi)! Innovation: Number of firms engaging R&D and innovative activities. Scope for innovation and diffusion. 36.! Manufacturing sectors are in turn grouped in five categories (Hallward-Driemeier and Nayyar 2017)): (i) commodity-based regional processing; (ii) capital intensive regional processing; (iii) low skill labor intensive tradables; (iv) medium skill global innovators; and (v) high skill global innovators. These groupings underlie the differences in how they are traded. For instance, the high skill global innovator industries are both highly traded and global value chain intensive with high GVC length followed by medium skill global innovators and low-skill labor intensive tradables. These have potential for providing more opportunities to low and middle-income countries. However, commodity processing sectors which are closely linked to the use of agricultural raw materials or mining products, are the least GVC intensive with low GVC length (Hallward- Driemeier and Nayyar, 2017). This typology also helps identify pro-development sectors across a range of skill levels (i.e. not all high skill intensive). 37.! A few sectors stand out as having potential for building a strong ladder of development for Turkey (Table 4 – darker shades of green signal stronger pro-development characteristics relative to other sectors in the group). Basic pharmaceuticals, chemicals, motor vehicles and transport equipment tend to have the greatest scope for productivity growth, innovation and tradability. Relatively more technology intensive sub-sectors (basic pharmaceuticals, other transport equipment, basic metals, motor vehicles and chemicals) tend to have higher scores in the productivity index. These sectors are more engaged in R&D and innovation and their products are highly traded. Except basic pharmaceuticals sector, these sectors generated significant employment in the recent period even though their employment shares are at low levels. With these characteristics, they stand out and seem to be good candidates for escalating growth. 48 Draws on a combination of Hallward- Driemeier and Nayyar (2017) and Amirapu, A and Subramanian, 2015. “Manufacturing or Services: An Indian Illustration of a Development Dilemma,” Center for Global Development (Working Paper 409). Page | 41 41 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 38.! There are other sectors with good potential but in Turkey are suffering from allocative inefficiencies. They include machinery and equipment, electrical equipment and computers, electronics and optical equipment. Though these sectors perform poorly in terms of productivity, they are highly traded in international markets and are associated with more innovative activities. Hallward-Driemeier and Nayyar (2017) suggest that these are likely to be the most competitive sectors to break in or maintain given both their high export concentration and high automation. The global competition is likely to be the most intense in electronics, computers and optical equipment, pharmaceutical products and transportation equipment sectors as they also embody a relatively high share of professional services input. 39.! Some sectors have strong job creation potential but little scope for innovation, and therefore less likely to contribute to Turkey’s longer-term ladder of development. They include food, textiles and wearing apparel sectors, leather and other manufacturing products. These are highly traded in international markets and labor intensive. But they are not R&D intensive and have limited scope for technology diffusion, which hinders their ability to act as growth escalators. 40.! Food, beverage, wood, furniture and leather sectors are relatively less traded sectors and have a high share of manual workers, thereby also unlikely to act as growth escalators. The food sector stands out in its high share of employment and value added, though it performs very poorly in allocative efficiency and across most pro-development characteristics. From a global perspective, Hallward- Driemeier and Nayyar (2017) highlight that global competition will likely be less intense in textile, apparel and footwear sectors. These sectors are both less automated and have the lowest trade concentration ratios so are likely to continue absorbing low-skill employment. However, they also highlight that in food, beverages, tobacco and coke and refined petroleum are among the subsectors of manufacturing that are relatively more professional services intensive. 41.! Turkey stands at higher risk of more technology intensive and complex sectors being disrupted given current levels of competitiveness, capabilities and connectedness (Chapter 1). The need for these qualities – competitiveness, capabilities, and connectedness – can vary across sectors depending on the sectors’ relative levels of automation, export concentration and services intensity (Hallward- Driemeier and Nayyar (2017). The idea is summarized below (Table 5) – for example more sophisticated and KI sectors (e.g. transport equipment; electronics; pharmaceutical; electrical machinery; machinery and equipment n.e.c; and manufacturing n.e.c) have high trade concentration, exposure to new technology and services intensity. They therefore will require high levels of competitiveness, capabilities and connectedness. High income countries are well placed for the growth and development of these sectors. In the case of Turkey, as discussed in chapter 1, though it is globally just above average on capabilities and connectedness, it is in a middling category on the competitiveness dimension. These more sophisticated and KI sectors are therefore at high risk of being disrupted in Turkey, unless it can address challenges across the “three Cs.” 42.! These are also Turkey’s main exporting sectors, so closing gaps across the 3Cs sustain the future growth of these sectors. Strengthening firm capabilities to facilitate the technology adoption and managerial practices and deregulating and improving the quality of professional services will be critical areas to develop (Chapters 3, 5 and 6). The second group of sectors (textiles, apparel and leather products) require connectedness and high competitiveness even though these sectors are low-skill labor intensive. Turkey’s position is relatively better in these sectors compared to the first group of sectors. However, policy actions are required to improve the business environment and labor market given the demands on connectedness and competitiveness. Page | 42 42 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY &?XB=!R.!*?;HE?F@;Q!GHXTG=FG]!X[!K>ATI=J=BAKL=;?F<=>@G<@FG! " in Share of " in Value Exporters R&D Innovative Manual TFP Emp. Value Added Foreign / Active / / Workers Index Emp. (2010- Added (2010- Sales / Number Number Number in Total Nace 2-digit Value Share 2015) Share 2015) Net Sales of Firms of Firms of Firms Empl. COMMODITY-BASED REGIONAL PROCESSING ! ! Food 24.0 13.0% 7.2% 7.7% 4.1% 17.5% 13.2% 0.8% 1.3% 83% Beverages 20.6 0.4% 10.0% 1.2% 5.0% 4.1% 35.9% 1.1% 3.3% 69% Wood 69.4 1.6% 10.3% 1.6% 11.2% 8.8% 12.3% 0.6% 0.7% 90% Fabricated Metal Products 67.8 8.7% 11.1% 6.3% 15.2% 18.7% 21.7% 1.2% 2.6% 88% Paper 126.2 2.0% 10.1% 2.4% 10.9% 20.3% 39.7% 0.5% 2.5% 83% Printing & Repr. of Recorded Media 88.5 1.2% 4.4% 1.0% 8.4% 7.7% 13.8% 0.3% 0.8% 76% Rubber and Plastic 128.9 5.9% 10.7% 5.5% 11.8% 20.1% 28.6% 1.2% 4.3% 89% Other Non-Metalic. Mineral 112.8 7.0% 9.8% 6.9% 10.4% 11.6% 20.2% 1.1% 1.4% 87% Basic Metals 139.8 4.0% 8.3% 8.9% 7.3% 28.3% 31.8% 1.7% 2.4% 84% CAPITAL-INTENSIVE REGIONAL PROCESSING Chemicals 118.3 2.3% 7.7% 4.2% 7.2% 17.9% 42.9% 3.4% 2.8% 73% LOW-SKILL LABOR INTENSIVE TRADABLES! Textiles 113.4 11.3% 10.7% 8.6% 10.3% 17.4% 30.5% 0.7% 2.5% 90% Wearing Apparel 125.4 12.9% 8.8% 4.9% 10.3% 35.6% 25.6% 0.3% 1.2% 84% Leather 97.7 1.8% 8.8% 0.7% 7.9% 12.9% 27.4% 0.3% 3.0% 87% Furniture 56.6 4.6% 11.2% 1.4% 9.0% 11.9% 21.0% 0.3% 5.4% 87% Other Manufacturing 97.3 1.7% 7.2% 1.3% 13.0% 30.7% 31.2% 2.0% 3.6% 77% MEDIUM-SKILL GLOBAL INNOVATORS! Machinery and Equipment n.e.c. 64.8 6.6% 12.3% 5.8% 18.3% 27.6% 43.0% 4.1% 4.5% 83% Motor Vehicles, Trailers &Semi-trailers 126.9 6.5% 9.5% 13.2% 6.5% 41.2% 41.4% 5.2% 5.6% 65% Other Transport Equipment 184.2 1.0% 7.7% 1.9% 14.0% 59.4% 40.2% 5.1% 3.8% 75% Electrical Equipment 40.6 5.0% 8.8% 6.5% 3.9% 38.5% 35.7% 3.4% 5.0% 81% HIGH-SKILL GLOBAL INNOVATORS! Computer, Electronic & Optical Products 65.3 1.2% 8.0% 2.3% 8.6% 45.0% 42.8% 11.3% 6.3% 41% Basic Pharmaceuticals 191.0 0.9% 0.2% 3.5% 14.1% 12.2% 53.3% 17.8% 7.6% 28% Sources: EIS database, WB Staff estimates. Notes: Darker shades of green signal stronger development characteristics relative to sectors with lighter green within one of the 5 categories above. Page | 43 43 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey &?XB=!^.!'@GD!AE!I@G>HK<@A;!?F>AGG!L?;HE?F@;Q!G=FG!@;!&H>D=[! Competitiveness, capabilities and Risk of disruption in Turkey Sectors connectedness priorities Transport equiment, electronics, pharmaceutical, electrical machinery, machinery and euipment All 3Cs needed Higher risk n.e.c and manufacturing n.e.c Higher competitiveness and high Medium to high risk Textiles, Apparel and Leather Products connectedness needed Rubber and Plastics, Fabricated Metals High capabilites needed Low risk Food and Beverages, coke and refined petroluem Higher competitiveness needed Medium to high risk Wood, paper, basic metals, non-metallic mineral No risk No significant change anticipated products Sources: Adapted from Hallward- Driemeier and Nayyar (2017). 43.! The services sector will also need a significant boost if Turkey is to contain risks of disruption, particularly in more sophisticated manufacturing sectors. Though the absorption of low skill labor in the services sector has played a major role in reducing poverty (Box 5), as discussed above, it will not be possible to sustain this unless better jobs within manufacturing and services are created in the future given Turkey’s current levels of per capita income. Moreover, manufacturing sectors use services either as an input for their production or as a complementary item bundled with goods in pre- or post-manufacturing phase and add greater value to the products. The growing interdependence of these sectors underscores the importance of the pro-development characteristics of services, which include: tradability; and source of innovation and technology diffusion. In Turkey, traditional services dominate. But high-end services such as IT and scientific sectors are more likely to help mitigate risks to disruption in deeper manufacturing sectors. This will require reforms that help reduce market distortions in these more sophisticated service sectors (Chapter 6). CAZ!^"!+;<=>G=F?B!LAJ=L=;!?;I!KAJ=><[!>=IHF<@A;! The contribution of intersectoral labor movement (productive, unproductive sectors) to poverty reduction is assessed by using the Survey of Income and Living Conditions panel dataset (2012-2015). The sectors are categorized into three groups: (i) Manufacturing; (ii) Construction; and (iii) Services (Retail and Wholesale, Transportation and Storage, Accommodation and Food, Real Estate and Administrative and Support). According to this definition, one third of individuals in the sample work in the manufacturing sector, 13 percent work in construction, and 58 percent in services. Some sectors, including agriculture are not included in this analysis to match the rest of the report. Inter-sectoral movement show significant variation across sectors (Table 6). The diagonal elements refer to the percentage of workers employed in the same sector both in 2012 and 2015. Some sectors, manufacturing, retail and wholesale and transportation and storage, displayed high persistence compared to other sectors. Those sectors also attracted workers from other sectors i.e. 16 and 13 percent of unemployed individuals in 2012 found employment in manufacturing and retail and wholesale in 2015, respectively. ! ! ! Page | 44 44 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY &?XB=!_.!&N=!I@G<>@XH<@A;!AE!=LKBA[L=;!B=J=B!be!AE!?;G@<@A;!L?<>@Z!AE!=LKBA[L=;?;G@<@A;!L?<>@Z!AE!=LKBA[L=;AHKGc! 2015 Employed Unemployed Manufacturing Construction Services Total B40 38 13 16 33 100 Unemployed T60 30 24 3 44 100 B40 6 83 3 8 100 Manufacturing T60 6 85 2 7 100 2012 Employed B40 11 5 74 9 100 Construction T60 9 11 69 12 100 B40 5 5 4 85 100 Services T60 4 7 2 87 100 Persistence Job destruction Job creation Transition Source: WB staff calculations by using Survey of Income and Living Conditions panel data 2012–2015. Page | 45 45 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey While the unemployed Bottom 40 is less likely to find employment, they are more likely to find employment in construction sector (Table 8). Among unemployed only 13 percent of bottom 40 could start working in manufacturing and 33 percent in services sector while 24 and 44 percent of top 60 could find employment in manufacturing and services sector, respectively. The transition from employment to employment in all sectors however is very similar for Bottom 40 and Top 60, although B40 is less likely than T60 to move out of construction. A decomposition analysis of the contribution of those working in manufacturing, construction and services to the reduction in poverty between 2012 and 2015 shows that poverty reduction derives from higher incomes in services (Table 9). Higher incomes in the sector accounts for about 45 percent of the poverty reduction between both years, followed by construction (27 percent) and manufacturing (8 percent). Lower incomes of those unemployed slightly increased poverty since 2012. Movements between sectors explain the rest; about 18 percent of the reduction in poverty responds to population shifts. &?XB=!:.!4HKK@!'?J?BB@A;!I=FALKAG@<@A;!T!#$%&'()!*&!+,-(.%/(!0,1(./2!*&!34.5(2!6786!9!678:! Poverty in period 1 (headcount) 13.6 Poverty in period 2 (headcount) 10.5 Change in poverty (HC) -3.1 By Groups Absolute change Percentage change Manufacturing -0.2614 8.46 Construction -0.8399 27.19 Services -1.4153 45.82 Unemployed 0.0213 -0.69 Total Intra-sectoral effect -2.4954 80.79 Population-shift effect -0.5554 17.98 Interaction effect -0.0381 1.23 By Sector Absolute change Percentage change Manufacturing -0.2614 8.46 Construction -0.8399 27.19 Retail and Wholesale -0.2886 9.34 Transportation and Storage -0.1618 5.24 Accommodation and Food -0.5683 18.4 Real Estate -0.0225 0.73 Administrative and Support -0.3556 11.51 Unemployed 0.0213 -0.69 Total Intra-sectoral effect -2.477 80.19 Population-shift effect -0.5448 17.64 Interaction effect -0.0671 2.17 Note: Huppi Ravallion decomposition with variable low productive if zero and high productive if 1 and Huppi Ravallion with sector variable. Huppi Ravallion decomposes changes in poverty over time into intrasectoral effects. Page | 46 46 CHAPTER 2 / MANUFACTURING, CONSTRUCTION AND SERVICES PRODUCTIVITY "A;FBHG@A;!?;I!KAB@F[!@LKB@F?<@A;G! 39.! The main messages that emerge from the analysis on productivity in manufacturing, construction and services are as follows: (i)! Manufacturing: some breakout sectors but overall productivity has remained flat in the past decade: Tradable sectors that have benefited from foreign investment perform well. But allocative efficiency has been declining since 2011. Within firm productivity growth, much of which was driven by large firms, has helped offset declining allocative efficiency. The largest firms on the other hand have experienced a slight decline in TFP growth, which may be due to capital deepening. (ii)! Growing but low productivity construction sector that is suffering from allocative inefficiency: The sector expanded rapidly over the past decade, which coincided with falling productivity. A sector characterized by low skill intensity, with little scope for learning-by-doing and innovation. (iii)! Services experiencing declining allocative efficiency and within sector productivity growth: Services are dominated by less knowledge intensive, low productivity sectors. Allocative inefficiencies have risen rapidly since 2009, which is exacerbated by declining within firm productivity. (iv)! A few sectors stand out as having potential for acting as growth escalators, but progress is needed on competitiveness, capabilities, and connectedness: Basic pharmaceuticals, chemicals, motor vehicles and transport equipment tend to have the greatest scope for productivity growth, innovation and tradability. Except pharmaceuticals, they also have potential for employment creation. Issues Policy options Manufacturing sector productivity is weighed Review and address constraints to growth of large down by resource misallocation. This is manufacturing firms. Strengthen their links to local partially offset by productivity gains within and international value chains including SMEs. large manufacturing firms. Wind down supply subsidies that keep inefficient SMEs in business and prevent reallocation of resources to more productive firms. Review quality of capital deepening by large manufacturing firms to promote productive investments. Construction sector is absorbing increasing Reduce incentives and credit for expansion of amount of resources though productivity construction activities levels are low. The services sector will need a significant Liberalize foreign investment regime for services boost if Turkey is to contain risks of sector (Chapters 3, 6) disruption, particularly in more sophisticated manufacturing sectors. ! Page | 47 47 CHAPTER 1 Chapter 3 Economic integration and productivity Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Economic integration and productivity 48 Global integration deepened Turkey’s 48 participation in GVCs enabling more and better exports More integrated firms and sectors tend to 55 be more productive Accelerating within firm productivity: 62 Trade costs and vertical FDI spillovers Conclusion and policy options 71 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY +++W +++W ! ) ! ) FA;AL@F!@;<=Q>?<@A;!?;I!K>AIHF<@J@<[!! FA;AL@F!@;<=Q>?<@A;!?;I!K>AIHF<@J@<[!! !! 1.1. With With increased increased global global integration, integration, the the scope scope for for productivity productivity growth growth also also increased. increased. Evidence Evidence suggests suggests that that open open economies economies tend tendtoto grow grow faster faster than than closed closed ones. ones. There There are are three three channels channels linking linking integration integration and and growth. growth. First, First, integration integration leads leads toto increased increased competition competition from from abroad, abroad, which which shrinks shrinks margins margins and and benefit benefit consumers consumers through through higher higher purchasing purchasing power power ––the the pro-competitive pro-competitive effect effect of of integration. integration. The The second second isis aa selection selection effect: effect: asasprofits profits shrink, shrink, only only the the most most productive productive survive survive or grow, or grow, while while the the least least productive productive shrink shrink or or exit the exit the market. market. That That increases increases aggregate aggregate productivity. productivity. The The third, third, isis the innovation innovation the and and learning learning effect: effect: more more integration integration increases increases the thereturns returns totoinnovation innovation investment investment and and increases increases exposure exposure to better to better management management practices practices and and know-how, know-how, increasing increasing the the scope scope for for learning, learning, and, and, leading leading toto productivity productivity gains. gains. !! 2.2. Turkey Turkey has has embraced embraced global global integration integration and and gained gained from from it. it. Turkey Turkey isis strategically strategically located, located, serving serving asasaa link link between between ‘Factory ‘Factory Europe’ Europe’ –– inin which which aa cluster cluster of of sophisticated sophisticated firms firms connected connected through through international international production production networks networks operate, operate, and and ‘Factory ‘Factory Asia’, Asia’, thethe most most dynamic dynamic region region inin recent recent years. years. InIn this this context, context, this this chapter chapter looks looks atat three three important important issues: issues: (i)(i) the the evolution evolution ofof Turkey’s Turkey’s global global integration integration since since the the early early 2000s, 2000s, andand its its outcomes outcomes inin terms terms ofof exports exports and and participation participation of of Turkish Turkish firms firms inin Global Global Value Value Chains Chains (GVCs); (GVCs); (ii) (ii) evidence evidence onon whether whether global global integration integration hashas ledledtoto higher higher productivity productivity within within firms firms and and sectors; sectors; and and (iii) (iii) factors factors that that can can help help accelerate accelerate within-firm within-firm productivity productivity gains gains arising arising out out ofof global global integration. integration. TheThe chapter chapter ends ends with with policy policy implications implications for for further further deepening deepening Turkey’s Turkey’s global global integration. integration. 2BAX?B!@;<=Q>?<@A;!I==K=;=I!&H>D=[dG!K?><@F@K?<@A;!@;!21"G!=;?XB@;Q!LA>=!?;I!X=<<=>! 2BAX?B!@;<=Q>?<@A;!I==K=;=I!&H>D=[dG!K?><@F@K?<@A;!@;!21"G!=;?XB@;Q!LA>=!?;I!X=<<=>! =ZKA>AVD=[.!&H>;!AE![!?;I!KAGAVD=[.!&H>;!AE![!?;I!KAG=!R:.!&H>D=[dG!L=>FN?;I@G=!=ZKA>=V!>=B?<@J=B[! /@QH>=!^8.!)ZKA>AV@=;F=I!E?G@AI!7888T789^! =?>B[!K=>@AI!?;I!GBHQQ@GN!Q>AV!G@E@=I!=ZKA>AV=!^9.!*)%,!GN?>=!AE!=ZKA>=?G=I!VN@BG=!^7.!&H>D@GN!=ZKA><=>G!>=?FN!?BLAG@=G!@;! )$!GN?>=!N?G!I=F>=?G=I! BI! Turkey’s export shares by destination (%) Number of export destinations reached, 2002 (2002-2016) and 2016 100% ZAF EU28 TUN 80% MENA RUS Non EU ECA POL 60% North America MYS MEX East Asia & Pacific 40% KOR Others HUN Sub-Saharan Africa 20% BRA LAC TUR South Asia 0 50 100 150 200 250 0% 2002 2016 2016 2002 ! ! Sources: UN Comtrade. CAGR stands for compound annual Sources: UN Comtrade, WB Staff estimates. growth rate that is used to measure growth over a 15-year period using a compounding formula. 52Based on sample from EIS data; according to TURKSTAT data, the number of exporters increased by 52 percent from 2006 to 2016. Page | 51 51 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY /@QH>=!^S.!)ZKA>?<@A;!@G!BAV!?;I!I=F>=?G@;Q! /@QH>=!^R.!&H>D@GN!=ZKA>=!N@QNB[!I@J=>G@E@=I!@;!<=>LG!AE! AJ=>!<@L=! K>AIHF=!^^.!&N=!;HLX=>!AE!E@>LG!=;<=>@;Q!D==?G=I!AJ=>!<@L=!! 60% Share of exporters in 2006 Share of exporters in 2016 50% 40% 30% 20% 10% 0% Be cal… … t c Ru asic es an tals g E l er s er er l od -m e y llic al tic d s l ac s re eta or cl e ge l ni r er tin oo ica tic N ni t u p ath th il as pa Fo t ri M ansp eta t ro n ra xt Pa e m hi in W O eu em hi m pl ec Fa g ap Le ve Te ve Pr r ec d ac Fu d Ch te tr El M or on m ri n i ca er ar ot B er th ea br Ph bb O W ! Sources: EIS, WB Staff estimates. bFc! +;F>=?G@;QB[!GAKN@G<@F?<=I!=ZKA>=!^_.!&H>D@GN!=ZKA>?IH?B!GHXG<@=!^U.!&H>D@GN!@LKA>?<=I!@;! BAVT<=FN!EA>!L@IT<=FN!=ZKA>=!<=FNT@;<=;G@J=!G=FG! Share of exports (%) by technological content, Share of imports (%) by technological content, 2002 and 2016 2002 and 2016 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2002 2016 2002 2016 High tech Mid-high tech Mid-low tech Low tech High tech Mid-high tech Mid-low tech Low tech ! ! Sources: OECD Stat data, WB Staff estimates. Sources: OECD Stat data, WB Staff estimates. bIc! +;<=Q>?<@A;!A;!=!^O.!&H>D=[dG!K?><@F@K?<@A;!@;=!^:.!/A>=@Q;!J?BH=!?II=I!@G!@;F>=?G@;QB[!LA>=! XH[=>!?;I!G=BB=>!N?G!@;F>=?G=I!G@;F=!=?>B[!788^! @LKA>AIHF=!&H>D@GN!=ZKA>AIHF=!_8.!/-+!G=?G=I!?F>AGG!?BB!G=FG! 0,9 Ratios of FDI stocks to output by sector, 2006 and 2013 0,8 0,7 2006 2013 FDI Stock to output ratio 0,6 0,5 0,4 0,3 0,2 0,1 0 tru e g Ru llic tals F o e s & pe r on l qu ry ic les rade les s E l e nt c ve n pe cals Tr ion Te d & m B e he r ies W tail t an era m ce e io vice R e por ge d Re isin ni o t t e oo l e u Co tra as ta c cti ilit i i c a c hi n t ro m t ra at pa t e c o e rv in i hi M talla lea t a n em na al es s pl M rt r W tro od ve an Ut ip an m rm l se ale ec ve Ac n s d a h ns s M Ad or C a in le m i I n nc ot & d ho eta er at d il xt bb od -m t ri Fi ke ir fo ec on pa Co El N Re ! Sources: OECD, WB Staff estimates. Page | 54 54 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey *A>=!@;<=Q>?<=I!E@>LG!?;I!G=FG!<=;I!=!K>AIHF<@J=! b?c! )LK@>@F?B!B@<=>?=!KA@;A;Q!B@;D!X=?<@A;!?;I!K>AIHF<@J@<[! 15.! International evidence tends to confirm that firms exposed to the global marketplace are exceptional performers. Causality works in both directions. Within industries, integrated firms – exporters and/or importers of intermediates – tend to be more productive and more innovative, pay higher wages, use more skills and capital, and are less likely to exit. On the one hand, firms that participate in the international market are better at facing the high fixed costs associated with searching for clients abroad, learning about their tastes, quality and safety standards, and adapting their products to match them. This suggests a process of selection into exporting. 16.! There is also increasing evidence on integration (exporting and importing) improving firms’ productivity. ‘Learning by exporting’, for example, has been linked to productivity gains associated with exposure to demanding clients, or to highly productive competitors. ‘Learning by importing’ has been identified when access to a wider variety of intermediates and capital goods relax technological constraints of firms allowing them to produce in better conditions.55 Exposure to multinationals - that typically display greater productivity levels – through FDI, has also been linked with learning and with increased incentives to innovate (The evidence on openness and productivity for Turkish firms is reviewed in Box 6). 17.! Empirical evidence has validated the channels and mechanisms identified in the theoretical literature. For example, Amiti and Konings (2007) focus on the ‘learning by importing’ channel. They use data on Indonesia, estimate the productivity gains from reducing tariffs on final goods and from reducing tariffs on intermediate inputs, and find that, (a) lower output tariffs increase productivity by inducing tougher import competition, and (b) cheaper imported inputs also raise productivity via learning, variety, and quality effects. This latter effect is substantially stronger than the former. Similar results are found by Yu (2014) for China. Focusing on the FDI channel, there is some consensus on the positive spillover effects on productivity downstream, associated with increased participation of multinationals in upstream sectors. For example, Arnold et al. (2012) found sizable effects on productivity of increased foreign entry into upstream services in India, Fernandes and Paunov (2012) in Chile, and Duggan et al. (2013) for Indonesia. 18.! Yet, the effects of openness on productivity are far from being automatic or inevitable, and gains accrue heterogeneously to different types of firms.56 The extent to which these channels are at work depends on local characteristics, such as, market structure, the overall investment climate, or the rigidities that may exist in labor and capital markets to facilitate or impede structural adjustments.57 Firms’ absorptive capabilities, often measured by human capital and R&D investment, are also key determinants of whether a firm can benefit from increased exposure to multinationals and international trade. 55 See for example, Pavcnik, 2002; Amiti and Konings, 2007; Brandt, Van Biesebroeck, Wang, Zhang, 2017, Atkins et al., 2017. 56 In Lithuania, Javorcik (2004) provides evidence of positive productivity spillovers from FDI taking place through interactions between foreign affiliates and their local suppliers in upstream sectors. The same author finds evidence of vertical spillovers through backward linkages also in the cases of Czech Republic and Latvia through multiple channels. 57See, for example, Blalock and Simon (2009). The authors show that firms’ absorptive capabilities (i.e.: better trained workers, more investments in R&D) are key determinants of whether firms can benefit from FDI spillovers. Page | 55 55 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY CAZ!_.!Y@<=>?=!A;!?<@A;!A;!K>AIHF<@J@<[!@;!&H>D=[!! Exporting and productivity gains: Yasar et al. (2007b), provides a comprehensive analysis of international linkages, including FDI, exports, imports and licensing, for two Turkish manufacturing sectors—the textile and apparel and the motor vehicle and parts industries. They find that plants with international linkages have higher productivity levels, are larger, invest more, pay more, and hire more administrative and technical workers. Results also show that internal plant characteristics, such as the share of skilled labor, enhance the productive role of international linkages. Finally, they find that engaging in a range of multiple international activities (e.g. FDI and exporting) further enhance productivity. This is also confirmed in Dalgic et al. (2005) for the entire manufacturing sector in Turkey. The authors find that firms engaging in both exporting and importing perform better than those involved only in one side of trade. There is evidence of both a selection and a learning mechanism when considering the participation of Turkish firms into the export market. On one hand, Dalgic et al. (2005) find a self-selection effect for exporting since engaging in international trade is associated with ex-ante superior performance and entry sunk costs among Turkish firms. In another study on the same sectors mentioned above, Yasar et al. (2007a) and Yasar et al. (2005) find evidence of stronger learning- by-exporting effects in the textile and apparel industry. This was not the case for the motor vehicle industry. It is suggested that because this industry is more likely to be using cutting-edge technology and skills, given the larger presence of foreign ownership and intensive competition, there is less knowledge to be gained by exporting. Maggioni (2012) finds evidence of both self-selection into exporting and learning-by-exporting and highlights a link between export and import activity. Finally, Cebeci (2016) investigates the relevance of export destinations in affecting productivity, and also employment, and wages in Turkey. In particular, the author compares firms that export to low- income destinations and high-income destinations and finds that participation in the export market has a positive effect on firm TFP only when firms export to high-income destinations. Importing and productivity gains: The evidence on the effects of participation in the import market is more limited. Taymaz and Yilmaz (2006) study a period of substantial tariff reduction in Turkey (1984-200). The authors analyze the impact of import penetration from the EU and find a positive association with productivity in import competing industries. Yasar et al. (2007b) evaluate the relationship between productivity and imports of capital goods for Turkish manufacturing plants in the apparel, textiles, and motor vehicles industries. They find that firms acquiring capital goods in the international market (machinery and equipment) increases productivity but the effects are smaller than those from entering the export market. A 10 percent higher asset import share is associated with 0.6 percent greater productivity (1.5 percent in the case of exports). Dalgic et al. (2005) find higher sunk costs for importing activity than for exporting and show that diversification of imports (either in terms of numbers of products or countries) has a bigger impact on firm performance than diversification in terms of exporting. Foreign direct investment and productivity gains: Benli (2016) finds that Turkish firms can benefit from both horizontal and vertical FDI spillovers but benefits are conditional to firms’ absorptive capacity, measured by the gap in TFP. In addition, he also finds that gains accrue especially to firms that have medium and high TFP growth rates. This is also found in an earlier paper by Köymen Özer et al. (2012) that looks at the role played by human capital, technology gap and export status, in favoring the realization of productivity spillovers from horizontal and vertical FDI. They find that both measures of absorptive capacity are associated to greater productivity gains from FDI. Yasar et al. (2007b) evaluate the relationships between productivity and FDI in the Turkish apparel, textile, and motor vehicles industries and find that plants with a foreign ownership share are the most productive, Page | 56 56 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey followed by plants that export. This is especially evident for larger plants and plants with more skilled labor. Source: Author’s elaboration. bXc! +;<=Q>?<=I!E@>LG!@;!&H>D=[!<=;I!=!K>AIHF<@J= 19.! Turkish exporters enjoy a productivity premium comparable to what is observed in advanced economies such as the UK, US, and Germany. In Turkey, exporters are on average 7.5 percent more productive than non-exporters (Figure 61). This compares to the exporting premium that firms in the UK, Sweden, Slovenia and the USA display, although it is lower than that observed among firms in Spain, Denmark, Italy or Belgium. 20.! Turkish importers and exporters show higher productivity growth than non-importers or non-exporters. Average total factor productivity of importers has grown by more than 4 percent over the period 2006-2016 compared to 3 percent for non-importers (Figure 62). This was driven by a rapid increase in TFP in the second part of the period (since 2012), while the post financial crisis period was associated with stable or even declining TFP particularly for non-importers. A similar pattern is also observed for exporters as compared to non-exporters (Figure 63). /@QH>=!_9.!&H>D@GN!=ZKA><=>G!=;fA[!?!K>AIHF<@J@<[!K>=L@HL!FALK?>?XB=!=!_7.!&H>D@GN!@LKA><=>G!GNAV!N@QN=>!K>AIHF<@J@<[! /@QH>=!_S.!&H>D@GN!=ZKA><=>G!GNAV!N@QN=>!K>AIHF<@J@<[! Q>AV?<=G!<=>G! Q>AV?<=G!<=>G! TFP of importers & non-importers, 2006-2016 TFP of exporters & non-exporters, 2006-2016 105 105 104 104 103 103 102 102 TFP index TFP index 101 101 100 100 99 99 98 98 97 97 96 96 95 95 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Non-importers Importers Non-exporters Exporters ! ! Sources: WB Staff calculations based on EIS data. TFP index with Sources: WB Staff calculations based on EIS data. Manufacturing base 2006. Manufacturing firms only. firms only. 21.! Exporters are more productive in the first place, but their productivity further increases after entering the export market. Just like observed in the rest of the world, in Turkey, firms that export tend to be more productive (Figure 64). The gap is larger for large firms. A large exporter is 25 percent more productive than a large firm that has never entered the export market. The gap is of 8 percent among small firms. But in addition to this selection effect (most productive Turkish firms are those that export), in Turkey, there is also evidence of learning by exporting among firms. Average productivity of firms increases by about 3 percent after entering the export market (Figure 65). The benefits are larger for the pharmaceutical, paper, basic metal and motor vehicle sectors. Page | 58 58 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!_R.!)ZKA><=>G!?>=!LA>=!K>AIHF<@J=!@;I=K=;I=;=!_^.!/@>LGd!K>AIHF<@J@<[!@;F>=?G=G!?E<=>!=;<=>@;Q!!G@\=! =ZKA>D==!__.!&N=!G@\=!AE!E@>G@;Q!=ZKA><=>G!I=F>=?G@;Q!AJ=>!<@L=! Median firm employment of exporters, 2006-2015 11,0 10,0 Median number of employees 9,0 8,0 7,0 6,0 5,0 4,0 3,0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Manufacturing Services Sources: EIS, WB Staff estimates. /@QH>=!_U.!)ZKA>I!B@;D?Q=G!B@;D=I! /@QH>=!_O.!+LKA>I!B@;D?Q=G!B@;D=I!L!K>AIHF<@J@<[!Q>AVL!K>AIHF<@J@<[!Q>AV=!_:.!Y=GG!/-+!>=G<>@F<@A;G!@;IHF=I!>=?BBAF?<@A;!AE!@;KH=G<>@F<=I!G=FG! Restrictiveness Reallocation 20% 0% -20% -40% -60% -80% -100% cs od ls llic y d al er s ile rt r g ls ls r m al cl e he pe er tin oo eta eta ica po ri c tic bb ni l eu xt Fo eta in Pa ot hi in ro W eu Te t em ns M M Ru h tro ec ve -m Pr re t ac ac tra ec El b. Ch Pe itu M or on m El Fa er rn ar ot N th Ph Fu M O Sources: WB Staff estimates based on OECD and TURKSTAT Input-Output tables. The graph shows the change in the overall FDI restrictiveness of upstream sectors between 2003 and 2012. The overall restrictiveness is given by the weighted average of the restrictiveness of each upstream sector, where weights are represented by input-output (technical) coefficients. The change over the period is decomposed into a decrease in screening and approval restrictions on FDI in upstream sectors (the blue bars) and a change in weights. A negative light blue bar indicates that upstream sectors with lower FDI restrictiveness increased their weights. bFc! C=<<=>!@;<=Q>?<=I!G=FG!AHQN!21"G!=ZK=>@=;F=!V@L!K>AIHF<@J@<[!Q>AV=!U8.!P=FG!V@A;Q=>!EA>V?>I!B@;D?Q=G!GNAV!?! /@QH>=!U8.!P=FG!V@A;Q=>!EA>V?>I!B@;D?Q=G!GNAV!?! /@QH>=!U9.!P=FG!V@A;Q=>!X?FDV?>I!B@;D?Q=G!GNAV!?! /@QH>=!U9.!P=FG!V@A;Q=>!X?FDV?>I!B@;D?Q=G!GNAV!?! Q>=?<=>!>=GKA;G=!AE!V@L!&/0!Q>AV=?G=I! Q>=?<=>!>=GKA;G=!AE!V@L!&/0!Q>AV=?G=I! Q>=?<=>!>=GKA;G=!AE!V@L!&/0!Q>AV=?G=I! Q>=?<=>!>=GKA;G=!AE!V@L!&/0!Q>AV=?G=I! B@X=>?B@\?<@A;! B@X=>?B@\?<@A;! B@X=>?B@\?<@A;!E>AL!?B@\?<@A;!E>AL!?<@;Q!V@L!K>AIHF<@J@<[.!&>?I=!FAG<@F?B!/-+!GK@BBAJ=>G! ,FF=B=>?<@;Q!V@L!K>AIHF<@J@<[.!&>?I=!FAG<@F?B!/-+!GK@BBAJ=>G! 27.! ! Given 27. Given the above the above findings findings atatthe thesectorial sectorial level, level,how how does doesexposure exposure to foreign to foreignintermediates intermediates and investment impact on productivity growth in Turkey, through firm-level and investment impact on productivity growth in Turkey, through firm-level lenses? lenses? Productivity Productivity gains gainsof offirms firmsoperating operating downstream downstream are areassociated associated with reduced with reduced trade and trade andinvestment investment costs costs upstream inin upstream sectors. The section below tests for the presence of vertical spillovers by relating firm-level sectors. The section below tests for the presence of vertical spillovers by relating firm-level total total factor factor including By 59 59 By includingsector fixed-effects sector we fixed-effects wecan observe can how observe how TFP TFP growth growthevolves evolves over time over timeinin relation toto relation changes inin export changes export and andimport importparticipation participation without withoutthe confounding the confounding effect ofof effect sector-specific structural sector-specific characteristics. structural characteristics. Page Page 62 || 62 62 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey productivity to upstream measures of: (i) trade costs, and (ii) FDI. The analysis also looks at whether the effects vary by firm size and degree of absorptive capacity. Details on the estimation techniques are discussed in Box 7 below. CAZ!U.!)G<@L?<@;Q!<@F?B!GK@BBAJ=>G!E>AL!>=IHF=I!<>?I=!FAG=?G=I!/-+! In this section we describe the empirical approach used to investigate the presence of vertical spillover due to greater integration of upstream sectors. We consider both FDI inflows and imports in upstream sectors. Vertical spillovers from reduced trade costs in upstream sectors To establish whether there exists a causal relationship between firm performance and a reduction in trade costs in upstream sectors, we follow Amiti and Konings (2007) and regress a measure of total factor productivity of Turkish firms on a constructed measure of upstream import tariffs. Formally, we estimate the following equation: "+,-." # /012 ." 3 45-" 3 67" 3 8- 3 9-." . (1) where TFP is the measure of productivity used in this report (described in the Technical Appendix of this report) and upstream trade costs (TUP) are proxied by a weighted average of tariffs in upstream sectors where the weights are cost shares as described below. All our specifications include firm- and time-fixed effects. We also include some additional controls such as the tariff in the own sector, share of exports, sector-level Herfindahl-Hirschman Index, and sector-level time trends. The model is estimated as with standard linear fixed effects estimator where standard errors are clustered at sector level. We also provide quantile regression results to test for the presence of heterogenous effect along the distribution of productivity. To explore additional heterogenous effects and the role played by absorptive capacity we also interact TUP with indicators of export status, size and participation in R&D expenditure. Vertical spillovers from increased FDI in upstream sectors To establish whether there exists a causal relationship between firm performance and FDI in upstream sectors, we regress the productivity of Turkish firms on measures of FDI in upstream sectors that are described below. Formally, we estimate the following equation: "+,-." # /:;<12 ." 3 45-" 3 67" 3 8- 3 9-." . (2) where FDI in upstream sector (TUP) is measured by a weighted average of the FDI Stock and FDI restrictiveness of upstream sectors as described below. All our specifications include firm- and time-fixed effects. We also include some additional controls such as the FDI in the own sector, and sector-level time trends. As above, the model is estimated as with standard linear fixed effects estimator where standard errors are clustered at sector level. We also provide quantile regression results to test for the presence of heterogenous effect along the distribution of productivity. To explore additional heterogenous effects and the role played by absorptive capacity we also interact FDIUP with indicators of export status, size and participation in R&D expenditure. We run all specifications using alternative lag structures. The reported specification is the one maximizing the goodness of fit. Page | 63 63 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY Data sources For the empirical analysis that follows we combined plant-level data with sector-level measures of upstream trade costs and FDI. Plant level data are from the Entrepreneur Information System (EIS), which is compiled and administered by the Ministry of Industry and Technology (MOIT). To measure the restrictiveness of policies towards foreign direct investment we rely on the OECD’s FDI restrictiveness index over the 2006- 2013 period. We also constructed a measure of FDI position in upstream sectors, described below, by considering the ratio of FDI over sector-level output, both obtained from the OECD. Finally, we obtained MFN tariffs from UNCTAD. The weights used to construct upstream measures of trade costs and FDI are obtained from Turkish input-output tables provided by TURKSTAT. Upstream policy variables We look at vertical spillovers from integration through forward linkages by constructing weighted averages of conditions in all upstream sectors. Input-output tables are used to get a sense of the importance that each upstream sector has in terms of input costs. Hence, weights are based on input-output coefficients and are combined with measures of integration as shown in equation (1): X UP it = ! w ijX j j where XUP is a generic index of openness of upstream sectors and the weights, w, are given by, input-output coefficients, i.e. the share in the total input bill of a given manufacturing sector ‘s’ accounted for the upstream sector ‘ j.’ Coefficients are fixed over time and obtained as the average IO coefficients (2002-2012). We consider two main drivers of integration into the global marketplace of upstream sectors: low trade costs and a liberal FDI regime. Trade costs are measured by import tariffs (MFN) and FDI spillovers are captured in two alternative ways: using a policy variable – a policy restrictiveness measure of FDI inflows in the form of equity restrictions, restrictions to hiring personnel, screening and discriminatory business licensing (as done in Duggan et al. 2013), and by using an outcome variable – the stock of FDI in each relevant sector. Source: WB Staff. b?c! 0>AIHF<@J@<[!Q?@;G!IAV;G<>=?L!E>AL!>=IHF=I!<>?I=!FAG=?L!G=FG! 28.! Reduced trade costs – in the form of lower tariffs - upstream have resulted in productivity increases in manufacturing firms downstream. Using plant level data, we tested the effect of reducing tariffs upstream – effectively a reduction in upstream trade costs - on the productivity of firms operating downstream. There is a significant and negative effect of upstream tariffs on the productivity of manufacturing plants downstream (Figure 72). The effects are not only statistically significant, but also economically significant: a one-percentage point reduction in upstream tariffs increases productivity by 3 percent. The result is robust to the inclusion of additional controls, and to the inclusion of sector-time trends (although the magnitude of the coefficient is substantially reduced). This is likely related to the fact that reduced tariffs upstream relax technological constraints that firms operating downstream face, and therefore allows them to gain efficiency. Turkish firms also benefit from a fall in output tariffs and the gains are similar to those from a reduction in tariffs in upstream sectors. This effect, can be attributed to increased competition from abroad, that induces firms to reduce X-inefficiencies. Page | 64 64 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!U7.!0>AIHF<@J@<[!Q?@;G!E>AL!>=IHF=I!<>?I=!FAG[!?F>AGG!G=FG! Share of productivity gains accounted for reductions in upstream tariffs, %, 2006-2016 Fabricated metal Basic metals Electrical equipment Machinery Printing Paper Motor vehicles Electronic Other transport Rubber and plastic 0% 1% 2% 3% 4% 5% 6% 7% Sources: EIS, WB Staff estimates. 29.! Reduced tariffs in upstream sectors account for a reasonable share of productivity gains observed during the period of analysis. A back of the envelope calculation based on the estimated coefficients, the evolution of productivity by sector over the period, and the evolution of tariffs in upstream sectors shows some heterogeneity in the effects (Figure 72). In the fabricated metals sector, reduced input tariffs explain about 6 percent of total TFP gains. This is because it is in the manufacturing sector experiencing the largest decrease in upstream tariffs. Gains have also been substantial in the basic metals, and electrical equipment sectors. 30.! Results reveal substantial potential for productivity gains through reduced trade costs. The analysis carried out here uses tariffs as a main indicator of trade cost. Because tariffs have been relatively low to start with, and determined by the EU’s CET, the reduction observed during the sample is limited, which results in a limited productivity gain through the upstream openness channel. However, tariffs are proxying for trade costs, and there are other sources of trade costs besides tariffs, such as trade defense instruments (TDIs) that have been widely used by Turkey. Hence, our results point to substantial scope for further productivity gains were these TDIs to be gradually phased out (Box 8). 31.! Productivity gains from reduced upstream tariffs do not accrue equally to all firms. Gains have been larger for firms with greater absorptive capacity. We measure absorptive capacity by whether a firm engages in R&D investment and assess the role of absorptive capacity for firms of different size (Figure 73). First, we observe that large manufacturing firms benefit the most from access to foreign intermediates than medium and small-size firms. Second, firms that engage in R&D investment are better positioned to benefit from reduced upstream tariffs than those who do not invest, in particular if they are small. Page | 65 65 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY /@QH>=!US.!0>AIHF<@J@<[!Q?@;G!E>AL!>=IHF=I!<>?I=!FAG=!N@QN=>!EA>!B?>Q=>!E@>LG!?;I! E@>LG!V@=?<=>!?XGA>K<@J=!F?K?F@<[! Estimated effects of average decrease in input tariffs on productivity by size and absorptive capacity of firms Estimated effect of a decrease in input tariffs on 0,6% 0,5% 0,4% producitviy 0,3% 0,2% 0,1% 0,0% Small Medium Large Low absorptive capacity High absorptive capacity ! Source: Authors’ calculations. The plot shows the effect on productivity of a decrease in upstream tariffs corresponding to the average annual decrease observed over the period of analysis (-0.15 percentage points). Small firms have less than 10 employees, medium firms have between 10 and 50 employees and large firms have 50 or more employees. High absorptive capacity firms are those that have undertaken R&D investment. Page | 66 66 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey CAZ!O.!&N=!HG=!AE!&>?I=!-=E=;G=!+;G<>HL=;D=[! Although tariffs have been largely reduced after Turkey joined the Custom Union, other forms of protection have been adopted. The number of antidumping measures imposed by Turkey has increased over time (Table 10). These measures mostly involve chemical products but are also increasingly used in the rubber and plastic and metal sectors, and are typically in the form of ad valorem duties. Most antidumping measures are imposed on China, Malaysia, South Korea, Taiwan and Thailand. However, more recently antidumping measures have also been imposed on some EU members, mainly on Germany and Romania (Table 11). There are only very few countervailing duties in place against China. &?XB=! 98.! ,;<@IHLK@;Q! L=?GH>=G! @LKAG=I! X[! &?XB=!99.!*AG=G!?>=!?Q?@;GD=[!N?J=!X==;!@;F>=?G@;Q!AJ=>!<@L=W (Number of ?;I! PAH@=G! XH=! >=F=;GW (Number of antidumping duties by year and targeted country) Sector 2002 2006 2010 2016 Country 2002 2006 2010 2016 Belarus 1 1 0 0 Textiles 2 10 12 33 Brazil 1 1 1 1 Wearing apparel 2 2 2 4 Canada 0 0 3 3 Wood 0 0 21 25 China 86 150 1 282 EU member 0 8 4 9 Paper 0 0 0 3 Hong Kong 0 0 0 1 Chemicals 376 391 417 476 India 1 5 14 15 Rubber and plastic 0 42 59 60 Indonesia 1 1 16 16 Israel 0 1 0 2 Non-metallic 0 1 7 20 Malaysia 73 74 80 84 Metals 21 42 31 49 Moldova 6 6 0 0 Electronic 1 1 2 1 Pakistan 0 0 0 4 Russia 6 12 6 6 Electrical 0 9 10 4 Saudi Arabia 0 0 1 0 Machinery 0 0 1 6 Serbia 0 0 1 1 Motor vehicles 0 2 2 2 South Korea 76 77 78 75 Sri Lanka 0 2 2 2 Other 4 9 9 9 Taiwan 76 81 82 84 Total 406 509 573 692 Thailand 73 78 81 84 USA 0 1 4 7 Ukraine 6 7 1 1 Vietnam 0 4 11 15 Note: The Table reports the number of antidumping Note: The Table reports the number of antidumping measures measures in place at the year reported in the header. by targeted country. Changes over time are the results of new Changes over time are the results of new measures been measures been introduced and some measures been revoked. introduced and some measures been revoked. Source: Source: Bown, Chad P. (2016) "Global Antidumping Database,” Bown, Chad P. (2016) "Global Antidumping Database,” The World Bank, available at The World Bank, available at http://econ.worldbank.org/ttbd/gad/ EU members refer to http://econ.worldbank.org/ttbd/gad/. Belgium, Bulgaria, Finland, Germany, Greece, Hungary, Italy, Netherlands and Romania. Source: WB Staff calculations based on Bown, Chad (2016) “Global Antidumping Database”. Page | 67 67 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY 32.! Productivity gains from reduced upstream tariffs are larger for firms with a smaller “technology gap”. A firm's technology gap, considered as the distance between its technology level and that of imported products, could affect a firm’s ability to absorb foreign technology. The relationship could go both ways: the larger the gap, the more scope for learning. However, it is also plausible that the larger the gap, the lower the absorptive capabilities of the firm. Ultimately, it is an empirical question which mechanism predominates. When we estimate the impact of reduced trade costs by quartile of the productivity distributions we find larger effects for firms at higher levels of productivity (Table 12). In line with our previous findings, we expect these firms to be technically proficient and have greater capacity to absorb the foreign technology embodied in imported inputs. Firms with a large technology gap, at the lower end of the productivity distribution, could lack the technical competency needed to absorb external knowledge. &?XB=! 97.! 0>AIHF<@J@<[! Q?@;G! E>AL! >=IHF=I! HKG<>=?L! @EEG! X[! `H?><@B=! AE! AIHF<@J@<[!I@G<>@XH<@A;( Manufacturing sector Lowest quartile Median Upper quartile Dep. Var.: log of TFP (1) (2) (3) Upstream tariffs -0.019*** -0.017*** -0.020*** (0.001) (0.000) (0.001) Output tariff -0.030*** -0.031*** -0.033*** (0.002) (0.002) (0.002) Year FE Yes Yes Yes Observations 515,902 Plants 108,275 Note: The table reports the estimates of quantile regressions where all variables have been demeaned. Standard errors clustered at the sector level are reported in parenthesis. Data on TFP and R&D are from EIS, upstream tariffs are obtained combining UNCTAD data on tariffs with IO coefficients from TURKSTAT IO tables. bXc! 1=><@F?B!/-+!GK@BBAJ=>G!?LA;Q!&H>D@GN!E@>LG! ! 33.! FDI provides another channel through which the integration of Turkey into the global economy may have contributed to firms’ increased productivity. As mentioned above, international evidence points to the domestic firms’ productivity positively responding to increased FDI in upstream sectors (vertical spillovers).60 This section presents the evidence on vertical spillovers from FDI for the case of Turkish firms. 34.! Turkish firms have gained from FDI in upstream sectors but the benefits do not accrue equally to all firms. The increase in FDI in upstream sectors has led to an increase in productivity in sectors downstream. Vertical spillovers through forward linkages differ between service and manufacturing firms so we will consider them separately. We also observe considerable differences across firms in terms of size, export participation and absorptive capacity. 60The evidence for horizontal spillovers, that is, domestic firms’ productivity increasing when there in increased FDI in the same sector is less conclusive. Page | 68 68 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 35.! Manufacturing 35.! Manufacturing firms firms benefit benefit mostly mostly from from FDI FDI in upstream in upstream service service sectors sectors and and thethe benefits benefits accrue mainly to medium and larger firms. accrue mainly to medium and larger firms. Unconditional on firm capabilities we find no effects of of Unconditional on firm capabilities we find no effects upstream upstream FDI on on FDI productivity. productivity. ThisThis thethe is because is because effects effects accrue accrue onlyonly to medium to medium and and large large firms firms thatthat areare better positioned to benefit from FDI in upstream sectors. The benefits mainly accrue better positioned to benefit from FDI in upstream sectors. The benefits mainly accrue from FDI in upstream from FDI in upstream service sectors. A service sectors. A 1 percent increase in the FDI stock to output ratio in upstream service sector increases thethe 1 percent increase in the FDI stock to output ratio in upstream service sector increases productivity productivity of large of large by by firms firms almost almost 0.30.3 percent percent (Figure (Figure 74).74). ThisThis is line is in in line with thethe with findings findings of Fernandes of Fernandes andand Paunov Paunov (2012) (2012) in Chile in Chile andand suggests suggests thatthat manufacturing manufacturing firms firms cancan benefit benefit fromfrom thethe interaction interaction with with foreign foreign services services suppliers suppliers and and gaingain managerial, managerial, organizational, organizational, marketing, marketing, and and technological technological knowledge. knowledge. /@QH>=!UR.!*?;HE?F@;Q!G=F.!K>AIHF<@J@<[!Q?@;G!E>AL! /@QH>=!UR.!*?;HE?F@;Q!G=F.!K>AIHF<@J@<[!Q?@;G!E>AL! /@QH>=! /@QH>=! U^.! U^.! *?;HE?F@;Q! *?;HE?F@;Q! G=F.! G=F.! K>AIHF<@J@<[! K>AIHF<@J@<[! Q?@;G! Q?@;G! E>AL! E>AL! /-+!@;!HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!L=I@HL!?;I!B?>Q=! /-+!@;!HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!L=I@HL!?;I!B?>Q=! /-+!@;!HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!=ZKA><=>G! /-+!@;!HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!=ZKA><=>G! E@>LGW! E@>LGW! /@>LG! /@>LG! Q?@;! Q?@;! LAGAL! E>AL! /-+! /-+! @;! HKG<>=?L! @;! HKG<>=?L! G=>J@F=! G=>J@F=! G=FG! G=FG! Estimated Estimated effects effects a 1%a 1% increase increase in upstream in upstream Estimated Estimated effects effects a 1%a 1% increase increase in upstream in upstream FDIFDI by size by size upstream sectors on producitviy downstream FDI FDI by export by export status status upstream sectors on producitviy downstream Estimated effect of a 1% increase in FDI in Estimated effect of a 1% increase in FDI in 0,29% 0,29% 0,10% 0,10% upstream sectors on producitviy downstream upstream sectors on producitviy downstream Estimated effect of a 1% increase in FDI in Estimated effect of a 1% increase in FDI in 0,24% 0,24% 0,08% 0,08% 0,19% 0,19% 0,06% 0,06% 0,14% 0,14% 0,04% 0,04% 0,09% 0,09% 0,02% 0,02% 0,04% 0,04% -0,01% -0,01% 0,00% 0,00% Small Small Medium Medium Large Large Non-exporters Non-exporters Exporters Exporters FDIFDI in in all all upstream upstream sectors FDIFDI sectors in upstream in upstream service service sector sector FDIFDI in in all all upstream upstream sectors FDIFDI sectors in upstream in upstream service service sector sector ! ! ! ! Note: Note: TheThe estimates estimates shown shown in the in the graphgraph are are obtained obtained fromfrom plant- plant- Note: TheThe Note: estimates estimates shown shown in the in the graph graph are obtained are obtained fromfrom the the plant- plant- level level regressions regressions of TFP of TFP on upstream on upstream FDI FDI stocks, stocks, by firm by firm size. size. TheThe level level regressions regressions of TFP of TFP on upstream on upstream FDI FDI stocks, stocks, by export by export status. status. impact impact of overall of overall FDI FDI is not is not statistically statistically different different for for zero. zero. TheTheTheThe impact impact of overall of overall FDI FDI is not is not statistically statistically different different for for zero. zero. TheThe samesame applied applied to the to the impact impact on small on small firms. firms. samesame applied applied to the to the impact impact on exporters. on exporters. 36.! Manufacturing 36.! Manufacturing exporters exporters gaingain more more fromfrom FDIFDI in upstream in upstream services services sectors sectors thanthannon-non- exporters, independently of their size. Our results show that manufacturing firms that exporters, independently of their size. Our results show that manufacturing firms that participate in the participate in the export market experienced twice as big gains from FDI in services sectors than firms export market experienced twice as big gains from FDI in services sectors than firms that do not, that do not, independently independently of their of their sizesize (Figure (Figure 75).75). Firms Firms thatthat engage engage in the in the export export markets markets areare more more likely likely to have to have built built absorptive capacities to be able to adapt to international quality and standards, and are more absorptive capacities to be able to adapt to international quality and standards, and are more likely to demand likely to demand high high quality quality services services inputs, inputs, forfor which which FDIFDI presence presence upstream, upstream, maymay have have proven proven useful. useful. 37.! Firms 37.! Firms in the in the services services sector sector benefit benefit mostly mostly fromfromFDIFDI in upstream in upstream non-service non-service sectors sectors and thethe and benefits are greater for medium and larger firms. Medium and large firms are better benefits are greater for medium and larger firms. Medium and large firms are better positioned to benefit positioned to benefit from from FDI FDI in upstream in upstream sectors. sectors. TheThe benefits benefits mainly mainly accrue accrue fromfrom FDI FDI in upstream in upstream manufacturing manufacturing sectors. sectors. A1 A1 percent percent increase increase in the in the FDIFDI stock stock to output to output ratio ratio in upstream in upstream sector sector increases increases thethe productivity productivity of large of large service service firms firms by by almost almost 0.40.4 percent percent (Figure (Figure 76). 76). The The effects effects areare alsoalso larger forfor larger firms firms thatthat participate participate in the in the export export market market (Figure (Figure 77). 77). Page Page | 69 | 69 69 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY 38.! Overall 38.! Overall productivity productivity gains gains from from upstream upstream FDI FDI have have been been larger larger forfor firms firms withwith greater greater absorptive absorptive capacity. capacity. The The above above results results point point towards towards important important differences differences in terms in terms of firms’ of firms’ ability ability to to benefit benefit from from vertical vertical spillovers. spillovers. A firm A firm sizesize and and export export status areare status important important indicators indicators of of a a firm firm ability ability to to absorb absorb external external knowledge. knowledge. When When we we measure measure absorptive absorptive capacity capacity moremore directly directly by by considering considering whether whether a a firm firm engages engages in R&D in R&D investment investment we we obtain obtain similar similar results. results. Firms Firms that that engage engage in R&D in R&D investment investment areare better better positioned positioned to benefit to benefit fromfrom increased increased FDI FDI in upstream in upstream sectors. sectors. /@QH>=! /@QH>=! U_.! U_.! P=>J@F=G.! P=>J@F=G.! K>AIHF<@J@<[! K>AIHF<@J@<[! Q?@;G! Q?@;G! E>AL! E>AL! /-+! /-+! @;! @;! /@QH>=! /@QH>=! UU.! UU.! P=>J@F=! P=>J@F=! G=F.! G=F.! K>AIHF<@J@<[! K>AIHF<@J@<[! Q?@;G! Q?@;G! E>AL! E>AL! /-+! /-+! @;! @;! HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!L=I@HL!?;I!B?>Q=!E@>LGW! HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!L=I@HL!?;I!B?>Q=!E@>LGW! HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!=ZKA><=>GW! HKG<>=?L!G=FG!?>=!Q>=?<=>!EA>!=ZKA><=>GW! /@>LG! /@>LG! Q?@;! Q?@;! LAGAL! E>AL! /-+! /-+! @;! HKG<>=?L! @;! HKG<>=?L! ;A;TG=>J@F=! ;A;TG=>J@F=! G=FG! G=FG! Estimated Estimated effects effects a 1%a 1% increase increase in upstream in upstream Estimated Estimated effects effects a 1%a 1% increase increase in upstream in upstream FDIFDI by size by size FDIFDI by export by export status status 0,4%0,4% 0,2%0,2% upstream sectors on producitviy downstream upstream sectors on producitviy downstream upstream sectors on producitviy downstream Estimated effect of a 1% increase in FDI in upstream sectors on producitviy downstream Estimated effect of a 1% increase in FDI in Estimated effect of a 1% increase in FDI in Estimated effect of a 1% increase in FDI in 0,3%0,3% 0,2%0,2% 0,1%0,1% 0,1%0,1% 0,0% 0,0%0,0% 0,0% Small Medium Large Non-exporters Non-exporters Exporters Exporters Small Medium Large FDIFDI in in all all upstream upstream sectors FDIFDI sectors in upstream in upstream service service sector sector FDIFDI in in all all upstream upstream sectors FDIFDI sectors in upstream in upstream service service sector sector ! ! ! ! TheThe Note: Note: estimates estimates shown shown in the in the graphgraph are obtained are obtained fromfrom plant- Note: plant- TheThe Note: estimates estimates shown shown in the in the graphgraph are obtained are obtained fromfrom the the plant- plant- level level regressions regressions of TFP of TFP on upstream on upstream FDI FDI stocks stocks by firm by firm size. level size. level regressions regressions of TFP of TFP on upstream on upstream TFP TFP stocks stocks by export by export status. status. TheThe impact impact of service of service FDI FDI is not is not statistically statistically different different fromfrom zero. TheThe zero. impact impact of service of service FDI FDI is not is not statistically statistically different different fromfrom zero. zero. 39.! Another 39.! Another way way of identifying of identifying vertical vertical FDIFDI spillovers spillovers through through forward forward linkages linkages is focus is to to focuson on how restrictive policies have been towards FDI in upstream sectors, rather how restrictive policies have been towards FDI in upstream sectors, rather than on the actual stock than on the actual stock of FDI in upstream sectors. In Turkey, restrictiveness toward FDI has been of FDI in upstream sectors. In Turkey, restrictiveness toward FDI has been falling over time. At thefalling over time. At the beginning beginning of 2000 of 2000 all Turkish all Turkish sectors sectors had had somesome limited limited restrictions restrictions forfor FDI. FDI. A more A more open open regime regime hashas been been associated with an increased stock of FDI across all sectors of the economy. In this associated with an increased stock of FDI across all sectors of the economy. In this analysis we focus on analysis we focus on screening screening restrictions restrictions thatthat affected affected foreign foreign investment investment in both in both manufacturing manufacturing and and service service sector, sector, as opposed as opposed to equity and personnel restrictions that were only applied to multinationals operating in to equity and personnel restrictions that were only applied to multinationals operating in the services sector. the services sector. By 2016, all screening and approval restrictions have been eliminated. Some equity and By 2016, all screening and approval restrictions have been eliminated. Some equity and personnel restrictions personnel restrictions remain remain and and affect affect only thethe only service service sector, sector, in particular in particular in the in the communication, communication, business business andand transport transport sectors. sectors. 40.! The 40.! The reduction reduction of restrictiveness of restrictiveness to FDI to FDI in upstream in upstream sectors sectors hashas positively positively impacted impacted firms’ firms’ productivity. The effects are spread out across both manufacturing and service firms and productivity. The effects are spread out across both manufacturing and service firms and accrue from accrue from liberalization liberalization in both in both manufacturing manufacturing andand service service upstream upstream sectors. sectors. TheThe reduction reduction hashas benefitted benefitted mostly mostly firms firms that participate in the export market and larger that participate in the export market and larger firms. firms. Page Page | 70 | 70 70 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 41.! Turkish manufacturing firms that have a large “technology gap”, benefit more from reduced FDI restrictions than firms with higher levels of productivity. We observe this by estimating quantile regressions that estimate the effect of upstream FDI for firms at different levels of the productivity distribution. We find that least productive firms have gained the most from reduced FDI restrictiveness in upstream sectors. These findings are in line with those from Blalock and Gertler (2009) on Indonesia and suggest that the marginal return to new knowledge is greater for firms that have more room to “catch up” than it is for already competitive firms. Manufacturing firms with low initial technology are more likely to encounter new processes that yield high returns at low cost. "A;FBHG@A;!?;I!KAB@F[!AK<@A;G! 42.! The main messages that emerge from the empirical analysis on economic integration and productivity are as follows: 44.! Turkish firms have gradually become key players in global markets. Turkish firms’ market shares have grown internationally, as have their interdependence with foreign firms: the import content of their production has increased substantially, but so has its domestic value added. More firms – including small ones - have become internationalized, with an overall increase in the level of sophistication of exports. 45.! Productivity of Turkish firms has grown thanks to increased integration. Specifically, firms have become more productive by gaining increased access to foreign intermediates, and by interacting with more multinational companies upstream. 46.! Gains from integration have not accrued equally to all firms. The absorptive capacity of firms matters in determining how much they gain from increased exposure to foreign intermediates, or from the presence of multinationals upstream. Larger, more R&D intensive firms tend to benefit more than smaller, less innovative ones. 47.! Increased integration on the back of GVCs helped markets allocate resources more efficiently. In addition to the effects at the firm-level, evidence shows that in Turkey, the reduction in trade barriers improved selection, contributing to the shrinking and eventual exit of inefficient firms, and allowing the most efficient to thrive and grow. 43.! Several policy implications emerge from these results. Turkey’s Customs Union with the EU mean that for a wide set of products, prevailing tariffs are governed by the CET, within policy control of the EU, there is ample space for policies to increase integration. Some are related to improving the way in which the CU works, and some are related to areas not covered by the CU. In addition, because absorptive capabilities matter to gain from integration, policies in these areas are also relevant. Issues Policy options Within the Customs Union Agriculture and services are key sectors in their Widen CU to cover agriculture and services: dual role of producers of final goods and services and of inputs into other activities. Yet, •! In agriculture, matching the EU’s CET would reduce both sectors remain highly protected. import protection and could stimulate productivity in the same sector as well as downstream. Page | 71 71 in trade barriers improved selection, contributing to the shrinking and eventual exit of inefficient firms, and allowing the most efficient to thrive and grow. 43.! Several policy implications emerge from these results. Turkey’s Customs Union with the EU mean that for a wide set of products, prevailing tariffs are governed by the CET, within policy control of the CHAPTER EU, there 3/ is ample space forECONOMIC INTEGRATION policies to increase ANDSome integration. PRODUCTIVITY are related to improving the way in which the CU works, and some are related to areas not covered by the CU. In addition, because absorptive capabilities matter to gain from integration, policies in these areas are also relevant. Issues Policy options Within the Customs Union Agriculture and services are key sectors in their Widen CU to cover agriculture and services: dual role of producers of final goods and services and of inputs into other activities. Yet, •! In agriculture, matching the EU’s CET would reduce both sectors remain highly protected. import protection and could stimulate productivity in the same sector as well as downstream. Issues Policy options •! For services, increased integration, is also potentially Page | 71 productivity enhancing, as results reported here reveal. In this area, the establishment of an FTA with GATS+ type of agreement would help secure market access and national treatment commitments. It is important for Turkey to participate in EU Work towards the formalized structures for appropriate committees to improve bilateral dialogue consultations: between parties for the design of a common commercial policy. Because of the way the CU •! Parallel track negotiations mirroring the main EU is designed, the EU can negotiate FTAs with negotiations with third countries will help both the third parties without consultation, and then EU and Turkey start and conclude negotiations with Turkey is required to provide duty free access potential FTA partners within a similar time frame. to products from EU’s FTA partners without Currently, this is an obstacle for Turkey, that has reciprocity. struggle to complete negotiations with several of the EU’s FTA partners, eroding Turkish firms’ trade preferences. Bilateral and transit road quotas imposed by Work towards the removal of bilateral and transit quotas, the EU countries restrain free movement of so that Turkey and the EU can benefit from the full goods between the EU and Turkey and potential of the Customs Union. impede development of the EU-Turkey trade relations. Beyond the Customs Union TDI such as antidumping, safeguards and Carefully assess the criteria for the use of trade defence countervailing duties are used to prevent or to instruments (TDI): remedy injury on the domestic industry that stems from imports. While TDIs are not •! To reduce the impact of TDIs on trade, and thus to intended to undermine bilateral trade, they reduce the costs that firms face of securing the best introduce uncertainty to firms that export possible intermediate input, a reduction of the stock these products (when these TDIs are imposed of TDIs is needed, particularly those that have been in by trading partners) and firms that use them as place for a long time. inputs (when they are imposed by Turkey). Most of the TDIs have targeted China and Asia. Over time, FDI restrictions have been reduced Reduce barriers to FDI: substantially in Turkey. However, some equity and personnel restrictions remain and affect •! These are crucial inputs into production, where, as mainly the service sector in the results reveal here, increased competition and communication, business and transport sophistication of suppliers will likely benefit sectors. downstream firms using these services intensively. Firms’ capabilities Not all firms equally benefited from the Support firms in building up absorptive capabilities: spillovers that integration offered. Those with stronger absorptive capacities – the ones with •! For smaller, less sophisticated firms, support to a greater stock of knowledge, able to learn connect to multinationals, for example, through well- from interaction with more sophisticated designed suppliers-development programs, that suppliers, or to incorporate cutting edge incorporate monitor and evaluation mechanisms can intermediates or capital goods in their be helpful. Page | 72 72 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Issues Policy options production processes fared better. The •! For more sophisticated firms, support to investment spillover from increased openness may warrant in research and development, also with monitoring public policy interventions (to help internalize and evaluation mechanisms attached, could help the externality). widening the range of firms that benefit from productivity spillovers through openness. Page | 73 73 CHAPTER 3 / ECONOMIC INTEGRATION AND PRODUCTIVITY Chapter 4 Innovation support and firm performance Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Innovation support and firm performance 74 Closing Turkey’s innovation gap and the 74 roles of KOSGEB and TUBITAK TUBITAK and KOSGEB impact positively on 86 firm performance R&D, Innovation and public support 92 programs impact positively on TFP Conclusions and policy implications 95 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE +1W! +;;AJ?<@A;!GHKKA>L!K=>EA>L?;F=! 4.! R&D investment is an important source of innovation and productivity growth. Besides 1.! having a Firms’ ability significant to adapt influence and innovate on firms’ innovationis performance, a big determinant of within investing in R&D firm alsoproductivity growth. facilitates learning and This is influenced adopting in big part more advanced by the investment technologies climate. that are already in Adaptation and innovation use by competitors. entail like In countries risks including thatlarge Turkey are fixed not costs, close located long gestation the world’speriods, and uncertain technology returns. frontier, one wouldA conducive investment expect higher climate returns to can mitigate productivity from those risks through lower administrative burden on businesses, an efficient tax system, access to long-term learning by exporting and adaptation of existing technologies. Despite of these higher returns to productivity, finance, andshare the agglomeration economies. of investments In addition, in productivity public incentive enhancing activities programs in the GDP canissupport firms generally low directly to invest in for developing technology, countries; spend a fact on that isR&D, known and innovate. as the Incentives innovation can(Cirera paradox be in theandform of tax2017). breaks, preferential credit, Maloney, targeted public procurement, and direct grants. They can help overcome market failures for productivity enhancing 5. ! Evidence fromor investments; can distort Turkey also markets points to that R&Dleadand to allocative innovation inefficiencies. impacting positively on firm productivity. In their analysis of Turkish manufacturing firms (2003-2007), Dayar and Pamukçu (2014) find 2.! R&D that has chapter This a positiveassesses the impact and significant effectof on two large labor public incentive productivity. programs For a selection in Turkey of OECD on firm countries performance and productivity. The programs are delivered including Turkey, Erdil et al. (2013) show that the impact of R&D on labor productivity is positive intwo to firms in the form of grants through the leading long run.public institutions: Kılıçaslan KOSGEB et al.’s (2015) andof analysis TUBITAK. Together they Turkish manufacturing are the firms largest providers (2003-2010) ofinvestment find that grants for in business support, Information including for investments and Communication Technology in (ICT) R&D has andainnovation. The chapter positive impact does notthat on productivity cover other forms is around 25 of public incentives relating to tax, credit and procurement, which may have similar objectives. to 50 percent greater than investment in conventional capital. The World Bank (2010, 2014) also finds The chapter a close starts with an association overview between innovation investment in Turkey. capabilities, in innovation TFP(Figure It then analyzes 78): firm levels, and (i) the impact level of KOSGEB and productivity. TUBITAK grants on firm performance in terms of employment creation, R&D investment, and innovation; and 6. ! (ii) how R&D Turkey investment, has prioritized innovation, and KOSGEB R&D spending and TUBITAK and is gradually catchinggrants, up impact with itson TFP growth. peers. R&D expenditure grew rapidly from a low base, nearly doubling between 2005 and 2010 standing at just over 1 percent of GDP (Figure 79). This is the average for Trapped MICs today, and the average for high /@QH>=!UO.!#HL!K=>EA>L?;F=!?;?B[G@G performers during their transition from UMIC to HIC, though understandably below the average for HICs. Turkey’s Tenth National Development Plan (2014-2018) targeted R&D expenditures to reach 1.8 percent of GDP by 2018, which is the average for High Performers today (Figure 80). It also aimed to raise the private Employment KOSGEBto 60 percent, which it was already close to achieving in 2015. The scale sector’s share in R&D expenditure 62 up in R&D expenditure has sharply increased demand for researchers, particularly in manufacturing and (3) Üçdo#ruk, 2013). (1) R&D industry (Taymaz and Starting from a low base, annually-averaged number of researchers TUBITAK (2) TFP to the period 1996- per million employees has more than doubled over the period 2005-2014 in comparison 2004 (Figure 81). Innovation 7.! The boost of (1)! Impact in KOSGEB R&D spending has been and TUBITAK associated support with intellectual on employment creation, R&D property applications, investment and innovation though also(2)! Impact from aof low R&Dbase. investment and innovation In general, TFP on to applications growth the European Patent Office (EPO) in the 2013- 2017 period(3)! has Impact of KOSGEB declined relativeand support 82). TUBITAK (Figure to 2008-2012 on TFP growth Though this also applies to Turkey, the pace of applications has been greater than some European countries – not controlling for GDP or population size – "BAG@;Q!&H>D=[dG!@;;AJ?<@A;!Q?K!?;I!AB=G!AE!5#P2)C!?;I!&$C+&,5! in recent years. The gap between the largest applicants to the EPO – US, Germany, Japan – and the rest is very large. They are big sources of innovation that can help shift the global technology frontier. Intellectual charges 3.! for is a key property use of intellectual Innovation driver ofrights (IPR) are productivity however growth . Itvery low for Turkey can improve even compared firms’ production to efficiency, Trapped MICs (Figure enable product 83); both differentiation, in terms and extendof receipts product for Turkish lines. IPR, however, Innovation, but also payments forthat is a process foreign has aIPR. partially stochastic nature and can be influenced, to some degree, by R&D investment. The evidence on the linkages between R&D and innovation or productivity is mostly positive (e.g. Lee and Kang, 2007; Hegde and Shapira, 2007; Jefferson et al., 2006; Crespi and Zuniga, 2012), while in some cases, the effect occurs with a lag (e.g. Alvarez et al., 2010).61 61There is a large literature on whether public R&D crowds out private R&D investments. David et al. (2000) provide a critical view on the robustness of empirical results. Correa et al. (2013) find in general that public support does not crowd out private R&D investments. Görg and Strobl (2007) suggest that for domestic plants, small grants increase private R&D spending, while large grants may crowd out private R&D. For foreign establishments, public grants cause Page |additionality neither 74 nor crowding out effects. 62 For a more details on the allocation of R&D across different segments of the economy see link: TURKSTAT. Page | 75 74 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 4.! R&D investment is an important source of innovation and productivity growth. Besides having a significant influence on firms’ innovation performance, investing in R&D also facilitates learning and adopting more advanced technologies that are already in use by competitors. In countries like Turkey that are not located close the world’s technology frontier, one would expect higher returns to productivity from learning by exporting and adaptation of existing technologies. Despite of these higher returns to productivity, the share of investments in productivity enhancing activities in the GDP is generally low for developing countries; a fact that is known as the innovation paradox (Cirera and Maloney, 2017). 5.! Evidence from Turkey also points to R&D and innovation impacting positively on firm productivity. In their analysis of Turkish manufacturing firms (2003-2007), Dayar and Pamukçu (2014) find that R&D has a positive and significant effect on labor productivity. For a selection of OECD countries including Turkey, Erdil et al. (2013) show that the impact of R&D on labor productivity is positive in the long run. Kılıçaslan et al.’s (2015) analysis of Turkish manufacturing firms (2003-2010) find that investment in Information and Communication Technology (ICT) has a positive impact on productivity that is around 25 to 50 percent greater than investment in conventional capital. The World Bank (2010, 2014) also finds a close association between investment in innovation capabilities, TFP levels, and firm level productivity. 6.! Turkey has prioritized R&D spending and is gradually catching up with its peers. R&D expenditure grew rapidly from a low base, nearly doubling between 2005 and 2010 standing at just over 1 percent of GDP (Figure 79). This is the average for Trapped MICs today, and the average for high performers during their transition from UMIC to HIC, though understandably below the average for HICs. Turkey’s Tenth National Development Plan (2014-2018) targeted R&D expenditures to reach 1.8 percent of GDP by 2018, which is the average for High Performers today (Figure 80). It also aimed to raise the private sector’s share in R&D expenditure to 60 percent, which it was already close to achieving in 2015.62 The scale up in R&D expenditure has sharply increased demand for researchers, particularly in manufacturing and industry (Taymaz and Üçdo#ruk, 2013). Starting from a low base, annually-averaged number of researchers per million employees has more than doubled over the period 2005-2014 in comparison to the period 1996- 2004 (Figure 81). 7.! The boost in R&D spending has been associated with intellectual property applications, though also from a low base. In general, applications to the European Patent Office (EPO) in the 2013- 2017 period has declined relative to 2008-2012 (Figure 82). Though this also applies to Turkey, the pace of applications has been greater than some European countries – not controlling for GDP or population size – in recent years. The gap between the largest applicants to the EPO – US, Germany, Japan – and the rest is very large. They are big sources of innovation that can help shift the global technology frontier. Intellectual charges for use of intellectual property rights (IPR) are however very low for Turkey even compared to Trapped MICs (Figure 83); both in terms of receipts for Turkish IPR, but also payments for foreign IPR. crowd out private R&D investments. Görg and Strobl (2007) suggest that for domestic plants, small grants increase private R&D spending, while large grants may crowd out private R&D. For foreign establishments, public grants cause neither additionality nor crowding out effects. 62 For a more details on the allocation of R&D across different segments of the economy see link: TURKSTAT. Page | 75 75 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE /@QH>=!U:.!&H>D=[dG!'g-!=ZK=;I@=G!?>=!F?=!O8.!&N=!K>@J?<=!G=FdG!GN?>=!AE!'g-!=ZK=;I@=!@G! V@G! FBAG=!Q==!O9.!&N=!XAAG=?G=I! /@QH>=!O7.!0?<=;=L?@;!G<>A;Q]!!>=G=?>FN=>G!@;!&H>D=[! =BG=VN=>=!N?J=!GBAV=I!IAV;!KAG=!OS.!+0!HG=!FN?>Q=G!EA>!&H>D=[!?>=!BAV! /@QH>=!OR.!+;;AJ?<@A;!K=>EA>L?;F=!A;!I=FB@;@;Q!<>=;I! International charges for use of intellectual GCI Innovation score and rank property (share of GDP, 2017) 5,5 1,2% Payments Receipts GCI Innovation score (2007 - 2018 High performers 1,0% Trapped MICs 4,5 0,8% Turkey 0,6% 0,4% 3,5 0,2% 0,0% s s D D s s EU EU ey ey IC IC er er EC EC rk rk rm rm 2,5 M M Tu Tu O O rf o rf o ed ed 0 20 40 60 80 100 p p pe pe ap ap i gh i gh Tr Tr GCI Innovation rank (2007 - 2018) H H ! ! Sources: WDI, WB Staff estimates. Sources: World Economic Forum, Global Competitiveness Index. 8.! Beyond R&D spending and IP application, Turkey has important gaps in innovation capacity relative to comparators. In general, Turkey and its comparators were not the best innovators in the past ten years (Figure 84). Turkey’s innovation gaps relative to its comparators are greatest in the areas of university-company collaboration in R&D, private R&D spending, and the quality of research institutions (Figure 85). These gaps are reflected in Turkish firms’ ability to introduce new products/services and processes relative to firms in peer countries (Table 13). The EU’s Innovation Union Scoreboard 2016 classifies Turkey as a moderate innovator, better than modest innovators such as Bulgaria, Romania and Ukraine, but worse among moderate innovators. Despite these innovation gaps, Turkey still has a lot of “efficiency driving” reforms,63 which impact innovation capacity; and innovation will be as much about absorbing existing technology as developing new ones. 63The World Economic Forum’s GCI groups economies into three categories based on per capita GDP and natural resource dependence: (i) factor driven; (ii) efficiency driven; and (iii) innovation driven. At each stage, the relative importance of different reforms will vary. Turkey is classified as transitioning from efficiency to innovation driven though it still faces challenges in efficiency enhancing areas of reform as discussed in this report including well- functioning labor markets, developed financial markets. For more detail: GCI methodology. Page | 77 77 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE /@QH>=!O^.!Y?>Q=G=B?<@J=!G!@;FBHI=!H;@J=>G@<[TFALK?;[!FABB?XA>?<@A;!@;!'g-]!K>@J?<=! 'g-!GK=;I@;Q]!?;I!`H?B@<[!AE!>=G=?>FN!@;G<@G!E>AL!);<=>K>@G=!PH>J=[! Chile Poland Turkey Argentina Malaysia 2010 2013 2013 2017 2015 Percent of firms that spend on R&D 46.1 19.1 8 23.3 22.3 Firms that introduced a new product/service 59.9 44.5 13.5 51.7 8.3 Firms whose new product/service new to market 56.6 72.3 63.7 61.7 47.2 Firms that introduced a process innovation 58.8 34.7 12.6 43.2 58.6 Source: WB Enterprise Surveys. 9.! TUBITAK and KOSGEB are two leading public institutions that aim to accelerate R&D and innovation in Turkey. TUBITAK is the Scientific and Technological Research Council of Turkey, whilst KOSGEB is Department of SME Development and Support. Besides these two institutions that are specifically established to support firm performance, as of 2016, there were several other government bodies that provide different support mechanisms to firms operating in Turkey such as the Ministry of Industry and Technology, the Ministry of Development, the Ministry of Economy, the Ministry of Treasury and Finance, the Ministry of Energy and TURKPATENT. Of the 58 programs listed by TUBITAK64, 35 are financed through direct grants; other forms of support include technical assistance and subsidized credit. In terms of program beneficiaries: 36 percent are universities and research centers, as well as individual research projects; 25 percent are individuals with commercial projects, Non-Governmental Organizations (NGOs) and non- academic public institutions; and 39 percent are private companies. 64An overview of the public R&D, innovation, entrepreneurship and commercialization support programs in Turkey is published online by TUBITAK as an appendix at http://www.tubitak.gov.tr/ sites/default/files/ek_10_destekler_dagilimi.pdf (accessed on 26 Dec. 2017). Page | 78 78 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 10.! KOSGEB provides grants to SMEs not only for R&D and innovation, but also a broader range of business support services that could also accelerate within firm productivity. The Entrepreneur Information System (EIS) database used for the analysis in this chapter (Box 3) and report lists 13 different support programs beyond R&D and innovation. Under each main program title, there are multiple sub-program titles, and under each sub-program title, there is a large set of eligible expenditures for projects and specific cost items. KOSGEB programs can also cover expenses that arise from projects with specific targets, for instance, raising energy efficiency. As of 2017, KOSGEB grants had an upper limit of TL 5 million for any one firm and therefore not designed for large scale projects. 11.! KOSGEB grants are relatively evenly distributed across firms regardless of which sub-sectors they operate in within manufacturing or services. In other words, the size of a KOSGEB grant for a wood manufacturing firm may not be very different to a firm operating in motor vehicles. This is likely due to the relatively small size of the grants, suggesting that probably most successful applicants apply for and receive the maximum grant. This can be seen by the high correlation between the number of grant recipients in a sub-sector and the volume of grants to that sub-sector (Figures 86). For example, although the machinery and equipment sub-sector receives the largest volume of grants within manufacturing, it also has the largest number of grantees (Figure 86). 12.! TUBITAK Programs on the other hand are more targeted to specific projects or firm level upgrades. TUBITAK’s National Science, Technology and Innovation Strategy 2010-2016 shifted focus from research to innovation. It also identified several priority sectors for support, namely automotive, ICT, defense, space, health, energy, water, food, machinery and production technologies. TUBITAK’s Department of Technology and Innovation Support Programs (TEY-DEB) provides the largest portion of the R&D and innovation grants. Although firms in different size categories are eligible for TUBITAK grants, there are TUBITAK programs that target specific firm groups such as SMEs and start-ups. TUBITAK also targets industries that have stronger links with international markets. Contrary to KOSGEB programs, TUBITAK provides support primarily to technology intensive industries. TUBITAK also supports large-scale projects in manufacturing and in services and construction sector (Figure 87). Page | 79 79 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE /@QH>=!O_.!5#P2)C!Q>?;@XH<=I!?F>AGG!E@>LG!>=Q?>IB=GG!AE!GHXTG=F! Mach. & Equip. n.e.c. KOSGEB Grants in Manufacturing 120 Number of grantees in sub-sector 100 R" = 0,9272 Fabricated Metals 80 Other Motor Manufacturing Vehicles Rubber and Plastic 60 Chemicals Food Printing Computer, Electric Other Non- 40 Elec. And Basic Equip Met. Minerals Furniture Optic. Metals 20 Other Paper Textiles Transp Wearing Equip. Wood Leather Apparel 0 0 1000 2000 3000 4000 5000 6000 7000 8000 Volume of grants (TL Million, Base Year 2016) KOSGEB Grants in Services 120 Number of grantees in sub-sector 100 80 Archit. And Engineering Legal and Act. Accounting Act. Comp. Prog. And 60 Consultancy Specialised Construction R" = 0,8471 40 Sci. Research and Dev. Construction of Buildings 20 Activities of Telecoms Wh. And Ret. Trade of Head Offices Motor V. 0 Civil Engineering 0 1000 2000 3000 4000 5000 6000 7000 8000 Volume of grants (TL Million, Base Year 2016) Sources: EIS, WB Staff estimates. Notes: Wholesale trade except motor vehicles is an outlier with 22,600 grantees and around TL 225 million in grants. 13.! TUBITAK grants are as a result less evenly distributed across firms and industries within manufacturing than KOSGEB grants. Industries within manufacturing that are considered less technology intensive (e.g. leather, wood and paper) receive a smaller share of TUBITAK grants. Conversely, more technology intensive industries (e.g. motor vehicles, machinery and equipment and computer, electronic and optical products) receive a relatively largely share of TUBITAK grants. The variance in grant size across firms is bigger than in the case of KOSGEB. For three industries in particular – motor vehicles; computer, electronic and optical products; and other transport equipment – the number of grantees is relatively low, whilst the total volume of grants is large. The findings indicate that for the TUBITAK programs, priority sectors within manufacturing are more prominent and there are some larger scale projects supported in certain high-tech industries. This pattern is less evident in services; architectural and engineering services, and scientific research and development receive proportionately more grants but overall grants are low. Page | 80 80 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!OU.!&$C+&,5!Q>?;@;Q!?>=!LA>=!Q=<=I!@=G TUBITAK Grants in Manufacturing 500 400 Motor Vehicles Number of grantees in sub-sector 300 Other Transport Equip. R" = 0,354 Mach. And Equip. n.e.c. Wearing… Computer, Elec. And 200 Optic. Electrical Printing Wood Equipment 100 Rubber… Fabricated Furniture Chemicals Metals Paper Leather Textiles Food Other Non-Met. Minerals Basic Metals Other Manufacturing 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Volume of grants (TL Million, Base Year 2016) TUBITAK Grants in Services 500 Construction Legal and of Buildings Accounting Activities of 400 Head Offices Act. Civil Engineering Specialised Number of grantees 300 Construction Wh. And Ret. Comp. Prog. And Trade of Telecoms Consultancy Motor V. 200 R" = 0,9155 Sci. Research and Dev. Wh. Trade Except Motor 100 V. Archit. And Engineering Act. 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Volume of grants (TL Million, Base Year 2016) Sources: EIS, WB Staff estimates. 14.! In addition to average grant size and sector distribution, the success of TUBITAK and KOSGEB will also depend on whether firms with high growth potential are grant recipients. Studies find that firms’ growth potential can be a function of their size, age and innovativeness; in general, younger firms regardless of size tend to have higher growth potential, which is linked to their innovativeness (Box 9). 15.! In Turkey also the youngest firms exhibit significantly higher innovativeness – as measured by the total patent, trademark, model and design applications divided by the total number of employees (Figure 88). Trademark applications make up the bulk for youngest firms. The second youngest firm group (5-10 years) account for most patent applications (Figure 89); these are new firms that survive the start-up period. Taymaz and Üçdo#ruk (2009) find that smaller firms that overcome the first obstacle of conducting R&D tend to invest more in R&D than other firms in Turkey. Though patent applications for older firms picked up by the end of the sample period, they generally performed worse than younger firms. Page | 81 81 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE CAZ!:.!/@>L!FN?>?F<=>@G<@FG!?;I!Q>AV=!OO.!(AH;Q=GLG!=ZN@X@! /@QH>=!O:.!(AH;Q!E@>LG!J@J=!@AI!<=;I! @;;AJ?<@A;!K=>EA>L?;F=! =!@;;AJ?<@J=!!E@>LG! Innovation per Employee by Firm Size Patents (only) per Employee by Firm Size 0,012 0,0008 0,01 0,0006 0,008 0,0004 0,006 0,0002 0,004 0 0,002 2010 2011 2012 2013 2014 2015 2016 2010 2011 2012 2013 2014 2015 2016 5$ 10$,5< 20$,10< 5$ 10$,5< 20$,10< 40$,20< 40< 40$,20< 40< ! ! Sources: EIS, WB Staff estimates. Page | 82 82 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 16.! Turkey’s innovative firms in turn display higher employment growth than non-innovative firms across services and manufacturing (Figure 90). This is consistent with evidence from elsewhere (Box 9, Van Reenen 1997). The type of innovation matters – product innovation generally creates more jobs, whilst process innovation may not (e.g. Pianta, 2005; Harrison et al., 2014). The innovation proxy in this chapter (Box 9) would mostly reflect product type innovation. The results below are based on the PSM-DiD analysis (Box 9). The length of the bars in the figures reflect the estimated value of the treatment effect (in Figure 90 below, it is the effect of innovation on employment). The values are the average employment growth rate differences between the innovating firms and the non-innovators in the matched sample. A blue bar indicates that the estimated effect is significant at 10 percent or lower level; a green bar indicates that the estimation is insignificant. /@QH>=!:8.!+;;AJ?<@J=!E@>LG!@;!XA@;Q!?;I!G=>J@F=G!I@GKB?[!N@QN=>!=LKBA[L=;AV@;Q!E@>L!FN?>?F<=>@G<@FG!JGW!@;;AJ?<@J=;=GG!?;I!KHXB@F!GHKKA>J@F=!?;I!FA;G<>HF<@A;!E@>L!FN?>?F<=>@G<@FG!JGW!@;;AJ?<@J=;=GG!?;I!KHXB@F!GHKKA>L!K=>EA>L?;F=! 20.! They key question is what impacts have TUBITAK and KOSGEB had on firm performance and productivity? How has the distribution of grants across industries and firm types impacted on employment, R&D, innovation and productivity growth? Do the impacts differ across the two programs and why? This section and the next look at these issues using firm level data (Box 10). They apply two types of micro-econometric methods to capture: (i) the impact of public support programs on firm innovativeness, employment and R&D expenditures; and (ii) the effects of innovation, R&D and public support on TFP (Box 10). CAZ!98.!-?!?GG=GG@;Q!=B?<@J=B[!B?>Q=>!=!:9.!5#P2)C!N?G!?!>=B?<@J=B[!B?>Q=!KAG@<@J=!=LKBA[L=;@;Q! KOSGEB Support TUBITAK Support Other Transport Equipment Other Transport Equipment Computer, Elec. & Optic. Other Manufacturing Furniture Food Motor Vehicles & Trailers Wood Electrical Equipment Chemiclas Wood Fabricated Metals Food Computer, Elec. & Optic. Fabricated Metals Wearing Apparel Textiles Textiles Basic Metals Paper Paper Rubber & Plastic Rubber & Plastic Basic Metals Printing Electrical Equipment Machinery & Equip. n.e.c. Motor Vehicles & Trailers Chemiclas Machinery & Equip. n.e.c. Wearing Apparel Other Non-Metalic Minerals Leather Insignificant Printing Insignificant Furniture Significant at 10% Other Manufacturing Significant at 10% Other Non-Metalic Minerals Leather 0 0,1 0,2 -0,05 0 0,05 0,1 0,15 0,2 0,25 Sources: EIS, WB Staff estimates. 22.! The relatively larger positive employment impacts of KOSGEB across more industries is likely related to the nature of the two programs. TUBITAK grants are not necessarily aimed at increasing the firm size but innovation, while the KOSGEB grants are mainly targeted to accelerate the growth of the SMEs. Since the sample of firms used in the estimations for the KOSGEB and TUBITAK grants are different, a direct comparison of the impacts of the two institutions’ support programs on firm performance may be, to some degree, misleading. Nevertheless, comparing of the signs of the impacts of the two institutions’ support programs is meaningful regardless of the differences in the sample size. 23.! Results are similar in the services sector. KOSGEB support impact on employment is significantly positive in every service or construction industry, while the TUBITAK support does not have a significant effect on employment in 4 out of 11 industries (Figure 92). The computer programming and consultancy and the wholesale trade except motor vehicles sectors, which receive the largest portions of the grants, the employment effects are significantly positive but relatively low in comparison to other industries where the TUBITAK effect is significant. In scientific research and development, which receives the third largest portion of the TUBITAK grants, the effect of grants on employment is insignificant and negative Page | 88 88 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!:7.!5#P2)C!=LKBA[L=;A;Q!@;!G=>J@F=G KOSGEB Support TUBITAK Support Legal & Accounting Act. Construction of Buildings Architect. & Engineering Act. Legal & Accounting Act. Civil Engineering Specialised Construction Telecommunications Telecommunications Construction of Buildings Architect. & Engineering Act. Specialised Construction Computer Prog. & Consult. Computer Prog. & Consult. Wholesale Tr. Except M. Vehi. Activities of Head Offices Wh. & Ret. Trade of Motor V. Wholesale Tr. Except M. Vehi. Activities of Head Offices Scientific R&D Scientific R&D Wh. & Ret. Trade of Motor V. Civil Engineering 0 0,1 0,2 -0,2 0 0,2 Insignificant Significant at 10% Insignificant Significant at 10% Sources: EIS, WB Staff estimates. bXc! &$C+&,5dG!@LK?F=B?<@J=B[!B?>Q=>!=!:S.!&$C+&,5!N?G!?!B?>Q=!?;I!G@Q;@E@F?;AGG!L?;HE?F@;Q KOSGEB Support TUBITAK Support Computer, Elec. & Optic. Leather Chemiclas Food Motor Vehicles & Trailers Paper Printing Basic Metals Machinery & Equip. n.e.c. Other Manufacturing Paper Wearing Apparel Furniture Other Non-Metalic Minerals Textiles Fabricated Metals Basic Metals Chemiclas Electrical Equipment Machinery & Equip. n.e.c. Other Manufacturing Textiles Rubber & Plastic Insignificant Motor Vehicles & Trailers Insignificant Wearing Apparel Furniture Significant at 10% Significant at 10% Leather Computer, Elec. & Optic. Other Non-Metalic Minerals Printing Fabricated Metals Electrical Equipment Food Rubber & Plastic Wood Other Transport Equipment Other Transport Equipment Wood -0,2 0,3 0,8 0 1 2 3 Sources: EIS, WB Staff estimates. /@QH>=!:R.!&$C+&,5!?BGA!@LK?FJ@F=G]!VN=>=?G!5#P2)C!GHKKA>=?<=>!@LK?F<@FHB?>B[!@;!L?;HE?F@;Q! 26.! Measuring the impact of public policy on innovation is a challenge, partly because innovation is difficult to measure. Firms’ R&D expenditures are often used as a proxy, but this is not always correlated with innovation. There is also survey data with firms’ own assessments of their innovativeness, but this is constrained by respondents’ knowledge and interests. The survey type innovation data such as the Community Innovation Survey for Turkey often lacks the time dimension which restricts a dynamic analysis to capture firms’ innovation performance over time. Page | 90 90 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 27.! The innovation measure adopted in this study (Box 10) addresses some of the weaknesses of other approaches though also has its own shortcomings. The innovation measure based on firms’ patenting activities can be considered as more reliable than firms’ self-assessment of their innovativeness, mainly because firms incur costs and spend time to make an application to TURKPATENT. This way of measuring innovation, however, does not fully account for some types of innovative activities such as process innovation that constitutes an important part of innovative activities. Patenting activities only capture innovations that are new, but not the adaptation of existing technologies that are not new to the market but new to the adopting firms. 28.! The impact of TUBITAK support on innovation in the manufacturing sector is larger and significant in more industries than the impact of KOSGEB (Figure 95). Except for two industries – wearing apparel and paper – which receive a small share of total TUBITAK grants, the impact of TUBITAK support on innovation is significantly positive. The estimated effect is highest in low-tech industries such as leather, furniture and wood. The impact of KOSGEB grants on the other hand is not significant in 5 industries, and is negative and significant for the activities of head offices and management consultancy. /@QH>=!:^.!&$C+&,5!N?G!?!>=B?<@J=B[!B?>Q=!?;I!G@Q;@E@F?;@;Q! KOSGEB Support TUBITAK Support Chemiclas Leather Furniture Furniture Electrical Equipment Wood Other Manufacturing Other Manufacturing Machinery & Equip. n.e.c. Other Non-Metalic Minerals Basic Metals Food Rubber & Plastic Other Transport Equipment Computer, Elec. & Optic. Motor Vehicles & Trailers Food Computer, Elec. & Optic. Motor Vehicles & Trailers Textiles Other Transport Equipment Electrical Equipment Fabricated Metals Chemiclas Insignificant Insignificant Wearing Apparel Machinery & Equip. n.e.c. Other Non-Metalic Minerals Significant at 10% Rubber & Plastic Significant at 10% Wood Fabricated Metals Paper Wearing Apparel Leather Basic Metals Printing Printing Textiles Paper 0 0,02 0,04 0 0,1 0,2 Sources: EIS, WB Staff estimates. 29.! The impact of both TUBITAK and KOSGEB on innovation in the services sector is more variable (Figure 96). TUBITAK support is significantly positive in 4 industries, which includes the largest industry – wholesale trade except motor vehicles. The results for the service industries is sensitive to the type of innovation measure used. As mentioned earlier, firms’ applications to TURKPATENT do not reflect process innovation adequately which can be an important component of innovation in the service sector. This may explain the observed weak link of public support programs with innovation in services. Page | 91 91 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE /@QH>=!:_.!+LK?FJ@F=G!@G!GL?BB!?;I!;A[!G@Q;@E@F?;AQ>?LG!@LK?F@;Q!?;I!J@F=G! ! 30.! The impact of R&D and innovation on TFP is estimated jointly to assess their relative importance. R&D and innovation indicators are introduced into the TFP estimation as the mean shifters for the dynamic first order Markov process that is assumed to represent the evaluation of TFP over time. Unlike in the previous parts, the R&D variable used in this estimation is an indicator variable that takes the value of 1 if the firm makes any R&D investments in that year and 0 otherwise. Using an indicator variable for R&D makes the coefficients of innovation and R&D more easily comparable, since the innovation is also proxied by an indicator variable. 31.! The impact of R&D on TFP in the manufacturing is significantly positive across all industries. Unlike in the previous section, the coefficient estimates in this section represent the percentage change in the TFP following innovation or investment in R&D in the previous period. There is some degree of heterogeneity in the productivity gains from R&D across the manufacturing industries, but this heterogeneity does not seem to be linked with the technology intensity of the industries. In some low-tech industries such as the manufacturing of printing and paper, the R&D effect is the largest, but in some other low-tech industries such as the manufacturing of wearing apparel and leather, the R&D impact on the TFP is the smallest. The estimated impact of the R&D is rather similar across the high-tech industries. Page | 92 92 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 32.! The impact of innovation on the TFP is significant and positive in all except two manufacturing industries (Figure 97). It is worth mentioning that the joint estimation of the coefficients of the R&D and innovation may suffer, to some degree, from multicollinearity and the estimates can be downwards biased. Interpreting the two sides of the graph jointly reveals that in 4 manufacturing industries – namely other non-metallic minerals; other transport equipment; computer, electronics and optical equipment; and wood – there are larger returns to productivity from R&D and innovation. /@QH>=!:U.!'g-!?;I!@;;AJ?<@A;!N?J=!?!KAG@<@J=!?;I!G@Q;@E@F?;@;Q R&D Effect on TFP Innovation Effect on TFP Printing Other Transport Equipment Paper Other Non-Metalic Minerals Other Non-Metalic Minerals Computer, Elec. & Optic. Other Transport Equipment Wood Rubber & Plastic Printing Computer, Elec. & Optic. Wearing Apparel Wood Food Other Manufacturing Other Manufacturing Motor Vehicles & Trailers Textiles Electrical Equipment Rubber & Plastic Food Furniture Chemiclas Insignificant Machinery & Equip. n.e.c. Insignificant Machinery & Equip. n.e.c. Significant at 10% Electrical Equipment Significant at 10% Textiles Paper Furniture Leather Fabricated Metals Fabricated Metals Basic Metals Motor Vehicles & Trailers Wearing Apparel Basic Metals Leather Chemiclas 0 0,2 -0,02 0,08 Sources: EIS, WB Staff estimates. /@QH>=!:O.!&N=!@LK?FJ@F=!E@>LGd!&/0!@G!?BGA!G@Q;@E@F?;< R&D Effect on TFP Innovation Effect on TFP Scientific R&D Scientific R&D Architect. & Engineering Act. Telecommunications Activities of Head Offices Construction of Buildings Legal & Accounting Act. Architect. & Engineering Act. Wh. & Ret. Trade of Motor V. Computer Prog. & Consult. Wholesale Tr. Except M. Vehi. Telecommunications Specialised Construction Wholesale Tr. Except M. Vehi. Civil Engineering Wh. & Ret. Trade of Motor V. Computer Prog. & Consult. Specialised Construction Activities of Head Offices Civil Engineering Legal & Accounting Act. Construction of Buildings -0,06 0,04 0,14 0 0,2 0,4 Insignificant Significant at 10% Insignificant Significant at 10% Sources: EIS, WB Staff estimates. Page | 93 93 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE 33.! In the services sector, both R&D and innovation have positive TFP effects across most industries (Figure 98). The impact of R&D on the TFP is positive and significant in every service industry. The effect of innovation, however, is not positive or significant in 3 service industries. This may be due to the innovation measure used in this study that does not capture all types of innovative activities in services. In addition, some of the innovation effect on the TFP may be captured by R&D which would explain non- positive effects observed in some service industries. Overall, the results indicate strong links between R&D or innovation and the TFP of firms in the main sectors of Turkey. ! bXc! &$C+&,5!N?G!?!G<>A;Q!?;I!G@Q;@E@F?;=B?<@J=!=!::.!&$C+&,5!@LK?F@;Q!&/0]!H;B@D=!5#P2)C KOSGEB Effect on TFP TUBITAK Effect on TFP Other Transport Equipment Other Transport Equipment Other Non-Metalic Minerals Other Non-Metalic Minerals Computer, Elec. & Optic. Computer, Elec. & Optic. Wood Wood Printing Printing Wearing Apparel Wearing Apparel Food Food Other Manufacturing Other Manufacturing Textiles Textiles Rubber & Plastic Rubber & Plastic Furniture Furniture Machinery & Equip. n.e.c. Insignificant Machinery & Equip. n.e.c. Insignificant Electrical Equipment Electrical Equipment Significant at 10% Significant at 10% Paper Paper Leather Leather Fabricated Metals Fabricated Metals Motor Vehicles & Trailers Motor Vehicles & Trailers Basic Metals Basic Metals Chemiclas Chemiclas -0,06 -0,03 0 0,03 0,06 0 0,2 0,4 0,6 Sources: EIS, WB Staff estimates. Page | 94 94 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 36.! A similar pattern in the services sector. For KOSGEB, the impact on TFP in four service industries is either insignificantly low or negative. In the largest 2-digit industry, the wholesale trade except the motor vehicles, which also receives the largest share of KOSGEB support in services, the impact of KOSGEB support is significantly positive. For TUBITAK, the impact is particularly large in the legal and accounting service, though the amounts allocated as well as the number of supported firms are low. In the two industries that receive the largest shares of the TUBITAK grants in services (computer programing and consultancy, and wholesale trade except the motor vehicles) the TFP effects of the TUBITAK programs are significant; TUBITAK grant lead to a 20 percent increase in firm productivity in the consecutive year. The impact on the TFP is even larger in the architectural and engineering activities and in the scientific research and development which are the two sectors that receive relatively large shares of the TUBITAK grants. "A;FBHG@A;G!?;I!KAB@F[!@LKB@F?<@A;G! 37.! The main messages that emerge from the analysis on innovation support and firm performance in Turkey are as follows: (i)! Turkey has prioritized R&D spending though has important gaps in innovation capacity relative to peers: Boost in R&D spending is associated with a big increase in intellectual property applications and demand for researchers. Turkey’s innovation gaps are greatest in university-company collaboration, private R&D spending, and quality of research institutions. (ii)! KOSGEB impact is stronger on employment than TFP: KOSGEB impact on employment is stronger than that of TUBITAK. Although the effects of KOSGEB grants on innovation and R&D is positive, the estimated impact on TFP is mostly non-positive. (iii)! TUBITAK has a strong and positive impact on R&D, innovation and TFP: The impact is particularly strong in the manufacturing sector. R&D and innovation impact positively on TFP in manufacturing and to some extent in services. (iv)! Improved targeting of TUBITAK and KOSGEB grants could potentially improve impact: Though younger firms are likely to be more innovative, they are not more likely to received TUBITAK grants. Issues Policy options Young firms exhibit stronger capacity to To enhance the positive impact on productivity and innovate and become more productive. But innovation, TUBITAK programs should target firms being young does not improve chances of with high growth potential and support their R&D receiving a TUBITAK grant in 10 out of 11 related activities. services sectors, and for KOSGEB grant in 5 Shifting the focus of public support programs, out of 11. Being young does not improve the particularly TUBITAK ones, in service industries, chance of receiving a TUBITAK grant in 13 from the SMEs towards start-ups or young firms out of 19 manufacturing industries. In 4 would be more successful in accelerating and industries, being young significantly decreases sustaining productivity growth. Targeting young firms the probability of receiving a TUBITAK grant. would also lead to more number of entries which in turn enhances competitive pressure on incumbents and can motivate them to be more innovative and efficient Page | 95 95 CHAPTER 4 / INNOVATION SUPPORT AND FIRM PERFORMANCE Issues Policy options Young firms exhibit stronger capacity to To enhance the positive impact on productivity and innovate and become more productive. But innovation, TUBITAK programs should target firms being young does not improve chances of with high growth potential and support their R&D receiving a TUBITAK grant in 10 out of 11 related activities. services sectors, and for KOSGEB grant in 5 Shifting the focus of public support programs, out of 11. Being young does not improve the particularly TUBITAK ones, in service industries, chance of receiving a TUBITAK grant in 13 from the SMEs towards start-ups or young firms out of 19 manufacturing industries. In 4 would be more successful in accelerating and industries, being young significantly decreases sustaining productivity growth. Targeting young firms the probability of receiving a TUBITAK grant. would also lead to more number of entries which in turn enhances competitive pressure on incumbents and can motivate them to be more innovative and efficient KOSGEB programs have been successful in KOSGEB eligibility criteria should include employment creation. But KOSGEB support productivity enhancing interventions. does not always impact TFP positively. Grant allocation mechanism to take productivity as Therefore, employment creation may be criteria to avoid poorly productive firm to be temporary i.e. supported firm is likely to shrink supported and protected from exiting back to its pre-support size. Alternatively, grants are sustaining less productive firms that would otherwise go out of business. Support to SMEs can indirectly help newly Support to SMEs should distinguishing between created establishments to survive during the young and/or high growth potential firms from fragile start-up period. SME support programs older/lower growth potential firms. The evidence in can motivate entry, because new firms tend to this chapter shows that young firms generally have be small and eligible for the funds allocated to higher potential to grow and innovate. the SMEs. The group of SMEs, however, does not only contain young firms. They include old firms that could not grow. So long as there is no significant obstacle to firm growth, the older SMEs generally stay small and exhibit poor productivity performances due to low managerial quality, lack of incentives to innovate or inefficient organizational structure. The impact of public support on firm Research into the underlying reasons behind policy performance vary considerably across failures in the specific industries could help design industries in Turkey. This study provides some more effective interventions. insights into how to detect those industries where there are limited or no effect. There are large returns to productivity from The authorities should sustain and grow budget R&D in Turkey. This is an area that received allocations for incentive programs to support firms’ priority over the past 10 years. R&D related activities. The impact of alternative financing mechanisms (credits, grants and tax incentives) should also be reviewed. Page | 96 96 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey CHAPTER 1 Chapter 5 Human capital and productivity Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Human capital and productivity 97 More education not always translating 97 into better jobs Demand side factors relative to supply side 104 factors deter deepening of skills No significant relationship between 110 skill composition and firm productivity Conclusions and policy implications 114 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY 1W!4HL?;!F?K@AIHF<@J@<[!! 1.! New technologies and business lines from economic integration and innovation discussed in chapters 3 and 4 can create new and better jobs that will require human capital. New technologies and business lines boost the demand for more capable workers, that in turn enhance firms’ absorptive and technical capacities. These factors work together to accelerate within firm productivity growth. Investing only in new technologies and business lines without human capital would create unutilized, excess capacity; whilst investing only in human capital without new technologies and business lines would create an excess supply of workers with redundant qualifications. Both outcomes are inefficient and will not accelerate productivity. 2.! Building on the above, this chapter analyzes labor market supply and demand dynamics in understanding the challenges to accelerating within firm productivity growth (i.e. is firm productivity constrained more by the lack of supply of qualified workers or is demand for qualified workers not keeping pace with supply?) This has implications for education policy and the participation of industry in building human capital. The analysis includes: (i) a review of labor supply in terms of its educational attainments, skills, and gender, and its absorption in the labor market; (ii) the types of skills most demanded in the labor market, which can guide policy and enable labor markets to clear effectively; and (iii) the relationship between the skills composition of the workforce and firms’ job creation and productivity. *A>=!=IHF?<@A;!;A?;GB?<@;Q!@;!fAXG! b?c! +;F>=?G=I!B?XA>!GHKKB[!@;!=!988.!&H>D=[!N?G!?!>?K@IB[!Q>AV@;Q!VA>D@;QT?Q=! /@QH>=!989.!,;I!>=B?<@J=B[!BAV!?Q=TI=K=;I=;F[]!VN@FN! KAKHB?<@A;! V@BB!B?G 15 years, Age dependency ratios (% of working age 2010-2030) population) Tajikistan 44,3 100 Turkey 30,0 80 60 Czech Republic 1,5 40 Poland -2,0 20 0 60 67 74 81 88 95 02 09 16 23 30 37 44 19 19 19 19 19 19 20 20 20 20 20 20 20 Bulgaria Total (Children + Elderly) Elderly (Age 65 +) -15 5 25 45 Children (Age 0-14) Sources: Arias et al. (2014). Sources: UN Population projects, WB Staff estimates. bXc! )IHF?<@A;?B!?<!L?>D=X!LA>=!=IHF?<=I!VA>D=>G! 5.! There are signs of a gradual improvement in the education profile of the labor force in recent years thanks to high employment growth among the more educated, including women. While employment increased by 5 percent overall (2014-2016) for the population aged 25 years and over, employment grew more quickly for those with tertiary education (19.4 percent) (Figures 102, 103). Those with primary education or lower experienced the slowest growth rate. This is in part a base effect given the large share of employed with relatively low education. But for both men and women, those with university education were the biggest drivers of employment growth between 2014 and 2016. This may therefore reflect an increased demand for more qualified workers, met by increased supply from recent progress in improving educational attainments of workers. 6.! At the same time, however, the share of unemployed workers with vocational and university education is relatively large, particularly among women. For men, around 27 percent of those unemployed have higher qualifications (i.e. vocational and university); for women it is just over 40 percent (Figure 104, 105). These figures show that unemployed women are more educated than men, and that – in addition to being exposed to higher unemployment risk than men for childbearing – they might suffer more severely from the mismatch problem between skills demanded and supplied 7.! In addition, unemployment among the more educated is also growing very quickly, particularly among women. Between 2014 and 2016, there was a big increase in the number of unemployed with lower-secondary and university education. This could be because new university graduates are being more selective about taking up their first job (the so-called phenomenon of “wait unemployment”). From the labor demand perspective, this could also signal a mismatch between the skills acquired at university and the skills demanded by employers (see below). Page | 98 98 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY /@QH>=!987.!PN?>=!AE!VA>D=>G!V@=!98S.!PAL=!>=X?B?;F@;Q!>=F=;AHQN!N@QN! ?<=L?@;G!N@QN! =LKBA[L=;AV=!=IHF?<=I! Employment levels by education attainment Growth in employment and contribution by and gender (> 25 years old, 2016) education attainment and gender (> 25 years old, 2014-2016) 8% 20 Employment levels (Million) 6% 15 4% 10 2% 5 0% 0 -2% Overall Male Female Overall Male Female Primary and less than primary Secondary Primary and less than primary Secondary High school Vocational high school High school Vocational high school University University Total Sources: LFS 2014 and 2016, WB Staff estimates. /@QH>=!98R.!$;=LKBA[L=;AGG!B=J=BG!AE!=IHF?<@A;!@G! /@QH>=!98^.!CH=!=IHF?<=I! >=B?<@J=B[!X>A?I!X?G=I! VA>D=>G!@G!Q>AV@;Q!LA>=!>?K@IB[! Unemployment levels by education attainment Growth in unemployment and contribution by and gender (> 25 years old, 2016) educational attainment and gender (> 25 years old, 2014-2016) 28% Unemployment levels (Million) 2 23% 18% 13% 1 8% 3% 0 -2% Overall Male Female Overall Male Female Primary and less than primary Secondary Primary and less than primary Secondary High school Vocational high school High school Vocational high school University University Total Sources: LFS 2014 and 2016, WB Staff estimates. 8.! Another concern for the education and skills mix of the labor force is the large share of people not in employment, education, or training (NEET) particularly among the young. In 2016 the NEET rate among the 15–24 age group was 24 percent and 28 percent among the 15–29 age group (Figure 106). This is almost double the average NEET rate in OECD countries (13.9 percent), indicating that a significant share of the Turkish labor force is not contributing to the economy, or undergoing any training to be able to contribute to the economy in the future. Page | 99 99 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!98_.!,!B?>Q=!GN?>=!AE!9^T7:T[=?>TABIG!?>=!;A=!98U.!*AG!<>?@;@;Q! @;FBHI=!LA>=!<=FN;@F?B!GHXf=F!=IHF?<@A;?B!?<!=IHF?<@A;?B!?<=!98O.!0>AE@F@=;F[!Q?KG!X=D=[!?;I!#)"-!?>=!GL?BB=>!?LA;Q![AH;Q!?IHB=!98:.!P<=?I[!@LK>AJ=L=;=G]!XH=!998.!0=>EA>L?;F=!A;!>=?I@;Q!GF?B=!BAV!XHK!I>AK!@;!789^! V@=;I!VN=;!FA;<>ABB@;Q!EA>!K=>!F?K@=!999.!&+*PP!L?=G!N?J=!@LK>AJ=I!XH=!997.!&+*PP!GF@=;F=!GFA>=G!N?J=!@LK>AJ=I!VN@BGD=[dG!>?;D@;Q!N?G!E?BB=;! &H>D=[dG!>?;D@;Q!N?G!>=L?@;=I!G!>=B?<@J=!G!I=<=>!I==K=;@;Q!AE!GD@BBG! ! b?c! P<=?I[!XH=?G=!@;!!LA>=!GD@BB=I!VA>D=>G! ! 17.! The demand for cognitive skills that are associated with better quality jobs is gradually increasing in Turkey. Building on the skills assessment above, it is possible to look in more detail at the skills content of the economy to measure whether jobs in Turkey are moving towards or away from a knowledge-based economy.73 A knowledge-based economy is one where cognitive and technical skills are common features of most occupations. The economy is increasing its use of cognitive skills associated with better quality of jobs compared to the manual skills associated with lower quality jobs; the reliance on lower level manual skills is declining over time. 18.! Breaking this down further, two distinct patterns are observed. First, for the national sample, both non-routine manual physical and routine manual skills are becoming less dominant, a sign of decreasing dependence on the type of manual skills which are typical in lesser quality jobs (Figures 113, 114). Non- routine cognitive (analytical and interpersonal) and routine cognitive skills are dominant, which is expected in an economy with relatively high levels of good jobs, but jobs intense in routine cognitive skills have seen the largest increase in the last years. These include jobs such as those performed by bookkeepers or call center operators. Unfortunately, these jobs are at high risk of automation. Second, for workers with low education levels, employment is predominantly in occupations with high levels of routine tasks. This indicates that in the medium or longer term this segment of the population is at risk of losing jobs to automation. 73In the methodology, first, ISCO 88 2-digit occupations were converted to ISCO 08, and then each occupation was assigned with an average per skill (e.g., Non-routine cognitive: Analytical), which is associated with a scale such as importance. Each scale has a minimum and maximum value (e.g., Importance: 1 – 5). Then, to get standard values, z- scores were computed by year and skill. Page | 104 104 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY /@QH>=!99S.!%A;T>AH<@;=!L?;H?B!KN[G@F?B!?;I!>AH<@;=! /@QH>=!99R.!,LA;Q!B=GG!=IHF?<=I]!=LKBA[L=;=!X=FAL@;Q!B=GG!IAL@;?;< >AH<@;=!@GD!AE!L=FN?;@\?<@A; Turkey: National Sample Turkey: Low education 0,15 0,5 0,05 Z-score Z-score 0 -0,05 -0,15 -0,5 2010 2011 2012 2013 2014 2015 2016 2010 2011 2012 2013 2014 2015 2016 Non-routine cognitive: Analytical Non-routine cognitive: Analytical Non-routine cognitive: Interpersonal Non-routine cognitive: Interpersonal Routine cognitive Routine cognitive Routine manual Routine manual Non-routine manual physical ! Non-routine manual physical ! Source: WB Staff estimates. Note: Skills classification follows Autor (2014).! 19.! Inadequacy of skills is often cited as a constraint to doing business in Turkey.74 Around 7 percent of firms perceive inadequately educated labor force as an obstacle; this is regarded as the biggest obstacle for firms located in the western parts of Turkey. Firms in large labor markets such as Istanbul, Ankara, and Izmir find the inadequacy of education in the labor force as a top constraint. These provinces have relatively diversified economies, indicating that this obstacle is encountered by firms across sectors. For firms in surrounding sub-regions (Manisa-Afyonkarahisar-Kutahya-Usak and Denizli-Aydin-Mugla) where wholesale and retail, and services in general dominate the economy, this obstacle is listed as a top constraint as well. 20.! Though firms can play a big role in developing workers’ skills, firms in Turkey invest relatively less on this compared to countries at similar levels of income (Figure 115). On-the-job learning is an investment that can increase workers’ productivity and wages through the accumulation of skills (Becker 1964; Heckman 1976; Mincer 1962, 1968). Both High Performer and Trapped MIC comparators have a larger share of firms offering formal training programs for their permanent, full-time employees. Turkey ranks 63rd out of 94 countries in this dimension. 21.! The Turkish Employment Agency (ISKUR) provides training to support people to find jobs more quickly. It has many Active Labor Market Policies (ALMPs) aimed at enhancing the skills of the working-age population and increasing their employability. This is a priority that has been matched with an increase in budget allocations; the number of programs and beneficiaries has also increased substantially in recent years. Many of the programs are skills training and on-the-job training programs. While nearly 393 million Turkish Lira was spent for 212,000 beneficiaries in 2010, the number of beneficiaries increased to 593,633 and expenditures rose to TL 5.3 billion in 2016 (Republic of Turkey Ministry of Development 2017). 74WBG Enterprise Survey (2015). This is the fifth biggest obstacle: others are tax rates, access to finance, informal competitors and political instability. Page | 105 105 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!99^.!&N=!K>AKA><@A;!AE!E@>LG!AEE=>@;Q!<>?@;@;Q!@;! /@QH>=!99_.!&N=!;?=!AE!I=L?;I!?;I!GHKKB[! &H>D=[!@G!>=B?<@J=B[!BAV L@GL?AGG!G=FG!N@QNB[!N=<=>AQ=;=AHG Firms offering formal training for their ISKUR vacancy growth vs. placement growth (2014- employees (2016) 2016, %) Customer Service Officer / Assistant Accounting Professionalist 100 Office Staff Nurse Growth in job Growth in job 90 Data Entry Control Operator vacancies less vacancies 80 Quality Controller-Textile than growth in greater than Assembler, Metal Products placements i.e. growth in 70 Secretary excess supply placements i.e. Call Centre Customer Representative excess demand 60 Driver, Coach 50 Mechanic, Sewing Customer Representative 40 Cnc Turning Machine Operator 30 Sales Representative Security Officer 20 Tourism And Hotel Business Staff Sales Consultant 10 Electrician 0 Marketers Store Attendant Brazil Poland Indonesia China Malaysia Argentina Mexico Turkey Front Accountants Automotive Assembly Worker -0,55 -0,35 -0,15 0,05 0,25 0,45 ! ! Source: Global Innovation Index.! Sources: ISKUR, WB Staff estimates.! 22.! Most of the jobs listed in the ISKUR portal are low- and medium-skill-demanding jobs; like in many countries, higher-skill-demanding jobs are usually more commonly available in private job portals. The difference between the growth in job vacancies and the growth in job placements reflects a mismatch between vacancies and job seekers with the relevant skills (Figure 116). A positive difference means that the growth in job vacancies has been greater than the growth in job placements – so demand for workers exceeded the supply with the relevant skills. This was particularly the case in areas such as automotive assembly workers, front accountants, electricians, but also somewhat surprisingly store attendants and sales representatives. This may therefore not only reflect skills mismatch but also the desirability of some jobs. A negative difference – where the growth in job vacancies was below the growth in placements – suggests the supply of workers exceeded demand. This was more prevalent among customer services, accounting, office staff, and nurses. 23.! With technological developments and their impacts on the labor market, the main actor ISKUR focuses its attention on rendering policies especially for the employment of students and white-collar personnel by diversifying its services. Creating modern mobile applications and web interfaces, establishing internship matching systems, giving introductory information to the students about Agency’s services and labor market under ISKUR Kampüste link and by this way making closer contacts with the students. These are some of the measures taken on ensuring harmonization of supply and demand sides of the labor in the last period. 24.! In sum, though there is a steady increase in the demand for skills needed to deepen technological capacity and accelerate productivity, progress is relatively slow. The relative importance of supply side factors (i.e. workers not having appropriate skills) and demand side factors (i.e. employers not seeking higher skills) is difficult to assess. The public sector through ISKUR is working on improving the matching between the demand and the supply of skills, by providing better services through its web interface, and by promoting career days and job market fairs. Further, the Vocational Qualifications Authority assessed and certified the skills of almost 600,000 persons as of April 2019. However, the role of the private sector in Page | 106 106 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY fostering the demand for more skills is still limited. This is evident in the relatively low levels of firm investment in training of employees. This is also consistent with the relatively low levels of FDI, which would otherwise help accelerate the demand, and likely supply, for more skilled workers. If demand for more and better skills does not accelerate, the incentives to acquire more and better skills on the supply side will decrease. bXc! 3?Q=!Q>AV=L@?!GHQQ=G!LA>=!?;I!X=<<=>!GD@BBG 25.! Wage developments across skill levels suggest that the demand for low and mid-level skills relative to high level skills has remained strong. High wage growth and premia for specific types of skills can be reflective of higher demand pressures for those skills. Between 2009 and 2016, real wages for low and middle-skills occupations has grown more quickly than those for high-skill occupations (Figure 117). This may not be so unusual given the base effect of significantly higher real wages at the upper end of the skills spectrum. Nevertheless, the large difference in real wage growth rates nevertheless point to strong relative demand for low skill employees. 26.! Part of the sharp increase in real wages for low-skill workers could be related to minimum wage developments. The increasing minimum wage and salaries in the public sector between 2014 and 2016 might have affected the positive gains in real wages and wage premiums particularly for middle and low skill demanding occupations. Public subsidies for minimum wages in recent years might have further reinforced distortions on the demand side. /@QH>=!99U.!'=?B!V?Q=!Q>AV!BAV!GD@BB!VA>D=>G!;=?>B[!IAHXB=!D=>G 3,0 100% 2,7 88% 90% Wage growth from 2009 to 2016 2,5 80% Ln of real hourly wage 71% 70% 2,0 1,7 60% 1,9 1,5 50% 40% 40% 1,0 30% 20% 0,5 10% 0,0 0% High skilled occupations Middle-skilled occupations Low-skilled occupations (intensive in nonroutine (intensive in routine (intensive in nonroutine cognitive and interpersonal cognitive and manual skills) manual skills) skills) 2016 Growth rate, 2009-2016 Source: An adaptation for Turkey of Frey and Osborne 2013. Note: Arrows show the share in low, medium, and high probability categories in the US, 2010. 27.! These trends are confirmed by wage developments across workers’ educational attainments. Considering unconditional averages, secondary graduates see a higher (and increasing over time) return to their educational investment than others, who face low (and flat) returns (Figure 118). On the other hand, Page | 107 107 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey workers in the informal sector derive minimal or no benefit from achieving a higher education (Figure 119). /@QH>=!99O.!P=FA;I?>[!Q>?IH?<=Gd!=?>;@;QG!@;F>=?G@;Q! /@QH>=!99:.!3A>D=>G!@;!L?B!G=F!I=>@J=!B@<!<@L=!@;!L?B!G=F X=;=E@AL!?FN@=J@;Q!N@QN=>!=IHF?<@A; Average wage by educational attainment Average wage by educational attainment (formal jobs, 2005-2016) (formal jobs, 2005-2016) 1200 1200 1000 1000 800 800 600 600 400 400 200 200 0 0 05 06 07 08 09 10 11 12 13 14 15 16 05 06 07 08 09 10 11 12 13 14 15 16 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Below high school, formal High school, formal Below high school, informal High school, informal Higher education, formal Higher education, informal ! ! Source: LFS, WB Staff estimates.! ! 28.! To ascertain the effects of education on wages, it is however important to control for other characteristics. Even though education is important determinant of wage, workers with the same education level might not been paid the same wage level. There are other factors that affect wage such as work experience, labor market status, gender, geographic area of residence etc. Wage regressions can be used to estimate the effect of education on earnings controlling for other individual characteristics. 29.! Few studies in Turkey measure returns to education controlling for these other factors. One study finds that an extra year of schooling increases wages by 17 percent in Turkey, which is much higher than average OLS estimates typically found in advanced OECD economies (Aydemir and Kirdar 2013). Another study on wage inequality in Turkey (focusing however on a period pre-2011) finds that returns to education are not constant, but increase over different levels of education, and that the highest returns are achieved with university-level education (Tansel and Bircan 2010). 30.! Building on earlier studies, real hourly wage is estimated below as a function of education, work experience, labor market status and other characteristics: =>?$@ABCDE F # GH 3 $ GI J KLMNAOPQRD 3 GS J KTUCVPCRNCDE 3 $ GW J XYZOAOMZDE 3 T [ DE \ 3 MDE Regional CPIs were used for wages at constant prices. Education includes: primary and less than primary, secondary, high school (inc. vocational high school) and university. Labor market includes interactions for: (i) permanent, paid and casual and formal workers, (ii) temporary, paid and casual and formal workers, (iii) paid and casual and formal workers. Remaining control variables (x) include dummies for sector of economic Page | 108 108 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY activity and geographic region of residence. The sample includes permanent wage and casual workers aged between 25 and 64. 75 31.! The results show that premiums for university-educated workers increased slowly from 2009 to 2012 and then fell (albeit slightly) from 2012 to 2016 compared to primary educated workers (Figure 120). It should be noted that the decline is very sharp in 2016. A similar trend is observed for high-school educated workers. On the other hand, wage premium remains roughly at the same level for secondary school educated workers compared to primary educated workers. There might be several explanations for these trends. /@QH>=!978.!3?Q=!K>=L@HL!EA>!H;@J=>G@<[T=IHF?<=I!VA>D=>G!N?G!E?BB=;!LAG=F=;=I!@L?>[!=IHF?<=I!VA>D=>G Wage premium and real minimum wage growth over time 1,1 12 0,9 Minimum wage growht Wage premium 0,7 8 0,5 0,3 4 0,1 -0,1 0 2009 2010 2011 2012 2013 2014 2015 2016 Secondary High school University Minimum wage growth (right axis, 2009=1) Sources: LFS 2009-2016, WB Staff estimates. Note: Gross real daily minimum wage was used. The figure presents the evolution of the wage premium by education level compared to the primary school, as measured by the coefficient on the education variable in the Mincer-style earnings equations. 32.! One reason might be that the increase in the supply of more-educated workers might be running ahead of increases in demand. As educational attainment has been steadily increasing in Turkey, increases in the relative supply of more educated workers may put pressure on the earnings premiums to fall. This could be accompanied both by increasing growth in demand for skilled labor or by declining growth in demand for skilled labor. In the former case, the supply is expected to win out over increases in demand. The latter one could be related to a slowdown in the process of skill-biased technological change or to a shift in the structure of production away from high-skill-intensive sectors toward low-skill-intensive sectors (World Bank, 2012). 75Several specifications of the model (including Heckman correction) by using different Turkish micro data (Survey of Income and Living Conditions and Household Budget Survey) were tested. Results are robust across different models. Page | 109 109 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 33.! Other plausible explanations might be the increases in minimum wage, as noted above. The increase in minimum wages leads to a rise in relative wages of low-skilled workers. The minimum wage in Turkey increased very sharply after 2014 (Figure 120). Several studies find that minimum wages compress the distribution of wages among workers. Because minimum wage has a bigger impact on the wages in the bottom part of the distribution than on those higher up (World Bank, 2012). This is in line with the findings on real wage growth, which show higher real wage increases for medium and low-skill-intensive occupations than high-skill-intensive ones. 34.! In addition, Turkey has become one of the world’s largest refugee-hosting countries. The large influx of Syrian refugees, particularly after 2011, has the potential to affect employment and wages in Turkish labor market. Findings from earlier studies (Del Carpio and Wagner, 201576) are consistent with the above trends. They find that the refugees inflow was associated with a large-scale displacement of domestic workers in the informal sector. Those displaced are men that did not complete high school education. This means that low-educated men in Turkey moved from the formal to the informal sector, following the increasing absorption of Syrian refugees in the Turkish labor market. However, more research is needed to assess whether these findings indeed reflect a causal relationship. 35.! The analysis of wage developments and premia therefore provide some additional evidence on the relative importance of demand side factors deterring deeper skills. This is reflected in the high wage growth for low skill workers relative to high skill workers, and the decline (albeit most) in wage premia for university educated workers relative to primary educated workers. Whilst the impact of minimum wages certainly plays a role in driving up earnings of those at the lower end of the wage spectrum, these developments together with the findings from the earlier sections suggest that a demand side boost is needed to deepen the skills of the Turkish labor force. %A!G@Q;@E@F?;=B?<@A;GN@K!X=L!K>AIHF<@J@<[! 36.! This final section looks at the interaction between firm-level skills and productivity. As noted at the start, skills are essential determinants of firms’ ability to innovate and absorb new technology. To assess whether the skill mix across firms has or has not impacted on productivity growth, firms are first grouped in three categories based on the prevalence of occupation-related skills, namely routine manual (low), non- routine manual (medium) and non-routine cognitive (high).77 In sum: (i)! Low skilled firms are those whose share of employment for routine manual workers exceeds the national average; 76 Del Carpio, X., and Wagner, M. C. (2015). The impact of Syrians refugees on the Turkish labor market. Policy Research working paper; no. WPS 7402. Washington, D.C. World Bank Group. 77 This section uses EIS data for categorizing skills by firms and assessing the impact of this on productivity. Given the absence of information on workers’ educational attainment in the EIS data, skills of the workforce are proxied by the information available on occupations ISCO-codes. Occupations are classified therefore into three broad categories: (i)! Occupations intensive in non-routine cognitive tasks and interpersonal skills, such as: professionals (doctors, lawyers, architects, dentists, journalists), academics, managers, etc.; (ii)!occupations intensive in routine tasks (both manual and cognitive), such as: clerks, secretaries, administrative assistants, book-keepers, mechanic workers, call-center attendants, etc. (iii)!Occupations intensive in non-routine manual tasks, e.g. typically involving repairing, attendance, care, or some form of interactions with clients, such as: plumbers, electricians, care-givers, caretakers, security personnel, sales workers, cleaners, waiters, bartenders, etc. Page | 110 110 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY (ii)! Medium-skilled firms are those whose share of employment for non-routine manual workers exceeds the national average; (iii)! High skilled firms are those whose share of employment for non-routine cognitive workers exceeds the national average. 37.! Firms characterized mostly by non-routine manual work (medium-skill) absorb nearly half of all employment (Figure 121). Low skill firms (mostly routine manual) absorb around 35 percent of employment, whereas high skill firms (mostly non-routine cognitive) account for around 16 percent of employment. /@QH>=!979.!/@>LG!FN?>?F<=>@\=I!X[!BAV!GD@BB!B?XA>!L?D=!HK!?!B?>Q=!GN?>=!AE!=LKBA[L=;<]! VN@FN!N?G!Q>AV;!G@;F=!788_ Employment % composition (by type of firm based on skills-intensity in 2016) 50 45 40 35 30 25 20 15 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Non-routine manual Routine Non-routine cognitive Sources: EIS, WB Staff estimates. Note: Includes only firms registered over the full period (2006-2015). 38.! Firms that today have largest share workers in non-routine manual occupations (medium skills) compared to the national average, have increased slightly their employment share over time (Figure 121). On the other hand, the share of employment in firms that today absorb higher than average workers in non-routine cognitive occupations (high skill), has slightly declined. There is no indication of a “hollowing out” of routine occupations over time, as observed for instance in the US (Acemoglu and Autor, 2010). This suggests that the Turkish economy is likely to maintain its demand prevalently for occupations intensive in non-routine manual and routine type of skills in the coming years. Employment in key sectors such as manufacturing and wholesale and retail trade, remain prevalently intense in routine and non-routine manual occupations, respectively (Figures 122, 123). 39.! Interestingly, there is no significant correlation between skill incidence and firm productivity. One might expect labor productivity to be higher in firms that have higher incidence of workers engaged in non-routine cognitive (high skill) work. However, there is little evidence of a clear association (Figure 124). In addition, over the last 10 years, TFP in firms which today have higher incidence of employment in non-routine cognitive occupations (high skill), has dropped by 12 percent compared to the 2006 levels (Figure 125). Surprisingly, the productivity decline is less pronounced for firms with higher intensity in routine occupations and non-routine manual occupations. Page | 111 111 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!977.!)LKBA[L=;@;Q! /@QH>=!97S.!)LKBA[L=;==IAL@;?;AH<@;=!L?;H?B!bBAV!GD@BBc ;A;T>AH<@;=!L?;H?B!bL=I@HL!GD@BBc Manufacturing firms' skill intensity and their Wholesale and retail firms' skill intensity and share of employment (%) their share of employment (%) 65 70 55 60 50 45 40 35 30 25 20 15 10 06 07 08 09 10 11 12 13 14 15 20 20 20 20 20 20 20 20 20 20 06 07 08 09 10 11 12 13 14 15 20 20 20 20 20 20 20 20 20 20 Non-routine manual Routine Non-routine cognitive Non-routine manual Routine Non-routine cognitive ! ! Source: EIS, WB Staff estimates.! ! /@QH>=!97R.!%A!FB=?>!?GGAF@?<@A;!X=AIHF<@J@<[! /@QH>=!97^.!/@>LG!V@!@;F@I=;F=!AE!=LKBA[L=;D=>G @;!;A;T>AH<@;=!FAQ;@<@J=!E?F=!GN?>K=GAK Productivity levels and growth by incidence of TFP , by firm type based on skills-intensity in 2016 workers in non-routine tasks 70 80000 5 Productivity growth % 2012-2016 60 70000 60000 Productivity in 2016 50 4,8 50000 40 40000 30 4,6 30000 20 20000 10 10000 4,4 0 0 ] [1 10% 0% ) 0% ) 0% ) 0% ) 0% ) 0% ) 0% ) -1 ) % [2 20% [3 30% [4 40% [5 50% [6 60% [7 70% [8 80% 0% % 4,2 00 [9 -90 an - - - - - - - 0% th 06 07 08 09 10 11 12 13 14 15 ss 20 20 20 20 20 20 20 20 20 20 Le % of employment in non-routine Cognitive tasks Non-routine manual Routine Productivity (VA per worker) in 2016 Productivity growth (%) 2012-2016 Non-routine cognitive ! ! Source: EIS, WB Staff estimates.! ! 40.! These counter-intuitive results are likely linked to the prevalence of small-micro firms in the economy. These firms absorb the bulk of employment and suffer from declining productivity (Figure 126). Larger firms exhibit constant productivity over the last 10 years across all skill incidences (Figure 127). In manufacturing–while firms with high-skill employees register the highest TFP level during the whole period– for all 3 types of firms, TFP trends have been flat over the past ten years (Figure 128). On the contrary, in wholesale and retail trade, all type of firms experienced a decline in TFP (Figure 129). Page | 112 112 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY /@QH>=!97_.!PL?BB!E@>LG!=ZN@X@=!97U.!Y?>Q=!E@>LG!=ZN@X@!K>AIHF<@J@<[! K>AIHF<@J@<[!?F>AGG!?BB!GD@BB!@;<=;G@<@=G K?><@FHB?>B[! TFP trends of small firms (20-49) across skill TFP trends of large firms (250+) across skill intensity intensity 6 6 5,5 5,5 5 5 4,5 4,5 4 4 3,5 3,5 06 07 08 09 10 11 12 13 14 15 06 07 08 09 10 11 12 13 14 15 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Non-routine manual Routine Non-routine cognitive Non-routine manual Routine Non-routine cognitive ! ! Source: EIS, WB Staff estimates.! ! /@QH>=!97O.!*?;HE?F@;Q!E@>LGd!K>AIHF<@J@<[!>=B?<@J=B[! /@QH>=!97:.!3NAB=G?B=M>=LGd!K>AIHF<@J@<[!A;! EB?AGG!?BB!GD@BB!@;<=;G@<@=G I=FB@;@;Q!<>=;I!?F>AGG!?BB!GD@BB!@;<=;G@<@=G TFP trends of manufacturing firms across skill TFP trends of wholesale/retail firms across intensity skill intensity 6 6 5,5 5,5 5 5 4,5 4,5 4 4 3,5 3,5 06 07 08 09 10 11 12 13 14 15 06 07 08 09 10 11 12 13 14 15 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Non-routine manual Routine Non-routine manual Routine Non-routine cognitive Non-routine cognitive ! ! Source: EIS, WB Staff estimates.! ! 41.! Small firms are not the ones that will drive innovation and technological changes, or train their employees; they are therefore unlikely to contribute significantly to deepening skills. The impact of firm characteristics on productivity and innovation are discussed further in chapters 2 and 4 respectively. In other words, small firms will not drive demand for higher skills. This means that even as the supply of skilled workers increases in the future, the prevalence of small-micro firms limits the labor market’s capacity to absorb fully the next cohort of skilled graduates. Page | 113 113 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey ! "A;FBHG@A;G!?;I!KAB@F[!@LKB@F?<@A;G! 42.! The main messages that emerge from the analysis on human capital and productivity in Turkey are as follows: (i)! Turkey’s demographic dividend is less constrained by education than skills: A relatively young population can sustain growth in labor supply for years to come. But education attainment is outpacing the labor market’s ability to absorb more educated workers. Despite higher educational attainments, workers need to upgrade skills. (ii)! Demand side factors relative to supply side factors deter deepening of skills: Despite a steady increase in the demand for skills needed to accelerate productivity, progress is relatively slow. Wage growth and premia suggest slow demand for more and better skills. Plus, there is growing unemployment and NEET rates among the more educated. (iii)! Turkey needs a boost in the demand for skills: Without this, rising labor supply will not be absorbed in good quality jobs. This means accelerating FDI levels, which can increase the demand for more qualified workers. (iv)! No significant relationship between skill composition and firm productivity: These counter-intuitive results are likely linked to the prevalence of small-micro firms in the economy. These firms absorb the bulk of employment and suffer from declining productivity. Larger firms exhibit constant productivity over the last 10 years across all skill incidences. Issues Policy options There is no significant correlation between Options that could strengthen the relationship between skill incidence and firm productivity. One skills of the workforce and workers’ and firms’ might expect labor productivity to be higher in productivity could be: firms that have higher incidence of workers engaged in non-routine cognitive (high skill) •! Investing in improving the foundational skills work. (numerical, literacy and problem solving) of future cohorts, starting from early years. Over the last 10 years, TFP in firms which today have higher incidence of employment in •! Revitalizing vocational training institutions, which are non-routine cognitive occupations (high skill), currently under-utilized, and lacking resources and has dropped by 12 percent compared to the investments; 2006 levels. •! Monitoring regularly the demand for skills (by sector and geographic area) to inform policy makers about the demands of the market, through a data-driven information system. Issues Policy options The prevalence of small-micro firms in the Same options as in chapter 2: economy might limit the capacity of firms to innovate, adopt new technology, and especially •! Review and address constraints to growth of large provide their own training to workers. manufacturing firms. Strengthen their links to local Therefore, even if the supply of skilled and international value chains including SMEs. Page | 114 114 CHAPTER 5 /HUMAN CAPITAL AND PRODUCTIVITY workers increases in the future, given the •! Wind down supply subsidies that keep inefficient structure of the Turkish economy and the SMEs in business and prevent reallocation of prevalence of small-micro firms, the labor resources to more productive firms. market might have a limited capacity to absorb fully the next cohort of graduates. A persisting challenge to firms’ investment in •! Better labor market intermediation mechanisms, to workers’ human capital, is the increasing maximize the quality of the match between firms and mandatory cost of labor (e.g. the Minimum workers/skills; Wage hike registered in 2016). Greater labor costs and more costly compliance with labor •! Fiscal incentives for firms willing to invest in on the regulations, might discourage formal job job training policies, that can improve the school- and creation, especially among small firms, and university-to work transition of the youth, reduce create important trade-offs for firms, making “wait” unemployment, and therefore the incidence of financially constraint firms even less-likely to NEET. invest in training policies. ! Page | 115 115 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey CHAPTER 1 Chapter 6 Competition, regulation and productivity Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Competition, regulation and productivity 116 More competitive sectors though 116 manufacturing concentration remains high Impact of competition on mark-ups, labor 121 share of revenues and productivity Regulatory framework affecting competition 126 improves but still restrictive Conclusion and policy options 137 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY 1+W! "ALK=<@<@A;]!>=QHB?<@A;!?;I!K>AIHF<@J@<[! 1.! Effective and open competition puts pressure on firms to become more efficient and allows high-productivity firms to increase their market share at the expense of low- productivity firms (Olley and Pakes 1996; OECD, 2014). What is more, these latter, low-productivity firms will exit openly competitive markets, which also allows for the entry of more productive competitors (Aghion and Howitt 1990; Melitz 2003). In this way, competition may free up the efficient reallocation of market resources from low- to high- productivity firms as well as from low- to high-efficiency sectors. Industries where competition is more intense also tend to produce more innovation, further improving productivity (Blundell et al. 1999). 2.! Turkey has over time reformed its business regulatory framework, which has gradually helped increase market competition. The big wave of reforms in the early 2000s targeted a reduction of state intervention in the economy and establishment of independent regulators, and through these aimed to encourage more private investment and competition in key sectors of the economy.78 This chapter looks at: (i) the evolution and degree of competition across different sectors over the past ten years; (ii) the extent to which competition has impacted on the labor share of revenues, firms’ ability to charge markups and firm- level productivity; and (iii) ongoing regulatory constraints to greater competition.79 *A>=!FALK=<@<@J=!G=FG!@;Q!FA;F=;<>?<@A;!>=L?@;G!N@QN! 3.! Turkey performs well among comparator economies in terms of intensity of local competition, whilst there is room to improve performance on competition policy. On the intensity of local competition, Turkey ranks 6th among OECD economies in the 2017-2018 Global Competitiveness Report (Figure 130). This indicates that Turkish firms are subject to significant domestic market rivalry compared to most high and upper-middle income economies. The indicators on competition policy, on the other hand, present a different picture (Figure 131). Turkey ranks 25th out of 35 OECD economies on the effectiveness of antimonopoly policy, recording a score that is high for upper-middle income economies but low relative to high income ones. 4.! Intense local competition is borne out in Turkish firms’ experiences. This can be seen from the nature of competition as reported by firms—namely whether competition is against the informal sector, or against those whose main market is local as opposed to national or international (Table 16). To make sense of some underlying trends, sub-groupings in manufacturing and services are also presented, based on varying levels of technology or knowledge intensity (KI).80 Overall, one-third of firms compete with the informal sector, though this decreases with technological intensity. Similarly, only in high-tech manufacturing do firms report that they generally operate beyond local markets. In all other sub-groupings firms’ local market is predominantly their local city or municipality. 78 WBG, “Turkey’s Transitions: Integration, Inclusion, Institutions,” (December 2014). 79 A comprehensive competition assessment should evaluate (i) whether market regulation and sector policies are conducive to competition; (ii) whether antitrust rules and enforcement are effective; and (iii) whether there is competitive neutrality and non-distortive public aid support. The analysis presented in this chapter is limited to the first point and aims to explain how government regulation might be influencing the observed market outcomes. 80 Using an OECD/Eurostat division. Manufacturing is divided into high-tech, high-medium-tech, low-medium-tech, and low tech. Services are divided into knowledge-intensive (KI) market services, high-tech KIS, other KIS, less KI market services, and other less KIS. Construction is still presented separately. The end notes from this chapter include the full list of sectors in each sub-sector. Page | 116 116 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey /@QH>=!9S8.!YAF?B!FALK=<@<@A;!@;!&H>D=[!@G!N@QN!>=B?<@J=! /@QH>=!9S9.!#;!FALK=<@<@A;!KAB@F[!NAV=J=>]!&H>D=[!E?BBG! ?!FAH;<>@=G! X=N@;I!FALK?>?!FAH;<>@=G! Local competition vs. GDP per capita Effectiveness of anti-monopoly policy vs. GDP 6,5 per capita 6 Turkey Local competition index (GCI) 5 Anti monopoly policy (GCI) 5,5 Turkey 4 4,5 3 UMIC UMIC HIC HIC 3,5 2 4 4,5 5 3,8 4,3 4,8 Log of GDP pc (2017), PPP (2011 int $) Log of GDP pc (2017), PPP (2011 int $) Sources: Global Competitiveness Report. Notes: Index score 1-7, with 7 best. Dotted lines show a fitted, quadratic function. &?XB=!9_.!"ALK=<@<@A;!V@L?B!E@>LG]!?;I!@;!BAF?B!L?>D==!9S7.!*?;HE?F@;Q! /@QH>=!9SS.!P=>J@F=G!FA;F=;<>?<@A;!@G! /@QH>=!9SR.!"A;G<>HF<@A;!N?G!=J=;! FA;F=;<>?<@A;!I=FB@;@;Q!!?;I!N?G!>=L?@;=I!>=B?<@J=B[! BAV=>!FA;F=;<>?<@A;!XH=L?@;G!N@QN!AJ=>?BB! EB?!@AI! Manufacturing concentration Services concentration Construction concentration 75% 75% 75% 65% 65% 65% 55% 55% 55% 45% 45% 45% 35% 35% 35% 25% 25% 25% 15% 15% 15% 5% 5% 5% 06 08 10 12 14 16 06 08 10 12 14 16 06 08 10 12 14 16 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 CR20 sales CR20 sales CR4 Sales CR20 employment CR20 employment CR4 sales CR4 sales CR4 Employment CR4 employment CR4 employment CR20 Sales Source: EIS, WB Staff estimates. Notes: Figures are modeled after those shown in Autor et al. (2017) and show the concentration of the top four (CR4) and top 20 (CR20) firms in a four-digit Nace2 sector in a given year. These sectors are averaged together weighted by their relative contribution to value added in 2006, the start of the period. Nace2 74 “Other professional, scientific, and technical activities” omitted as it is an extreme outlier value. 81 Shares are averaged by four-digit Nace 2 activity sub-sector, weighted by their contribution to value added in 2006 82 They separate out different services sub-sectors and so these figures are less comparable. Their manufacturing data are available for the period 1981 to 2011. 83 Note that the scaling of graphs varies. To avoid interpretation based on these scales, gridlines are shown in 10% increments and Technical Appendix includes the actual figures shown. Page | 118 118 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 8.! Manufacturing remains the most concentrated, even though the trend is toward a more openly competitive market. Just thirty-one firms (across both manufacturing and services) account for over one fifth of value added on average over the period shown, while only accounting for 5 percent of total employment in 2016, indicating a much higher market concentration in value added than in the labor market.84 As reflected in the figures above, similar concentrations in sales are highest in manufacturing, which despite experiencing a notable decline over the period, remains highly concentrated. Such a persistent high degree of concentration further highlights potential risks from near-monopolistic market shares, underlining Turkey’s relative performance on antimonopoly policy shown above. 9.! More high-tech and skill intensive sectors tend to be more concentrated, with gradual increases in competition over time. This is to be expected as high tech and more skill intensive sectors more difficult to break into. They involve greater risk for example in terms of high start-up costs and long gestation periods. For this reason, public subsidies and incentives are often targeted to these sectors to overcome market failures that lower investment in high-tech and skill-intensive sectors (Chapter 4). 10.! In Turkey, there is some evidence of declining concentration in more sophisticated manufacturing and services sectors between 2006 and 2016. All manufacturing sub-sectors have exhibited decreasing concentration (i.e. more competition) between 2006-2016 (Table 17).85 High and medium-tech sub-sectors remain highly concentrated in value added, though far less so in terms of employment. In the services sector, knowledge intensive market services (see Technical Appendix) exhibited increased concentration in value added over time, though a decrease in employment concentration. All other sub-sectors showed declined concentration. On the employment side, only less knowledge intensive market services increased concentration, though as shown below, this is the predominant market sub-sector. &?XB=!9U.!"A;F=;<>?<@A;!?F>AGG!L?;HE?F@;Q!?;I!G=>J@F=G!X?G=I!A;!GD@BB!?;I!<=FN;ABAQ[! @;<=;G@<[! CR 20, Value Added CR 20, Employment 2006 2016 Diff. 2006 2016 Diff. High tech 79.6 72.2 -7.4 62.3 52.9 -9.4 Manuf. Medium-high tech 56.2 42.0 -14.3 23.2 18.4 -4.8 Medium-low tech 42.9 28.7 -14.2 10.4 6.9 -3.5 Low tech 22.8 15.3 -7.5 9.2 5.9 -3.3 KI Market services 40.1 50.5 10.4 20.9 19.3 -1.6 Services High-tech KIS 61.6 35.0 -26.6 28.4 23.0 -5.3 Less KI market services 12.6 10.2 -2.4 5.2 6.8 1.7 Construction 23.1 18.0 -5.1 9.9 6.1 -3.8 Sources: EIS, WB Staff estimates. Note: Other KIS and Other less KIS are omitted as the two groups together account for approximately 1% of value added or employment. 84 Based on calculations from EIS data. The decline in real value-added share of these firms over the period is from 29% in 2006 to 21% in 2016; and from 7% of employment to 5% over the same period. These firms are identified as having accounted individually for 0.5% of value added in a given year over the period. 85 The two residual “other” services sub-sectors are omitted hereafter as they collectively account for roughly 1% of value added and employment. Page | 119 119 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY 11.! There has been a general shift in the economy towards services; all manufacturing sub-sectors (except low-tech, which accounts for the biggest share in value added and employment) lost value-added and employment shares between 2006 and 2016 (Table 18). In services, less knowledge intensive market services—predominantly commercial trade—dominate in terms of value added and employment shares. Though this sub-sector shows little movement in shares, it did drop substantially in labor productivity, dampening overall labor productivity of services (chapters 1 and 2). In the professional-services-heavy sub- sector of knowledge market services, shares and labor productivity grew notably; construction accounted for the greatest change in shares toward services, but with declining labor productivity. Therefore, sectors that are absorbing more labor are becoming less concentrated, though labor productivity in those sectors is also declining (Figures 135, 136). &?XB=!9O.!*?;HE?F@;Q!?;I!G=>J@F=G!GHXTG=FGd!GN?>=G!@;!=!9S^.!YAV=>!=LKBA[L=;?<@A;!XH=!9S_.!+;F>=?G=I!=LKBA[L=;?<@A;!@;!BAV! BAV=>!B?XA>!K>AIHF<@J@<[!@;!L?;HE?F@;Q! GD@BB=I!G=>J@F=G!I>?QG!IAV;!AJ=>?BB!K>AIHF<@J@<[! Manufacturing employment concentration, Services employment concentration, share and share and labor productivity, 2006-2016 labor productivity, 2006-2016 80 30 High-tech Employment concentration (% of sector 70 High tech KIS KI Market 2006 25 services Employment concentration (% of sector 60 2016 Bubble size = 50 employment share in total 20 employment) 2006 40 Medium- 2016 15 emloyment) high tech Bubble size = 30 employment share in total 10 20 Construction Low tech Medium-low Less KI tech market 10 5 services 0 0 5,5 6 6,5 7 7,5 8 5,5 6 6,5 7 7,5 8 8,5 -10 Labor productivity (Log scale) Labor productivity (log scale) Sources: EIS, WB Staff estimates. +LK?FDHKG]!B?XA>!GN?>=!AE!>=J=;H=G!?;I!K>AIHF<@J@<[! b?c! 3@=IHF=I!FA;F=;<>?<@A;]!B?XA>!GN?>=!@G!>@G@;Q!VN@BGDHKG!?>=!I=FB@;@;Q! 13.! Additional measures of market power help illuminate competitive market forces. The concentration measures presented in the first part of this chapter, while illustrative, are also limited in cases of high product differentiation (Bresnahan 1989), which allows firms to command higher prices and revenues relative to their marginal costs. The most granular data presented in this chapter is for a given four-digit sector, within a year, and thus may mask even finer gradations in competition. What is more, concentration measures do not always move in tandem with increased competition, per se: reduced concentration does not necessarily and always imply reduced market power for a handful of firms. A useful framework that illustrates this point is one where opening of competition results in a ‘winner take most’ scenario, whereby a few firms command larger and larger market shares (Autor et al. 2017). This scenario may result as new technology and availability make consumers more sensitive to price differences, resulting in those consumers purchasing from only a small sub-set of firms (Akerman et al. 2017).86 14.! There is a global concern over the secular decline of the once-stable share of revenues going to labor (Karabarbounis and Neiman 2013; Dao et al. 2017; Autor et al. 2017). One issue in such a scenario, is that firms are able to command greater prices and revenues, with lower total labor compensation accruing to workers, as a proportion of total firm revenues—what is known as “labor share”. In fact, several analyses have noted that globally, labor share has been declining for the past several decades. This decline has been attributed to foreign competition, technological change, and predominant-firm market power. Market power enables firms to charge a higher markup (price above marginal cost of producing one unit), leading to higher revenue due to inelastic demand, but a proportionately smaller increase in labor share. 86 As elaborated in Autor et al. (2017). Page | 121 121 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY 15.! In Turkey, the labor share has been on a consistent, positive trend over the past decade, in stark contrast to trends in several high-income economies (Figure 137). This is consistent with lower concentration (and the strong contribution of labor incomes to Turkey’s impressive record of poverty reduction).87 In general, if market power is declining, firms cannot command prices too far above the marginal cost required to produce a good or service (i.e. a markup); this will also result in a larger share of the bottom line accruing to labor.88 Likewise, a rising labor share would occur due to either increased wage values or due to a shift toward labor-intensive production; with some evidence for the latter following in the later parts of this chapter. /@QH>=!9SU.!Y?XA>!GN?>=!@;!J?BH=!?II=I!@G!>@G@;Q!?F>AGG!?BB!G=FG! Labor Share of Value Added 50% 40% 30% 20% 2006 2008 2010 2012 2014 2016 Manufacturing Services Construction Source: EIS, WB Staff estimates. Notes: Labor share is the total wage bill divided by value-added, which is adjusted to remove measurement error form the first-stage of the Ackerberg-Caves-Frazer TFP estimate, as in De Loecker and Eekhout’s (2017) estimate of markups. Sector level averages weight four-digit Nace2 sectors by value added contribution in 2006. 16.! Service sectors across all skill levels dedicate a greater share of value added to labor compared to manufacturing (Figures 138 ,139). High labor share in high-tech knowledge intensive services is likely driven by wages, whereas in construction and low skill areas by employment levels. Similar wage and employment-driven trends may explain the relative movement of low and high-tech manufacturing. What is clear across both figures, nonetheless, is the general rise in labor share across all sub-sectors. 87 WBG, “Turkey’s Future Transitions: Towards Sustainable Poverty Reduction and Shared Prosperity,” (2017). ]^_ bc ]e 88 Formally, this can be expressed by defining the share of labor of firm i as `a_ $ #d$3 , where fgD $ is firm-level _ `a_ $ value added and hD is a firm-level markup. Wages, w, are set at the market level, for both variable labor and a fixed amount of labor, F, which is required for production. The elasticity of labor is i ^ . Note that for labor share to increase, the comparative statics of wages, fixed labor, and the elasticity of labor are upward; they are downward for value added and markups. Thus, increased labor intensity either through wages, the labor requirement F, or i ^ would be consistent with an increasing labor share. See Autor et al. (2017) for a framework. Page | 122 122 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 17.! Consistent with the above trends, firm-level markups have also declined relative to their 2006 values. Firms can charge markups when they have market power, but this can cause inefficiencies by embedding non-competitive firms and practices. Therefore, a reduction in markups suggests that firms’ market power may be loosening with greater competition. 18.! The median firm-level markups (relative to labor share) have fallen precipitously across all sub-sectors in services and manufacturing. In services, the sharpest drops have been at either extreme of the skills spectrum (Figure 140): (i) among high-tech knowledge intensive services, which is likely due a base effect given that markups are likely to be high to start with, and more intense competition (Figure 140); and (ii) among low knowledge intensive market services, which is likely due to stronger competition. In the manufacturing sector, the sharpest drop in markups has been in the high-tech sub-sectors. This is also likely driven in part by a base effect; but this is also the one sub-sector within manufacturing that experienced lower concentration and higher competition (Figure 141). /@QH>=!9SO.!Y?XA>!GN?>=!@;!N@QN!GD@BB!G=>J@F=G!?;I! /@QH>=!9S:.!Y?XA>!GN?>=!@;!N@QN!?;I!BAV!GD@BB! FA;G<>HF<@A;!I>@J=;!X[!V?Q=!?;I!fAXG!>=GK=F<@J=B[! L?;HE?F@;Q!I>@J=;!X[!V?Q=!?;I!fAXG!>=GK=F<@J=B[! Services labor share Manufacturing labor share 55% 55% 45% 45% 35% 35% 25% 25% 15% 15% 2006 2008 2010 2012 2014 2016 2006 2008 2010 2012 2014 2016 KI Market services High-tech KIS High tech Medium-high tech Less KI market services Construction Medium-low tech Low tech Sources: EIS, WB Staff estimates. Notes: Labor share is the total wage bill divided by value-added, which is adjusted to remove measurement error form the first-stage of the Ackerberg-Caves-Frazer TFP estimate, as in De Loecker and Eekhout’s (2017) estimate of markups. Sector level averages weight four-digit Nace2 sectors by value added contribution in 2006. Page | 123 123 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY /@QH>=!9R8.!PN?>K!I>AK!@;!L?>DHKG!?F>AGG!?BB!G=>J@F=G! /@QH>=!9R9.!PN?>K=GAK!@;!L?>DHKG!@;!G! GHXTG=FG!G=FG!V@@;Q! Services markups (2006 = 1) Manufacturing markups (2006 = 1) 1,1 1,1 1,0 1,0 0,9 0,9 0,8 0,8 0,7 0,7 0,6 0,6 0,5 0,5 2006 2008 2010 2012 2014 2016 2006 2008 2010 2012 2014 2016 KI Market services High-tech KIS High tech Medium-high tech Less KI market services Construction Medium-low tech Low tech Sources: EIS, WB Staff estimates. Notes: Values shown are the weighted-average of four-digit-level median markups based, following e.g. De Loecker and Eekhout (2017). Figures are from a Cobb-Douglas estimate and thus are not conditional on other firm-level inputs, like capital. Sector level averages weight four-digit Nace2 sectors by value added contribution in 2006. ! bXc! *A>=!FALK=<@<@A;!XHAIHF<@J=!G=FG!N?LK=>G!AJ=>?BB!K>AIHF<@J@<[!Q>AV>=B?<@A;!X=F=;?<@A;! ?;I!AJ=>?BB!K>AIHF<@J@<[! Concentration in Sales CR 4 CR 20 HHI TFP -1.1% *** -0.5% -1.2% ** Labor Productivity -1.4% *** -0.8% ** -1.3% *** Labor Share 0.1% *** 0.2% *** 0.1% * Concentration in Employment CR 4 CR 20 HHI TFP 0.1% 0.3% -0.5% Labor Productivity -0.1% 0.1% * 0.0% Labor Share 0.0% 0.2% -0.5% Note: Calculations based on EIS data and show regression coefficients multiplied by -1% to reflect a linear, marginal effect of a change in a concentration measure. Stacked five-year change regressions of change within four-digit Nace2 sector; n=2568 for all regressions. Observations are weighted by their contribution to total value added in 2006. Year fixed effects included. Robust standard errors clustered on the Nace2 four-digit sector level. Nace2 74 omitted as an extreme outlier. Bold results indicate significant results at * 10%, ** 5%, and *** 1%. 20.! This should not be taken to say that more competitive markets result in declining productivity. The above results should be given their appropriate caveats. First, as illustrated above, sectors, particularly manufacturing remain notably concentrated. Second, under a model of ‘winner take most’ competition, where a few highly productive firms capture increasing market share, such trends are consistent with the lack of highly productive market superstars and the presence of increasing numbers of less productive (and possibly more labor-intensive) firms. That is, the prevailing market conditions are such that there has been declining market concentration at the same time as a re-allocative shift toward less productive activities. This is consistent with findings in chapter 1 and 2, and is analyzed further below. Interestingly, the lower pane of the table shows little movement correlated with changes in market concentration measures for employment. This question of the allocative forces are taken up in the last part of the chapter. 21.! Some interesting trends are unmasked at sub-sector level. In manufacturing, the correlation between decreasing concentration and lower productivity is driven by medium-high-tech and low-tech manufacturing (Table 20). High-tech manufacturing, in fact, shows signs that decreased concentration is correlated with notably high gains in productivity. Page | 125 125 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY &?XB=!78.!"A>>=B?<@A;!X=F=;?<@A;!?;I!K>AIHF<@J@<[! ?F>AGG!GHXTG=FG CR4 C420 HHI High-tech TFP -0.9% 7.7% *** 1.2% *** Manufacturing Labor Productivity -1.9% 7.9% ** 0.9% *** n=72 Labor Share 0.2% *** 0.0% 0.0% Med.-high-tech TFP -1.8% *** -1.7% ** -1.6% *** Manufacturing Labor Productivity -2.2% *** -2.5% *** -2.5% *** n=348 Labor Share 0.1% 0.4% *** 0.1% Med.-low-tech TFP -1.6% *** -0.8% -0.4% Manufacturing Labor Productivity -0.7% -0.2% -0.3% n=354 Labor Share 0.1% 0.3% *** 0.0% Low-tech TFP -0.6% -1.0% -2.0% ** Manufacturing Labor Productivity -1.3% ** -1.4% *** -2.1% *** n=474 Labor Share 0.1% 0.2% * 0.3% *** KI Market Services TFP -5.1% ** -2.8% * -6.5% * n=132 Labor Productivity -5.5% *** -2.3% * -7.4% *** Labor Share 0.2% 0.0% 0.4% *** High-tech KIS TFP -4.5% -6.7% -2.0% n=132 Labor Productivity -4.5% * -5.4% -2.1% * Labor Share 0.4% *** 0.6% *** 0.0% Less KI Market TFP 0.6% 0.8% 1.0% Services Labor Productivity -0.3% -0.2% -0.5% n=822 Labor Share 0.0% 0.1% 0.1% Construction TFP -1.1% -0.8% -1.5% n=132 Labor Productivity -1.5% -1.3% -1.7% Labor Share 0.3% 0.4% 0.2% Note: calculations based on EIS data and show regression coefficients multiplied by -1% to reflect a linear, marginal effect of a change in a concentration measure. Stacked five-year change regressions of change within four-digit Nace2 sector. Observations are weighted by their contribution to total value added in 2006. Year fixed effects included. Robust standard errors clustered on the Nace2 four-digit sector level. Nace2 74 omitted as an extreme outlier. Bold results indicate significant results at * 10%, ** 5%, and *** 1%. '=QHB?[!E>?L=VA>D!?EE=F<@;Q!FALK=<@<@A;!@LK>AJ=G!XH=G<>@F<@J=! b?c! &H>D=[!F?=QHB?[!>=EA>L!XHG! ! 22.! Competition-enhancing regulations enable the market to select the most efficient firms, thereby creating incentives for firms to reduce costs and for new, more efficient firms to enter the market (Aghion and Schankerman, 2004). By contrast, the presence of dominant players in the market with too little competition can drive up firm markups and can decrease the level of innovation (Aghion et al. 2001; Impulliti and Licandro 2018). Effective antitrust policies should avoid the creation of dominant positions while safeguarding economies of scale and incentives for innovation, especially in resource and technology- intensive sectors. Regulations directly impact the entry and exit of firms. An efficient bankruptcy settlement law can ensure that investors are able to close a failing entrepreneurial experience and move on to new challenges. Similarly, a streamlined business entry legislation allows competitive firms to enter the market. Page | 126 126 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 23.! There is a vast amount of empirical evidence suggesting that pro-competitive regulations enhance productivity growth. Hsieh and Klenow (2009) show that eliminating idiosyncratic distortions to United States levels could increase manufacturing TFP by 30 to 50 percent in China and by 40 to 60 percent in India. Nguyen et al. (2016) show that such improvements could boost manufacturing TFP in Turkey by 24.5 percent. Aghion et al. (2008) point out that increased competition—resulting in a 10 percent reduction in markups—can lead to an increase in productivity growth in South Africa by between 2 and 2.5 percent yearly. Finally, Nicoletti and Scarpetta (2003) find that reforms intended to promote private governance and competition tend to also boost productivity across OECD economies. 24.! A comprehensive competition policy framework relies on two complementary pillars: enabling markets by removing sector specific competition constraints; and economy wide enforcement of competition policies (Kitzmuller and Licetti, 2012). Both pillars rely on an effective institutional set up that can foster and guarantee healthy market conduct. 25.! In the past two decades, Turkey made progress in regulatory reform. This includes modernization of institutions, improvement in the quality of regulations and simplification of administrative procedures (World Bank, 2010). The OECD’s Product Market Regulations Indicators (PMR) summarize information on economy-wide and industry-specific regulatory provisions that have the potential to restrict competition in areas where competition is viable (Figure 142). The analysis conducted in this chapter relies on 2013 data in the absence of more updated figures. PMR data are extremely informative and updating such data set represents a priority for countries that intend to monitor the effects of their regulatory framework on competition. Since the 1998 edition of the indicators Turkey’s score has steadily improved, implying an overall reduction in the regulatory barriers to competition. The scores of high performing countries that recently crossed the HIC bar improved more quickly, but Turkey caught up quickly with selected Trapped MICs, whose progress has been relatively flat (Figure 143). /@QH>=!9R7.!#)"-!0>AIHFD==!9RS.!&H>D=[!@G!F?=QHB?[!>=EA>LG! /@QH>=!9RR.!&NAHQN!>=L?@;G!X=N@;I!>=F=;=I!X[!0*'! Q>?IH?<=G!?;I!A!#)"-!=FA;AL@=G! OECD Product Market Regulation scores 1997, OECD PMR Index 2013 2003, 2008, 2013 0,5 1,0 1,0 1,5 1,5 PMR score (2013) PMR score 2,0 2,0 2,5 2,5 Turkey 3,0 3,0 3,5 3,5 3,7 4,2 4,7 1997 2002 2007 2012 Log of GDP pc (2017), PPP (2011 Int US$) High Performers Trapped MICs Turkey OECD Non-OECD Sources: OECD. Notes: PMR Score 1 is best. High Performers as in other parts of the report include Chile, Czech Republic, Korea and Poland. Trapped MICs in the above chart include Brazil, Mexico, South Africa. /@QH>=!9R^.!3@AQ>=GG!@;!LAG=F=;G!N?G!GBAV=I!IAV;! World Governance Indicators: Regulatory Quality score, (1998-2016) 1,5 0,5 -0,5 -1,5 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 High performers Trapped MICs Turkey Linear (High performers) Sources: World Governance Indicators 26.! Despite the improvements achieved in the past decade, regulatory restrictions to competition are still more prevalent in Turkey than comparator countries. Turkey's PMR score is almost twofold (1.9) more restrictive (higher score in the index) than the average OECD best-practice economy. Even when compared to non-OECD members Turkey performs poorly (Figure 144). The World Governance Index Regulatory Quality score also suggests some slippage in most recent years (Figure 145). Page | 128 128 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 27.! Disaggregated PMR data allow to identify specific dimensions where regulation could be most hampering to competition in Turkey. Barriers to entrepreneurship, such as burdensome market- entry procedures and restrictive permitting systems, are high. Moreover, restrictions in the services sector are common. These are typically regulations stipulate a fixed number of suppliers, grant exclusivity in the provision of services and impose restrictions to practice. Finally, state control on economic activities, such as price regulation mechanisms and rules that grant large state participation in markets, are prevalent. 28.! The next sub-sections look at two aspects of the regulatory framework that affect competition and productivity. The first relates to regulatory barriers for firm entry and exit, which are implicit constraints on competition. Barriers to firm entry also fall under the PMR barriers to entrepreneurship (Figure 143, 144), where Turkey performs lags comparator countries. The second, are regulatory restrictions on services, as discussed in chapters 3 and 6, which impact services and manufacturing sectors’ productivity. Turkey is also trailing comparator countries in the PMR barriers to the service sector. bXc! -=GK@<=!>=QHB?[!>=EA>L]!L!=;<>[!?;I!=Z@AIHF<@J@<[!@G!;=Q?<@J=! 29.! Regulatory barriers to firm entry and exit can dampen within sector productivity gains as more efficient entrants are prevented from displacing less efficient incumbents. This can be aggravated by supply side subsidies discussed in chapter 1 or poor targeting of business incentives, which can distort markets and enable less productive firms to survive. The World Bank Group’s Doing Business indicator on Starting a Business can be a proxy for regulatory barriers to firm entry.90 Turkey performs well on this dimension, in-line with OECD, high-income averages (Figure 146): in Turkey, starting a business takes 7 procedures (compared to an OECD average of 5), but these can be accomplished in under a week, which is 2 days shorter than the OECD average 8.5 days. /@QH>=!9R_.!&H>D=[!K=>EA>LG!V=BB!A;!XHG@;=GG!=;<>[! /@QH>=!9RU.!&NAHQN!;A[!V=BB!A;!>=QHB?<@A;G!EA>! >=QHB?<@A;G! I=?B@;Q!V@=!9RO.!&N=!K>@J?<=!G=F!@;!&H>D=[!@G!X>A?IB[!?II@;Q!E@>LG! Continuing and Entering Firms in Turkey (Thousands) 450 Construction 400 (continuing) 350 Construction (entering) 300 Manufacturing 250 (continuing) 200 Manufacturing (entering) 150 Services (continuing) 100 50 Services (entering) 0 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Sources: EIS, WB Staff estimates. 91 Based on the 2018 Doing Business Report. Page | 130 130 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 33.! At the same time, the rate of firm entry is decelerating, whilst the rate of firm exit is accelerating. Over the period 2007 to 2015 (the period for which exit and entry rates can be calculated in the EIS data), the net entry rates of firms (new entrants plus exiting firms over the stock of continuing firms) dropped sharply—a trend that continued even after an initial plunge between 2007 and 2008 (Figure 149). These trends point two developments: (i) over the period entry rates far outpace exit rates – only in recent years have former slightly decreased, including in 2015 when the net rate dropped to zero; (ii) the declining trend in net entry rates over more recent years has been due to a pick-up in the exit rate of firms. /@QH>=!9R:.!&N=!;=?<=!AE!E@>L!=;<>[!@G!I=FB@;@;Q! Net Entry Rate of Firms (2007-15) 23% Net Rates Manufacturing 18% Net Rates Services Net Rates Construction 13% 8% 3% -2% 2007 2008 2009 2010 2011 2012 2013 2014 2015 Sources: EIS, WB Staff estimates. Notes: Net entry rates defined as new entrants plus exiting firms over the stock of continuing firms. 34.! While net entry rates are generally decelerating, trends differ by sub-sectors. The average net entry rates for two periods (2008 to 2011 and 2012 to 2015)92 by sub-sectors show that net entry rates decelerated most in high-skill sub-sectors (Table 21). As noted above, these are also sectors that generally more difficult to break into. &?XB=!79.!%=[!>?<=G]!788OT789^ Avg. 08-11 Avg. 12-15 Diff. High tech 5.3% 2.3% -2.9% Medium-high tech 4.7% 2.8% -1.9% Manufacturing Medium-low tech 6.3% 4.3% -2.0% Low tech 5.5% 3.8% -1.8% KI Market services 9.4% 5.4% -4.0% High-tech KIS 12.9% 5.7% -7.2% Services Less KI market services 11.1% 4.9% -6.2% Construction 6.5% 3.1% -3.4% Sources: EIS, WB Staff estimates. Rates for 2007 are omitted as that year appears to be somewhat of an outlier. Including these rates maintains (and in 92 many cases increases the magnitude) of results. Page | 131 131 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY 35.! The next step is to understand how firm entry and exit have impacted on within sector productivity. If competitive forces allow more productive, new firms to enter the market as less productive firms exit, productivity will increase. If stagnant or less productive firms remain in operation, while new, dynamic firms are dissuaded from entering, productivity will flounder. Just the same, if more market share is gained by more efficient firms (or more efficient sectors), there will be a productivity gain. Researchers have been interested in breaking down the various potential sources of productivity growth, and several have adopted various decompositions to make sense of various trends. Generally, these decompositions separate out the effects of different types of firms, their measure of productivity, and their share of the market. 36.! The analysis shows that only manufacturing experienced net gains in Total Factor Productivity (TFP), and that this was due to TFP gains of incumbents rather than net entry dynamics.93 This is illustrated by decomposing the TFP (and labor productivity) effect(s) of both continuing firms (which includes both productivity growth within firms, as well as or market share shifts between firms) and the effect of net entry (Figure 150).94 Both services and construction experienced declining productivity in both measures. What is more, the effect of net entry is negative across all broad sectors and for both measures; this indicates that on average, less productive firms are entering the market than those leaving it. /@QH>=!9^8.!&N=!=EE=F[!A;!K>AIHF<@J@<[!N?G!X==;!;=Q?<@J=! Decomposition of TFP Growth Decomposition of Labor Productivity Growth 0,4 0,4 0,3 Continuing Firms 0,3 0,2 Net Entry 0,2 Change 0,1 0,1 0,0 0,0 -0,1 -0,1 -0,2 -0,2 -0,3 -0,3 -0,4 -0,4 -0,5 -0,5 Manufacturing Services Construction Manufacturing Services Construction Sources: EIS, WB Staff estimates. Notes: Figures are additive components of a log-difference of productivity and approximate the percentage change growth over the 2007-2016 period. 93 TFP is calculated using a value-added measure of output using material inputs as a proxy for underlying productivity according Levinsohn and Petrin (2003). The correction proposed by Ackerberg et al. (2015) is used to allow firm-level labor inputs to affect their subsequent productivity levels. 94 This section will rely on dynamic Olley-Pakes productivity decompositions with entry and exit of Melitz and Polanec (2015). Figures showing “continuing” effects combine both within and between effects; “Net entry” combines the effect of entry and exit. Page | 132 132 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 37.! The relative change of labor productivity can also be informative, recalling that labor productivity can grow (shrink) either due to increased (decreased) TFP or due to firms increasing (decreasing) more capital per worker.95 Each of the broad sectors showed an entry effect that was positive for TFP (Table 22) i.e. even though entry effects are productivity-enhancing, they’re generally offset by productive firms leaving the market. This would be consistent with expansions of new firms in low-value- added, labor-intensive industries, particularly if these effects represent a shift across sectors. Likewise, the exit effect for all three sectors was negative for TFP, particularly so for services and construction, indicating a loss from the market of more efficient, comparatively capital-intensive firms. 38.! Decomposing for manufacturing and services sub-sectors, labor productivity increased in only high- and low-tech manufacturing, with TFP increasing in the former but decreasing in the latter. In high-tech manufacturing, net entry (the added effect of entry plus exit) had a dampening effect on productivity growth, meaning that productivity gains came from shifts toward more productive continuing firms. In low-tech manufacturing, net entry effects were positive for labor productivity (LP), but were negative for TFP, largely due to the exit of more productive firms. &?XB=!77.!);<>[!?;I!=Z@!K>AIHF<@J@<[! Cont. Entry Exit Change LP 115.6% -16.7% -52.2% 46.7% High tech TFP 84.4% 6.7% -59.5% 31.6% LP -3.7% -18.1% 12.3% -9.5% Medium-high tech TFP -14.3% -5.2% 8.0% -11.5% Manufacturing LP -8.6% 32.9% -32.1% -7.8% Medium-low tech TFP -9.1% 3.2% -18.0% -23.8% LP 17.3% 16.8% -15.5% 18.6% Low tech TFP -10.6% 1.9% 4.9% -3.8% Cont. Entry Exit Change LP -52.3% 24.3% -57.1% -85.1% KI Market services TFP 36.1% -18.0% -6.8% 11.3% LP -94.0% 13.3% -85.1% -165.9% High-tech KIS TFP 7.1% -23.0% -32.2% -48.2% Services LP -12.5% 3.8% -16.8% -25.5% Less KI market services TFP -12.1% -13.4% -7.0% -32.5% LP -10.3% 60.7% -70.3% -19.9% Construction TFP -11.0% 47.2% -59.4% -23.2% Sources: EIS, WB Staff estimates. Notes: Figures are additive to the total change over the 2007-2016 period for each sub-sector. LP: Labor productivity. Cont.: continuing firms. 39.! In services, all productivity measures, except for TFP in KI market services, showed decreases. However, the net effect of firm entry in all services sub-sectors was negative, for both labor productivity and TFP. Thus, the TFP gains for KI market services were from shifts of market shares toward more productive, continuing firms. 95From a Cobb-Douglas production function, the first-order change in labor productivity (Y/L)’ = (A)’ + ! (K/L)’ , where (A)’ is the change in TFP and ! (K/L)’ is the change in capital per worker, scaled to the elasticity of capital, !. Page | 133 133 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY bFc! '=QHB?<@A;G!@;!J@F=!G=F!KAG=!AXGAIHF<@J@<[:_! 40.! Appropriately regulated and productive services are a major determinant of TFP growth. Services can be direct outputs or an input that is embodied in another service or manufacturing (Chapters 1 and 2). They are important drivers of Global Value Chains (GVCs) (Chapter 3). Firm-level studies (Arnold et al. (2015; 2011), Van der Marel et al. (2016)) show that regulatory reform in services has a positive impact on downstream manufacturing and services firms’ productivity. When service inputs are effectively supplied, this has a knock-on effect on industries that use services. In other words, services sector deregulation can lead to a more productive allocation of resources, thereby generating economy-wide productivity effects. Yet, to optimize benefits from deregulation of services, good institutions are needed to guide this process of services liberalization (Beverelli, et al., 2015; Van der Marel, 2016). 41.! A recent World Bank study finds that Turkey has much potential to increase the impact of services on manufacturing productivity (Haven and Van Der Marel, 2018). It finds that the bulk of services inputs in Turkish manufacturing exports come from (relatively less productive) transport and distribution; OECD firms, in contrast, rely on a much higher share of (relatively more productive) business services. It also finds, surprisingly, that manufacturing firms with service affiliates tend to be less productive.97 42.! Part of the reason for the above results is because most service activities within Turkish manufacturing are in traditional service sectors. Services can be categorized into six production process stages (Figure 151). Most service activities within Turkish manufacturing firms are in the post-manufacturing stage (e.g. transportation and distribution), which are also fast-growing sectors. Back-office services and establishment services are the next largest categories. Establishment services and post-manufacturing services are in large part comprised of more traditional services. Modern services include telecom, computer, R&D, intellectual property, banking, insurance; these tend to be more prevalent in pre-manufacturing, post-sales and back-office stages. /@QH>=!9^9.!0>AIHF<@A;!G=!G=>J@F=G!?>=!?II=I! Establishment Pre-manufacturing Manufacturing Post - Post-sales Back-office manufacturing • License, • R&D, special • Engineering; • Transport, • Repair, • Accounting, construction, design services, cleaning; distribution, maintenance, insurance, real estate tech services technical testing; whole sale, finance and finance, legal, env. mgt. marketing renting management Sources: Haven and Van Der Marel (2018) based on high-level categorization from Low and Pasadilla (2015). 43.! A key difference between traditional and modern services is that modern services can be delivered at arm’s length (e.g. using the internet), without simultaneous production and consumption in the presence of producers and consumers. As a result, modern services often are associated with higher levels of value added and hence productivity (see Ghani et al. 2012; Ghani 2010). Moreover, traditional services are often in sectors that many countries have found difficult to reform, e.g. transportation services which still face many restrictions globally. 96This section is based entirely on Haven, T, and E Van Der Marel, “Servicification of manufacturing and boosting productivity through services sector reform in Turkey,” The World Bank (2018). 97 For instance, non-exporter manufacturing firms with a services affiliate are 18 percent less productive than firms without a services affiliate. Exporters with a services affiliate are 9 percent less productive than those without one. Page | 134 134 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey 44.! Consistent with the above, the type of services that Turkish manufacturing firms perform is correlated with their productivity. Overall, manufacturing firms engaged in post-manufacturing and establishment services are the least productive in Turkey, whereas those engaged in pre-manufacturing and back-office service are relatively more productive. Regression analysis shows that firms with establishment services are around 20 percent less productive than non-servicified firms. Similarly, manufacturing firms that also produce post-manufacturing services are on average 13.8 percent and 18.7 percent less productive in the case of exporters and non-exporters respectively compared to their non-servicified peers. 45.! Turkey has scope to further reduce services regulatory restrictions. The discussion below covers three regulatory areas in particular: i) FDI restriction in services; ii) domestic regulatory barriers in services that affect both Turkish and foreign firms; and iii) discriminatory services trade barriers that prevent foreign service providers from entering and operating in the Turkish economy. 46.! First on FDI in services, Turkey generally has an open policy regime (Figure 69), though there are restrictions within the establishment and post-manufacturing stages of production. High FDI restrictions still exist in services sectors such as maritime and air transport, which are part of the post- manufacturing stage. Although FDI for construction services (establishment stage) is not restricted, the real estate sector has high FDI barriers. Similarly, while there are few restrictions for FDI in overall business (back-office) services, accounting services face high barriers. 47.! FDI restrictions in the post-manufacturing and establishment stages can have a negative impact on the productivity of manufacturing firms. Evidence of positive spillovers from FDI is discussed in chapter 3. FDI barriers in Turkey not only limit the entry of more productive foreign service providers, but also productivity-enhancing spillovers to local suppliers of those service providers. This could have a negative productivity impact on manufacturing firms, particularly those that rely more heavily on efficient establishment and post-manufacturing services. Establishment services are more prevalent in Automotives, Plastics and Textiles, and post-manufacturing services stand out for Agribusiness, Plastics, Textiles, and Chemicals. 48.! Secondly, domestic regulatory barriers affect both domestic and foreign service suppliers. Based on the 2013 OECD PMR results, higher levels of domestic regulatory restrictions in Turkey are mainly in post-manufacturing and back-office services. The largest restrictions are in rail and road transport (post- manufacturing stage), as well as accounting and legal services (back-office stage). These sectors may be indicative of the broader restrictiveness of services within each production stage, as many more services fall under each stage. 49.! Turkey shows a mix of entry and conduct barriers. Rail transport services are completely restricted, affecting both firm entry and conduct. Air transport services mainly have restrictions related to public ownership, which affects the entry of foreign firms, but some conduct barriers also exist. The road transport sector has high entry barriers, while operations regulations are significantly lower. Gas and electricity services (manufacturing stage) have high conduct regulations, such as vertical integration policies, public ownership regulations, and uncompetitive market structures, whereas they are relatively open to firm entry. Architectural and engineering services (manufacturing stage) have high entry barriers while accounting and legal services (back-office services) have both restrictive entry and conduct regulations. Page | 135 135 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY 50.! Thirdly on services trade barriers, these are more prevalent in Turkey in the post- manufacturing and back-office stages. Regulatory policies can have an explicit discriminatory nature preventing entry from foreign services suppliers.98 These can be distinct from restrictions on foreign firms investing in a host country. Trade-related restrictions in services in the post-manufacturing stage are particularly high for air transportation, followed by logistics cargo handling (OECD STRI).99 In the post-sales stage, courier services appear to be the most restrictive. Back-office restrictions are led by accounting services (where Turkey is still completely closed), followed by legal services. 51.! The most restrictive components of trade barriers in air transport and cargo-handling services are related to foreign entry and competition. In air transport, there are still equity restrictions. Turkey also has some restrictions on mergers and acquisitions by foreign firms and nationality restrictions on boards of directors. Regarding competition policies, Turkey has regulations related to slots (i.e. no auction takes place to give away slots, all carriers are allowed to retain slots from one season to another, and no administrative slot system is in place) and minimum capital requirements, all of which affect foreign service suppliers. In logistics cargo handling services, there are barriers to competition, such as price and fee regulations, contracts awarded without competitive bidding processes, and minimal capital requirements. 52.! Services trade barriers tend to affect firm operations more than entry. Although entry barriers are still present in services markets, more restrictions affect firm operations. On the other hand, no clear pattern arises when looking at the differences between discriminatory and non-discriminatory barriers. In most services sectors (apart from air transport and insurance or logistics freight and maritime services), restrictions affect both foreign and domestic firms. 53.! Restrictions on firm operations can dampen growth. Cross-country differences in the post-entry performance of all firms tend to be more marked than differences in entry and exit patterns, which suggests an important role for services conduct regulations (Van der Marel et al. 2016). Furthermore, the OECD (2016) finds that post-entry growth patterns between countries explain firms’ ability to achieve sufficient scale to reach global markets, and the share of small firms that are relatively old negatively affects aggregate productivity and employment growth. Put differently, in countries with high restrictions on firm operations, there are not sufficient selection mechanisms or up-or-out dynamics. Regulations on the operations of the firm are in effect barriers to up-scaling after firm entry. 54.! In Turkey, services firms are relatively small, and there is little evidence of post-entry firm growth in both goods and services. The recent Turkey enterprise survey (World Bank, 2017) shows that services firms are overwhelmingly small on average, with a few stand-out large firms. This reflects the general patterns in Turkey that both goods and services firms tend to be smaller than in comparator countries. More importantly, the vast majority of firms in Turkey start with very few employees and show no significant pattern of growth. No matter the age group, about three quarters of firms in Turkey that start small remains small. Only a handful of Turkish firms see a substantial growth pattern. Moreover, medium-sized firms (20- 99 employees) do not show signs of scaling-up over time. Reducing conduct regulation could therefore help young firms to grow or exit (i.e. “up-or-out”). 98Although regulatory services barriers can be de jure non-discriminatory, some non-discriminatory regulatory barriers may still have an effect on the operating costs of the foreign entrant making them de facto discriminatory (Francois and Hoekman, 2010). This is because complying with specific domestic regulations might be costlier for foreign suppliers than for local ones who have better access to information and lower red tape costs (Crozet and Milet, 2016). 99 While there are more services sectors under each production stage, only 22 services sectors are covered by the OECD’s STRI. Page | 136 136 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey "A;FBHG@A;!?;I!KAB@F[!AK<@A;G! 44.! The main messages that emerge from the empirical analysis on competition, regulation and productivity in Turkey are as follows: (i)! Reduced concentration across the board though manufacturing concentration remains high: More high-tech manufacturing and skill intensive services tend to be more concentrated as these sectors are more difficult to break into. But even here, the trend, particularly in terms of employment shares of large firms, has been on a downward path. (ii)! With reduced concentration, labor share is rising whilst markups are declining: Total labor compensation as a share of firm revenues has consistently risen over the past decade largely due to a shift towards labor-intensive production. Firm-level markups (i.e. price above marginal cost of producing one unit) has been declining across all sectors. (iii)! A shift towards less productive sectors hampers overall productivity growth: This is not to say that more competitive markets result in declining productivity. Declining concentration has been accompanied by reallocation towards less productive sectors, which is related to business regulations that have an impact on competition. (iv)! Despite regulatory reform, the impact of firm entry and exit on productivity is negative: Starting a business is relatively easy, though dealing with insolvency is difficult. Net entry remains positive, though rate of firm exit in recent years has outpaced firm entry. But net entry has had a negative impact on productivity (i.e. on average, less productive firms are entering the market than those leaving it). (v)! Regulations in the services sector pose obstacles to competition and productivity: Scope to reduce restrictions on FDI in services; domestic regulatory barriers in services that affect Turkish and foreign firms; discriminatory services trade barriers that prevent foreign service providers from entering and operating in Turkey. Issues Policy options Firm entry and exit Dynamism in high value-added, particularly While policy barriers to entry (such as the procedure capital-intensive industries is low. By and costs to opening a business) are not comparatively extension, many of these sectors remain highly high, other barriers to entry in these sectors persist, concentrated and closed to competition. including the predominance of a few superstar firms. Measures, such as anti-trust policies, further opening competition in these sectors are important. Moreover, aligning capital risk, via policies that, for instance, would increase creditor recovery rates would help support more activity in these sectors. Turkey fares poorly on the time and cost Implement new ‘concordat’ procedure, which enables required for firms to resolve insolvency. This authorities to set timelines for the procedure, and puts creates inefficiencies in the economy by a heavy focus on business continuation rather than its keeping “zombie” firms alive, and hampers liquidation through new financing, confirmation of reallocation of resources to more productive contracts and sale of essential assets in bankruptcy. firms. Page | 137 137 CHAPTER 6 /COMPETITION, REGULATION AND PRODUCTIVITY Issues Policy options Competition in the services sector Productivity gaps appear in areas where Remove FDI restrictions in: maritime and air- services are more restricted. Restrictions on transport (post-manufacturing services); real estate competition in services, particularly from (though not construction); overall business services, foreign firms may lead Turkish manufacturers and accounting services. to provide more services in-house than would Professional services should be opened to be optimal from a productivity perspective. competition. This includes reform to Turkey’s Minimum prices are in place for accounting, regulatory framework governing the liberal architecture, legal, and engineering services, professions (professions requiring special training in while notaries regulate the exact level of the liberal arts or sciences, such as notaries, lawyers, pricing. Limitations on advertising apply to engineers, or accountants) and the regulations some professions and access to several imposed by professional associations lessen professions is closed to foreign nationals. Fees competition by either restricting entry (such as for key professional services, such as legal and stipulating a fixed number of suppliers, exclusivity in notaries, are among the highest in the OECD the provision of services, and restrictions to practice) and the EU for comparable transactions or aiding members in coordinating prices (such as (World Bank, 2013). establishing minimum prices). Page | 138 138 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey );I!;A<=G.!"B?GG@E@F?<@A;!AE!G=FG!X[!F?K@F=! We rely on Entrepreneur Information System (EIS) which is an administrative dataset and which contains data on all registered firms in Turkey. EIS is compiled and administered by the Ministry of Industry and Technology (MOIT) and spans 2006-2016. Data are collected from various sources and matched. Table 23. &?XB=!7S.!#J=>J@=V!);<>=K>=;=H>!+;EA>L?<@A;!P[G<=L! Data Source Coverage Unit of Observation Variables Turkish Statistical 2006-2016 Plant Level Enterprise: ID Institute Plant: ID, Activity Code (Nace Rev.2 4 Digit), Location (City), Social Security ID, Employment Social Security 2006-2016 Enterprise Level Enterprise: ID, Activity Code (Nace Rev.2 4 Institution Digit), Location of the Administrative Center (City), Social Security ID, Year of Establishment Social Security 2006-2016 Worker Level Plant: ID Institution Worker: ID, Gender, Age, Wage, Days Worked, Occupation (2014-2016) Ministry of 2006-2016 Enterprise Level Enterprise: ID, Balance Sheet Data, Income Finance1 Statement Data Ministry of 2006-2016 Enterprise-Product Enterprise: ID, Exports (Quantity, Value, Customs and Level Product Code - HS 6 Digit), Imports (Quantity, Trade Value, Product Code – HS 6 digit) Turkish Patent 2009-2016 Enterprise Level Enterprise: ID, Number of Patents, Number of Institute Designs, Number of Brands KOSGEB2 and 2011-2016 Enterprise Level Enterprise: ID, Type and Amount of support TUBITAK3 received Ministry of 2006-2016 Enterprise Level Enterprise: Reporting ID, Partner ID, Value of Finance Goods/Services Sold/Purchased (above 5,000 TL), Time of Transaction Notes: 1 This information is only available for the specific types of firms which are obliged to provide the Ministry of Finance with their balance sheet and income statement data. Namely, firms that either make purchases for amounts above TL160000 or make sales for more than TL 220000. These firms constitute around 20-25% of all enterprises in Turkey but account for more than 90% of the total economic activity. 2 Small and Medium Enterprises Development Organization 3 The Scientific and Technological Research Council of Turkey ! "AJ=>?Q=! The analysis covers the period 2006-2016 and the manufacturing, construction and services sectors. Manufacturing sector includes 2-digit Nace Rev.2 sectors 10-32. We exclude 33 - Repair and installation of machinery and equipment. Page | 140 140 TECHNICAL APPENDIX &?XB=!7R.!*?;HE?F@;Q!%?F=!'=JW7!7I@Q@G! 10 Manufacture of food products 11 Manufacture of beverages 12 Manufacture of tobacco products 13 Manufacture of textiles 14 Manufacture of wearing apparel 15 Manufacture of leather and related products 16 Manufacture of wood and of products of wood and cork, except furniture; manufacture of articles of straw and plaiting materials 17 Manufacture of paper and paper products 18 Printing and reproduction of recorded media 19 Manufacture of coke and refined petroleum products 20 Manufacture of chemicals and chemical products 21 Manufacture of basic pharmaceutical products and pharmaceutical preparations 22 Manufacture of rubber and plastic products 23 Manufacture of other non-metallic mineral products 24 Manufacture of basic metals 25 Manufacture of fabricated metal products, except machinery and equipment 26 Manufacture of computer, electronic and optical products 27 Manufacture of electrical equipment 28 Manufacture of machinery and equipment n.e.c. 29 Manufacture of motor vehicles, trailers and semi-trailers 30 Manufacture of other transport equipment 31 Manufacture of furniture 32 Other manufacturing Construction sector includes 2-digit Nace Rev.2 sectors 41-43. &?XB=!7^.!"A;G<>HF<@A;!%?F=!'=JW7!7I@Q@G! 41 Construction of buildings 42 Civil engineering 43 Specialised construction activities Services sector, on the other hand, includes retail and wholesale (45-47), transportation and storage (49- 53), accommodation and food (55-56), ICT (58-63), real estate (68), scientific, occupational and technical activities (69-75), administrative and support services (77-82) and culture (90-93). We leave finance and insurance (64-66), public administration (84), education (85), health (86-88) and other services (94-99) either because there are too few firms or because these sectors are mostly state dominated. ! ! ! ! ! ! ! ! Page | 141 141 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey &?XB=!7_.!P=>J@F=!%?F=!'=JW7!7TI@Q@G!?;I!G=F!FB?GG@E@F?<@A;G! 45 Wholesale and retail trade and repair of motor vehicles and motorcycles Retail and 46 Wholesale trade, except of motor vehicles and motorcycles Wholesale 47 Retail trade, except of motor vehicles and motorcycles 49 Land transport and transport via pipelines 50 Water transport Transportation 51 Air transport and Storage 52 Warehousing and support activities for transportation 53 Postal and courier activities Accomodation 55 Accommodation and Food 56 Food and beverage service activities 58 Publishing activities Motion picture, video and television programme production, sound recording and music Information and 59 publishing activities Communication 60 Programming and broadcasting activities Technologies 61 Telecommunications 62 Computer programming, consultancy and related activities 63 Information service activities 64 Financial service activities, except insurance and pension funding Finance 65 Insurance, reinsurance and pension funding, except compulsory social security 66 Activities auxiliary to financial services and insurance activities Real Estate 68 Real estate activities 69 Legal and accounting activities 70 Activities of head offices; management consultancy activities Scientific, 71 Architectural and engineering activities; technical testing and analysis Occupational and 72 Scientific research and development Technical 73 Advertising and market research Activities 74 Other professional, scientific and technical activities 75 Veterinary activities 77 Rental and leasing activities 78 Employment activities Administrative 79 Travel agency, tour operator and other reservation service and related activities and Support 80 Security and investigation activities Services 81 Services to buildings and landscape activities 82 Office administrative, office support and other business support activities Public Administration 84 Public administration and defence; compulsory social security Education 85 Education 86 Human health activities Health 87 Residential care activities 88 Social work activities without accommodation 90 Creative, arts and entertainment activities 91 Libraries, archives, museums and other cultural activities Culture 92 Gambling and betting activities 93 Sports activities and amusement and recreation activities 94 Activities of membership organisations 95 Repair of computers and personal and household goods 96 Other personal service activities Other Services 97 Activities of households as employers of domestic personnel Undifferentiated goods- and services-producing activities of private households for own 98 use 99 Activities of extraterritorial organisations and bodies ! ! ! ! 142 Page | 142 TECHNICAL APPENDIX 1?>@?XB=!-=GF>@K<@A;G!?;I!0>AIHF<@J@<[!"?BFHB?<@A;G! For the labour productivity and total factor productivity (TFP) calculations, we create many variables using information from the balance sheets and income statements of the firms. Below is a table which provides a detailed description of the variables created for productivity calculations: Variable Definition Employment Simple average of number of employees in each quarter Total Wages Total wages paid to all employees within a year Capital Stock Tangible Assets + Intangible Assets (both depreciated values) Real Capital Stock Capital stock deflated with PPI for capital goods Output Net Sales + Change in Stocks of Finished and Semi-Finished Goods Real Output Output deflated with sectoral deflators Total Operating Profits + Total Wages + Depreciation of Tangible and Intangible Value Added Assets Real Value Added Value added deflated with sectoral deflators Material Inputs Output - Value Added Real Material Inputs Material inputs deflated with PPI for intermediate inputs We have two different measures of labour productivity and three different measures of TFP. First measure of labour productivity is real value added per worker which is defined as: ?XAnQMV$oQZOZDE 3 pUCVAOPRB$qVQrPOZDE 3 sCUVCNPAOPQRDE F utCNOQVAj$sCrjAOQVE fAjMC$gLLCL$UCV$kQVlCVDmE # vMwnCV$Qr$KwUjQxCCZDE Depreciation of the tangible and intangible assets is taken as a proxy for the cost of capital. We use two- digit sectoral PPI for the manufacturing industry to get the sectoral deflators, whereas for the services sector we use CPI at a more aggregated level. Second measure of labour productivity is real output per worker which is calculated as follows: ?yQOAj$tAjCZDE 3 ozARBC$PR$tOQNlZDE F utCNOQVAj$sCrjAOQVE pMOUMO$UCV$kQVlCVDmE # vMwnCV$Qr$KwUjQxCCZDE Our main TFP measure is estimated using the methodology of Ackerberg, Caves and Frazer (2015). This methodology requires a proxy variable to control for unobservables when estimating the production function. To estimate our main measure of TFP, we calculate intermediate inputs as the difference between output and value added and use this as the proxy variable. For robustness checks, we estimate two more measures of TFP: one with changing the proxy from intermediate inputs to exporting status dummy and another one with changing the methodology to the one of Levinsohn and Petrin (2003). We estimate these three measures of TFP by using the real value added as the outcome variable. The table below provides the correlations of various productivity measures for the manufacturing, services and construction sectors. As can be seen from the table, value added based measures of productivity are more robust to alternative estimation methods indicated by the high correlations. Page | 143 143 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey TFP - ACF (Raw TFP - ACF (Export TFP - Value Added per Material) Dummy) LevPet Worker Manufacturing TFP - ACF (Raw Material) 1.00 TFP - ACF (Export Dummy) 0.79 1.00 TFP - LevPet 0.70 0.71 1.00 Value Added per Worker 0.75 0.74 0.90 1.00 TFP - ACF (Raw TFP - ACF (Export TFP - Value Added per Material) Dummy) LevPet Worker Services (excl. Construction) TFP - ACF (Raw Material) 1.00 TFP - ACF (Export Dummy) 0.73 1.00 TFP - LevPet 0.52 0.15 1.00 Value Added per Worker 0.70 0.20 0.88 1.00 TFP - ACF (Raw TFP - ACF (Export TFP - Value Added per Material) Dummy) LevPet Worker Construction TFP - ACF (Raw Material) 1.00 TFP - ACF (Export Dummy) 0.99 1.00 TFP - LevPet 0.74 0.78 1.00 Value Added per Worker 0.93 0.93 0.83 1.00 -?AF=IH>=! EIS contains data on all the firms in Turkey. The table below provides the number of observations in the population of firms along with the total employment and total output of these firms for manufacturing, construction and services sectors as defined above. At this point, we should note that employment data exists for all registered firms in Turkey. Therefore, total employment figures in the table below reflect total formal employment. Output, on the other hand, is the total for the subset of firms that are obliged to provide the Ministry of Finance with financial information. Year Sector Number of Firms Employment Output 2006 Manufacturing 267,926 2,203,769 381,484,547,573 2007 Manufacturing 319,882 2,354,990 438,169,620,583 2008 Manufacturing 331,622 2,446,322 508,302,681,956 2009 Manufacturing 331,967 2,296,449 462,181,177,176 2010 Manufacturing 334,957 2,522,344 559,099,220,925 2011 Manufacturing 342,996 2,793,105 745,091,585,354 2012 Manufacturing 355,846 3,078,413 807,639,470,175 2013 Manufacturing 383,254 3,257,779 908,614,524,777 2014 Manufacturing 389,879 3,415,472 1,057,706,269,075 2015 Manufacturing 385,847 3,602,990 1,175,175,520,971 2016 Manufacturing 383,308 3,542,384 1,291,590,596,422 2006 Construction 127,410 637,858 63,585,008,129 2007 Construction 150,820 745,953 82,087,142,229 2008 Construction 159,680 796,175 93,416,589,108 2009 Construction 164,216 734,872 93,517,959,675 2010 Construction 171,057 864,691 107,768,741,994 2011 Construction 182,190 1,030,222 137,901,485,473 2012 Construction 192,094 1,171,863 164,973,293,213 Page | 144 144 TECHNICAL APPENDIX Year Sector Number of Firms Employment Output 2013 Construction 209,761 1,249,725 202,678,948,872 2014 Construction 223,682 1,332,040 245,312,309,860 2015 Construction 231,083 1,502,219 279,881,015,389 2016 Construction 244,735 1,509,878 340,201,762,189 2006 Services 1,483,808 3,312,349 933,671,256,569 2007 Services 1,902,416 3,667,125 1,080,981,905,567 2008 Services 2,002,545 4,058,336 1,202,255,889,927 2009 Services 2,042,054 4,096,422 1,203,891,023,079 2010 Services 2,082,690 4,500,560 1,451,251,274,056 2011 Services 2,108,170 5,070,099 1,807,401,477,872 2012 Services 2,142,723 5,653,675 2,069,377,335,648 2013 Services 2,384,274 5,998,540 2,352,089,826,866 2014 Services 2,407,658 6,377,970 2,689,111,070,769 2015 Services 2,401,843 6,999,001 3,076,907,964,502 2016 Services 2,401,205 6,941,987 3,293,805,270,503 As stated above, balance sheet and income statement data from the Ministry of Finance are available for firms that that either make purchases for amounts above TL160,000 or make sales for more than TL 220000. In the first step of the data cleaning procedure, we drop all the observations for which the balance sheet and the income statement information are missing. In this first step, we also drop the observations where net sales, real output, real capital stock, real value added or real material inputs are negative in a given year. This first step of the cleaning procedure produces the following table: Sector Number of Firms Employment Output Number of Firms (%) Employment (%) Output (%) 2006 Manufacturing 76,828 1,655,195 327,295,046,811 28.68% 75.11% 85.80% 2007 Manufacturing 84,800 1,838,007 382,675,157,112 26.51% 78.05% 87.33% 2008 Manufacturing 87,771 1,913,908 447,654,278,750 26.47% 78.24% 88.07% 2009 Manufacturing 86,767 1,778,958 400,490,470,369 26.14% 77.47% 86.65% 2010 Manufacturing 89,102 2,023,683 495,295,946,515 26.60% 80.23% 88.59% 2011 Manufacturing 93,847 2,310,315 678,796,073,807 27.36% 82.71% 91.10% 2012 Manufacturing 99,351 2,539,927 744,593,433,681 27.92% 82.51% 92.19% 2013 Manufacturing 102,649 2,697,784 832,101,606,344 26.78% 82.81% 91.58% 2014 Manufacturing 108,267 2,891,280 977,803,868,806 27.77% 84.65% 92.45% 2015 Manufacturing 113,886 3,063,823 1,088,666,215,515 29.52% 85.04% 92.64% 2016 Manufacturing 117,869 3,060,625 1,196,991,674,424 30.75% 86.40% 92.68% 2006 Construction 38,091 408,003 55,851,461,210 29.90% 63.96% 87.84% 2007 Construction 44,273 506,480 72,578,884,216 29.35% 67.90% 88.42% 2008 Construction 47,155 551,357 82,744,127,156 29.53% 69.25% 88.58% 2009 Construction 47,376 505,496 82,374,103,619 28.85% 68.79% 88.08% 2010 Construction 49,601 597,456 94,378,004,254 29.00% 69.09% 87.57% 2011 Construction 54,060 718,297 120,915,059,412 29.67% 69.72% 87.68% 2012 Construction 58,939 822,237 142,512,995,468 30.68% 70.16% 86.39% 2013 Construction 63,363 897,931 176,890,848,811 30.21% 71.85% 87.28% 2014 Construction 68,626 952,685 208,978,434,449 30.68% 71.52% 85.19% 2015 Construction 75,032 1,056,392 243,941,285,320 32.47% 70.32% 87.16% Page | 145 145 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Sector Number of Firms Employment Output Number of Firms (%) Employment (%) Output (%) 2016 Construction 81,116 1,073,162 294,483,396,813 33.14% 71.08% 86.56% 2006 Services 316,629 2,375,164 825,210,598,587 21.34% 71.71% 88.38% 2007 Services 360,768 2,746,784 969,609,735,983 18.96% 74.90% 89.70% 2008 Services 378,462 3,060,685 1,055,772,035,697 18.90% 75.42% 87.82% 2009 Services 381,947 3,106,746 1,066,822,861,149 18.70% 75.84% 88.61% 2010 Services 394,556 3,454,367 1,292,977,557,779 18.94% 76.75% 89.09% 2011 Services 415,011 3,891,208 1,612,106,962,107 19.69% 76.75% 89.19% 2012 Services 443,008 4,368,498 1,843,676,798,300 20.68% 77.27% 89.09% 2013 Services 459,883 4,655,128 2,095,184,485,916 19.29% 77.60% 89.08% 2014 Services 487,378 4,943,885 2,422,939,792,370 20.24% 77.52% 90.10% 2015 Services 517,646 5,375,442 2,711,346,139,981 21.55% 76.80% 88.12% 2016 Services 537,488 5,386,675 2,956,905,054,471 22.38% 77.60% 89.77% In the second step of the data cleaning procedure, we drop all the observations of a firm if the firm always has less than 2 employees. This second step of cleaning provides the following sample: Year Sector Number of Firms Employment Output Number of Firms (%) Employment (%) Output (%) 2006 Manufacturing 61,675 1,651,624 315,447,550,819 23.02% 74.95% 82.69% 2007 Manufacturing 69,453 1,835,037 368,132,693,401 21.71% 77.92% 84.02% 2008 Manufacturing 73,575 1,911,148 434,097,427,052 22.19% 78.12% 85.40% 2009 Manufacturing 73,688 1,776,838 390,345,374,297 22.20% 77.37% 84.46% 2010 Manufacturing 77,051 2,021,681 485,078,285,833 23.00% 80.15% 86.76% 2011 Manufacturing 82,320 2,308,120 669,162,307,704 24.00% 82.64% 89.81% 2012 Manufacturing 87,803 2,537,529 735,450,131,124 24.67% 82.43% 91.06% 2013 Manufacturing 91,807 2,695,243 821,751,060,005 23.95% 82.73% 90.44% 2014 Manufacturing 96,427 2,888,429 965,839,613,510 24.73% 84.57% 91.31% 2015 Manufacturing 100,517 3,059,880 1,074,826,448,973 26.05% 84.93% 91.46% 2016 Manufacturing 101,136 3,054,279 1,178,892,904,729 26.39% 86.22% 91.27% 2006 Construction 28,087 406,051 50,508,231,611 22.04% 63.66% 79.43% 2007 Construction 33,701 504,766 66,707,678,295 22.35% 67.67% 81.26% 2008 Construction 36,584 549,754 76,217,137,416 22.91% 69.05% 81.59% 2009 Construction 37,221 504,237 76,609,882,282 22.67% 68.62% 81.92% 2010 Construction 39,762 596,266 87,973,068,634 23.24% 68.96% 81.63% 2011 Construction 44,195 716,918 113,716,117,710 24.26% 69.59% 82.46% 2012 Construction 48,768 820,699 132,792,102,500 25.39% 70.03% 80.49% 2013 Construction 52,531 896,204 166,075,416,164 25.04% 71.71% 81.94% 2014 Construction 56,542 950,700 196,503,525,337 25.28% 71.37% 80.10% 2015 Construction 61,337 1,053,582 229,819,385,385 26.54% 70.14% 82.11% 2016 Construction 63,595 1,068,357 270,704,760,867 25.99% 70.76% 79.57% 2006 Services 203,573 2,346,944 764,096,584,054 13.72% 70.85% 81.84% 2007 Services 235,129 2,719,863 897,424,771,553 12.36% 74.17% 83.02% 2008 Services 254,016 3,032,882 980,719,942,950 12.68% 74.73% 81.57% 2009 Services 261,308 3,081,698 995,547,404,110 12.80% 75.23% 82.69% 2010 Services 276,533 3,429,010 1,213,176,414,333 13.28% 76.19% 83.60% 2011 Services 295,968 3,862,752 1,521,869,113,097 14.04% 76.19% 84.20% Page | 146 146 TECHNICAL APPENDIX Year Sector Number of Firms Employment Output Number of Firms (%) Employment (%) Output (%) 2012 Services 316,890 4,335,782 1,751,726,857,324 14.79% 76.69% 84.65% 2013 Services 332,208 4,619,243 2,000,572,529,973 13.93% 77.01% 85.06% 2014 Services 348,061 4,902,945 2,310,253,541,335 14.46% 76.87% 85.91% 2015 Services 362,267 5,323,395 2,579,791,543,100 15.08% 76.06% 83.84% 2016 Services 359,876 5,320,166 2,787,083,546,317 14.99% 76.64% 84.62% In the third step, we drop the service sectors as explained in the first section and marked grey in Table 4. This produces the following sample: Year Sector Number of Firms Employment Output Number of Firms (%) Employment (%) Output (%) 2006 Manufacturing 61,675 1,651,624 315,447,550,819 23.02% 74.95% 82.69% 2007 Manufacturing 69,453 1,835,037 368,132,693,401 21.71% 77.92% 84.02% 2008 Manufacturing 73,575 1,911,148 434,097,427,052 22.19% 78.12% 85.40% 2009 Manufacturing 73,688 1,776,838 390,345,374,297 22.20% 77.37% 84.46% 2010 Manufacturing 77,051 2,021,681 485,078,285,833 23.00% 80.15% 86.76% 2011 Manufacturing 82,320 2,308,120 669,162,307,704 24.00% 82.64% 89.81% 2012 Manufacturing 87,803 2,537,529 735,450,131,124 24.67% 82.43% 91.06% 2013 Manufacturing 91,807 2,695,243 821,751,060,005 23.95% 82.73% 90.44% 2014 Manufacturing 96,427 2,888,429 965,839,613,510 24.73% 84.57% 91.31% 2015 Manufacturing 100,517 3,059,880 1,074,826,448,973 26.05% 84.93% 91.46% 2016 Manufacturing 101,136 3,054,279 1,178,892,904,729 26.39% 86.22% 91.27% 2006 Construction 28,087 406,051 50,508,231,611 22.04% 63.66% 79.43% 2007 Construction 33,701 504,766 66,707,678,295 22.35% 67.67% 81.26% 2008 Construction 36,584 549,754 76,217,137,416 22.91% 69.05% 81.59% 2009 Construction 37,221 504,237 76,609,882,282 22.67% 68.62% 81.92% 2010 Construction 39,762 596,266 87,973,068,634 23.24% 68.96% 81.63% 2011 Construction 44,195 716,918 113,716,117,710 24.26% 69.59% 82.46% 2012 Construction 48,768 820,699 132,792,102,500 25.39% 70.03% 80.49% 2013 Construction 52,531 896,204 166,075,416,164 25.04% 71.71% 81.94% 2014 Construction 56,542 950,700 196,503,525,337 25.28% 71.37% 80.10% 2015 Construction 61,337 1,053,582 229,819,385,385 26.54% 70.14% 82.11% 2016 Construction 63,595 1,068,357 270,704,760,867 25.99% 70.76% 79.57% 2006 Services 188,344 2,123,436 593,296,721,454 12.69% 64.11% 63.54% 2007 Services 217,376 2,451,876 693,799,392,235 11.43% 66.86% 64.18% 2008 Services 234,685 2,724,574 802,349,725,427 11.72% 67.14% 66.74% 2009 Services 241,278 2,757,207 799,295,518,353 11.82% 67.31% 66.39% 2010 Services 255,529 3,077,153 967,194,681,025 12.27% 68.37% 66.65% 2011 Services 273,686 3,479,282 1,232,903,717,154 12.98% 68.62% 68.21% 2012 Services 293,079 3,901,863 1,424,963,713,947 13.68% 69.01% 68.86% 2013 Services 306,880 4,145,943 1,632,882,159,993 12.87% 69.12% 69.42% 2014 Services 321,370 4,397,964 1,894,671,533,187 13.35% 68.96% 70.46% 2015 Services 334,401 4,756,479 2,105,497,600,655 13.92% 67.96% 68.43% 2016 Services 331,700 4,770,722 2,296,136,459,143 13.81% 68.72% 69.71% Page | 147 147 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey In the final step of the cleaning procedure, we do an outlier cleaning at the year and Nace 2-digit level using value per worker as the reference variable. All observations, which are more than 3 interquartile ranges away from the mean within a year & nace 2-digit cell, are marked as outliers and all the observations of a firm are dropped if less than 75 percent of the observations of that firm are marked as outliers. We keep the threshold at the 75 percent as we regard those firms which always have outlier observations as those that are on the frontier or lower end of the productivity distribution and we want to keep them in the dataset for the aggregations. Those, which do not always but sometimes have outlier observations, are marked as those with measurement errors and dropped. The table below provides the final sample after this final step of the cleaning procedure: Year Sector Number of Firms Employment Output Number of Firms (%) Employment (%) Output (%) 2006 Manufacturing 53,301 1,467,175 277,273,516,435 19.89% 66.58% 72.68% 2007 Manufacturing 61,650 1,658,766 324,166,855,968 19.27% 70.44% 73.98% 2008 Manufacturing 66,471 1,763,689 385,911,286,791 20.04% 72.10% 75.92% 2009 Manufacturing 68,278 1,697,623 356,066,767,639 20.57% 73.92% 77.04% 2010 Manufacturing 71,394 1,908,077 438,978,119,603 21.31% 75.65% 78.52% 2011 Manufacturing 76,365 2,169,064 609,427,710,599 22.26% 77.66% 81.79% 2012 Manufacturing 81,242 2,384,452 669,739,021,600 22.83% 77.46% 82.93% 2013 Manufacturing 85,119 2,542,263 753,079,286,009 22.21% 78.04% 82.88% 2014 Manufacturing 89,375 2,709,055 878,560,336,183 22.92% 79.32% 83.06% 2015 Manufacturing 93,020 2,870,666 980,750,888,240 24.11% 79.67% 83.46% 2016 Manufacturing 91,555 2,842,799 1,071,827,857,349 23.89% 80.25% 82.99% 2006 Construction 21,113 304,383 36,149,631,985 16.57% 47.72% 56.85% 2007 Construction 26,942 407,393 50,887,637,049 17.86% 54.61% 61.99% 2008 Construction 30,365 470,780 58,741,493,967 19.02% 59.13% 62.88% 2009 Construction 32,024 439,706 57,121,395,057 19.50% 59.83% 61.08% 2010 Construction 34,360 518,204 67,917,562,024 20.09% 59.93% 63.02% 2011 Construction 38,215 630,245 87,343,922,121 20.98% 61.18% 63.34% 2012 Construction 42,297 722,368 102,916,791,797 22.02% 61.64% 62.38% 2013 Construction 45,933 778,679 128,906,842,040 21.90% 62.31% 63.60% 2014 Construction 50,147 846,047 155,411,684,022 22.42% 63.52% 63.35% 2015 Construction 54,135 948,638 188,469,487,593 23.43% 63.15% 67.34% 2016 Construction 53,306 905,204 215,635,199,758 21.78% 59.95% 63.38% 2006 Services 160,275 1,838,721 475,709,938,256 10.80% 55.51% 50.95% 2007 Services 191,857 2,180,685 575,439,106,969 10.08% 59.47% 53.23% 2008 Services 210,161 2,451,542 666,896,230,366 10.49% 60.41% 55.47% 2009 Services 220,431 2,522,783 672,963,014,007 10.79% 61.59% 55.90% 2010 Services 233,965 2,824,811 820,614,245,493 11.23% 62.77% 56.55% 2011 Services 251,867 3,215,386 1,045,438,312,199 11.95% 63.42% 57.84% 2012 Services 268,866 3,602,298 1,207,620,216,758 12.55% 63.72% 58.36% 2013 Services 282,571 3,818,126 1,381,797,414,222 11.85% 63.65% 58.75% 2014 Services 296,345 4,048,960 1,609,824,809,666 12.31% 63.48% 59.86% 2015 Services 306,600 4,370,236 1,809,593,240,446 12.77% 62.44% 58.81% 2016 Services 295,017 4,300,296 1,932,455,082,940 12.29% 61.95% 58.67% As can be seen from the table above, in our final sample, we keep 19 percent to 24 percent of all firms in the manufacturing industry over the period of analysis. Although low in numbers, these firms capture 66- 80 percent of the total employment and 72-82 percent of the total output. For the construction sector, 16- Page 23 | 148of firms are kept in the final sample and these firms account for 47-63 percent of the total percent employment and 56-67 percent of the total output of the construction sector. For the services sector, percentages are fairly lower. We keep 10-12 percent of all the firms in the services sector in our sample and these firms account for 60-63 percent of total employment and 50-59 percent of total output. 148 TECHNICAL APPENDIX P@\=]!,Q=!?;I!&=FN;ABAQ[!"B?GG@E@F?<@A;G! Manufacturing Services Construction Number of Firms % Number of Firms % Number of Firms % 2-9 employees 51,387 56.1% 224,489 76.1% 34,068 63.9% 10-19 employees 16,211 17.7% 38,207 13.0% 10,035 18.8% 20-49 employees 14,319 15.6% 22,283 7.6% 6,398 12.0% 50-99 employees 4,719 5.2% 5,373 1.8% 1,714 3.2% 100-249 employees 3,255 3.6% 2,978 1.0% 764 1.4% 250-499 employees 1,018 1.1% 939 0.3% 199 0.4% 500+ employees 646 0.7% 748 0.3% 128 0.2% TOTAL 91,555 100.0% 295,017 100.0% 53,306 100.0% ! Manufacturing Services Construction Number of Firms % Number of Firms % Number of Firms % 0-1 year 7,210 7.9% 19,837 6.7% 5,630 10.6% 2-5 years 22,720 24.8% 72,205 24.5% 19,891 37.3% 6-10 years 21,006 22.9% 73,221 24.8% 13,600 25.5% 11-15 years 13,385 14.6% 46,011 15.6% 5,247 9.8% 16-20 years 12,304 13.4% 39,065 13.2% 4,512 8.5% 21-25 years 8,982 9.8% 25,553 8.7% 3,093 5.8% 26+ years 5,831 6.4% 18,458 6.3% 1,262 2.4% TOTAL 91,438 99.9% 294,350 99.8% 53,235 99.9% ! Manufacturing Services Construction Number of Firms % Number of Firms % Number of Firms % Exporter 24,364 26.6% 31,521 10.7% 2,135 4.0% Non-exporter 67,191 73.4% 263,496 89.3% 51,171 96.0% TOTAL 91,555 100.0% 295,017 100.0% 53,306 100.0% ! Manufacturing Services Construction Number of Firms % Number of Firms % Number of Firms % Importer 17,487 19.1% 22,040 7.5% 1,295 2.4% Non-importer 74,068 80.9% 272,977 92.5% 52,011 97.6% TOTAL 91,555 100.0% 295,017 100.0% 53,306 100.0% ! ! ! ! ! ! Page | 150 149 Country Economic Memorandum: Firm Productivity and Economic Growth in Turkey Nace 2-digit Technology Class 10 Low Technology 11 Low Technology 12 Low Technology 13 Low Technology 14 Low Technology 15 Low Technology 16 Low Technology 17 Low Technology 18 Low Technology 19 Medium Low Technology 20 Medium High Technology 21 High Technology 22 Medium Low Technology 23 Medium Low Technology 24 Medium Low Technology 25 Medium Low Technology 26 High Technology 27 Medium High Technology 28 Medium High Technology 29 Medium High Technology 30 Medium High Technology 31 Low Technology 32 Low Technology ! Nace 2-digit Sub Sector Technology Class 41 1-Construction Not Classified 42 1-Construction Not Classified 43 1-Construction Not Classified 45 2-Retail and Wholesale Less KI Market Services 46 2-Retail and Wholesale Less KI Market Services 47 2-Retail and Wholesale Less KI Market Services 49 3-Transportation and Storage Less KI Market Services 50 3-Transportation and Storage KI Market Services 51 3-Transportation and Storage KI Market Services 52 3-Transportation and Storage Less KI Market Services 53 3-Transportation and Storage Other Less KIS 55 4-Accomodation and Food Less KI Market Services 56 4-Accomodation and Food Less KI Market Services 58 5-Information and Communication Tech. Other KIS 59 5-Information and Communication Tech. HiTech KIS 60 5-Information and Communication Tech. HiTech KIS 61 5-Information and Communication Tech. HiTech KIS 62 5-Information and Communication Tech. HiTech KIS 63 5-Information and Communication Tech. HiTech KIS 64 6-Finance and Insurance KI Financial Services 65 6-Finance and Insurance KI Financial Services 66 6-Finance and Insurance KI Financial Services 68 7-Real Estate Less KI Market Services 69 8-Scientific, Occupational and Technical Activities KI Market Services 70 8-Scientific, Occupational and Technical Activities KI Market Services 71 8-Scientific, Occupational and Technical Activities KI Market Services 72 8-Scientific, Occupational and Technical Activities HiTech KIS 73 8-Scientific, Occupational and Technical Activities KI Market Services 74 8-Scientific, Occupational and Technical Activities KI Market Services 75 8-Scientific, Occupational and Technical Activities Other KIS 77 9-Administrative and Support Less KI Market Services 78 9-Administrative and Support KI Market Services 79 9-Administrative and Support Less KI Market Services 80 9-Administrative and Support KI Market Services Page | 151 150 TECHNICAL APPENDIX Nace 2-digit Sub Sector Technology Class 81 9-Administrative and Support Less KI Market Services 82 9-Administrative and Support Less KI Market Services 84 10-Public Other KIS 85 11-Education Other KIS 86 12-Health Other KIS 87 12-Health Other KIS 88 12-Health Other KIS 90 13-Culture Other KIS 91 13-Culture Other KIS 92 13-Culture Other KIS 93 13-Culture Other KIS 94 14-Other Services Other Less KIS 95 14-Other Services Less KI Market Services 96 14-Other Services Other Less KIS 97 14-Other Services Other Less KIS 98 14-Other Services Other Less KIS 99 14-Other Services Other Less KIS ! ! 151 Page | 152 Productivity Report 2019 Uğur Mumcu Caddesi, No:88, 2. Kat 06700, Gaziosmanpaşa, Ankara/ Türkiye turkeywebfdbk@worldbank.org