In S b tu fr pe an r U ruc as r ci iz e D t ResearchDigest al at e Fo io ve World Bank cu n a lop s nd m on e nt VOLUME 13 NUMBER 3 SPRING 2019 Building Sustainable Cities and Efficient Infrastructure O ur future is in cities. By 2050 managed well. Moreover, with younger IN THIS ISSUE nearly 70 percent of human- and more-skilled populations, cities ity will be living in cities and can capture demographic dividends. Building Sustainable Cities and Efficient towns, up from around 50 percent And cities can play a vital part in Infrastructure … page 1 today. Much of that urban growth will tackling climate change. Some cities take place in the developing world, today are building efficient housing Who Benefits from Investments in Bus especially Sub-Saharan Africa and and improving road transport infra- Rapid Transit Systems? Evidence from Asia. The rapid expansion of cities is structure, critical in reducing energy Colombia … page 2 expected to transform the way we live, consumption as well as greenhouse Beyond Faster Travel: Assessing the work, and commute, with serious con- gas emissions. Wider Economic Benefits of Transport sequences for economic development This issue features World Bank re- Corridors … page 3 and climate change. search aimed at supporting national Much of the world’s urban expan- and city-level policy makers in think- A Big Push or a Backwash? sion may occur in slum-like condi- ing strategically about the opportuni- Economic Effects of Connecting tions, exacerbating economic dispari- ties and challenges of urbanization. a Hinterland … page 4 ties and exposure to health risks. And The first three studies look at who the speed and scale of city growth will benefits from large public investments Highway Politics in Mexico … page 5 mean accelerated demand for basic in transport infrastructure. They also services, affordable housing, and jobs. reveal that realizing the benefits can How Does the Shape of Latin American Infrastructure development usually depend on local and national condi- Cities Affect Their Productivity? … page 6 lags behind urban population growth. tions. The fourth study looks at the Congested roads, inadequate public part that politics can play. Getting to Jobs in Nairobi … page 7 transport, and inefficient ports can un- The next two studies highlight the dermine the competitiveness of cities. importance of city planning, land poli- Public Infrastructure and Structural Overburdened schools, hospitals, and cies, and efficient transit systems in Transformation … page 8 public safety net delivery systems can fostering growth and improving job Buses, Houses, or Cash? Choices and hamper poverty reduction efforts. opportunities for poor urban resi- Consequences in Reforming Public Rapid urban growth is also linked dents. The following two, focusing on Transport Subsidies … page 9 to environmental concerns. Cities public intervention in the provision of consume close to 70 percent of the infrastructure services, remind us of Improving the Resilience of Peru’s Road world’s energy and account for more the pitfalls in distortionary policies. Network to Climate Events … page 10 than 70 percent of global greenhouse The last two underscore the impor- gas emissions. As cities expand, their tance of ensuring that infrastructure is The Vulnerability of Manufacturing exposure to climate and disaster risk resilient to extreme climate events. Firms to Unreliable Water increases. Almost half a billion urban Building cities that are safe, inclu- Infrastructure … page 11 residents live in coastal areas or on sive, resilient, and sustainable will riverbanks, vulnerable to floods, storm require sound investment choices Recent Policy Research Working Papers surges, and sea-level rise. and intensive policy coordination. on Urbanization and Infrastructure Yet urbanization also offers the po- Researchers are exploring creative Development … page 12 tential to transform economies. With solutions to the challenges of urban- more than 80 percent of global GDP ization—to help ensure that the deci- generated in cities, urbanization can sions we make today shape healthy contribute to sustainable growth if cities for tomorrow. 2 World Bank ResearchDigest Who Benefits from Investments in Bus Rapid Transit Systems? Evidence from Colombia Knowing how new urban transport midsize cities or developing countries. families replacing poorer ones. First, affects household welfare is key to A recent study by Pfutze, Rodríguez- proximity to feeder stations increased designing policies to ensure that its Castelán, and Valderrama-González the number of smaller households seeks to help fill this gap in knowledge (with one or two members) in the benefits reach the poorest by investigating the distributional ef- area and reduced the number of R fects of Barranquilla’s Transmetro sys- large households (with more than apid urbanization is among the tem, which was originally intended to five members). This change was re- defining characteristics of de- benefit the poorest households in the lated to a reduction in the number veloping economies. In Latin metropolitan area through better con- both of young adults ages 16–22 and America 80 percent of people live in nectivity and lower fares. of children. Second, the character- urban areas today, up from 49 percent The analysis focuses on composi- istics of adults living in proximity to in the 1960s; in Colombia 77 percent tional changes that may have occurred Transmetro stations also changed, do, up from 45 percent. The rapid in neighborhoods as a result of prox- with more-educated adults replac- population growth in urban areas imity to the opening of Transmetro ing less-educated ones. These joint requires adequate infrastructure to stations. It combines geocoded in- effects translated into a reduction support efficient provision of essential formation on the expansion of the in the number of households in the services. Urban transport services are Transmetro system with household poorest two quintiles of the income particularly important; where they are survey data from the 2008–15 rounds distribution and, at the same time, an lacking, the result can be congested of the main labor survey in Colombia increase in the number of those in the roads, limited urban mobility, and (Gran Encuesta Integrada de Hogares). richest quintile. misallocation of productive resources. The survey is representative at the The study’s main results sug- One response to these challenges metropolitan-area level for 13 cities gest that the development of the has been the introduction of bus rapid including Barranquilla and provides Transmetro system in Barranquilla had transit (BRT) systems, which have information on an ample set of wel- positive effects relating to transport multiplied in recent decades. BRT fare, housing, and labor character- efficiency and urban mobility—but systems consist of a network of buses istics. To assess the exposure of that despite the original objective of using dedicated lanes that allow rapid households to the Transmetro system, benefiting poorer households through mobility. These systems operate in including both main lines and feeder better connectivity, it also had a re- 205 cities around the world, transport- line stations, the analysis aggregates gressive distributional impact. This ing 34 million passengers a day. households into blocks and then cal- regressive effect was driven by signifi- In Latin America BRT systems were culates the proximity of each block to cant displacement of poorer house- pioneered by Curitiba in Brazil and bus stations as the distance from its holds and their replacement by mid- Quito in Ecuador. They soon spread to center. The study analyzes 2,088 blocks dle- and upper-income households. large capital cities, including Mexico over the 2008–15 period. The study thus contributes to the City and Bogotá. After introducing The findings show that after imple- literature in two ways, both by inves- the BRT system in Bogotá in 2001, mentation of the Transmetro system, tigating the overlooked distributional Colombia later expanded it to other housing character- cities. One of these is Barranquilla, istics improved and where a system known as Transmetro housing prices rose. has been serving the metropolitan Proximity to stations area, Colombia’s fourth largest had a strong effect population center, since April 2010. on the quality of the The Transmetro system now has 18 housing stock, as reflected 2016 PASSENGER 2010 TRIPS stations on its main lines and more in an increase in the number of 36.4M 3.7M than 600 stops along its feeder lines. apartments, rooms, and exclusive Passenger trips on the system rose toilets. In addition, home values and effects of urban transport infrastruc- from nearly 3.7 million in 2010 to rents increased, though not as a direct ture and by questioning the often- more than 36.4 million in 2016. effect of the proximity to stations but unstated assumption that households The benefits of BRT systems for as a consequence of the improvement do not relocate in response to public transport efficiency and the environ- in the quality of housing. transport policies. ment have been widely documented. Compositional changes in house- Tobias Pfutze, Carlos Rodríguez-Castelán, and But there is scant literature on how holds living in proximity to Transmetro Daniel Valderrama-González. 2018. “Urban urban transport infrastructure af- stations followed in response to the Transport Infrastructure and Household Welfare: fects household welfare—and few better and more expensive housing Evidence from Colombia.” Policy Research Work- urban transport studies focusing on stock, with better-off individuals and ing Paper 8341, World Bank, Washington, DC. World Bank Research Digest 3 Beyond Faster Travel: Assessing the Wider Economic Benefits of Transport Corridors Better highways bring economic as welfare (district-level GDP and this effect was not widely shared. It did benefits beyond those from faster household consumption); inequality not have significant positive effects travel. But the gains can depend on (district-level poverty); social inclu- on household consumption, poverty, sion (total employment and female or employment, including of women. local conditions employment in regular wage jobs); Moreover, the evidence suggests sig- I and environmental quality (density of nificant trade-offs in effects, with GDP ndia’s Golden Quadrilateral, a net- aerosol particles). The authors also increasing while air quality decreased work of four- to six-lane express look at structural transformation in significantly. highways totaling 5,846 kilometers employment (the shift from farm to The authors also examine how the in length, connects the country’s four nonfarm work), an intermediate out- effects of the Golden Quadrilateral top metropolitan cities—New Delhi, come of interest. could have depended on local condi- Mumbai, Chennai, and Kolkata (map Applying the “differences in differ- tions in a district—that is, on how well 1). Its construction was launched in ences” method to district-level data the land, labor, capital, and product 2001, two-thirds complete by 2005, from 1994 to 2011, the authors esti- markets function and on the quality and mostly finished in 2007. By mate that the Golden Quadrilateral in- of local institutions such as public January 2012 its entire length was creased the share of nonfarm employ- governance. To estimate this pos- operational. ment among women in India by 2.4 sible dependence, they use interac- Large highway projects like this percentage points, with a correspond- tion effects between the connectivity one are expected to benefit busi- ing proportional decline in their farm to the highway network and different nesses through greater connectivity. employment. It had a similar, though measures of initial conditions in local But what about the potential wider smaller, effect on male employment. markets. They find, for example, that economic benefits that may also Overall, the authors find that the the Golden Quadrilateral had a bigger matter in evaluating such projects? Golden Quadrilateral contributed sig- positive effect on nonfarm employ- Building on earlier research, a paper nificantly to the structural transforma- ment in districts with a greater share by Melecky, Sharma, and Subhash tion of employment. The network also of secondary education. Moreover, examines the effect of the Golden had a significant positive effect on the highway network appears to have Quadrilateral on such measures district-level per capita GDP, though boosted total employment more— particularly for women—in districts Map 1. India’s Golden Quadrilateral Highway Network with local governance of above aver- age quality. HARYANA DELHI NewDelhi DELHI New Delhi The results suggest that initial conditions in local markets are im- UTTAR PRADESH portant in determining the outcomes RAJASTHANJaipur Lucknow Ajmer of highway upgrades and the ways Kanpur BIHAR and extent to which wider economic Varanasi Patna benefits are shared across geographic Aurangabad Udaipur units. The central mechanism through JHARKHAND which highways are expected to affect WEST Gandhinagar Bhopal BENGAL development is increased trade across GUJARAT Vadodara MADHYA PRADESH connected locations. This requires Porbandar CHHATTISGARH Kolkata Kolkata a reallocation of resources to more Surat Haridaspur Nagpur Raipur productive firms and to sectors of ODISHA Bhubaneshwar comparative advantage. Conditions in MAHARASHTRA land, labor, capital, and product mar- Brahmapur kets could matter because they affect Mumbai Mumbai this process of reallocation. TELANGANA Vishakhapatnam Hyderabad Rajahmundry PUDUCHERRY ANDAMAN Panaji Hubli Ongole AND NICOBAR GOA ANDHRA PRADESH Nellore KARNATAKA Martin Melecky, Siddharth Sharma, and Hari Subhash. 2018. “Wider Economic Benefits of In- Bangalore Chennai Chennai vestments in Transport Corridors and the Role of Kanchipuram PUDUCHERRY TAMIL NADU Complementary Policies.” Policy Research Working Paper 8350, World Bank, Washington, DC. LAKSHADWEEP LAKSHADWEEP KERALA 4 World Bank ResearchDigest A Big Push or a Backwash? Economic Effects of Connecting a Hinterland estimators in a “differences in differ- ences” design in which the compari- son hinterland is in a region that was The Jamuna bridge in Bangladesh to comparative advantage. But a large supposed to be connected to the core connected a poor hinterland to literature shows that they may instead (the capital city) by the proposed but growth centers. How did the new cause a “backwash effect” in which not yet constructed Padma bridge. resources leave the hinterland and Results show that in the short run the links affect the local economy? high-productivity manufacturing con- Jamuna bridge leads to significant re- C centrates in cities. allocation of labor from agriculture to an greater connectivity act as The paper develops a spatial gener- services in the connected (treatment) a catalyst for economic growth al equilibrium model with a core and, hinterland but has no perceptible and structural transformation in at opposite sides of it, two hinterlands effects on the employment share of poorer regions? One test case might separated from the core by rivers and manufacturing, on population den- be the construction of a major bridge allows for productivity gains through sity, or on night-lights, a measure of in Bangladesh. Spanning the Jamuna agglomeration in agriculture and man- overall economic activity (figure 1). In River, one of the largest rivers in the ufacturing. The model yields insights the long run the labor share of manu- world, the 4.8-kilometer bridge con- that differ from those of the standard facturing declines in the connected nected about 30 million people living core-periphery and trade models: hinterland and increases in the core, in chronically poor areas in the north- The newly connected hinterland may raising the possibility of backwash. west to growth centers in the east, experience higher population density But significant positive effects on pop- including the capital city of Dhaka and agricultural productivity despite ulation density and agricultural yields and the port city of Chittagong. Since significant deindustrialization. Even in the connected hinterland contradict opening in 1999 the bridge has re- with greater specialization in agricul- backwash effects of the bridge. The ef- duced freight costs by 50 percent, and ture, the share of agricultural employ- fects of the bridge on the allocation of travel time by three to four hours, from ment may decline when interregional labor among sectors vary across areas, areas in the northwest to Dhaka. trade requires local services (such with relatively weak effects in those Using quasi-experimental study as processing and trading). And the close to the bridge. of the Jamuna bridge, a paper by strongest effects on the structure of Blankespoor, Emran, Shilpi, and Xu employment may be felt not in the Brian Blankespoor, M. Shahe Emran, Forhad looks at how a large reduction in trade areas next to the bridge but in those Shilpi, and Lu Xu. 2018. “Bridge to Bigpush or costs affects the pattern of structural that move out of market isolation as a Backwash? Market Integration, Reallocation, and change and agricultural productivity. result of the bridge. Productivity Effects of Jamuna Bridge in Ban- In classic trade models lower transport For the empirical estimation of the gladesh.” Policy Research Working Paper 8508, costs lead to specialization according model the authors use doubly robust World Bank, Washington, DC. Figure 1. Short- and Long-Run Effects of the Jamuna Bridge on Employment and Population Density Industry share of employment Services share of employment 0.06 0.04 0.04 0.03 Size of effect Size of effect 0.02 0.02 0.01 0 0 −0.02 Connected hinterland Short run Long run Short run Long run Short run Long run Short run Long run Core Population density Rice yield 95% confidence interval 0.20 0.10 0.15 0.05 Size of effect Size of effect 0.10 0 0.05 0 −0.05 −0.05 −0.10 Short run Long run Short run Long run Short run Long run Short run Long run Note: Results from regression analysis comparing average differences in outcomes between areas connected by the Jamuna bridge (hinterland as well as core) and hinterland areas similarly remote but still unconnected. The short run is 2 years, and the long run 12 years, after the opening of the bridge. World Bank Research Digest 5 Highway Politics in Mexico elections more than doubled com- pared with those in locations lost to the opposition. Alignment with the Where federal highway spending is In a recent study Selod and presidential party’s legislative candi- directed in Mexico can depend on Soumahoro address these challenges dates increased federal highway provi- electoral outcomes—and more so by applying a quasi-experimental de- sion but had no significant effect on sign to a data set comprising election state and rural roads that are not fed- under divided government outcomes and georeferenced road erally funded. Interestingly, there is no R data from Mexico. To guard against evidence of allocation bias in federal oad investments can have big missing or poor road expenditure highways when the winner of the leg- impacts on economic growth data, the authors extract from sheet islative race was from the majority in and development. This is be- maps spatially explicit measures of the chamber of deputies or when the cause they facilitate trade through actual roads that were built (map 1). authors exclusively examine presiden- lower transaction costs and better Combining these road statistics with tial election outcomes rather than the access to markets, or because they data from nearly 10,000 legislative legislative election outcomes. So it is may stimulate the spatial agglomera- elections in more than 2,000 Mexican the interaction between the executive tion of firms and local specialization, municipalities between 1993 and 2012, and legislative branches that gener- increasing local employment and they compare changes in newly built ated political bias in highway build- productivity. roads across locations with different ing. Overall, the study underscores The placement of roads, however, voting outcomes. how the relationships among different may not be optimal when it arises Specifically, the authors examine branches of government could distort from political rather than economic whether municipalities that voted for resource allocation. incentives—as with costly white el- candidates from the ruling presiden- To shed light on why this distortion ephant projects. While this problem tial party afterward received a dispro- occurs, the study discusses two poten- seems to emerge mainly in autocratic portionate share of federally funded tial explanations. First, politicians may settings, electoral competition can highways. To identify political bias in target resources toward constituencies also generate resource misallocation, road building, they compare changes with common ideological interests especially when politicians seek to in highway length between municipali- and policy priorities. For example, improve their reelection prospects. ties where the presidential party bare- evidence from the United States sug- Still, empirical efforts to under- ly won and those where it barely lost gests that Democrats are more in- stand the role of political preferences the legislative race. The rationale for clined to spend heavily on education in the subnational distribution of road this approach is that elections decided while Republicans tend to prioritize investments face two major challeng- with razor-thin margins involve a ran- defense spending. The study could not es: Disaggregated road expenditures dom component that makes it reason- substantiate this prediction, as none are difficult to track, especially in de- able to assume similarities between of the dominant parties in Mexico veloping countries. And subnational locations with highly contested races. exhibited a differentiated preference administrative jurisdictions may vary Results suggest that in the four for road investments. Second, legisla- in ways that potentially correlate with to five years after an election, federal tive instability in the form of divided both political preferences and public highways built in municipalities back- investment choices. ing the presidential party in legislative (continued on page 12) Map 1. Road Network in Mexico, 1993 and 2012 1993 2012 Federally funded Multilane roads Not federally funded Two-lane roads Divided roads Other paved roads 6 World Bank ResearchDigest How Does the Shape of Latin American Cities Affect Their Productivity? A city’s spatial structure matters for closer people and firms are located The authors characterize city shape its productivity—suggesting that city together. But while the links are well with three key interrelated dimensions planning and land policies can serve established through a theoretical of urban form: the border’s shape and framework, few empirical studies have perimeter; the internal structure of as tools to help foster growth examined the relationship between ur- urban area; and the land use patterns D ban form and economic performance; observed within city boundaries and ifferent cities can use land and moreover, there is still little evidence reflecting the use of space and the dis- space in very different ways. linking the presence of agglomera- tribution of people within the city. Local governments, through tion economies with the form that Using a series of empirical strate- land use regulations and other urban cities take. gies, the authors estimate the rela- policy instruments, can influence In Latin America, where more than tionship between measures of urban where economic activities, urban infra- 80 percent of the people live in cit- productivity and urban shape. The structure, and households are located. ies, understanding the links between results show that the shape of the ur- And the spatial structure of cities, or city form and productivity is critical ban extent, the inner-city connected- urban form, is thought to have impor- for policy makers. The challenges of ness, the level of urbanization, and the tant influence on the emergence of topography may have prevented dense population level have a statistically agglomeration economies and conges- development in some cases, as many significant influence over the produc- tion costs—and thus on a city’s level Latin American cities are located in tivity level of the city. The authors also of productivity—but also on the sus- rugged topographies with natural bar- find evidence of the presence of a con- tainability of cities and their quality riers. Moreover, rapid urbanization gestion effect on productivity as cities of life. The same channels—matching, combined with limited infrastructure become denser. According to these learning, and sharing—that are used investments may have led to urban results, a compact, dense, and well- to explain the emergence of agglom- forms that pose a barrier to the rise of connected city meets important condi- eration economies are also thought agglomeration economies, limit the tions for being highly productive. to be at the core of the links between interactions among firms, and make it An important conclusion from this urban form and city productivity. hard for workers to reach their jobs. A work is that a city can grow in differ- Differences in the occupation of ur- better understanding of the links be- ent shapes and still achieve high pro- ban space have consequences for the tween urban form and productivity can ductivity by guaranteeing a high rate transport modes used (for example, shed light on whether urban policy of inner-city connectedness (equally, by favoring the use of private vehicles has a part to play in supporting city a compact but poorly connected city over public transport), for commuting productivity and, through that, eco- can show low productivity). The re- times in cities, and for the type and nomic growth. sults also underscore the fact that intensity of human interaction. Denser In a recent paper Duque, Lozano- urban form tends to persist over time, cities are thought to improve labor Gracia, Patino, and Restrepo examine requiring policy makers to think far productivity through better matching the links between urban form and city ahead and ensure good accessibility of firms and workers and enhanced productivity using different urban form within cities. This has important im- interactions that facilitate the spread metrics for a comprehensive sample of plications for policy makers because of tacit knowledge, both of which Latin American cities with more than mayors often ask what they can do to are thought to occur more easily the 50,000 inhabitants in 2010. One of the improve the productivity of their cit- main contributions of this ies. City planning and land manage- Characterizing City Shape work is the creation of a com- ment policies are not often regarded The authors use seven indicators of urban form to characterize city shape. The parable series of city shape as instruments to foster productivity examples here illustrate high and low values for two of these: Roundness measures the share of the urban area that is inside the equal-area circle about its center of indicators for 919 cities in and growth in cities. The results in this gravity. Smoothness of perimeter is the ratio of the perimeter of the equal-area Latin America, used to char- work suggest otherwise: these are im- circle to the perimeter of the shape. acterize urban structure and portant tools that local governments High Low create a proxy measure of a have at hand to increase productivity city’s economic performance. in Latin American cities. Roundness These were derived from city extents created from a con- sistent time series of night- time lights imagery from Juan C. Duque, Nancy Lozano-Gracia, Jorge E. the Defense Meteorological Patino, and Paula Restrepo. 2019. “Urban Form Smoothness of perimeter Satellite Program’s and Productivity: What Is the Shape of Latin Operational Linescan System American Cities?” Policy Research Working Paper (DMSP-OLS). 8697, World Bank, Washington, DC. World Bank Research Digest 7 Getting to Jobs in Nairobi Traveling to a job can be costly and limiting their employment opportuni- residents of formal residential areas. time-consuming in Nairobi. This is ties, could trap them in poverty. As the authors demonstrate especially so for poorer residents, A recent paper by Nakamura and through econometric analysis, part of Avner looks at these issues through an the reason that poorer households in limiting their job opportunities empirical analysis of the spatial dis- Nairobi have more limited job accessi- I tribution of job accessibility, housing bility is that living in housing with bet- n cities where people can easily rents, and poverty in Nairobi, Kenya. ter job accessibility is costly, imposing reach job opportunities, whether To disentangle them, the authors constraints on their choice of residen- by foot or by public transport, the combine multiple data sources on job tial location. Living in areas closer to labor market can be more productive. locations, road and minibus networks, job opportunities is costly not only That benefits both the urban economy and household consumption and because housing quality and living and workers. housing costs. The overall level of job conditions tend to be better in such Cities create agglomeration econo- accessibility in a city depends on the areas, but also because job accessibil- mies, the productivity gains that result locations of jobs and workers’ resi- ity itself is valued as an amenity in the when enterprises are clustered togeth- dences as well as transport networks. housing market. er. Poor job accessibility can inhibit And which households have good How to improve job accessibility these productivity gains by making it access to job opportunities hinges among low-income households? There difficult to match jobs and job seek- on the trade-offs they face between are several possible ways. On the sup- ers. By contrast, good job accessibility job accessibility, living conditions, ply side, promoting the development —with city residents able to reach a and housing costs when choosing an of affordable transport networks is es- large share of job opportunities within area to live. sential over the long term. For exam- a reasonable amount of time—can The analysis finds that workers and ple, facilitating the extension of mini- boost the productivity gains. jobs are poorly connected in Nairobi. bus services to underserved areas that For workers, job accessibility plays On average, residents can access fewer are home to many low-income house- a part in employment outcomes in than 10 percent of existing jobs by holds would help reduce their job two ways: by affecting the probability foot within an hour. Even when using search and commuting costs. In large, of finding a job and by affecting the a minibus they can reach only about a high-density informal settlements that quality of the matching (more or less quarter of the jobs within that time. may lack adequate space for transit, a well paying, more or less formal). Poorer households and those transport right-of-way could be intro- Just as important, limited job acces- living in informal settlements face duced as part of slum upgrading proj- sibility may disproportionately affect even more limited job accessibility. ects, to support the economic integra- economic outcomes for disadvantaged Compared with richer households tion of these settlements with the rest workers. In many African cities reliable (those in the fourth quartile of per of the city. On the demand side, such and affordable transport is lacking. capita consumption), poor house- policies as cash transfers could be ef- Low-income workers may live far from holds (those in the first quartile) can fective in the short term. clusters of jobs and thus face higher reach on average 20 percent fewer jobs job search and commuting costs, or by foot within 60 minutes (figure 1). they may be too poor to use transport Similarly, residents of informal settle- and thus have no option other than to ments can reach 30 percent fewer jobs walk to work. These circumstances, by by foot within 60 minutes than can Figure 1. Accessibility of Jobs for Different Groups in Nairobi, 2013 40 Share of jobs that can be reached 30 within 60 minutes (%) 20 10 Shohei Nakamura and Paolo Avner. 2018. “Spa- 0 tial Distributions of Job Accessibility, Housing Poorest Q2 Q3 Richest Informal Formal Rents, and Poverty in Nairobi, Kenya.” Policy Consumption quartile Type of neighborhood Research Working Paper 8654, World Bank, By foot By minibus Washington, DC. 8 World Bank ResearchDigest Public Infrastructure and Structural generates a change in share larg- er than that obtained from the data Transformation and accounts for 59 percent of the combined contribution of public infrastructure and relative productiv- A recent study shows that public To answer these questions, the ity growth. infrastructure is an important driver study develops a multisectoral gen- The study also conducts experi- of the structural transformation of eral equilibrium model of unbalanced ments relating the model simulations economic growth and analyzes it to public infrastructure policies. Two developing economies quantitatively. Public infrastructure of these consider what might have E enters the model as a complementary happened if Brazil, rather than allow- conomists recognize certain pro- factor that increases the productivity ing public capital formation to fall cesses as key features of success- of private inputs such as labor or capi- from an average 5.7 percent of GDP in ful economic development. One tal. The quality of public infrastructure the 1970s to an average 3.4 percent in of these is structural transformation— affects firms’ operating costs. Lack of 1985–2009, had instead maintained it the reallocation of inputs to more public capital adds to those costs— at the level of the 1970s. productive activities, such as from such as when poor electricity networks In the first experiment the increase agriculture to manufacturing or, later, require firms to rely on generators— in the ratio of public capital formation from agriculture and manufacturing to and thus acts as a barrier to firms’ to GDP comes as a consequence of services. Another is public infrastruc- entry into markets. Conversely, good stronger partisan incentives in favor ture (or public capital) formation. public infrastructure is conducive to of public spending. The result is an Moreover, cross-country data sug- the creation of large business clusters increase in the share of manufacturing gest that these two processes are sig- that generate increasing returns and and declines in the shares of agricul- nificantly correlated. A recent study by lower barriers to firms’ entry. ture and services. There is also a sub- Perez Sebastian and Steinbuks, using The analysis uses data for Brazil, stantial positive effect on total GDP World Bank data for 263 economies, focusing on the post-hyperinflation and production levels. GDP per capita shows that the quality of electricity period between 1995 and 2013. Brazil rises by 10 percent. And the increase infrastructure (as proxied by average is widely considered to be in deep in output is especially strong in manu- monthly power outages) is worse in need of infrastructure investment. And facturing, thanks to its important role economies depending more heavily on the country offers high-quality data for in capital accumulation and to the agriculture—and that it improves as validating the model predictions. greater efficiency of smaller firms that manufacturing accounts for a growing The results show that the accu- operate under diminishing returns to share of GDP (figure 1). mulation of infrastructure accelerates private inputs. But the understanding of how structural transformation. According In the second, the increase in pub- these two important processes inter- to the model simulations, public capi- lic capital formation occurs because act is limited. For example, how much tal formation explains 5 percent of the of improvement in the management does public infrastructure formation total variation in the share of manu- of public investment, so that a larger matter in structural transformation, facturing (as measured by gross value share of public investment spending and vice versa? And how do different added) observed in Brazil’s economy ends up being converted into public constraints to public capital accumu- over the period of the study—and lation affect economic growth and 15 percent of that for services. For structural transformation? agriculture, public capital formation (continued on page 12) Figure 1. Quality of Electricity Infrastructure and Income Shares of Agriculture and Manufacturing in 263 Economies, 2006–15 55 55 Average power outages per month experienced by firms Average power outages per month experienced by firms 50 50 45 45 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 Average share of agriculture in GDP (%) Average share of manufacturing in GDP (%) World Bank Research Digest 9 Buses, Houses, or Cash? Choices and Consequences in Reforming Public Transport Subsidies Subsidizing public transport can the system has since been reformed, commuting patterns. The most ob- have complex effects in a city. So including as recently as January 2018, vious consequence is that higher can other policy approaches aimed looking back a few years provides fares would lead to a decline in pub- interesting insights and could shed lic transport ridership and to more at achieving the same goals light on possible effects of recent fare carbon-intensive trips. But rising P increases. transport costs would also encourage olicy makers use urban transit Using a simple urban economics people to move closer to their jobs, subsidies mainly to help im- model, the authors run four different substantially reducing commuting prove equity—by making it eas- scenarios: distances. As a result, the increase in ier for people, especially poor people, •  Removing the transit subsidy commuting-related emissions of car- to access jobs and other opportuni- with no compensation bon dioxide (CO2) would remain lim- ties. Yet some studies show that such •  Replacing the subsidy with a ited under most scenarios. subsidies fail to reach the poor and lump sum transfer Third, replacing the transit subsidy instead benefit middle-income groups •  Replacing the subsidy with a with a lump sum transfer would have the most. Even if they did reach those general construction subsidy significant short-term redistributive in need, would transit subsidies be •  Replacing the subsidy with a effects for the heaviest users of public the best way of improving welfare, construction subsidy targeting areas transport. For tenants in the hardest- or would other policy tools be more with high accessibility to jobs hit areas, lower accessibility would efficient? Moreover, by artificially re- What does the model show about trigger a decrease in housing rents ducing commuting costs, don’t policy likely effects? First, replacing the tran- over the medium term that would makers run the risk of encouraging sit subsidy with a cash transfer could partially offset the rising fares. This urban sprawl with all its negative generate welfare and efficiency gains adjustment in rents would benefit ten- consequences? (table 1). For Buenos Aires in 2012, ants and newcomers to the urban area A paper by Avner, Mehndiratta, this would have translated into an ad- but would penalize homeowners, who Viguié, and Hallegatte tackles these ditional $325 per household in equiva- would face both higher commuting questions by focusing on the urban lent income, or 1.75 percent of the costs and lower resale values for their area of Buenos Aires, Argentina. The average household budget in 2014. property. authors use data from 2012, a time Second, removal of the transit These results paint a complex pic- when public transport fares in the city subsidy would lead to only moder- ture of public transport subsidies in were heavily subsidized. Transit subsi- ate reductions in urbanization over the urban area of Buenos Aires. An dies that year amounted to more than time. This suggests that the argument ideal solution would be to channel $5 billion (0.7 percent of Argentina’s that low transit fares exacerbate ur- lump sum transfers only to the poor- GDP) and covered two-thirds of the ban sprawl is likely overplayed. But est households. This would require de- cost of public transport trips. While removal of the subsidy would affect tailed registries recording households’ incomes and their evolution over time. Second best would be to have transit Table 1. Effects under Prospective Alternative Scenarios Compared with the subsidies benefit only the poorest Baseline Scenario of Public Transport Subsidies in Buenos Aires in 2050 households. This solution would pro- Percentage change tect low-income transit users without With no With a With a dramatic negative effects on efficiency public general targeted or the environment. It would also ease transport With a lump construction construction the pressure on public resources. subsidy sum transfer subsidy subsidy Urban sprawl Average distance to city center −6.97 −3.65 −7.10 −8.73 Urbanized area −0.40 1.71 −0.51 −0.76 Average density in urban area 0.40 −1.68 0.51 0.77 CO2 emissions Paolo Avner, Shomik Raj Mehndiratta, Vincent Public transport share −19.15 −18.95 −19.14 −19.18 Viguié, and Stephane Hallegatte. 2017. “Buses, Houses or Cash? Socio-Economic, Spatial and CO2 emissions from commuting 1.28 5.11 1.12 −0.69 Environmental Consequences of Reforming Public Welfare Transport Subsidies in Buenos Aires.” Policy Re- search Working Paper 8166, World Bank, Wash- Household utility −3.73 1.30 −2.67 −2.69 ington, DC. 10World Bank ResearchDigest Improving the Resilience of Peru’s Road Network to Climate Events A new model helps identify where of removing one particular transport possible future risks to guide invest- investments can have the greatest link, such as a bridge or a road. They ment decisions. impact in keeping traffic flowing run this model for each link in the net- How does one choose whether to work, one by one. After considering ev- increase maintenance or to invest in during flood disruptions ery possibility, they identify the most bigger culverts and elevate roads? Or I critical links as those whose removal perhaps building new roads is a bet- n Peru nearly all cargo travels by from the network would force many ter option, to ensure that people have road. That can mean big challenges users to travel for much longer or at an alternative route if the main link is for businesses, because Peru’s dif- a much higher cost or would prevent disrupted? ficult topography and climate some- them from reaching certain destina- This is where the “decision mak- times lead to closures of important tions (figure 1). ing under deep uncertainty” method roads. In 1982–83, for example, a se- With the critical links identified, comes in. For each critical link the au- vere El Niño event caused flash floods the authors overlay a map of the trans- thors “stress-test” many different strat- and landslides, and most bridges port network with maps of extreme egies by calculating their economic on the northern parts of the Pan events such as floods and landslides. performance across a range of future Americana road, along the Peruvian After identifying the links exposed to conditions. They identify those that coast, were destroyed. Not all have such events, they estimate their vul- will yield acceptable economic returns been rebuilt, and many temporary nerability: What does it take to disrupt whatever happens to the climate, the structures remain. or shut down a road? How high can traffic, or the vulnerability of the link. How can countries like Peru im- the water rise before a bridge becomes Unsurprisingly, the most robust prove the resilience of their transport impassable for trucks? strategy depends on the context. networks in a context of scarce re- The next step is to determine how Increasing maintenance always yields sources and growing climate uncer- likely those climate events are. That’s positive returns and is often a robust tainty? A study by Rozenberg, Briceno- where things get tricky. For many plac- option, but it does not prevent cata- Garmendia, Lu, Bonzanigo, and Moroz es there are not enough historical data strophic impacts in places where the seeks to identify where investments to properly estimate the return period risk can be very high. In high-risk plac- can have the biggest impact on the of a climate event. And even when es, increasing redundancy is a better functionality of Peru’s transport net- there are, climate change is making option despite the high cost of build- work. The authors create a geospatial projections much less reliable—the ing new roads. model to simulate the consequences flood risks of yesterday are no longer a The stress tests also allow identi- good guide for fication of the main conditions that Figure 1. Critical Links in Peru’s Road Network: Estimated Effect of Their those of tomor- will make an investment succeed or Removal on Road User Cost and Distance, and Their Exposure to Floods row. Moreover, fail—and usually it’s only a few factors 600 vulnerabilities that matter. For example, building a Increase in cumulative road user cost (US$ per vehicle) Pan Americana are changing tunnel on the Carretera Central, one cluster 500 as mainte- of the most critical roads, would have V1 nance practices positive net benefits only under some 400 evolve. And conditions of discount rate and the finally, current frequency of annual disruptions on 300 traffic patterns that road. V2 are hard to Importantly, the authors find Carretera Central V3 identify, and that factoring climate risk into the 200 cluster future ones economic analysis of projects could Piura 100 cluster even harder change the prioritization of a port- to predict. folio and allow projects initially ne- All these fac- glected to get funds for completion or 0 tors together implementation. 0 50 100 150 200 250 300 350 400 Additional kilometers make it hard Daily economic cost Flood exposure to foresee the $0.0–2.0 million Moderate floods risk that the $2.0–4.0 million Severe floods transport net- Julie Rozenberg, Cecilia Briceno-Garmendia, Xijie > $4.0 million Severe floods with climate change work will face Coastal floods or vulnerable to sea-level rise Lu, Laura Bonzanigo, and Harry Moroz. 2017. in a few years. “Improving the Resilience of Peru’s Road Network Note: A moderate flood is more than 30 centimeters (cm) with a return period of 10 years or less, a severe flood is more than 50 cm with a return period of 10 years or less, and a coastal flood has a return period So the result to Climate Events.” Policy Research Working Pa- of 100 years or less. V1, V2, and V3 are segments on the Pan Americana. is a range of per 8013, World Bank, Washington, DC. World Bank Research Digest11 The Vulnerability of Manufacturing Firms to Unreliable Water Infrastructure Firms in poorer economies suffer In exploring the complex path- manufacturing firms are vulnerable to more from the effects of drought on ways through which water affects the the effects of drought on water sup- water supplies, and the problems private sector, a first step is to estab- plies, with the overall impact driven lish a set of narrower relationships. by the effect in low- and lower-middle- may worsen with climate change A recent paper by Islam and Hyland income economies. While wealthier W starts by asking two questions. First, economies appear to have water infra- ater affects the private sector what are the potential determinants of structure in place that can buffer the through a myriad of chan- poor water infrastructure? One issue effects of unanticipated droughts, this nels, direct and indirect. This is the extent to which water infrastruc- is not the case in these poorer econo- is evident from its prominent role in ture can cope with extreme variability mies (figure 1). the global Sustainable Development in rainfall. Climate change will result Beyond this, the study finds that Goals, nearly all of which are linked to in more frequent wet spells in more unreliable water supplies have a water. And consider the manufacturing humid areas—and more frequent dry negative effect on firm output that is sector, an important source of growth spells in more arid ones. This may both statistically and economically especially in developing economies. It have asymmetric effects across econo- significant. Moreover, this effect is provides jobs. It drives modernization. mies, as water infrastructure in the more acute in low- and lower-middle- And it can be hurt by inadequate water developing world may be ill equipped income economies and in water- infrastructure. Where the public water to cope with weather extremes. intensive sectors, further evidence that supply is unreliable, it can drive firms Second, how does the quality of the consequences of a changing cli- to costly alternative sources. water infrastructure affect firm perfor- mate will not be equally borne. These Yet few studies have given water mance? It could be that the reliability economies and sectors may need to infrastructure an important role in of water supply has little effect on firm consider policies to improve the reli- the private sector. The World Bank productivity and that the lack of atten- ability of water infrastructure or to Enterprise Surveys, which provide tion given to water as a factor in the boost their resilience to interruptions the most consistent coverage of firm- business environment has not been in water supplies. level data globally, do contain infor- a mere oversight. But it also could be With a changing climate and in- mation on water outages. But when that firms in many economies have creasing urbanization, the problems respondents are asked to rank the adapted production strategies to poor of poor water infrastructure in many top obstacles they face, water is not water infrastructure. economies may worsen in the foresee- among the options, though power, The study is the first to explore able future. Unless steps are taken to communication, and transportation both questions using a data set that improve the infrastructure, the effect are. Similarly, the World Bank’s Ease is truly global. The analysis draws on on the private sector will only grow. of Doing Business Index includes elec- firm-level data from the World Bank And in the face of looming water chal- tricity but does not cover water infra- Enterprise Surveys, covering more lenges, there is a greater need than structure or access to water. But all than 16,000 manufacturing firms in a ever for high-quality data on water in- this could change as the growing focus cross-section of 103 economies be- frastructure quality and private sector on climate change continues to push tween 2009 and 2015. water use to support future research. looming water challenges to the front The study brings to light sev- lines of policy. eral new insights. Across the globe, Figure 1. Severity of Water Infrastructure Problems by Income Group, Various Years, 2009–15 20 16 Percent or number 12 8 4 Asif Islam and Marie Hyland. 2018. “The Drivers 0 Low income Lower middle Upper middle High income and Impacts of Water Infrastructure Reliability: A income income Global Analysis of Manufacturing Firms.” Policy Firms experiencing a water shortage in previous year (% of total) Research Working Paper 8637, World Bank, For these firms, average number of water shortages per month Washington, DC. 12World Bank ResearchDigest Recent Policy Research Working Papers on Urbanization and Infrastructure Development 8496 Funding and Financing Infrastructure: The 8746 Assessing Rural Accessibility and Rural Roads Joint-Use of Public and Private Finance Investment Needs Using Open Source Data result in allocative inefficiency. This Marianne Fay, David Martimort, and Stephane Mehdi Mikou, Julie Rozenberg, Elco Koks, may in turn constrain returns to in- Straub Charles Fox, and Tatiana Peralta Quiros 8530 Urbanization in Kazakhstan: Desirable Cities, 8750 Are They Really Being Served? Assessing frastructure and exacerbate regional Unaffordable Housing, and the Missing Rental Effective Infrastructure Access and Quality in disparities, an important question for Market 15 Kenyan Cities William Seitz Sumila Gulyani, Andrea Rizvi, and Debabrata future research. 8546 Mobility and Congestion in Urban India Talukdar Prottoy A. Akbar, Victor Couture, Gilles 8762 Growing a Developing City: A Computable Duranton, Ejaz Ghani, and Adam Storeygard Spatial General Equilibrium Model Applied Harris Selod and Souleymane Soumahoro. 2019. 8547 Infrastructure, Value Chains, and Economic to Dhaka Julia Bird and Anthony J. Venables “Highway Politics in a Divided Government: Upgrades Xubei Luo and Xuejiao Xu 8764 Migration and Urbanization in Post-Apartheid Evidence from Mexico.” Policy Research Working 8548 Heterogeneous Impacts of Main and Feeder South Africa Paper 8710, World Bank, Washington, DC. Road Improvements: Evidence from Ethiopia Jan David Bakker, Christopher Parsons, and Ferdinand Rauch Atsushi Iimi, Haileyesus Mengesha, James Markland, Yetmgeta Asrat, and Kefargachew 8800 Rural Roads, Poverty, and Resilience: Public Infrastructure and 8552 Kassahun Water and Sanitation in Dhaka Slums: Access, Evidence from Ethiopia Shohei Nakamura, Tom Bundervoet, and Structural Transformation Quality, and Informality in Service Provision Mohammed Nuru Yurani Arias-Granada, Sabrina S. Haque, 8801 Common Transport Infrastructure: A George Joseph, and Monica Yanez-Pagans Quantitative Model and Estimates from the (continued from page 8) 8584 Incorporating Resilience in Infrastructure Belt and Road Initiative Prioritization: Application to the Road François de Soyres, Alen Mulabdic, and capital. Compared with the previ- Transport Sector Michele Ruta Darwin Marcelo, Schuyler House, and Aditi 8806 Who Wins, Who Loses? Understanding the ous scenario, the sector shares do Raina Spatially Differentiated Effects of the Belt not change much, because the same 8602 Prioritizing Infrastructure Investments: A and Road Initiative Somik V. Lall and Mathilde Lebrand amount of investment goods can now Comparative Review of Applications in Chile Darwin Marcelo, Schuyler House, and Aditi generate stronger capital accumula- Raina Papers can be downloaded at www.worldbank.org/prwp. tion. But the effects on GDP and sec- 8614 How Much Will the Belt and Road Initiative Reduce Trade Costs? toral production remain substantial. François de Soyres, Alen Mulabdic, Siobhan Murray, Nadia Rocha, and Michele Ruta Highway Politics in Mexico Finally, the study looks at the prac- 8628 Identifying Urban Areas by Combining Data tice in many developing economies from the Ground and from Outer Space: An (continued from page 5) of subsidizing electricity. By limiting Application to India Virgilio Galdo, Yue Li, and Martin Rama the government’s capacity to invest in 8641 Definition Matters: Metropolitan Areas government may be conducive to electricity infrastructure, this practice and Agglomeration Economies in a Large Developing Country greater distributive spending. In fact, can have a significant negative effect Maarten Bosker, Jane Park, and Mark the study finds evidence of greater dis- on GDP per capita and the share of Roberts 8660 States Diverge, Cities Converge: Drivers of tortion in highway provision when the manufacturing in the economy. Local Growth Catch-up in India president’s party held fewer legislative Yue Li, Martin Rama, and Qinghua Zhao seats than the opposition. Fidel Perez Sebastian and Jevgenijs Steinbuks. 8678 Measuring Urban Economic Density J. Vernon Henderson, Dzhamilya Thus distributive politics fueled by 2017. “Public Infrastructure and Structural Nigmatulina, and Sebastian Kriticos divided government, itself a manifes- Transformation.” Policy Research Working Paper tation of the democratic game, could 8285, World Bank, Washington, DC. The World Bank Research Digest is a quarterly publica- The Research Digest is financed by the World Bank’s Editorial Committee: Shanta Devarajan, David tion disseminating findings of World Bank research. Research Committee and managed by DECSO, the Rosenblatt, Aslı Demirgüç-Kunt, Roberta Gatti, The views and interpretations in the articles are those Strategy and Operations unit of the Development Hiau Looi Kee, Chorching Goh, and Shiva S. Makki of the authors and do not necessarily represent the Economics Senior Vice Presidency (DEC). The Research (managing editor). Editor: Alison Strong; production: views of the World Bank, its Executive Directors, or the Digest is not copyrighted and may be reproduced with Roula Yazigi; workflow: Pauline Kokila. For free countries they represent. appropriate source attribution. subscriptions email research@worldbank.org or visit http://www.worldbank.org/researchdigest. The World Bank 1818 H Street, NW Washington, DC 20433, USA Printed on Recycled Paper