057 057 This paper is a product of the Poverty and Equity Global Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and contribute to development policy discussions around the world. The authors may be contacted at jazevedo@worldbank.org or jyang4@worldbank.org or oinan@worldbank.org. The Poverty & Equity Global Practice Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. ‒ Poverty & Equity Global Practice Knowledge Management & Learning Team This paper is co-published with the World Bank Policy Research Working Papers. What Are the Impacts of Syrian Refugees on Host Community Welfare in Turkey? A Subnational Poverty Analysis Joao Pedro Azevedo Judy S. Yang Osman Kaan Inan December 7, 2015 The World Bank1 JEL Codes: F22, I3, O15, R23 Keywords: Turkey, Syria, Poverty, Refugees, Forced Migration This paper has benefitted from comments by participants from presentations made in World Bank Washington, DC and Ankara offices, more specifically from the Poverty and Equity Global Practice’s Europe and Central Asia Team and the Turkey Country Office as well as inputs by our colleagues in the Social Protection and Labor Global Practice and the Social, Urban, Rural, and Resilience Global Practice. The team thanks Martin Raiser, William Wiseman for their guidance and support. We are also thankful for comments and assistance with data received from JoAnna P. De Berry, Ximena Del Carpio, Doreen Triebe, Zeynep Durnev Darendeliler, Jose Montes, and Minh C. Nguyen. The usual disclaimer applies. This paper is a product of the FY2015 Turkey Poverty Team in the World Bank’s Poverty and Equity Global Practice. This piece is part of a collaboration with the Urban, Rural, and Social Development GP supporting Turkey’s Regional Development Strategy (RDS) 2014-2023. 1Contacts: Joao Pedro Azevedo (Lead Economist, jazevedo@worldbank.org); Judy S. Yang (ET Consultant, jyang4@worldbank.org); Osman Kaan Inan (Junior Professional Associate, oinan@worldbank.org) This paper has benefitted from comments by participants from presentations made in World Bank Washington, DC and Ankara offices, more specifically from the Poverty and Equity Global Practice’s Europe and Central Asia Team and the Turkey Country Office as well as inputs by our colleagues in the Social Protection and Labor Global Practice and the Social, Urban, Rural, and Resilience Global Practice. The team thanks Martin Raiser, William Wiseman for their guidance and support. We are also thankful for comments and assistance with data received from JoAnna P. De Berry, Ximena Del Carpio, Doreen Triebe, Zeynep Durnev Darendeliler, Jose Montes, and Minh C. Nguyen. The usual disclaimer applies. This paper is a product of the FY2015 Turkey Poverty Team in the World Bank’s Poverty and Equity Global Practice. This piece is part of a collaboration with the Urban, Rural, and Social Development GP supporting Turkey’s Regional Development Strategy (RDS) 2014-2023. INTRODUCTION In 2015, an estimated 2.2 million Syrians Under Temporary Protection (SUTPs) were residing in Turkey, the majority arriving in the country over the last 4 years.2 Turkey’s national population is roughly 75 million; recent refugees account for approximately 3 percent of the population. For a country that has never experienced such a large-scale, sudden inflow of foreigners, demographic changes in the composition of the population and labor force will yield unprecedented implications. This paper examines, as data allows, the relationship between the size of the foreign-born population and host community poverty rates in Turkey. First, this paper finds the poverty rates of ‘recent migrants’ near the Syrian border (NSB) significantly increased from 2009 to 2013. Second, the number of foreign-born households being captured by the Labor Force Survey (LFS) is expanding, which suggests a growing number of foreign households that are likely to be Syrians. Third, with respect to poverty, the results show no negative impacts on the host community as a result of the increasing size of the foreign-born population. The impact of SUTPs has been both positive and negative. Overall, a significant negative impact on host communities’ welfare is not observed in the data. This paper’s scope of analysis includes the country as a whole using a nationally representative survey. While regional case studies may reveal salient stresses on public services and job displacement, nationally, there is no significant impact. Over the period of 2009 to 2013, the poverty rates of host community households have stayed relatively stable near the Syrian border; despite the high poverty rates experienced among the recent migrants. By country of origin, the displacement of Syrians is one of the largest in recent history. As a result of the civil war that began in 2011, Syrians started to leave their homes and look for safety in neighboring countries across the region. By November 2015, about 4.3 million Syrians were seeking refuge in primarily Turkey, Lebanon, Jordan, Iraq, and the Arab Republic of Egypt.3 The only other time in the last half century that the world experienced a larger group of refugees from a single country is the case of Afghan refugees during the 1980s to 1990s. Refugee displacements of this size are rare. Consequently, they are not well studied and their impacts are not well understood. Moreover, the case of Afghan refugees in Pakistan is different, since they were stigmatized to a larger extent, which limited their movement in Pakistan. While there is a large literature on the role of immigrants on the native-born population in terms of labor market competition, there is a limited amount of studies that examine the effect from displaced populations. Many conclusions from the traditional literature on the study of immigrants’ impact on natives cannot be applied to the case of Syrians in Turkey. There are many differences between the inflow of Syrians and other flows of extended family and economic immigrants. First, the sheer volume of Syrian refugees and the short time- frame in which they entered Turkey is unprecedented. For the case of Syrians in Turkey, or displaced populations in general, large movements of refugees are not restricted due to humanitarian reasons. Second, formal immigration processes are controlled, limited, and regulated by destination countries. Therefore, results from literature on “immigrants” are very different than a focus on displaced or refugee populations. Recent literature on the labor market effects of SUTPs estimates negative impacts on host community employment rates. The negative displacement results are largest for the young, women, informal workers, 2 United Nations High Commissioner for Refugees (UNHCR) – Syrian Regional Refugee Response, Inter-agency Information Sharing Portal 3 (UNHCR) – Syrian Regional Refugee Response, Inter-agency Information Sharing Portal 2 and the less educated (Ceritoglu, Tunculer, Torun, and Tumen, 2015; Del Carpio and Wagner, 2015). The economic effects of SUTPs not only vary across different segments of the labor market, there are also strong regional differences in their economic effects. Using synthetic modelling methods, Ozturkler and Goksel (2015) estimate the impact of Syrian refugees on local prices, wages, inflation, and services in 10 cities with large refugee populations. Some of the salient negative effects have been increases in rental prices, increases in inflation at border cities, illegal hiring by small business, and decreases in wages. However, in some cities (Gaziantep, Adana, Kahramanmaras, and Mardin), the presence of refugees has improved the trade balance, and economic activity in these areas are projected to increase as economic integration with MENA deepens. Orhan and Gundogar (2015) also note both positive and negative aspects of SUTPs. A primary contribution of this paper is the estimation of poverty at the sub-national level and among population groups of interest. Since migration, geographic, and welfare variables do not exist in a single data set, imputation techniques are required to overcome these limitations and to compute household level poverty. The imputation of income poverty is done using the Turkish Labor Force Survey (LFS), and with information and modeling parameters determined from the Survey of Income and Living Conditions (SILC). More details and validation of this methodology is discussed throughout this paper. While explicit identification of Syrians in available surveys is not feasible, there is evidence of an increase in the amount of foreign-born individuals that is being captured in the LFS. The arrival year of foreign-born migrants is available in the data which allows for identification of “Settled Migrants” and “Recent Migrants”. The latter is used as a proxy for Syrian refugees for the purposes of this paper. National official surveys that are conducted under-report the refugee population. Yet, since about 10 percent of Syrian refugees are in camps and the remaining are residing throughout the country, it is not surprising that they are accessible to interviews by the LFS. Despite data limitations, there are strong and significant trends in the poverty rates for the recent foreign- born, especially for those near the Syrian border. In 2013, recent migrants near the Syrian border were the poorest group4 in Turkey. While this statistic in itself is not initially surprising, fluctuating welfare trends of recent migrants over time is noteworthy. In previous years, migrant households in Turkey tend to have much lower poverty on average than even the host community. Across comparison groups and time, the poverty rates of recent migrants is higher than among the host community in only one instance: in 2013 near the Syrian border. This sudden change in the historically stable pattern implies that the LFS is able to capture at least a part of the incoming SUTPs who have significantly different socioeconomic profiles in comparison the previous economic migrants. Throughout history, immigration to Turkey has been relatively limited and consisted mostly of those of Turkish heritage. In the early 20th century, immigration was encouraged by the government as a method to increase the population. Since 1970, immigration has slowed down and has been even discouraged at times. Many immigrants to Turkey are of Muslim Turkish background, since the government prioritized preserving a national identity. This is likely why “migrants” had very similar or even lower poverty rates than the host community. The sharp degradation of welfare among recent migrants in 2013 illustrates the severity of poverty that is arising very likely from a growing population of Syrian refugees. The Syrian refugee inflow to Turkey 4 Based on grouping by host community, established migrant, recent migrant. 3 began in April 2011 and has been continuing at an increasing pace as the conflict in Syria expands.5 The Turkish government has provided a tremendous amount of support in the form of shelter and essential items to help sustain the livelihood of large numbers of refugees. However, aid funds are not limitless and refugees face hardships that will persist over the long-term. The refugee camp population in Turkey has been stable since March of 2013 as the physical capacity of the camps have been exhausted.6 This saturation has resulted in a steep increase in the number of Syrians living outside camps across Turkey. The proportion of Syrian refugees living outside camps increased from 53 percent to 87 percent between March 2013 and November 2014.7 In addition, even though refugees living outside camps continue to be concentrated near the Syrian border (64 percent), the dispersion of Syrians across the country has expanded, especially in major urban centers such as Istanbul and Ankara. The results in this paper are limited to 2013 due to changes in the 2014 LFS that make poverty estimations incomparable to previous years.8 Therefore our results may provide only a partial insight into the impact of SUTPs, since the dispersion of Syrians across Turkey has increased greatly in 2014 and 2015. The descriptive characteristics of foreign-born and host community households are still comparable, and therefore allow for comparisons between 2013 and 2014. It is possible to make some inferences on welfare changes by looking at changes in host community employment rates and labor market characteristics. Overall, this paper finds no negative effects on host community welfare from an increasing population of SUTPs. As other authors have stated, the influx of SUTPs has had both positive and negative impacts. It seems on average, the host community has been strong and adaptive, and not negatively impacted. This is not to disregard that real strains do exist in some regions where the SUTP population is very large. Nor do these results undermine findings of displacement effects in the labor market that certain types of workers are experiencing. However, on average nationally, we do not see a systematic decline in the welfare of the host community between 2011 and 2013. The remainder of the paper is organized as follows. Section 2 outlines the data availability and technical issues. Section 3 discusses descriptive statistics of the foreign-born and host community. Section 4 explores the impact of the foreign-born population on host community welfare. 1. DATA AND TECHNICAL ISSUES Data Sets Data availability limits which data set can be used to identify the foreign-born population and geographic location while measuring poverty. The Turkish Statistical Institute (TUIK) has been conducting three nationally representative surveys annually since 2005; the Household Income and Consumption Expenditure Survey (HICES), the Survey on Income and Living Conditions (SILC) and the Labor Force Survey (LFS). However the HICES, which is the national survey that is used to measure official poverty, 5 UNCHR, 2013pg 13 6 UNHCR (22 November 2013), UNHCR (15 September 2014) 7 UNHCR (22 March 2013), Erdogan (2014), pg 14 8 This technical issue will be discussed further later in the paper. 4 does not have migration or geographic identifiers. The SILC contains geographic identifying variables, (at the NUTS1 level) but still lacks migration variables. The data set this study uses is the Labor Force Survey (LFS) since there is an adequate availability of both migration and geographic variables. The LFS is representative at the NUTS2 level which corresponds to 26 regions in Turkey. One caveat is that income in the LFS refers to only wage income from employment,9 and is an insufficient measure of income that should be used for welfare measurement. For example, important sources of income such as social assistance, asset liquidation, or remittances are missing. Therefore, income in the LFS is imputed with a few assumptions using information from the SILC. The NUTS1 spatial effects of the SILC are a good proxy for NUTS2 welfare dynamics in the LFS which increases the accuracy of the imputation model. However, since the original sample frame of the LFS does not account for the recent influx of foreign migrants in Turkey, the labor market characteristics of recent migrants might not be representative of the actual SUTP population. Therefore, results of the imputation could be interpreted as upper bound estimates for recent migrants. More details of survey techniques used to complete this exercise are available in the Annex. As a result, imputed poverty is measured as income poverty. Another advantage of using the LFS is the availability of CPI at the NUTS2 level in Turkey which allows for spatial deflation of different price levels across the country. Table 1. Survey Comparison and Data Availability Years Migration Income or Consumption Geographic Identifier Spatial Available Variables Deflation HICES 2003-2012 No Consumption, income National, urban/rural No SILC 2009-2012 No Income NUTS1 No LFS 2009-2013 Yes Imputed Income NUTS2 Yes However, there are other issues for consideration when using the LFS. Principally, there is a low number of sample points that are migrant households. Moreover, the study cannot identify migrant households and individuals that are specifically Syrian refugees. Foreign migrants are defined as those who were born abroad and have lived abroad for at least more than 12 months. Some Turkish-born households have also lived abroad for over a year, and these individuals are not considered to be migrants. Amongst foreign-born individuals, only the ones who have been in the country for more than 12 months are included in the sample which underrepresents the actual number of foreign migrants in the region. In addition, no specific procedure is adopted by the enumerators if the household does not speak Turkish. Given that a majority of Syrian refugees do not speak Turkish, the language barrier might result in the removal of Syrian households from the sample. Finally, refugee camps are not included in the sample frame, which limits the study to only examining recent migrants who do not live in refugee camps. 9Wage income is only available for regular and casual employees in the LFS which accounts for around 60% of total employment. There is no other monetary income value for the rest of the working population. 5 Comparison Groups Six population groups are constructed based on their migrant status and geographic location (Table 2). Five out of 26 regions are defined as Near Syrian Border (NSB) regions based on their proximity to Syria as well as their popularity as a destination for migrants (see Map 1 for details). These regions are Mardin (TRC3), Sanliurfa (TCR2), Gaziantep (TCR1), Hatay (TR63), and Adana (TR62).10 TRC3-Mardin, TCR2- Sanliurfa, TCR1-Gaziantep and TR63-Hatay are Southeastern regions of Turkey that border Syria. TR63- Adana does not border Syria but is a southern Mediterranean region that is a common destination for migrants due to abundant labor opportunities. The rest of the country includes the remaining 21 NUTS2 regions. Map 1. Near Syrian Border Regions TR62 Adana Subregion TR63 Hatay Subregion TRC1 Gaziantep Subregion TRC2 Sanliurfa Subregion TRC3 Mardin Subregion 10 NUTS2 regions are referred with name of the largest province in each regions. The full list of provinces in each region are; Mardin-Batman-Sirnak-Siirt (TRC3), Sanliurfa-Diyarbakir (TCR2), Gaziantep-Adiyaman-Kilis (TCR1), Hatay-Kahramanmaras- Osmaniye (TR63), and Adana-Mersin (TR62). 6 Host community households are those whose head of household were born in Turkey, or born outside Turkey but did not live abroad for more than a year. Conversely, migrant households are defined as those whose head of household was born abroad and has lived abroad for more than 12 months. The duration of a migrant household’s stay in Turkey is also based on when the head of household arrived in Turkey. Three thresholds are tested: 2, 3, and 4 years. The 4 year cut-off is preferred to maximize the sample size of recent migrant households. Table 2. Population Groups for Comparison Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 Geographic Near the Syrian Border The Rest of the Country Location Settled Migrant Recent Migrant Host Settled Migrant Recent Migrant Host Community Status Households Households Community Households Households Arrived in Arrived in Head of Arrived in Head of Arrived in Turkey more Turkey more Years in Turkey Household born Turkey less than: Household born Turkey less than: than: 2, 3, or 4 than: 2, 3, or 4 in Turkey 2, 3, or 4 years in Turkey 2, 3, or 4 years years years Are Syrian Refugees being captured using the LFS? While variables covering all topics of interest (migration, welfare, and geography) are available or can be imputed into the LFS, it may be unclear, ex-ante, if SUTPs are adequately included in the survey. Despite a small sample of foreigners and other concerns, there is evidence that the LFS sample does include some “recent foreign” migrants, especially in the border regions (NUTS2) [TRC1-Gaziantep, Adiyaman, Kilis, TRC2-Sanliurfa, Diyarbakir, TRC3-Mardin, Batman, Sirnak, Siirt TR63-Hatay, Kahramanmaras, Osmaniye] (Map 1). 7 Figure 1. Change in Recent Migration Recent Migrants (1 or less than 1 year in Turkey) Recent Migrants (2 or less years in Turkey) 60.00% 60.00% 50.00% 50.00% 40.00% 40.00% 30.00% 30.00% 20.00% 20.00% 10.00% 10.00% 0.00% 0.00% 2010 2011 2012 2013 2010 2011 2012 2013 -10.00% -10.00% Recent Migrants (3 or less years in Turkey) Recent Migrants (4 or less years in Turkey) 60.00% 60.00% 50.00% 50.00% 40.00% 40.00% 30.00% 30.00% 20.00% 20.00% 10.00% 10.00% 0.00% 0.00% 2010 2011 2012 2013 2010 2011 2012 2013 -10.00% -10.00% Regions Bordering Syria Rest Regions Bordering Syria Rest Source: LFS 2009-2014. World Bank author’s calculations. Figure 1 illustrates the percentage increase of recent foreign migrants when compared to the level of recent foreign migrants present in Turkey in 2009 (baseline year omitted). The dashed line shows the trends in the border regions and the solid line shows trends in the rest of the country. While there is an uptick in recent migrants near the Syrian border regions, it is important to remember that the sample size in the LFS is small, especially among recent migrants (Table 3). Notice, that the sample of recent migrants is relatively constant from 2009 to 2012, but in 2013 and 2014 there is a clear acceleration in the growth of the number of recent migrants being included in the survey. Since the number of recent migrant households begins to increase substantially in the 2013 LFS, there is evidence pointing to the representation of SUTPs in the survey. The analysis in this paper will be as transparent and rigorous as possible under the data conditions and limitations. This will translate into high standard errors in our statistical estimates, but estimates and especially trends of poverty rates can still be informative. 8 Table 3. Number of households (HH) in the sample Host Settled Migrant Recent Migrant Settled Migrant Recent Migrant HH Community HH HH HH (<=3 years in (>4 years in (<=4 years in (>3 years in Turkey) Turkey) Turkey) Turkey) 2009 133,710 2,127 54 2,142 39 2010 141,597 2,216 58 2,223 51 2011 142,203 2,101 57 2,111 47 2012 143,756 2,086 68 2,098 56 2013 143,572 2,363 120 2,374 109 2014 147,175 2.523 359 2.553 329 Source: LFS 2009-2014. World Bank author’s calculations. The rate of increase in the recent migrant population of labor force surveys between 2011 and 2013 is parallel to observed trends in the government and UNHCR data regarding the SUTPs (Figure 2 in the annex). At the level, the LFS is not comparable to the government and UNHCR data since the sampling structure was not planned to represent the entire SUTP population. However, Figure 2 shows that the change between 2012 and 2013 in the recent migrant population of the LFS has a similar slope to the increase in the SUTP population from administrative data. The trend in the recent migrant population near the Syrian border is almost identical to the movement in the UNHCR and government numbers. A caveat to the analysis in this paper is the small sample of foreign-born households in the LFS. In the 2009 survey, there are only 58 foreign-born households, and only three are located near the Syrian Border. Applying sampling weights, these three households represented 1,368 people in the population. By 2013, the number of surveyed foreign households increased substantially to 120, with 34 households in the NSB region. In 2014, the sample of foreign-born almost tripled from the year before and the estimated foreign- born population over the age of 15 reached 1.3 million. One very important change in the survey that cannot be emphasized enough is that the sampling of the LFS changed in 2014. Therefore the population estimates in 2014 are only of the population aged 15 and over. From 2009 to 2013, estimates reflect the entire population. However, estimates of the number of households should not be affected by this sampling change. Table 6 through Table 9 highlight statistics on host community and foreign-born households and population. An interesting trend is the movement of the local population away from the East. In 2013, 16.6 percent of host community households were located near the Syrian border. This value dropped to 15.7 percent in 2014. The recently (<=4 years in Turkey) arrived foreign households, are also mobile and able to move inland. In 2013, 25.7 percent of recent foreign households were located Near the Syrian Border compared to 22.2 percent in 2014. However, the sampling frame of the LFS also changed in 2014 which might have impacted population trends. 9 2. POVERTY TRENDS Imputation11 Household-level poverty is calculated using imputed income from the LFS. The LFS is ideal due to the availability of regional and migrant information, but only has wage data which is insufficient to capture total income for welfare measurement. Thus total income needs to be imputed. The SILC data set can also be used to study regional poverty rates, but only at the NUTS1 level. Another disadvantage of the SILC is that spatial deflation in Turkey is only available at the NUTS2 level (LFS data), and not at the NUTS1 level since TUIK computes regional CPIs only at the NUTS2 level. Income and household characteristics from the SILC are used to produce a model to impute total income into the LFS. The imputation model is an OLS with regionally deflated incomes at the NUTS2 level and time-invariant characteristics. Regressions are conducted using household-level data. For each year, the model specified from the SILC may vary. For every household, total household income is imputed 100 times. At the regional and comparison group level, poverty rate estimates and standard errors are computed from these 100 imputations. The international $5/day per capita 2005PPP poverty line is used. One way to interpret the poverty rates from imputed income is simply the conditional welfare based on observables. Observed income poverty can be computed using the SILC but only at the NUTS1 level and without differentiation between the host and foreign-born populations. As a validation exercise, observed poverty computed using the SILC and imputed poverty using the LFS can be compared. Figure 7 illustrates this comparison and shows that, at the NUTS1 level, imputed poverty rates using the LFS are very similar to observed poverty in the SILC. One caveat is that 2012 and 2013 LFS uses models computed in the 2011 SILC due to a lag in the SILC data availability and changes in the LFS survey methodology. Estimates of Poverty Table 4 lists results of imputed poverty rates by geographic location, comparison group, and year. Results are shown using the OLS imputation model with time-invariant characteristics and regionally deflated incomes. Recall that in the LFS, the number of recent migrant households in the sample is small when broken down to geographical groups. An implication of small sample size is high standard errors in our poverty estimates for this group, especially in 2009. However, in 2013, the number of foreign-born households increased significantly which allows for more precise estimates of migrant poverty rates. In principle, the migration and assimilation literature sets expectations that migrant households are generally worse off than host community households upon arrival. The rate of assimilation of migrants depend on their education, language, and opportunities. The children of migrant households also do better than their parents. However, in Turkey, migrant households traditionally have not been worse off than host community households. This is due to the homogeneity of migrants which tend to be of Turkish heritage. 11 See the Annex for further details on the survey-to-survey imputation methodology. 10 However, recent migrant households near the Syrian border in 2013 are distinctively poorer than any previous group of migrant households. Table 4. Imputed Poverty Rates (LFS) Near Syrian Border Rest of the Country Host Settled Recent Host Settled Recent Community Migrants Migrants Community Migrants Migrants (>4 years) (<=4 years) (>4 years) (<=4 years) 2009 Poverty Rate 44.2 27.5 15.6 21.1 9.1 7.3 S.E. 0.8 6.7 22.8 0.3 0.7 5.0 2010 Poverty Rate 42.6 23.8 27.1 20.8 10.7 5.2 S.E. 0.7 6.4 19.3 0.3 0.8 3.2 2011 Poverty Rate 40.9 12.5 17.5 19.3 9.3 13.3 S.E. 0.7 5.4 25.6 0.3 0.7 4.9 2012 Poverty Rate 41.1 21.3 37.0 18.4 7.0 13.2 S.E. 0.7 5.1 22.5 0.3 0.7 4.5 2013 Poverty Rate 40.3 17.7 46.7 17.9 6.6 16.1 S.E. 0.8 5.8 7.9 0.3 0.5 6.7 Notes: Regionally deflated, OLS imputation model, with time-invariant characteristics. The Near Syrian Border region (NSB) includes the NUTS2 areas TRC3, TCR2, TCR1, TR63, TR62. Time in Turkey is based on the date of arrival of the head of household. Source: LFS 2009-2014. World Bank author’s calculations. Figure 2 illustrates poverty trends in 2009 and 2013, by region and migrant status. Regions are divided into the near Syrian border region, and the rest of the country. The NSB includes the NUTS2 areas TRC3, TCR2, TCR1, TR63, and TR62. The population is also divided into host communities, recent migrant households, and settled migrant households. Poverty is higher in regions near the Syrian border, for both host communities and migrants. The South Eastern region of Turkey has been historically the poorest area of the country. However, between 2009 and 2013, poverty has declined for host communities and settled migrant households. On the other hand, poverty of recent migrant households in the NSB region has been increasing, and significantly so in 2013 (Figure 3). The sharp increase in the poverty rates of recent migrant households is evident. Near the Syrian border, recent migrant households had a poverty rate of 15.6 percent (caveat on the small sample size and large standard error) in 2009, and a rate of 46.7 percent in 2013. There was a parallel increase among the recent migrant households in rest of the country, from 7.3 percent to 16.1 percent over the four years. 11 Figure 2. Poverty Estimates, by year and migrant status 2009 80 70 Poverty Rate (%) 60 50 40 44.2 30 27.5 20 21.1 15.6 10 9.1 7.3 0 Host Community Settled Migrants Recent Migrants Host Community Settled Migrants Recent Migrants (>4 years) (<=4 years) (>4 years) (<=4 years) Near Syrian Border Rest of the Country 2013 80 70 Poverty Rate (%) 60 50 46.7 40 40.3 30 20 17.7 17.9 16.1 10 6.6 0 Host Community Settled Migrants Recent Migrants Host Community Settled Migrants Recent Migrants (>4 years) (<=4 years) (>4 years) (<=4 years) Near Syrian Border Rest of the Country Notes: Regionally deflated, OLS imputation model, time-invariant. Vertical line represents standard error. Poverty line is $5/day PPP per capita. The Near Syrian Border region (NSB) includes the NUTS2 areas TRC3, TCR2, TCR1, TR63, TR62. Time in Turkey is based on the date of arrival of the head of household. Results displayed at the 95% Confidence Interval Source: LFS 2009-2014. World Bank author’s calculations. It is important to understand which characteristics or circumstances may be driving the large differences in poverty between our six comparison groups. Changes over time are also worthy to note since the number of foreign households in the LFS increases dramatically in 2013 and 2014. Table 10 illustrates the mean summary statistics for our comparison groups in 2009, 2013, and 2014 that are located near the Syrian border.12 Table 11 shows complementary statistics for households located in the rest of the country. Recent migrant households are younger and have more children, and a higher dependency ratio. Dependency in migrant households result not from the elderly but from children. Families of recent migrants are much larger. In 2013, recent migrant households near the Syrian border had on average 3.1 adults, 2.4 children, and 0.4 elderly. The large number of children is indicative of entire families moving to Turkey from Syria. The composition is in contrast to working age individuals who are more likely to be economic migrants. These individuals would migrate without families, and send remittances home to their 12The years 2009 and 2014 are the earliest and latest available surveys in the LFS that are usable for measuring migration. However, due to data availability on Syrian refugees and changes to the survey in 2014, regression analysis will use only 2011 and 2013 LFS data. 12 origin countries. Compared to host community or settled migrant households, recent migrant households are younger. In 2013, heads of households in over half of recent migrant households were aged 25-39. In host communities, about a third of household heads are in this age group. Moreover, recent migrant households near the Syrian border are less educated. The heads of household in 76% of recent migrant households NSB had no education, and none of them had post-secondary education. The majority of recent migrant households are also not registered with the social security institution. In 2013, only 20.7 percent of recent foreign-born households near the Syrian border were registered. In 2014, this rate dropped to only 2.3 percent of households. Registration to social security is related to a person’s main job and the formality of the occupation. This rate decreased, indicating a lack of formal jobs. Moreover, the recent jobs were mostly blue-collar. From 2013, to 2014, there is a clear trend of increasing informality among these workers. To summarize, the high poverty rates observed among recent migrant households near the Syrian border can be related to a few descriptive characteristics: larger family size, higher dependency ratios, less education, younger families, informality, and less social protection. Interestingly, poverty rates of settled migrant households are actually lower than rates among host communities, and significantly so. In 2009, poverty rates for both settled and recent migrants (large sample error for recent migrants) are substantially lower than the host communities. In 2013, settled migrants’ poverty rate is less than half of the host community’s rate while recent migrants became the poorest group. This decoupling between the poverty rates of settled and recent migrants over the four years indicates that the type of recent migrants changed over that time span. The increase in the poverty rate of this new type of recent migrants suggest that the cohort, in 2013, is capturing the incoming SUTPs, particularly in the NSB regions. There is weak evidence that the welfare of migrant households in Turkey does improve over time in the long-run. Near the Syrian border regions, the poverty rates of migrant households who have entered Turkey <10 or 10-19 years ago are not significantly different. There could be a number of reasons for this, including wealthier families being able to move farther west to Istanbul or Europe in the long run. In 2013, the poverty rates of recent migrants do increase substantially, while this cannot be attributed directly to the influx of Syrian refugees, recent migrants in the NSB regions are very likely to be SUTPs. However, migrant households in Turkey are still very dynamic, with poverty rates generally lower than those of host communities. Since the nationalities of migrants from 20-30 years ago are much different than the composition in recent years, no clear conclusions on assimilations can be drawn yet. Figure 5 illustrates the poverty rate of only host community households. Poverty rates are illustrated by geographic area as well as characteristic groups. Among the selected characteristics, host community households whose head is a blue collar worker have higher poverty rates, while male headed households have lower poverty rates. Geographically, the host community in the NSB region have poverty rates almost twice as high as their counterparts in the rest to the country, and this does not vary significantly by their characteristics groups. Overall, no matter whether poverty rates are calculated for selected host community households whose head of household is a blue collar worker, female head, male head, or head with low education; poverty rates among these households do not increase in 2013. This pattern holds for both in the near Syrian border region and in the rest of the country. 13 3. ANALYSIS Has the increase in Syrian refugees impacted the welfare and socioeconomic conditions of the host community? Summary statistics in the previous section showed clear trends of increasing poverty among recent migrants throughout the country, both near the Syrian border and across rest of the country. While the poverty rates of recent migrant households spiked in 2013, poverty of host community households maintained a relatively constant level in the whole country. From these trends, it appears that there at least has not been an increasing trend in poverty among the host community over the latest years. The empirical model is shown in Equation 1. Regressions are estimated at the NUTS2-year level and using data from only the years 2011 and 2013. The dependent variable of interest is the host community poverty rate by region and year, where the poverty rate is based on spatially deflated imputed household income. Unlike the computation of the poverty rates, “recent migrant” information is not used for the analysis. Only the host community poverty rates are calculated using the LFS and the number of Syrians are taken from government sources. In that respect, the regression dos not suffer from the small sample size or any other limitations relating to the estimation of poverty rates for recent migrants. c ∗ + ∗ (1) The year 2013 is selected since there is available administrative information on the stock of Syrian refugees across regions. The choice of 2011 is rather arbitrary since we assume no foreign-born population in the year 2011 or before. The explanatory variable of interested is the proportion of SUTPs relative to the population ( ). The assumption is that the proportion equals zero for all regions in 2011. This is a simplistic but reasonably accurate estimate since the majority of SUTPs arrive in 2012. Moreover, official estimates of the number of Syrians in Turkey were also not available until starting in 2012. Overall, in 2013, Syrians were still very concentrated in only a few regions. Out of 26 regions, Syrians comprised more than 1 percent of the population in only 7 regions. Estimations cannot be conducted at the household-level data since the survey-to-survey imputation methods used to compute poverty rates yield only group level estimates of poverty rates, not household level poverty rates. The ideal methodology would be to measure the causal effects of the influx (shock) of Syrians under temporary protection as a proportion of the population onto the poverty rates of the host community. However, there are some concerns regarding endogeneity. The foreign-born in Turkey have freedom of movement and it is very likely that households prefer to settle in locations that are more welcoming or offer better economic opportunities. Regional differences in social and economic conditions can influence their decisions to locate. Del Carpio and Wagner (2015) discuss these concerns and propose an instrumental variable approach. Two different, but complementary IVs are proposed: distance and time from Syrian regions to Turkish regions. In this paper, regional aggregation and time dimension of the data limits the data to only 52 data points. There are not enough rank conditions and IV is not feasible. 14 Moreover, there is an added level of complexity of estimation error arising from imputation estimates as well as from the regression. Running regressions at this level of aggregation can only provide a general sense if host community poverty rates are correlated to the size of the foreign-born population. The most likely scenario is that refugees are attracted to locations with healthy economies and attractive wages, or locations that show promising growth and good job opportunities. These unobservable circumstances would be negatively associated with poverty rates of the host community. The variable of interest Rry would be negatively correlated to these types of unobservable conditions. Therefore would be positively biased. In other words, estimates are reasonably lower bounds. Results Table 5 illustrates results from the empirical estimation of Equation (1). There is no evidence that poverty rates of the host community are linked to the proportion of the foreign-born population. OLS regressions without any control variables do show a positive correlation between the proportions of foreign-born and host community poverty. However, this is primarily due to regions in the East and near the Syrian border historically being the poorest regions in the country. Once region and time controls are added to the regressions, the correlation between the host community poverty and the size of the foreign-born no longer exist. Similarly when controls on average characteristics of the regional population are included into the specification, there is no relationship between the explanatory variable of interest and the dependent variable. Table 5. OLS Regression Results, the Host Community (1) (2) (3) Proportion of Foreign-Born in the Population 0.957** -0.0910 -0.00385 (0.423) (0.126) (0.0753) Constant 0.203*** 0.108*** 0.505 (0.0386) (0.00177) (0.289) Year Dummy X X Region Dummies X X Labor and Demographic Controls X Observations 52 52 52 R-squared 0.091 0.944 0.992 Notes: * p<0.10, ** p<0.05, *** p<0.01. Source: LFS 2011 and 2013. World Bank author’s calculations. Del Carpio and Wagner (2015) find some negative impacts, displacement effects, from the increase in foreign-born from 2011 to 2014. The displacement effect is largest among low-skill, agricultural, and low- educated workers. There are various reasons why results in this paper are neutral compared with those of DCW (2015). Their analysis is based on the 2014 LFS, which exhibits a much larger foreign-born population than the 2013 survey that is used in this paper. Unfortunately, due to changes in survey collection 15 methods, particularly the exclusion of children from the LFS data set in 2014, this paper is not able to use the 2014 data set to impute household-level poverty rates.13 Our analysis is limited to 2013. Another difference is that this paper’s analysis is conducted at the regional level. Regional aggregation smooths away household differences. Therefore, regressions are run on poverty rates of different sub- samples of the host community. Table 13 and Table 14 present results of regressions on sub-samples based on the characteristics of the head of the household. We consider gender, education, and work type. Del Carpio and Wagner (2015) found the impact of SUTPs varies across different groups of the labor force. It is possible that some households are more susceptible or vulnerable than others in managing externalities from large in-flows of SUTPs. However, results from these regressions continue to show that across various groups, the proportion of foreign-born did not have a significant effect on poverty. A third reason is substitution of labor yielded no net effects at the household level. Not all impacts are negative to the host community. As Del Carpio and Wagner (2015) point out, the influx of refugees also generated better paying jobs that were filled by local Turkish workers. It is possible that at the household level, the net labor market effects from the influx of refugees remained insignificant as some low skilled members lost jobs but then better jobs became available. Summary statistics show that from 2009 to 2013, the share of employed adults in a host community household near the Syrian border actually increased from 24 to 27 percent, and the share of unemployed adults dropped from 5 to 4 percent (Table 10). The share of host community heads of household working in Agriculture, Forestry, Fishing, and Mining also significantly declined near the Syrian border from 2009 to 2013. In 2009, 16 percent of heads of household in this group worked in this sector, and declined to 12 percent in 2013. The share of workers in white collar jobs also significantly increased and blue collar jobs declined during this period. This would result in insignificant impacts from an increasing foreign-born population at the household level. In the shorter time period, from 2011 to 2013, the characteristics of host communities in near the Syrian border did not regress (Table 16). For example, the proportion of workers in a household working white collar jobs increased from 30.9 percent in 2011 to 34.4 percent in 2013. The share of workers registered with a social security institution also increased from 46.1 to 52.4 percent. One caveat is that over this period, Turkey experienced healthy growth, and direct impact of these effects whether through displacement effects or increased growth due to the influx of Syrians is not completely obvious. For example, in the rest of the country, which is a less popular destination for migrants, the share of adults employed also increased from 32 to 37 percent from 2009 to 2013. In other words, improved conditions of host community families could be due to a period of healthy growth. Would host community families have stopped working in agriculture anyway even if there were no Syrians or did Syrians start taking these jobs? Whether or not host communities would be better off is not clear, but what can be said is that the poverty rate has not gone up. While poverty rates in 2014 are not available, a comparison of the host community characteristics in 2013 and 2014 may give insights to trends. Looking at a more compressed period from 2013 to 2014, there is some evidence that host community families changing in their labor market characteristics. For example, the proportion of workers with a permanent contract type declined from 36.2 percent in 2013 to 35.1 percent 13 Household size is not reported for the entire household, only the 15+ segment. Dependency ratios and per capita income cannot be computed using the 2014 LFS. 16 in 2014. The share of workers registered with social security in host community households declined but the change is not significant. In terms of the number of hours worked, and the proportion of workers in blue or white collar jobs, there is no statistical difference between 2013 and 2014. 4. CONCLUSION The movement of Syrian refugees is one of the largest passages of refugee populations in recent history. With millions of people leaving Syria and settling in Turkey, concerns about externalities onto the native population are very salient. This paper addressed the poverty impacts of SUTPs on the host communities and found no evidence that the increase in foreign-born population from 2011 to 2013 resulted in higher poverty rates among the host community. As recent literature has noted, the SUTPs have both positive and negative impacts. While some types of people may be more likely to be displaced by Syrians in the labor market, Syrians are consumers and renters, they also open businesses and create jobs. Local Turkish citizens have also benefited as employers and sellers. In some border cities, the balance of trade has improved as exports to the Middle East increased. On the other hand, analysis of only the recent migrants clearly demonstrates that the group’s poverty profile is worsening between 2009 and 2013. Moreover, descriptive characteristics suggest that their conditions might have worsened in 2014. As this unprecedented event continues, the integration of Syrians into the Turkish labor market, access to public services, changing demographics, and socioeconomic impacts should be monitored closely. Especially with an increasing rate of SUTP migration to Turkey during 2014 and 2015 and the continued conflict in the region, the inflow of Syrians will be one of the most critical short, medium, and possibly long term policy issues in the country. In addition, Turkey’s role as a pathway to Europe for those escaping conflict in the Middle East makes the issue an international phenomenon. In this respect, the healthy incorporation of SUTPs that will protect the wellbeing of host communities while satisfying the humanitarian necessity of helping Syrians will be among the more important development issues of today and the foreseeable future. 17 REFERENCE AFAD (2013). “Syrian Refugees in Turkey, 2013. Field Survey Results”, Republic of Turkey, Prime Ministry, Disaster and Emergency Management Presidency. Ceritoglu, Evren, H., Burcu Gurcihan Yunculer, Huzeyfe Torun, and Semih Tumen. (2015). “The Impact of Syrian Refugees on Natives’ Labor Market Outcomes in Turkey: Evidence from a Quasi-Experimental Design”, MRPA Paper No. 61503, posted January 23, 2015. Del Carpio, Ximena V. and Mathis Wagner. (2015). “The Impact of Syrians Refugees on the Turkish Labor Market”, World Bank Working Paper 7402, August 2015. Del Carpio, Ximena V., Doreen Triebe, and Mathis Wagner. (2015). “Short-term Impacts of Syrians Refugees on the Turkish Labor Market”, Working Paper, May 2015. Erdogan, M. Murat. (2014). “Syrians in Turkey: Social Acceptance and Integration Research”, Executive Summary and Report, November 2014. Kirisci, Kemal. (2003). “Turkey: A Transformation from Emigration to Immigration” November 1, 2003. (http://www.migrationpolicy.org/article/turkey-transformation-emigration-immigration) Orhan Oytun, and Sabiha Senyucel Gundogar (2015). “Effects of the Syrian Refugees on Turkey”, ORSAM Report No 195, January 2015. Ozturkler, Harun, and Turkmen Goksel. (2015). “The Economic Effects of Syrian Refugees on Turkey: A Synthetic Modelling”, ORSAM Report No: 196, January 2015. UNHCR (2013). “Countries Hosting Syrian Refugees, Solidarity and Burden-Sharing, Background papers for the High Level Segment”. Provisional Release, September 2013. UNHCR, Turkey Syrian Refugee Daily Sitrep, (22 March 2013) UNHCR, Turkey Syrian Refugee Daily Sitrep, (22 November 2013) UNHCR, Turkey Syrian Refugee Daily Sitrep, (15 September 2014) 18 FIGURES Figure 2. SuTPs Living Outside Camps vs Recent Migrants in the LFS 1400000 90000 GOV 80000 1200000 UNHCR 70000 1000000 UNHCR (registered) 60000 800000 50000 LFS-Near Syrian Border 40000 600000 LFS-National 30000 400000 LFS-Other Regions 20000 200000 10000 0 0 12/31/2011 12/31/2012 12/31/2013 Sources: AFAD 2013, MOI, UNHCR, Turkey LFS 19 TABLES Table 6. Population, by Comparison Groups Host Settled Migrant HHs Recent Migrant HHs Community (>4 years in Turkey) (<=4 years in Turkey) 2009 Near the Syrian Border 13,645,091 32,177 1,368 Population Rest of the Country 55,441,281 956,218 27,345 2013 Near the Syrian Border 14,442,691 18,099 29,846 Population Rest of the Country 57,129,706 952,052 50,989 2014 Near the Syrian Border 9,942,906* 28,272* 99,243* Population Rest of the Country 45,700,589* 1,004,532* 210,658* Notes: *The population in 2014 is only of the population aged 15+. In 2014, the population weights exclude individuals under 15 due to changes in the LFS. Source: LFS 2009-2014. World Bank author’s calculations. Table 7. Population Shares, by Comparison Group Settled Migrant Recent Migrant HHs Host HHs (>4 years in (<=4 years in Community Turkey) Turkey) 2009 Population Near the Syrian Border 19.8% 3.3% 4.8% Share Rest of the Country 80.2% 96.7% 95.2% 2013 Population Near the Syrian Border 20.2% 1.9% 36.9% Share Rest of the Country 79.8% 98.1% 63.1% 2014 Population Near the Syrian Border 17.9% 2.7% 32.0% Share Rest of the Country 82.1% 97.3% 68.0% Notes: In 2014, the population weights exclude individuals under 15 due to changes in the LFS Source: LFS 2009-2014. World Bank author’s calculations. Table 8. Households, by Comparison Group Settled Migrant Recent Migrant HHs Host HHs (>4 years in (<=4 years in Community Turkey) Turkey) 2009 Near the Syrian Border 3,046,843 9,744 399 Population Rest of the Country 15,623,820 315,158 10,433 2013 Near the Syrian Border 3,384,455 6,439 5,064 Population Rest of the Country 16,947,974 349,414 14,678 2014 Near the Syrian Border 3,176,654 10,917 18,575 Population Rest of the Country 17,047,500 417,846 64,973 Notes: In 2014, the population weights exclude individuals under 15 due to changes in the LFS. However, this should not affect estimates of the number of households. Source: LFS 2009-2014. World Bank author’s calculations. 20 Table 9. Household Shares, by Comparison Groups Settled Migrant Recent Migrant HHs Host HHs (>4 years in (<=4 years in Community Turkey) Turkey) 2009 Population Near the Syrian Border 16.3% 3.0% 3.7% Share Rest of the Country 83.7% 97.0% 96.3% 2013 Population Near the Syrian Border 16.6% 1.8% 25.7% Share Rest of the Country 83.4% 98.2% 74.3% 2014 Population Near the Syrian Border 15.7% 2.5% 22.2% Share Rest of the Country 84.3% 97.5% 77.8% Notes: In 2014, the population weights exclude individuals under 15 due to changes in the LFS. However, this should not affect estimates of the number of households. Source: LFS 2009-2014. World Bank author’s calculations. 21 Figure 3. Poverty Estimates 2009-2013 Near the Syrian Border Host communities Near the Syrian Border 80 70 60 Poverty Rate (%) 50 44.2 42.6 40 40.9 41.1 40.3 30 20 10 0 2009 2010 2011 2012 2013 Settled Migrant Households Near the Syrian Border (>4 years in Turkey) 80 70 60 Poverty Rate (%) 50 40 30 27.5 23.8 21.3 20 17.7 10 12.5 0 2009 2010 2011 2012 2013 Recent Migrant Households Near the Syrian Border (<=4 years in Turkey) 80 70 60 Poverty Rate (%) 50 46.7 40 37.0 30 27.1 20 15.6 17.5 10 0 2009 2010 2011 2012 2013 Notes: Regionally deflated, OLS imputation model, with time-invariant characteristics. Vertical line represents standard error. Poverty line is $5/day PPP per capita. The Near Syrian Border region (NSB) includes the NUTS2 areas TRC3, TCR2, TCR1, TR63, TR62. Time in Turkey is based on the date of arrival of the head of household. Results displayed at the 95% Confidence Interval. Source: LFS 2009-2014. World Bank author’s calculations. 22 Figure 4. Poverty Estimates 2009-2013 Rest of the Country Host communities 80 70 60 Poverty Rate (%) 50 40 30 20 21.1 20.8 19.3 18.4 17.9 10 0 2009 2010 2011 2012 2013 Settled Migrant Households in the Rest of the Country (>4 years in Turkey) 80 70 60 Poverty Rate (%) 50 40 30 20 10 9.1 10.7 9.3 7.0 6.6 0 2009 2010 2011 2012 2013 Recent Migrant Households (Rest of the Country, <=4 years in Turkey) 80 70 60 Poverty Rate (%) 50 40 30 20 13.3 16.1 10 13.2 7.3 5.2 0 -10 2009 2010 2011 2012 2013 Notes: Regionally deflated, OLS imputation model, with time-invariant characteristics. Vertical line represents standard error. Poverty line is $5/day PPP per capita. The Near Syrian Border region (NSB) includes the NUTS2 areas TRC3, TCR2, TCR1, TR63, TR62. Time in Turkey is based on the date of arrival of the head of household. Results displayed at the 95% Confidence Interval. Source: LFS 2009-2014. World Bank author’s calculations. 23 Figure 5. Host Community Poverty Trends, by Geographic Region and Group All Host Community Households Rest of the Country Near the Syrian Border 80% 80% 70% 70% Poverty Rate (%) Poverty Rate (%) 60% 60% 50% 50% 44% 43% 41% 41% 40% 40% 40% 30% 30% 20% 21% 21% 19% 18% 18% 20% 10% 10% 0% 0% 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 Low Educated Households (Head has no or primary level education) Rest of the Country Near the Syrian Border 80% 80% 70% 70% Poverty Rate (%) Poverty Rate (%) 60% 60% 50% 50% 51.4% 48.3% 48.3% 48.4% 47.0% 40% 40% 30% 29.6% 28.7% 30% 26.2% 25.1% 25.1% 20% 20% 10% 10% 0% 0% 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 Female Headed Households Rest of the Country Near the Syrian Border 80% 80% 70% 70% Poverty Rate (%) Poverty Rate (%) 60% 60% 50% 50% 40% 40% 40% 37% 37% 36% 36% 30% 30% 20% 20% 16% 17% 14% 10% 13% 12% 10% 0% 0% 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 24 Male Headed Households Rest of the Country Near the Syrian Border 80% 80% 70% 70% Poverty Rate (%) Poverty Rate (%) 60% 60% 50% 50% 45% 43% 41% 42% 40% 40% 41% 30% 30% 20% 22% 21% 20% 19% 19% 20% 10% 10% 0% 0% 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 Head of Households: Blue Collar Rest of the Country Near the Syrian Border 80% 80% 70% 70% Poverty Rate (%) Poverty Rate (%) 60% 60% 50% 50% 50% 49% 48% 48% 47% 40% 40% 30% 28% 28% 30% 27% 26% 25% 20% 20% 10% 10% 0% 0% 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 Notes: OLS imputation method, regionally deflated. Results displayed at the 95% Confidence Interval. Source: LFS 2009-2013. World Bank author’s calculations. 25 Table 10. Descriptive Statistics, Near the Syrian Border (2009, 2013, 2014) 2009 2013 2014 Near the Syrian Border Near the Syrian Border Near the Syrian Border Host Settled Migrant HHs Recent Host Settled Recent Migrant Host Settled Recent Community (>4 years in Turkey) Migrant HHs Community Migrant HHs HHs Community Migrant HHs Migrant HHs (<=4 years in (>4 years in (<=4 years in (>4 years in (<=4 years in Turkey) Turkey) Turkey) Turkey) Turkey) Sample of Households 21,324 68 3 21,467 51 34 22,791 71 88 Households 3,046,842 9,744 398,844 3,384,455 6,439 5,063 3,176,654 10,917 18,575 Population* 13,645,090 32,177,378 1,367,570 14,442,691 18,099 29,846 9,942,906 28,272 99,243 N children 1.6 1.1 0.6 1.4 0.8 2.4 - - - N adults 2.4 1.4 2.6 2.3 1.4 3.1 2.5 2.3 4.7 N old 0.5 0.8 0.2 0.5 0.5 0.4 0.6 0.3 0.7 0-14 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 15-24 2.3% 0.0% 40.0% 1.6% 0.0% 0.0% 1.1% 1.0% 2.3% 25-39 36.2% 38.8% 0.0% 31.6% 34.9% 53.1% 29.0% 44.7% 29.8% 40-54 34.0% 6.4% 37.0% 35.6% 29.9% 37.7% 35.8% 30.5% 33.7% 55+ 27.5% 54.8% 23.0% 31.2% 35.2% 9.1% 34.1% 23.8% 34.2% 1-Male, 0-Female 84.9% 67.1% 100.0% 83.9% 54.4% 87.9% 86.1% 80.1% 82.7% None 25.8% 37.1% 60.0% 22.9% 19.4% 75.6% 22.1% 21.6% 51.3% Primary 44.9% 26.0% 0.0% 42.4% 29.9% 8.2% 43.8% 20.0% 23.5% Secondary 22.4% 31.4% 40.0% 24.7% 30.3% 16.2% 24.1% 24.6% 20.7% Post-Secondary 6.9% 5.5% 0.0% 9.9% 20.4% 0.0% 10.2% 35.6% 4.6% Share of Employed Adults 22.5% 26.7% 0.0% 25.0% 31.8% 20.6% 35.9% 47.5% 23.8% Share of Unemployed Adults 3.7% 2.1% 25.0% 2.3% 3.4% 4.6% 3.7% 1.2% 7.7% Total Hours Worked in the Whole Family 52.96 51.40 51.50 55.04 58.77 51.58 57.95 47.77 Mean Hours Worked out of all working individuals 52.09 52.62 50.93 50.46 50.41 58.81 50.88 56.21 51.98 Share Workers White Collar 29.1% 44.2% 33.4% 40.0% 40.8% 32.9% 51.3% 20.9% Share Workers Blue Collar 53.6% 38.1% 47.1% 35.1% 50.8% 44.7% 24.3% 48.4% Share with Permanent Contract Type 35.1% 33.3% 0.0% 36.4% 34.6% 33.8% 33.6% 45.6% 11.8% Regular or casual employee 60.3% 44.9% 65.3% 70.9% 80.5% 65.0% 79.1% 81.1% Employer 7.1% 9.4% 6.2% 15.6% 7.4% 6.5% 13.4% 6.1% Self employed 32.3% 43.0% 28.4% 13.6% 12.2% 28.2% 6.2% 12.8% Unpaid family worker 0.3% 2.7% 0.2% 0.0% 0.0% 0.4% 1.3% 0.0% 1-9 employees 51.8% 38.6% 45.3% 40.9% 76.3% 44.2% 49.6% 56.4% 10-49 employees 14.7% 15.9% 17.4% 15.2% 15.3% 14.1% 4.3% 14.9% 50+ employees 16.2% 27.8% 17.7% 19.0% 0.0% 19.2% 22.1% 0.0% Share Agriculture, forestry, fishing-Mining and quarrying 16.6% 8.2% 13.6% 0.0% 5.1% 12.1% 1.0% 15.3% Share Manufacturing Industry 13.8% 25.8% 12.7% 27.6% 21.8% 13.7% 4.6% 15.0% Share Electricity, Gas , Steam, Water and Sewerage 0.7% 0.0% 1.0% 0.0% 0.0% 1.1% 0.0% 0.0% Share Construction 8.2% 5.9% 10.0% 0.0% 18.6% 8.4% 7.7% 18.1% Share Commerce and services 25.1% 25.3% 22.7% 22.6% 35.0% 22.4% 30.7% 17.7% Share Others 18.4% 17.2% 20.4% 24.9% 11.1% 20.0% 31.6% 3.2% Registered with Any SS institution related to your main job 50.2% 56.5% 62.1% 64.6% 20.7% 63.4% 82.9% 2.3% Share of workers registered with any social security institution 43.0% 56.5% 51.4% 50.1% 20.7% 51.3% 61.1% 2.3% related to your main job Notes: Population estimates in 2014 are of the 15+ population. Source: LFS 2009-2014. World Bank author’s calculations. 26 Table 11. Descriptive Statistics, Rest of the County (2009, 2013, 2014) 2009 2013 2014 Rest of the Country Rest of the Country Rest of the Country Host Settled Recent Host Community Settled Migrant Recent Host Settled Migrant Recent Community Migrant HHs Migrant HHs HHs (>4 years Migrant HHs Community HHs (>4 years in Migrant HHs (>4 years in (<=4 years in in Turkey) (<=4 years in Turkey) (<=4 years in Turkey) Turkey) Turkey) Turkey) Sample of Households 112,386 2,059 51 122,105 2,312 86 124,384 2,452 271 Households 15,623,820 315,158 10,432 16,947,974 349,414 14,678 17,047,500 417,846 64,974 Population* 55,441,280 956,217 27,345 57,129,706 952,052 50,989 45,700,589 1,004,532 210,658 N children 0.9 0.5 0.6 0.8 0.4 0.9 - - - N adults 2.1 1.8 1.8 2.0 1.4 2.4 2.1 1.6 3.1 N old 0.5 0.8 0.2 0.6 0.8 0.2 0.6 0.8 0.2 0-14 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 15-24 1.9% 0.7% 17.3% 1.6% 0.5% 10.4% 1.5% 0.7% 5.1% 25-39 32.0% 25.9% 57.6% 30.0% 24.2% 55.4% 27.9% 22.8% 50.8% 40-54 35.4% 29.8% 15.8% 35.2% 26.0% 19.0% 35.3% 27.0% 36.1% 55+ 30.7% 43.6% 9.4% 33.1% 49.4% 15.2% 35.3% 49.6% 7.9% 1-Male, 0-Female 86.0% 80.3% 83.7% 84.6% 76.1% 84.9% 84.3% 76.3% 81.8% None 12.2% 10.4% 2.1% 12.2% 10.5% 5.4% 12.0% 11.5% 20.1% Primary 47.4% 25.5% 7.3% 43.9% 22.8% 12.5% 44.0% 21.7% 21.2% Secondary 27.6% 48.3% 63.8% 28.5% 49.4% 55.2% 28.7% 46.7% 40.8% Post-Secondary 12.8% 15.7% 26.8% 15.3% 17.2% 26.8% 15.8% 22.1% 20.4% Share of Employed Adults 28.7% 32.0% 39.9% 31.6% 35.0% 35.5% 41.4% 44.1% 37.3% Share of Unemployed Adults 2.8% 3.5% 10.3% 1.9% 4.0% 8.9% 2.5% 3.4% 8.7% Total Hours Worked in the Whole Family 51.69 50.59 49.45 50.40 49.26 54.73 50.03 48.68 51.59 Mean Hours Worked out of all working individuals 50.30 49.57 49.25 48.80 48.01 52.32 48.60 47.79 51.52 Share Workers White Collar 34.9% 42.0% 55.7% 35.6% 45.6% 28.2% 35.6% 43.9% 24.5% Share Workers Blue Collar 43.0% 34.4% 20.9% 39.4% 28.0% 48.1% 37.5% 29.2% 49.5% Share with Permanent Contract Type 39.5% 40.1% 38.6% 40.0% 39.9% 49.0% 38.4% 41.1% 40.6% Regular or casual employee 61.3% 75.7% 69.0% 64.5% 74.9% 89.3% 65.8% 78.7% 89.9% Employer 9.1% 8.5% 7.2% 7.5% 9.4% 0.9% 7.6% 7.4% 1.3% Self employed 29.2% 15.5% 13.3% 27.5% 15.0% 6.8% 25.9% 13.2% 8.7% Unpaid family worker 0.4% 0.4% 10.5% 0.5% 0.6% 3.0% 0.7% 0.7% 0.0% 1-9 employees 41.2% 30.2% 48.5% 37.8% 28.2% 30.8% 38.0% 28.1% 39.3% 10-49 employees 16.3% 18.1% 18.0% 16.7% 18.7% 29.4% 13.9% 16.7% 22.9% 50+ employees 20.5% 28.1% 10.1% 20.5% 26.7% 16.1% 21.1% 28.3% 11.5% Share Agriculture, forestry, fishing-Mining and quarrying 11.0% 2.8% 5.7% 11.0% 2.2% 0.0% 9.4% 2.1% 0.7% Share Manufacturing Industry 16.0% 24.6% 1.6% 15.4% 21.5% 45.0% 15.1% 21.0% 32.7% Share Electricity, Gas , Steam, Water and Sewerage 0.6% 0.4% 0.0% 0.9% 0.5% 0.0% 0.9% 0.4% 0.8% Share Construction 6.3% 4.2% 11.5% 6.3% 5.2% 1.7% 6.6% 6.2% 11.2% Share Commerce and services 25.5% 25.4% 45.4% 23.3% 24.4% 13.8% 22.7% 21.3% 9.9% Share Others 18.5% 18.9% 12.5% 18.2% 20.0% 15.8% 18.3% 22.0% 18.6% Registered with Any SS institution related to your main job 67.4% 76.4% 25.0% 73.9% 83.9% 48.5% 75.2% 83.4% 25.5% Share of workers registered with any social security institution related to 54.7% 58.9% 20.0% 57.5% 62.2% 37.1% 57.1% 62.2% 20.7% your main job Notes: Population estimates in 2014 are of the 15+ population. Source: LFS 2009-2014. World Bank author’s calculations. 27 Table 12. Proportion of SUTPs by NUTS2 regions in 2013 Percent of SUTPs in the total population TR10 Istanbul 2.8% TR21 Tekirdag 0.0% TR22 Balikesir 0.0% TR31 Izmir 0.6% TR32 Aydin 0.1% TR33 Manisa 0.0% TR41 Bursa 0.5% TR42 Kocaeli 0.5% TR51 Ankara 0.6% TR52 Konya 3.5% TR61 Antalya 0.4% TR62 Adana 4.9% TR63 Hatay 16.9% TR71 Kirikale 0.1% TR72 Kayseri 0.8% TR81 Zonguldak 0.0% TR82 Kastamonu 0.0% TR83 Samsun 0.1% TR90 Trabzon 0.0% TRA1 Erzurum 0.0% TRA2 Agri 0.0% TRB1 Malatya 0.1% TRB2 Van 0.1% TRC1 Gaziantep 22.4% TRC2 Sanliurfa 11.0% TRC3 Mardin 11.3% Sources: AFAD 2013, MOI, UNHCR 28 Table 13. OLS Blue Collar Households Households with Low Education (1) (2) (3) (1) (2) (3) Proportion of Foreign-Born in the Population 1.123* -0.128 -0.0154 1.098* -0.108 0.0367 (0.512) (0.0857) (0.0763) (0.517) (0.0890) (0.0707) Constant 0.279*** 0.190*** -0.724 0.258*** 0.161*** 0.238 (0.0447) (0.00121) (0.553) (0.0465) (0.00126) (0.459) Observations 52 52 52 52 52 52 R-squared 0.095 0.959 0.995 0.086 0.959 0.994 Year Controls X X X X Regional Controls X X X X Labor and Demographic Controls X X Notes: * p<0.10, ** p<0.05, *** p<0.01. Source: LFS 2011 and 2013. World Bank author’s calculations. Table 14. OLS Female Headed Households Male Headed Households (1) (2) (3) (1) (2) (3) Proportion of Foreign-Born in the Population 0.898** 0.00125 0.0348 0.969** -0.107 -0.0393 (0.325) (0.209) (0.0653) (0.436) (0.112) (0.0675) Constant 0.153*** 0.0520*** 0.555** 0.211*** 0.119*** 0.601** (0.0352) (0.00296) (0.247) (0.0388) (0.00159) (0.211) Observations 52 52 52 52 52 52 R-squared 0.095 0.910 0.973 0.092 0.948 0.993 Year Controls X X X X Regional Controls X X X X Labor and Demographic Controls X X Notes: * p<0.10, ** p<0.05, *** p<0.01. Source: LFS 2011 and 2013. World Bank author’s calculations. 29 Table 15. Host Community Near the Syrian Border, 2009 & 2013 Means Comparison Test mean(2013)- two-sided 2013- 2009- mean(2009) means test p- mean mean value N children -0.377 0.0000 1.28 1.66 N adults -0.176 0.0000 2.24 2.42 N old 0.060 0.0000 0.58 0.52 Characteristics of the head of household 0-14 0.0% . - - 15-24 -0.6% 0.0000 1.5% 2.1% 25-39 -5.6% 0.0000 27.6% 33.2% 40-54 0.9% 0.0491 35.4% 34.5% 55+ 5.3% 0.0000 35.5% 30.2% 1-Male, 0-Female -1.5% 0.0001 82.5% 84.0% None -4.7% 0.0000 23.6% 28.4% Primary -2.1% 0.0000 41.6% 43.7% Secondary 3.2% 0.0000 24.5% 21.4% Post-Secondary 3.7% 0.0000 10.3% 6.6% Al workers in the household Share of Employed Adults 3.4% 0.0000 25.5% 22.2% Share of Unemployed Adults -1.2% 0.0000 2.4% 3.6% Total Hours Worked in the Whole Family -1.62 0.0000 51.31 52.93 Mean Hours Worked out of all working individuals -1.71 0.0000 50.45 52.16 Share Workers White Collar 5.6% 0.0000 34.4% 28.8% Share Workers Blue Collar -7.8% 0.0000 45.9% 53.7% Share with Permanent Contract Type 2.9% 0.0000 36.2% 33.3% 1-9 employees -8.8% 0.0000 44.3% 53.1% 10-49 employees 3.1% 0.0000 17.5% 14.4% 50+ employees 3.5% 0.0000 18.5% 15.1% Share Agriculture, forestry, fishing-Mining and -5.9% 0.0000 12.2% 18.0% quarrying Share Manufacturing Industry 1.1% 0.0109 13.2% 12.1% Share Electricity, Gas , Steam, Water and Sewerage 0.2% 0.0470 1.0% 0.8% Share Construction 1.5% 0.0000 9.7% 8.2% Share Commerce and services -1.3% 0.0126 23.6% 24.9% Share Others 2.2% 0.0000 20.8% 18.5% Registered with Any SS institution related to your 15.7% 0.0000 63.5% 47.8% main job Share of workers registered with any social security 11.6% 0.0000 52.4% 40.8% institution related to your main job Notes: 2009 N = 21,324, 2013 N = 21, 467 Source: LFS 2009 & 2013. World Bank author’s calculations. 30 Table 16. Host Community Near the Syrian Border, 2011 & 2013 Means Comparison Test mean(2013)- two-sided 2013- 2011- mean(2011) means test mean mean p-value N children -0.097 0.0000 1.28 1.38 N adults -0.084 0.0000 2.24 2.33 N old 0.021 0.0048 0.58 0.56 Characteristics of the head of household 0-14 0.000 . - - 15-24 -0.002 0.0434 0.01 0.02 25-39 -0.014 0.0009 0.28 0.29 40-54 0.001 0.7810 0.35 0.35 55+ 0.016 0.0007 0.36 0.34 1-Male, 0-Female -0.009 0.0123 0.83 0.83 None -0.002 0.6283 0.24 0.24 Primary -1.2% 0.0124 41.6% 42.7% Secondary 0.4% 0.3095 24.5% 24.1% Post-Secondary 1.0% 0.0007 10.3% 9.4% Al workers in the household Share of Employed Adults 1.0% 0.0000 25.5% 24.6% Share of Unemployed Adults 0.1% 0.3775 2.4% 2.3% Total Hours Worked in the Whole Family -0.526 0.0085 51.31 51.83 Mean Hours Worked out of all working individuals -0.542 0.0014 50.45 50.99 Share Workers White Collar 3.5% 0.0000 34.4% 30.9% Share Workers Blue Collar -3.2% 0.0000 45.9% 49.1% Share with Permanent Contract Type -0.1% 0.7832 36.2% 36.4% 1-9 employees -2.6% 0.0000 44.3% 46.9% 10-49 employees 1.3% 0.0048 17.5% 16.3% 50+ employees 1.6% 0.0004 18.5% 16.9% Share Agriculture, forestry, fishing-Mining and quarrying -1.4% 0.0002 12.2% 13.6% Share Manufacturing Industry 1.3% 0.0009 13.2% 11.9% Share Electricity, Gas , Steam, Water and Sewerage -0.1% 0.3614 1.0% 1.1% Share Construction 1.2% 0.0005 9.7% 8.4% Share Commerce and services -1.4% 0.0055 23.6% 25.0% Share Others 0.7% 0.1739 20.8% 20.1% Registered with Any SS institution related to your main job 8.0% 0.0000 63.5% 55.5% Share of workers registered with any social security 6.3% 0.0000 52.4% 46.1% institution related to your main job Notes: 2011 N = 21,460, 2013 N = 21, 467 Source: LFS 2011 & 2013. World Bank author’s calculations. 31 Table 17. Host Community Near the Syrian Border, 2013 & 2014 Means Comparison Test mean(2014)- two-sided 2014- 2013- mean(2013) means test p- mean mean value N children* -1.284 - - 1.28 N adults 0.050 0.0006 2.29 2.24 N old 0.024 0.0017 0.60 0.58 Characteristics of the head of household 0-14 0.0% . 0.0% 0.0% 15-24 -0.1% 0.3616 1.4% 1.5% 25-39 -0.3% 0.5363 27.4% 27.6% 40-54 0.7% 0.1266 36.1% 35.4% 55+ -0.3% 0.4658 35.2% 35.5% 1-Male, 0-Female 2.4% 0.0000 85.0% 82.5% None -0.3% 0.4101 23.3% 23.6% Primary 1.3% 0.0054 42.9% 41.6% Secondary -0.6% 0.1661 23.9% 24.5% Post-Secondary -0.2% 0.4396 10.1% 10.3% Al workers in the household Share of Employed Adults 11.0% - 36.5% 25.5% Share of Unemployed Adults 1.5% 0.0000 3.8% 2.4% Total Hours Worked in the Whole Family 0.190 0.3610 51.50 51.31 Mean Hours Worked out of all working individuals 0.251 0.1554 50.70 50.45 Share Workers White Collar -0.7% 0.2275 33.8% 34.4% Share Workers Blue Collar -0.1% 0.8321 45.8% 45.9% Share with Permanent Contract Type -1.1% 0.0273 35.1% 36.2% 1-9 employees 1.6% 0.0042 45.9% 44.3% 10-49 employees -3.4% 0.0000 14.2% 17.5% 50+ employees 0.9% 0.0512 19.4% 18.5% Share Agriculture, forestry, fishing-Mining and quarrying 1.4% 0.0002 13.6% 12.2% Share Manufacturing Industry -0.2% 0.6378 13.0% 13.2% Share Electricity, Gas , Steam, Water and Sewerage 0.1% 0.4251 1.1% 1.0% Share Construction -1.2% 0.0005 8.5% 9.7% Share Commerce and services -1.3% 0.0113 22.3% 23.6% Share Others 0.4% 0.4353 21.1% 20.8% Registered with Any SS institution related to your main job -1.1% 0.0785 62.5% 63.5% Share of workers registered with any social security -0.8% 0.1496 51.6% 52.4% institution related to your main job Notes: *In 2014, the LFS no longer captures information about children 14 and under. N2013 = 21,467 for most variables, N2014=22,791 for most variables. Variables are at the household level. Source: LFS 2013 & 2014. World Bank author’s calculations. 32 ANNEX. SURVEY-TO SURVEY IMPUTATION METHODOLOGY 1. Methodology The estimates of poverty and welfare status of the households rely on survey data with a complex consumption module containing large set of detailed questions on prices and quantities consumed. Given its complexity the collection and analysis of that data involves significant investment of time, money and analytical efforts. On the other hand, there is a need for timely poverty estimates for evidence-based policies in the face of the high cost of fielding comprehensive surveys to track income and/or expenditure has led to the development of a variety of approaches for estimating poverty in the absence of consumption expenditure or income data. In this section, we explore the survey-to-survey approach, which used the common observed assets and household characteristics in order to impute a proxy for welfare. The survey-to-survey imputation requires that at least one previous comparable survey contains household-level income or consumption information. The method draws upon the imputation literature (see Brick and Kalton, 1996 for a discussion of various techniques), which utilizes non-missing data in a larger data set to predict the values for missing variables, and from the poverty targeting literature (see Grosh and Baker, 1995) which seeks proxies for poverty status from household characteristics. The survey-to-survey imputation method extends that to using common variables in two data sets, only one of which contains consumption to predict consumption values in the second data set. A common application of this method in poverty analysis is “poverty mapping,” which uses consumption and poverty estimates from a household survey imputed into census data to achieve very fine levels of geographic disaggregation. (see Rao, 2003, for a discussion of the general technique of small area estimation, and Elbers et al., 2002, specifically related to poverty mapping.) For examples of survey- to-survey imputation for poverty analysis, see Stifel and Christiaensen (2007), Tarozzi (2007), Grosse et al (2009), and Douidich et al (2013). A more formal presentation of the S2S model is as follows. There are surveys: in Survey 1 there are information on the income or consumption as well as the set of household characteristics . In the Survey 2, the same set of household characteristics are also observed, and those characteristics are comparable between the two surveys. The (log) of per capita household income or consumption is modeled for the first survey as: ′ where is the per capita income or consumption of household h residing in area c, are household and area/location characteristics, and is the residual composed of the area component and the household component , which have zero expectations and are independent of each other. In the second survey where there is no income or consumption information, a set of common variables will be use to impute the income or consumption for each household in the second survey using the estimated points and their distributions estimated from the survey 1. Since each estimated point is fluctuated from an assumed normal distribution with its standard errors, the imputation is done through a number of simulation in order to preserve the error structure of the correlates. 33 Any statistics on the imputed welfare will based on the set of imputed welfares for each household. The estimator takes the form, with R denotes the number of simulation: 1 where is a function that converts the vector y with (log) incomes for all households into a poverty measure (such as the head-count rate or bottom 40%), and where denotes the r-th simulated imputed welfare. Figure 6. Survey-to-Survey Imputation Methodology, an illustration For the case of Turkey, we use the Survey on Income and Living Conditions survey to impute to the Labor Force Survey. Income is used instead of consumption for this paper’s analysis. The model included variables related to: household demographics (age, gender, age composition, etc.), household characteristics (education, labor activity, etc.), household head’s characteristics (age, gender, labor, education, marital status, etc.) and household assets holding (both livestock and durables). Based on that model the simulated values of consumption (at household level) were imputed for the households in the corruption survey. This allowed for consistent ranking the households into welfare quintiles and cross- tabulation of welfare status with household characteristics and indicators derived from the survey data. The imputation was carried out using s2sc algorithm in STATA. 34 Inputs: 1. Household Survey with consumption or income welfare aggregates 2. Project data/Other survey data without welfare aggregates 3. Set of harmonized common variables in both surveys Outputs: 1. Set of imputed welfare variables for project data/other survey for each household in the data 2. Imputed welfare variables can be used for poverty, distributional analysis (quintiles or more), profiling of the poor or group of interest Models: 1. Ordinary Least Squares (OLS) 2. Probit 3. Multiple Imputation (MI) Table 18. Model Specification Variables Demographic Share of children, share of adults, share of adults squared and share of old (omitted) Characteristics of head Age, gender, and level of education Interactions with urban dummy variable Level of education of the head, age of the head Geography Dummies for regions at NUTS 1 level (12 regions) Interactions with Geography Level of education of the head, age of the head interacted with regions at NUTS 1 level (12 regions) and urban-rural division 1. Validation and Robustness Check Figure 7. External Validation, NUTS1 Level $5/day PPP– Observed (SILC) & Imputed (LFS), 2007 LFS SILC 35 Source: LFS, SILC, imputed OLS, regionally deflated, authors’’ calculations Table 19. Robustness Checking 2009 2013 2009 2013 2009 2013 Model Probit OLS OLS Specification Mean Time Invariant – Not Regionally Deflated Time Invariant – Regionally Deflated NSB - Host community 45.98 42.76 45.86 41.89 44.16 40.30 NSB - Recent migrants 17.81 49.01 16.44 47.82 15.63 46.70 NSB -Settled migrants 28.33 19.13 28.25 18.60 27.50 17.74 Rest - Host community 20.17 17.37 20.01 17.08 21.06 17.90 Rest - Recent migrants 6.06 13.23 5.88 13.23 7.30 16.13 Rest - Settled migrants 8.37 5.98 8.19 5.83 9.14 6.56 Figure 8. Robustness Checking Near Syrian Border - Near Syrian Border - Near Syrian Border - Rest of the country - Rest of the country - Rest of the country - Turkey Host community Recent migrants Settled migrants Host community Recent migrants Settled migrants 50 45 40 35 Poverty Rate ($5 PPP, %) 30 25 20 15 10 5 0 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 Type Probit - Not Regionally Deflated OLS - Not Regionally Deflated OLS - Regionally Deflated 36 Poverty & Equity Global Practice Working Papers (Since July 2014) The Poverty & Equity Global Practice Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. This series is co‐published with the World Bank Policy Research Working Papers (DECOS). It is part of a larger effort by the World Bank to provide open access to its research and contribute to development policy discussions around the world. For the latest paper, visit our GP’s intranet at http://POVERTY. 1 Estimating poverty in the absence of consumption data: the case of Liberia Dabalen, A. L., Graham, E., Himelein, K., Mungai, R., September 2014 2 Female labor participation in the Arab world: some evidence from panel data in Morocco Barry, A. G., Guennouni, J., Verme, P., September 2014 3 Should income inequality be reduced and who should benefit? redistributive preferences in Europe and Central Asia Cojocaru, A., Diagne, M. F., November 2014 4 Rent imputation for welfare measurement: a review of methodologies and empirical findings Balcazar Salazar, C. F., Ceriani, L., Olivieri, S., Ranzani, M., November 2014 5 Can agricultural households farm their way out of poverty? Oseni, G., McGee, K., Dabalen, A., November 2014 6 Durable goods and poverty measurement Amendola, N., Vecchi, G., November 2014 7 Inequality stagnation in Latin America in the aftermath of the global financial crisis Cord, L., Barriga Cabanillas, O., Lucchetti, L., Rodriguez‐Castelan, C., Sousa, L. D., Valderrama, D. December 2014 8 Born with a silver spoon: inequality in educational achievement across the world Balcazar Salazar, C. F., Narayan, A., Tiwari, S., January 2015 Updated on December 2016 by POV GP KL Team | 1 9 Long‐run effects of democracy on income inequality: evidence from repeated cross‐sections Balcazar Salazar,C. F., January 2015 10 Living on the edge: vulnerability to poverty and public transfers in Mexico Ortiz‐Juarez, E., Rodriguez‐Castelan, C., De La Fuente, A., January 2015 11 Moldova: a story of upward economic mobility Davalos, M. E., Meyer, M., January 2015 12 Broken gears: the value added of higher education on teachers' academic achievement Balcazar Salazar, C. F., Nopo, H., January 2015 13 Can we measure resilience? a proposed method and evidence from countries in the Sahel Alfani, F., Dabalen, A. L., Fisker, P., Molini, V., January 2015 14 Vulnerability to malnutrition in the West African Sahel Alfani, F., Dabalen, A. L., Fisker, P., Molini, V., January 2015 15 Economic mobility in Europe and Central Asia: exploring patterns and uncovering puzzles Cancho, C., Davalos, M. E., Demarchi, G., Meyer, M., Sanchez Paramo, C., January 2015 16 Managing risk with insurance and savings: experimental evidence for male and female farm managers in the Sahel Delavallade, C., Dizon, F., Hill, R., Petraud, J. P., el., January 2015 17 Gone with the storm: rainfall shocks and household well‐being in Guatemala Baez, J. E., Lucchetti, L., Genoni, M. E., Salazar, M., January 2015 18 Handling the weather: insurance, savings, and credit in West Africa De Nicola, F., February 2015 19 The distributional impact of fiscal policy in South Africa Inchauste Comboni, M. G., Lustig, N., Maboshe, M., Purfield, C., Woolard, I., March 2015 20 Interviewer effects in subjective survey questions: evidence from Timor‐Leste Himelein, K., March 2015 21 No condition is permanent: middle class in Nigeria in the last decade Corral Rodas, P. A., Molini, V., Oseni, G. O., March 2015 22 An evaluation of the 2014 subsidy reforms in Morocco and a simulation of further reforms Verme, P., El Massnaoui, K., March 2015 Updated on December 2016 by POV GP KL Team | 2 23 The quest for subsidy reforms in Libya Araar, A., Choueiri, N., Verme, P., March 2015 24 The (non‐) effect of violence on education: evidence from the "war on drugs" in Mexico Márquez‐Padilla, F., Pérez‐Arce, F., Rodriguez Castelan, C., April 2015 25 “Missing girls” in the south Caucasus countries: trends, possible causes, and policy options Das Gupta, M., April 2015 26 Measuring inequality from top to bottom Diaz Bazan, T. V., April 2015 27 Are we confusing poverty with preferences? Van Den Boom, B., Halsema, A., Molini, V., April 2015 28 Socioeconomic impact of the crisis in north Mali on displaced people (Available in French) Etang Ndip, A., Hoogeveen, J. G., Lendorfer, J., June 2015 29 Data deprivation: another deprivation to end Serajuddin, U., Uematsu, H., Wieser, C., Yoshida, N., Dabalen, A., April 2015 30 The local socioeconomic effects of gold mining: evidence from Ghana Chuhan-Pole, P., Dabalen, A., Kotsadam, A., Sanoh, A., Tolonen, A.K., April 2015 31 Inequality of outcomes and inequality of opportunity in Tanzania Belghith, N. B. H., Zeufack, A. G., May 2015 32 How unfair is the inequality of wage earnings in Russia? estimates from panel data Tiwari, S., Lara Ibarra, G., Narayan, A., June 2015 33 Fertility transition in Turkey—who is most at risk of deciding against child arrival? Greulich, A., Dasre, A., Inan, C., June 2015 34 The socioeconomic impacts of energy reform in Tunisia: a simulation approach Cuesta Leiva, J. A., El Lahga, A., Lara Ibarra, G., June 2015 35 Energy subsidies reform in Jordan: welfare implications of different scenarios Atamanov, A., Jellema, J. R., Serajuddin, U., June 2015 36 How costly are labor gender gaps? estimates for the Balkans and Turkey Cuberes, D., Teignier, M., June 2015 37 Subjective well‐being across the lifespan in Europe and Central Asia Bauer, J. M., Munoz Boudet, A. M., Levin, V., Nie, P., Sousa‐Poza, A., July 2015 Updated on December 2016 by POV GP KL Team | 3 38 Lower bounds on inequality of opportunity and measurement error Balcazar Salazar, C. F., July 2015 39 A decade of declining earnings inequality in the Russian Federation Posadas, J., Calvo, P. A., Lopez‐Calva, L.‐F., August 2015 40 Gender gap in pay in the Russian Federation: twenty years later, still a concern Atencio, A., Posadas, J., August 2015 41 Job opportunities along the rural‐urban gradation and female labor force participation in India Chatterjee, U., Rama, M. G., Murgai, R., September 2015 42 Multidimensional poverty in Ethiopia: changes in overlapping deprivations Yigezu, B., Ambel, A. A., Mehta, P. A., September 2015 43 Are public libraries improving quality of education? when the provision of public goods is not enough Rodriguez Lesmes, P. A., Valderrama Gonzalez, D., Trujillo, J. D., September 2015 44 Understanding poverty reduction in Sri Lanka: evidence from 2002 to 2012/13 Inchauste Comboni, M. G., Ceriani, L., Olivieri, S. D., October 2015 45 A global count of the extreme poor in 2012: data issues, methodology and initial results Ferreira, F.H.G., Chen, S., Dabalen, A. L., Dikhanov, Y. M., Hamadeh, N., Jolliffe, D. M., Narayan, A., Prydz, E. B., Revenga, A. L., Sangraula, P., Serajuddin, U., Yoshida, N., October 2015 46 Exploring the sources of downward bias in measuring inequality of opportunity Lara Ibarra, G., Martinez Cruz, A. L., October 2015 47 Women’s police stations and domestic violence: evidence from Brazil Perova, E., Reynolds, S., November 2015 48 From demographic dividend to demographic burden? regional trends of population aging in Russia Matytsin, M., Moorty, L. M., Richter, K., November 2015 49 Hub‐periphery development pattern and inclusive growth: case study of Guangdong province Luo, X., Zhu, N., December 2015 50 Unpacking the MPI: a decomposition approach of changes in multidimensional poverty headcounts Rodriguez Castelan, C., Trujillo, J. D., Pérez Pérez, J. E., Valderrama, D., December 2015 51 The poverty effects of market concentration Rodriguez Castelan, C., December 2015 52 Can a small social pension promote labor force participation? evidence from the Colombia Mayor program Pfutze, T., Rodriguez Castelan, C., December 2015 Updated on December 2016 by POV GP KL Team | 4 53 Why so gloomy? perceptions of economic mobility in Europe and Central Asia Davalos, M. E., Cancho, C. A., Sanchez, C., December 2015 54 Tenure security premium in informal housing markets: a spatial hedonic analysis Nakamura, S., December 2015 55 Earnings premiums and penalties for self‐employment and informal employees around the world Newhouse, D. L., Mossaad, N., Gindling, T. H., January 2016 56 How equitable is access to finance in turkey? evidence from the latest global FINDEX Yang, J., Azevedo, J. P. W. D., Inan, O. K., January 2016 57 What are the impacts of Syrian refugees on host community welfare in Turkey? a subnational poverty analysis Yang, J., Azevedo, J. P. W. D., Inan, O. K., January 2016 58 Declining wages for college‐educated workers in Mexico: are younger or older cohorts hurt the most? Lustig, N., Campos‐Vazquez, R. M., Lopez‐Calva, L.‐F., January 2016 59 Sifting through the Data: labor markets in Haiti through a turbulent decade (2001‐2012) Rodella, A.‐S., Scot, T., February 2016 60 Drought and retribution: evidence from a large‐scale rainfall‐indexed insurance program in Mexico Fuchs Tarlovsky, Alan., Wolff, H., February 2016 61 Prices and welfare Verme, P., Araar, A., February 2016 62 Losing the gains of the past: the welfare and distributional impacts of the twin crises in Iraq 2014 Olivieri, S. D., Krishnan, N., February 2016 63 Growth, urbanization, and poverty reduction in India Ravallion, M., Murgai, R., Datt, G., February 2016 64 Why did poverty decline in India? a nonparametric decomposition exercise Murgai, R., Balcazar Salazar, C. F., Narayan, A., Desai, S., March 2016 65 Robustness of shared prosperity estimates: how different methodological choices matter Uematsu, H., Atamanov, A., Dewina, R., Nguyen, M. C., Azevedo, J. P. W. D., Wieser, C., Yoshida, N., March 2016 66 Is random forest a superior methodology for predicting poverty? an empirical assessment Stender, N., Pave Sohnesen, T., March 2016 67 When do gender wage differences emerge? a study of Azerbaijan's labor market Tiongson, E. H. R., Pastore, F., Sattar, S., March 2016 Updated on December 2016 by POV GP KL Team | 5 68 Second‐stage sampling for conflict areas: methods and implications Eckman, S., Murray, S., Himelein, K., Bauer, J., March 2016 69 Measuring poverty in Latin America and the Caribbean: methodological considerations when estimating an empirical regional poverty line Gasparini, L. C., April 2016 70 Looking back on two decades of poverty and well‐being in India Murgai, R., Narayan, A., April 2016 71 Is living in African cities expensive? Yamanaka, M., Dikhanov, Y. M., Rissanen, M. O., Harati, R., Nakamura, S., Lall, S. V., Hamadeh, N., Vigil Oliver, W., April 2016 72 Ageing and family solidarity in Europe: patterns and driving factors of intergenerational support Albertini, M., Sinha, N., May 2016 73 Crime and persistent punishment: a long‐run perspective on the links between violence and chronic poverty in Mexico Rodriguez Castelan, C., Martinez‐Cruz, A. L., Lucchetti, L. R., Valderrama Gonzalez, D., Castaneda Aguilar, R. A., Garriga, S., June 2016 74 Should I stay or should I go? internal migration and household welfare in Ghana Molini, V., Pavelesku, D., Ranzani, M., July 2016 75 Subsidy reforms in the Middle East and North Africa Region: a review Verme, P., July 2016 76 A comparative analysis of subsidy reforms in the Middle East and North Africa Region Verme, P., Araar, A., July 2016 77 All that glitters is not gold: polarization amid poverty reduction in Ghana Clementi, F., Molini, V., Schettino, F., July 2016 78 Vulnerability to Poverty in rural Malawi Mccarthy, N., Brubaker, J., De La Fuente, A., July 2016 79 The distributional impact of taxes and transfers in Poland Goraus Tanska, K. M., Inchauste Comboni, M. G., August 2016 80 Estimating poverty rates in target populations: an assessment of the simple poverty scorecard and alternative approaches Vinha, K., Rebolledo Dellepiane, M. A., Skoufias, E., Diamond, A., Gill, M., Xu, Y., August 2016 Updated on December 2016 by POV GP KL Team | 6 81 Synergies in child nutrition: interactions of food security, health and environment, and child care Skoufias, E., August 2016 82 Understanding the dynamics of labor income inequality in Latin America Rodriguez Castelan, C., Lustig, N., Valderrama, D., Lopez‐Calva, L.‐F., August 2016 83 Mobility and pathways to the middle class in Nepal Tiwari, S., Balcazar Salazar, C. F., Shidiq, A. R., September 2016 84 Constructing robust poverty trends in the Islamic Republic of Iran: 2008‐14 Salehi Isfahani, D., Atamanov, A., Mostafavi, M.‐H., Vishwanath, T., September 2016 85 Who are the poor in the developing world? Newhouse, D. L., Uematsu, H., Doan, D. T. T., Nguyen, M. C., Azevedo, J. P. W. D., Castaneda Aguilar, R. A., October 2016 86 New estimates of extreme poverty for children Newhouse, D. L., Suarez Becerra, P., Evans, M. C., October 2016 87 Shedding light: understanding energy efficiency and electricity reliability Carranza, E., Meeks, R., November 2016 88 Heterogeneous returns to income diversification: evidence from Nigeria Siwatu, G. O., Corral Rodas, P. A., Bertoni, E., Molini, V., November 2016 89 How liberal is Nepal's liberal grade promotion policy? Sharma, D., November 2016 90 CPI bias and its implications for poverty reduction in Africa Dabalen, A. L., Gaddis, I., Nguyen, N. T. V., December 2016 91 Pro-growth equity: a policy framework for the twin goals Lopez-Calva, L. F., Rodriguez Castelan, C., November 2016 92 Building an ex ante simulation model for estimating the capacity impact, benefit incidence, and cost effectiveness of child care subsidies: an application using provider‐level data from Turkey Aran, M. A., Munoz Boudet, A., Aktakke, N., December 2016 93 Vulnerability to drought and food price shocks: evidence from Ethiopia Porter, C., Hill, R., December 2016 94 Job quality and poverty in Latin America Rodriguez Castelan, C., Mann, C. R., Brummund, P., December 2016 Updated on December 2016 by POV GP KL Team | 7 For the latest and sortable directory, available on the Poverty & Equity GP intranet site. http://POVERTY WWW.WORLDBANK.ORG/POVERTY Updated on December 2016 by POV GP KL Team | 8