86432 Reducing Poverty by Closing South Asia’s Infrastructure Gap Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 Acknowldgements The authors are grateful to those who contributed to the South Asia Infrastructure Needs Regional Study in various capacities. This paper is a synthesis of the findings of this study. Thus, there are many to be thanked: Sudeshna Banerjee, Ashma Basnyat, Cecilia Belita, Edgar C. Bouharb, Diana Cubas, Juan A. Echenique, Jorge J. Escurra, Céline Ferré, Kirsten Hommann, Atsushi Iimi, Ada Karina Izaguirre, Johannes G. P. Jansen, Rahul Kanakia, Pravin Karki, Bill Kingdom, Neetu Mihal, Pradeep Mitra, Diana Moreira, Mohua Mukherjee, Elisa Muzz- ini, John Newman, Shaheena Nisar, Sheoli Pargal, Mario Picon, Saurabh Naithani, Fernanda Ruiz Núñez, Ste- fanie Sieber, Rabin Shrestha, Govinda Timilsina, Gonzalo Vázquez Baré, and Tomoyuki Yamashita. The authors also thank Laura Wallace for her editorial support. The authors are grateful to the South Asia Regional Chief Economist, Martin Rama, and the former regional Chief Economist, Kalpana Kochhar, for their support and tech- nical input. The authors also greatly appreciate the guidance of Jack Stein, Gajan Pathmanathan, Jyoti Shuk- la, the South Asia Sustainable Development Department management team, and Country Management Units in the South Asia region. In addition, Marianne Fay (SDN, Chief Economist), Vivien Foster (SEGEN, Sector Manag- er), and José L. Guasch (Consultant) provided insightful and constructive comments through their capacities as peer reviewers. Financial support from SAR Chief Economist Office and the Australian Government is also grate- fully acknowledged. Senior authorship is not assigned. The findings, interpretations, and conclusions expressed in the reports and out- puts under this initiative are those of the authors, and do not necessarily reflect the views of the Executive Direc- tors of The World Bank, the governments they represent, or the counterparts that were consulted or engaged with during the informal study process. Any factual errors are the responsibility of the team. Photo credits: Power: Solar energy is used to light village shop. Sri Lanka. Photo: Dominic Sansoni / World Bank Telecom: Supervisors team of the SDO NGO in front of a ‘Basic Healt Clinic’ under construction in the village of Said Ahmad Qazi. Photo: © Nicolas Bertrand / TAIMANI FILMS / WORLD BANK Irrigation: An apricot nursery, supported by the HLP program. Afghanistan. 2008. Photo: © Sofie Tesson / TAIMANI FILMS / WORLD BANK Water: Girl getting water from community water pipe. Sri Lanka. Photo © Dominic Sansoni / World Bank Roads: 13 May 2012, Enjil District, Herat, Afghanistan :The 13 kilometer stretch of road leading to Nawin Enjil vil- lage outside of Herat that was developed under the auspices of the National Rural Access Program (NRAP). The villages of the area have benefitted from NRAP that has funded the completion of this road. The NRAP project aims to provide year -round access to basic services and facilities in the rural areas of Afghanistan to enhance the well being of the population and promote economic growth in the country. Under the project secondary roads are being rehabilitated by the Ministry of Public Works and tertiary roads by the Ministry of Rural Rehabilitation and Development.. Picture by Graham Crouch/World Bank Paulo Correa* and Irem Guceri† INTRODUCTION of infrastructure services. There are also enormous varia- Despite recent rapid growth and poverty reduction, the South tions within SAR countries. Districts with very low access to Asia Region (SAR) continues to suffer from a combination of infrastructure can be found in rich Indian states while dis- insufficient economic growth, slow urbanization, and huge in- tricts with high access can be found in poor states. More- frastructure gaps that together could jeopardize future prog- over, within the same district, high access rates to one ser- ress. It is also home to the largest pool of individuals living vice (for example electricity) can coexist with low access under the poverty line of any region, coupled with some of rates to other services, such as sanitation. the fastest demographic growth rates of any region. Between 1990 and 2010, the number of people living on less than It is commonly asserted that the poor have less access to US$1.25 a day in South Asia decreased by only 18 percent, infrastructure than the rich, similar to the case of private while the population grew by 42 percent.1 assets. In effect, a non-regressive access to infrastructure services would mean no correlation between actual access If South Asia hopes to meet its development goals and not and different poverty related measures (such as households risk slowing down—or even halting—growth and poverty al- below poverty lines, and certain income and consumption leviation, it is essential to make closing its huge infrastruc- levels). Whereas this may be desirable theoretically—espe- ture gap a priority. But the challenges on this front are mon- cially for infrastructures with high public good characteris- umental. Many people living in SAR remain unconnected to tics—it is virtually impossible to achieve anywhere in the a reliable electrical grid, a safe water supply, sanitary sewer- world. For example, location matters, and the choice be- age disposal, and sound roads and transportation networks. tween infrastructure access to all, regardless of where indi- This region requires significant infrastructure investment vidual households are located and quality access to where (roads, rails, power, water supply, sanitation, and telecom- most households are located, is a real policy challenge il- munications) not only to ensure basic service delivery and lustrated in its extreme case. While studies on the topic are enhance the quality of life of its growing population, but also scarce, it is clear that not all countries fare the same in their to avoid a possible binding constraint on economic growth infrastructure service provision, and SAR countries are no owing to the substantial infrastructure gap. different. Yet, do some SAR countries fare better with re- spect to providing infrastructure access to their poor? Are For the past two decades, SAR and East Asia and Pacific there infrastructure sectors that tend to be more regressive (EAP) have enjoyed similar growth rates, yet SAR lags signif- than others? What is happening with access to infrastructure icantly behind EAP, Latin America, and the Caribbean (LAC) services at the household and individual levels? when it comes to access to infrastructure services—with certain areas featuring access rates comparable only to In an effort to shed light on all of these questions, this re- Sub-Saharan Africa (SSA). At the same time, there are tre- port takes a critical look at the status of infrastructure in SAR mendous variations among countries in terms of access to compared with other regions, as well as among and within infrastructure services. Afghanistan, Nepal, and Bangla- desh have access rates that resemble the average Sub-Sa- 1 The proportion of people living on less than $1.25 a day de- haran country, while Sri Lanka and the Maldives are more creased from 54 percent to 31 percent (a 42 percent decrease), similar to Latin American countries in terms of average rates between 1990 and 2010, mainly due to the increase in population. SAR countries. It then explores inequality of access to infra- the most: women, the poor, and marginalized social groups. structure services across South Asia’s space (namely phys- There are no simple guidelines to follow however, given the ical space, poverty space, and income space) and across tremendous variations among and within SAR countries as to time (how access of the young will influence future oppor- who has access to infrastructure services (in terms of phys- tunities). Next, the report gives an estimate of the total cost ical location, income, and age). For example, while leading of regional infrastructure needs, along with the infrastruc- regions generally have better levels of access, many poor ar- ture investment trends in SAR countries, and proposes a eas enjoy levels of access that are similar to those of rich ar- framework on how to rank infrastructure needs. Finally, the eas. Plus some infrastructure services (like water) are more report examines ways to better use existing resources by re- equally distributed than others (like sanitation). That is why thinking infrastructure service provision—including the role providing some level of access is a start—even if those ser- of the private sector—and policy options to help the poorest vices are not of the highest quality. At the same time, poli- gain better access to infrastructure. cy makers should take into account which types of services best fit each population’s needs (such as septic tanks for a Our conclusion is that infrastructure deficiencies in South mountainous region but sewerage lines for a more accessi- Asia are enormous, and a mix of investment in infrastruc- ble urban area). ture stock and implementing supportive reforms will enable the region to close its infrastructure gap. As for the size of the infrastructure gap, we estimate that SAR needs to invest HOW SOUTH ASIA COMPARES WITH between US$ 1.7 and US$ 2.5 trillion (at current prices) in OTHER REGIONS infrastructure until 2020.2 In GDP terms, if investments are The demand for infrastructure has been growing global- spread evenly over the years until 2020, SAR needs to in- ly, especially in Asia, driven by a myriad of factors such as vest between 6.6 and 9.9 percent of 2010 gross domestic economic growth, technological progress, and urbaniza- product (GDP) per year—which would be an increase of up tion—putting greater and greater pressure on infrastructure to three percentage points from the 6.9 percent of GDP in- services that are already severely stretched. According to vested in infrastructure by SAR countries in 2009. the United Nations, five South Asian cities (Mumbai, Del- hi, Kolkata, Karachi, and Dhaka) are expected to surpass Faced with this enormous demand for infrastructure invest- the 15 million-person mark by 2015. Furthermore, accord- ment, and with only limited available financial resources, ing to the livability index produced by the Economist Intelli- it is critical for SAR to prioritize infrastructure investment gence Unit, four South Asian cities (Dhaka, Karachi, Kath- needs. The criteria used to accomplish this must be able mandu, and Colombo) are in the bottom 10 cities out of the to answer questions about short-term needs versus longer- 140 countries evaluated. term development needs, especially in developing coun- tries. For example, should infrastructure investment in the Yet structural change in South Asian countries has been rel- electricity sector be given priority over the transport sector? atively slow compared to that of East Asian countries—espe- Given substantial lock-ins associated with infrastructure in- cially since 1990, when they had similar urbanization rates vestments, should a country continue attempting to fill cur- (SAR, 25 percent; EAP 28 percent) and were close in terms rent gaps or direct investments to infrastructures that are of infrastructure service provision. While they both enjoyed likely large bottlenecks in the medium term? Moreover, it high growth rates over the next two decades, EAP has seen is not feasible to expect South Asian governments alone to rapid urbanization (50 percent in 2012) while SAR has re- shoulder the entire financial burden, underscoring the need mained the least urbanized region in the world (31 percent), for a bigger role for the private sector (such as through pub- well below the world urbanization rate (53 percent). In ef- lic-private partnerships). fect, departing from similar points, South Asian countries In addition, South Asian governments need to ensure that 2 The US$1.7 to US$ 2.5 trillion are at current prices, and they are infrastructure access is extended to the people who need it equivalent to US$1.4 to US$2.1 trillion at 2010 prices. 2 Reducing Poverty by Closing South Asia’s Infrastructure Gap TABLE 1: SAR LAGGING BEHIND ALL BUT SSA IN ACCESS TO INFRASTRUCTURE SERVICES Telecom Electricity Access to Access to Avg GDP Access (per Access (% Improved Improved Water Growth Urbanization 100 people) of pop.) Sanitation (% of (% of pop.) (2000–2012)a Rate (2012) (2011)b (2010)c pop.) (2011)d (2011)e East Asia and Pacific (EAP) 8.9% 50 98 92 67 91 Europe and Central Asia 4.4% 60 157 100 94 95 (ECA) Latin America and the 3.1% 79 125 94 81 94 Caribbean (LAC) Middle East and North Africa 4.2% 60 105 94 89 89 (MNA) South Asia Region (SAR) 6.7% 31 72 71 39 90 Sub-Saharan Africa (SSA) 4.7% 37 54 35 30 63 World 2.5% 53 103 78 64 89 Source: World Development Indicators, except when noted otherwise. Notes: a The average GDP growth for MNA is for the period 2000–2009; b Telecom access is defined as the number of fixed and mobile lines; c World Energy Outlook 2010 by International Energy Association; d Improved sanitation is defined as connection to a public sewer, a septic system, pour- flush latrine, simple pit latrine, and ventilated improved pit latrine; e Improved water is defined as household connection, public standpipe, borehole, protected dug well, protected spring, rainwater collection. are remarkably “under-urbanized” when compared to East quality and quantity of improved water may be in ques- Asian countries over the past half century. tion. Most of the access to water is through public stands; only 25 percent of the population has access to How large is SAR’s infrastructure gap compared with oth- piped water and 24/7 water supply is a rare exception er regions? At this point, its access to infrastructure services in South Asian cities.4 closely resembles SSA, even though its economic growth is ƒƒ Telecom access. Communication among people who second only to EAP (table 1). are not in close proximity is inefficient. In terms of tele- com access (measured as fixed and mobile lines per ƒƒ Electricity access. In SAR only 71 percent of the pop- 100 people), SAR and SSA rank at the bottom (72 and ulation enjoys the benefits of electricity access, ahead 54) with less than half the access found in ECA and of SSA at 35 percent, but way behind the rest of the re- LAC (157 and 125).This situation becomes even more gions at above 90 percent. According to businesses in dramatic given SAR’s low level of urbanization. South Asia, infrastructure is a major or severe hindrance ƒƒ Transport access. This other form of connectivity is also to their growth, and electricity is the largest problem. poor—a problem that troubles much of the developing ƒƒ Improved sanitation access. In this category, SAR world. Using total length of road network per 1,000 peo- (39 percent) is at the bottom with SSA (30 per- ple, SAR has 2.9 km—which is close to EAP (2.5 km), cent)—rates that are close to half the world average of SSA (2.5 km), and MNA (2.8 km), but well below the 64 percent population access. Open defecation seems world average (4.7 km), ECA (8 km), and North Amer- to be one of the most salient issues facing SAR—which ica (24 km). Furthermore, the transport infrastructure ranks as the region with the highest incidence of open suffers from serious shortcomings (such as lack of intra- defecation in the world—with 680 million people regional connectivity among the national road networks, (41 percent of the population) relying on it in 2011.3 ƒƒ Improved water access. This is the only indicator where South Asia is about even with the rest of the world and 3 WHO/UNICEF Joint Monitoring Program EAP, averaging 90 percent population access. Yet the 4 Ibid. Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 3 TABLE 2: BIG RANGE AMONG SAR COUNTRIES IN ACCESS TO INFRASTRUCTURE SERVICES Access to Infrastructure Services in SAR countries Access to Access to Telecom Access Electricity Improved Improved Water Total Road (per 100 Access (% of Sanitation (% of (% of pop.) Network (km per % Paved people) (2011)a pop.) (2010)b pop.) (2011)c (2011)d 1000 people)e Roadsf Afghanistan (AFG) 54 30 29 61 1.6 29 Bangladesh (BGD) 58 47 55 83 0.1 10 Bhutan (BTN) 69 65 45 97 9.7 40 India (IND) 75 75 35 92 3.5 50 Maldives (MDL) 173 95 98 99 0.3 100 Nepal (NPL) 47 76 35 88 0.8 54 Pakistan (PAK) 65 67 47 91 1.5 72 Sri Lanka (LKA) 104 77 91 93 5.5 81 Source: World Development Indicators, except when noted otherwise. Notes: a Telecom access is defined as the number of fixed and mobile lines; b World Energy Outlook 2010 by International Energy Association, except BTN and MDV, which are based on authors’ estimations; c Improved sanitation is defined as connection to a public sewer, a septic system, pour-flush latrine, simple pit latrine, and ventilated improved pit latrine; d Improved water is defined as household connection, public standpipe, borehole, protected dug well, protected spring, rainwater collection; e Varying data years: 2005 (MDV), 2006 (AFG), 2008 (IND, NPL), 2010 (BGD, BTN, PAK, LKA); f Varying data years: 2003 (LKA), 2005 (MDV), 2006 (AFG), 2008 (IND, NPL), 2010 (BGD, BTN, PAK). unrealized potential for rail and inland water freight improved sanitation, which is better than in LAC at 81 per- transport, and inadequate road and rail connectivity of cent. In terms of electrification, only Maldives (95 percent) ports with hinterlands). These limitations turn transport and Sri Lanka (77 percent) are above the average rate for infrastructure into a hindrance for regional and interna- developing countries (76 percent).6,7 On telecom, Sri Lanka tional trade, as indicated by investment climate surveys. and Maldives top the lists with 104 and 173 telephone lines per 100 people. This places Sri Lanka almost at the world average of 103 lines per 100 people and above EAP (98 HOW ACCESS TO INFRASTRUCTURE lines per 100 people). VARIES WITHIN SOUTH ASIA So who has access to each type of infrastructure? We be- Afghanistan, Nepal, and Bangladesh have the worst access gin with a cross country comparison among South Asian rates in the region. Nepal, with the lowest number of tele- countries and then we look at inequality of access across phone lines per 100 people in SAR (47), is behind Af- South Asia’s region, weighing physical (“spatial”), poverty, ghanistan (54) – which matches SSA (54). For electrifica- and income spaces—an assessment that has never been tion, Afghanistan, not surprisingly, is the worst; a meager done before for SAR.5 Mapping access to infrastructure with 30 percent of the population can rely on electricity powered characteristics of households provides a better understand- lighting at night. Moreover, Afghanistan and Bangladesh ing of the issues limiting access for some groups of the pop- ulation, and allows for better design and targeting of policies 5 See Biller et al. (2013) for a detailed discussion about the meth- to expand access to infrastructure services. odology in this section. 6 It should be noted that data sources are kept the same for con- sistency purposes when comparing countries. The Ceylon Electric- ity Board (CEB) estimates for example that over 90 percent of Sri Benchmarking within South Asia Lankan households were electrified in 2011. 7 World Energy Outlook/International Energy Association (IEA): Sri Lanka and Maldives have the best access rates in the re- http://www.worldenergyoutlook.org/resources/energydevelopment/ gion. More than 90 percent of their population has access to globalstatusofmodernenergyaccess/. 4 Reducing Poverty by Closing South Asia’s Infrastructure Gap (47 percent) are closer to the 35 percent found in SSA than What is surprising, however, is Pakistan’s relatively low levels to the 71 percent found in SAR. Total road network (km) per of spatial inequality. One possible explanation is the coun- 1000 people is also low in Nepal, Afghanistan, and Ban- try’s higher urbanization rate, with access to infrastructure gladesh—in Maldives it is also low, but it is explained by services more skewed to its cities relative to other countries geographical reasons. And only 29 percent of Afghanistan’s in the region. roads, and 10 percent of Bangladesh’s roads, are paved. We can also ask whether some infrastructure services fare The exception is high average access to improved water in SAR, better than others in terms of how they are distributed with- and not just in a few countries. Five out of the eight countries in countries to households. Again, the results show large in SAR (i.e., Bhutan, India, Maldives, Pakistan and Sri Lan- variations. The most unequally distributed service through- ka) have access rates to improved water of at least 90 per- out SAR is cooking gas (Liquefied Petroleum Gas (LPG))— cent, similar to the 94 percent rate found in LAC. likely reflecting its reliance on transport connectivity and its capital-intensive nature (it is mostly distributed in bot- tles). This heavy use of biomass for cooking, rather than the Inequality across physical space cleaner LPG, affects mostly children and women through One way of judging inequality is to analyze whether some indoor air pollution. In Sri Lanka, as Figure 1 shows, there countries in South Asia do a better job of making access to in- are “mountains” of firewood and poverty.9 frastructure more equal within the country spatially than oth- ers. We do this by focusing on a lower administrative level (such as a district or province) and measuring inequality with Gini coefficients.8 A Gini coefficient of zero represents per- 8 The first step is to estimate a Gini coefficient over total households that have access to a given type of infrastructure service at the ad- fect equality, while a coefficient of one represents maximal ministrative level. But given that differences in the distribution of inequality. Our goal is to come up with a country-level mea- the access could be determined by where the households are allo- sure of spatial inequality of infrastructure access adjusted by cated in the country, we actually estimate the Gini coefficient over the number of households of each administrative area. The next household spatial distribution, which is presented in table 3. step is to subtract the Gini coefficient of population from the Gini of connections to see if there are areas in a country that are not re- ceiving a rate of access proportional to their population. Access to The results show quite a varied picture, with the Maldives a particular infrastructure service is spatially evenly distributed if its having the lowest—and Afghanistan the highest—inequal- Gini coefficient is equal to the Gini coefficient of households—al- ity of access to infrastructure services in the region. These though this could also mean an equal absence of services. 9 Improved water is a noticeable outlier in the Maldives which may results are not surprising, especially in the case of Afghan- be explained by small atolls having more access in relative terms istan, given its level of development and years of conflict. because of the inclusion of rain water in improved water. TABLE 3: TREMENDOUS INEQUALITY OF ACCESS ACROSS SAR’S PHYSICAL SPACE Gini Coefficients of Access to Infrastructure in South Asia Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka   Access Gini Coefficients Adjusted by Household Distribution Improved Water 0.12 0.01 0.00 0.06 –0.10 0.04 0.02 0.01 Improved Sanitation — –0.01 0.18 0.29 0.01 0.06 0.02 0.01 Electricity 0.49 0.11 0.10 0.15 0.00 0.04 — 0.04 Cooking Gas 0.50 0.49 0.22 0.35 0.01 0.24 0.00 0.33 Phone 0.28 0.15 0.19 0.20 0.00 0.05 0.03 0.03 Source: Authors’ calculations based on surveys presented in Andres et al. (2013). Note: The Gini coefficients are estimated over a sample of administrative subdivisions selected on each country. An access Gini coefficient adjusted by household distribution of one expresses maximal inequality. An adjusted coefficient of zero expresses perfect equality. Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 5 FIGURE 1: SRI LANKA’S MOUNTAINS OF FIREWOOD line) and access move together. In this analysis, high re- AND POVERTY gressivity means that poor districts typically have less ac- Total Population that Uses Firewood for Cooking Relative to Poor cess compared to richer districts, whereas low regressivi- Households ty means access is more equitably distributed among poor and rich districts alike. High usage Color Our results show that overall the link is regressive, but the Low usage strength of that link varies greatly among the countries and High poverty among sectors. Height Low poverty ƒƒ India shows strong regressivity of infrastructure service access, except in water and phone services (Figure 2). Note: This is an inverted view of Sri Lanka for visual purposes. The red The water exception is similar to that in the other coun- peaks indicate both high usage and high poverty. The solid blue color area with no mountains indicates the absence of data for the north tries. and part of the east. ƒƒ Sri Lanka shows a relatively weak regressivity of infra- structure service access (Figure 3). The exceptions to this trend are cooking gas and phone, which show a much stronger link. The most equally distributed service throughout SAR is im- ƒƒ Afghanistan shows a relatively strong regressivity, ex- proved water, which is important in terms of welfare im- cept in water and in Kabul (its capital), an area with pacts given that no one can survive without it. The other one of the country’s lowest poverty rates (Figure 4). sectors—improved sanitation, electricity, and phones—fall Cooking gas seems less regressive than in India and somewhere in between the two extremes. Households, ap- Sri Lanka. propriately or not, generally solve their own sanitation needs, so the incentives to invest in adequate technologies are more limited than in the case of water. Like water, electric- Inequality across physical and poverty space ity is a direct benefit to the household as opposed to sani- Yet another question that we can explore, this time by tation. One would expect that households are willing to pay bringing together poverty data and physical access data more for adequate electricity, however appropriate electrici- for India and Sri Lanka, is whether poorer regions with- ty services can be costly. in a country have less access to key public services as a whole. We do this by constructing two infrastructure in- dexes10—which encompass only the basic infrastructure Inequality across poverty space services that have the highest impact on welfare (such as Another way of judging inequality is to analyze how pock- water, sanitation, and electricity)11—and then use these to ets of poverty fit into the picture, in effect, introducing a generate maps depicting how poverty and the location of socio-economic variable. We assume that a country with infrastructure services intersect. a higher poverty rate will have worse access to infrastruc- ture services than a country with a lower one; but how Our results show that leading regions generally mean better strong is that link? To do this, we correlate the district pov- access but lagging regions do not necessarily mean worse erty rate (percentage of people of each district that live under the poverty line) with the district rate of access to infrastructure (percentage of households that have infra- 10 The two methods are: (i) equal weights; and (ii) multicriteria de- structure in each district) for India, Afghanistan, and Sri cision-making approach assigning weights according to household Lanka. This correlation helps us understand the income level infrastructure service importance. 11 Maps could be generated for each infrastructure service, but this regressivity of access to these services by sector—that is, analysis is easier done via other means as discussed later in the to what extent income (being above or below the poverty paper. 6 Reducing Poverty by Closing South Asia’s Infrastructure Gap FIGURE 2: I NDIA’S INFRASTRUCTURE SERVICES ACCESS IS STRONGLY REGRESSIVE India Access to Infrastructure and Poverty Rates, in Percent a. Electricity b. Cooking Gas (LPG) 100 100 90 90 80 80 70 70 Rate of Access Rate of Access 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 20 40 60 80 0 20 40 60 80 Poverty Rate Poverty Rate c. Improved Sanitation d. Improved Water 100 100 90 90 80 80 70 70 Rate of Access Rate of Access 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 20 40 60 80 0 20 40 60 80 Poverty Rate Poverty Rate e. Phone 100 90 80 70 Rate of Access 60 50 40 30 20 10 0 0 20 40 60 80 Poverty Rate Source: Authors’ calculations based in infrastructure access from India DHLS-3 and poverty rates from Debroy and Bhandari (2003). Note: The size of each point is based on the population size. The coefficients associated with the scatter plots are –0.89 (significant at 99% of confidence) for electricity, –0.44 (significant at 99% of confidence) for cooking gas, –0.67 (significant at 99% of confidence) for improved sanita- tion, –0.02 for improved water, and –0.64 (significant at 99% of confidence) for phone. access. As expected, in India the lagging states (those with Western Province—enjoys better access and a lower pov- a higher poverty level) have higher access to basic infra- erty rate. Yet, for the lagging provinces, the story is more structure, in contrast with the leading states (those with a mixed, except for those where the country’s 30-year conflict lower poverty level). This is intuitively expected. The curi- was more present.12 ous exception is the northeast area bordering Bangladesh, Bhutan, China, and Myanmar. The exception found in In- dia is more prevalent in Sri Lanka, where basic infrastruc- ture seems to be more inclusive. Access is widely spread, and the quality of these services in the country is known to be generally good. It is clear that the leading region—the 12 See Biller et al. (2013) for further discussion on this issue. Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 7 FIGURE 3: SRI LANKA’S INFRASTRUCTURE SERVICES ACCESS IS ONLY WEAKLY REGRESSIVE Sri Lanka Access to Infrastructure and Poverty Rates, in Percent a. Electricity b. Cooking Gas (LPG) 100 100 90 90 80 80 70 70 Rate of Access Rate of Access 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 20 40 60 80 0 20 40 60 80 Poverty Rate Poverty Rate c. Improved Sanitation d. Improved Water 100 100 90 90 80 80 70 70 Rate of Access Rate of Access 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 20 40 60 80 0 20 40 60 80 Poverty Rate Poverty Rate e. Phone (Mobile and Landlines) 100 90 80 70 Rate of Access 60 50 40 30 20 10 0 0 20 40 60 80 Poverty Rate Source: Authors’ calculations based in Sri Lanka HIES 2010. Note: The size of each point is based on the population size. The coefficients associated with the scatter plots are: –1.51 (significant at 99% of confidence) for electricity, –2.34 (significant at 99% of confidence) for cooking gas, –0.25 for improved sanitation, –0.9 for improved water, and for phone –1.69 (significant at 99% of confidence. Inequality across income space Our results show that in Sri Lanka and Afghanistan, the One question that still remains is how equitably access is rich enjoy better access than the poor, but the countries distributed across different income levels—that is, for those differ greatly in how equal that access is across incomes. with access, is it the richer individuals who have the bulk of In Sri Lanka, the difference in access across quintiles is the access or do richer and poorer individuals tend to have small—all quintiles are close to the mean—meaning that more similar access? This matters because it allows policy there is an almost equal share of access to infrastructure makers to better target policies to expand access. To answer regardless of income quintile. The opposite story is true in this question, we compared income quintiles and access Afghanistan. rates for Afghanistan and Sri Lanka. 8 Reducing Poverty by Closing South Asia’s Infrastructure Gap FIGURE 4: AFGHANISTAN’S INFRASTRUCTURE SERVICES ACCESS IS REGRESSIVE Afghanistan Access to Infrastructure and Poverty Rates, in Percent a. Electricity b. Cooking Gas (LPG) 100 100 90 90 80 80 70 70 Rate of Access Rate of Access 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 20 40 60 80 0 20 40 60 80 Poverty Rate Poverty Rate c. Improved Water d. Phones 100 100 90 90 80 80 70 70 Rate of Access Rate of Access 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 20 40 60 80 0 20 40 60 80 Poverty Rate Poverty Rate Source: Authors’ calculations based in Afghanistan NVRA 2008. Note: The size of each point is based on the population size. The coefficients associated with the scatter plots are –0.03 for improved water, –0.39 (significant at 90% of confidence) for electricity, –0.36 for cooking gas and –0.66 (significant at 95% of confidence) for phone. However, some services (like water) are more equitably dis- HOW ACCESS TO INFRASTRUCTURE tributed than others (like cooking gas) among those with ac- AFFECTS OPPORTUNITIES FOR YOUTH cess. In Afghanistan, the equality of access across income So how unequal is access to infrastructure among children? quintiles is particularly striking for improved water. Whether After all, infrastructure investment choices to fill the infrastruc- poor or rich, the shares of quintile over the total connection ture gap made today affect current and future generations. in the country hardly deviate from the mean. This is particu- Moreover, not addressing the infrastructure gap threatens both larly remarkable given that access to improved water is very welfare and economic growth in the medium and long term. low in the country—significantly lower than all other coun- tries in South Asia and the region’s average. Regardless of years of conflict and scarcity of service, it seems that the Af- The Human Opportunity Index: Access to ghani society has emphasized sharing household access to Infrastructure as Opportunity water. Whether in Sri Lanka or Afghanistan, the use of cook- Our main instrument for measuring the inequality of ac- ing gas is particularly prevalent for the highest quintile, mak- cess to infrastructure across time is the Human Opportuni- ing it the rich’s form of cooking. The reason, as discussed, ty Index (HOI), which was first published in 2008, and was is the capital intensive nature of LPG, its reliance on network used to evaluate access in Latin America. It essentially mea- connectivity, and the easy available of cheaper, albeit inferi- sures how personal circumstances impact a child’s proba- or, alternatives.13 bility of accessing the services that are necessary to suc- ceed in life. This is critical because the opportunities a child 13 This has been underscored in the literature as well; see Kojima et gets throughout life are determined directly by the circum- al. (2011) and Kojima (2011). stances related to access to infrastructural services during Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 9 their formative years—not necessarily to their personal deci- Can we ascertain how much individual circumstances drive sions or level of effort. the HOIs for each type of infrastructure? We explore this by looking at all 9 types of infrastructure, and then calculating This study calculates an HOI that is focused on basic infra- both the dissimilarity/inequality index and HOI for each indi- structure as opportunities, and the importance of both im- cator, and then the contribution of each circumstance to the proving overall access to it and ensuring its equitable allo- HOI. A few patterns stand out: cation to achieve key socio-economic outcomes—such as early childhood development, education completion, good ƒƒ Indicators linked to a higher quality of access register health, and access to information. It can be interpreted as significantly higher inequality of opportunity than more a composite indicator of two elements: (i) the level of cov- standard, general indicators of access (improved water erage of basic opportunities necessary for human devel- source versus piped water, improved sanitation versus opment (such as access to primary education, water and sewerage). Hence, general indicators of access hide sanitation, or electricity); and (ii) the degree to which the differences in types of access among different circum- distribution of those opportunities is conditional on circum- stance groups. stances children are born into (such as gender, income, or ƒƒ At the country level, two factors—the location of the household characteristics). For this study we have select- household (urban versus rural) and the education of ed four circumstances: (i) household size, (ii) location (ur- the household head—explain over 70 percent of the ban versus rural), (iii) education of household head, and (iv) HOIs across countries and across indicators (most of gender of household head.14 the times, over 80 percent). ƒƒ In the case of India, at the state level location is still a key factor, but the role of education of the household head, Inequality of Opportunity in the Access to and in a few cases, household size, gain in importance in Infrastructure Services explaining HOIs. And while the contribution of gender of Our results show that typically, South Asian countries with the household head seems negligible at the country lev- better infrastructure coverage also provide more equitable el, there are a number of states where access is unevenly access for households with children under 15, thereby of- distributed among female and male headed households. fering higher HOIs. Take the case of improved sanitation (ta- ble 4). As expected, countries with the highest coverage (the Maldives and Sri Lanka) feature the lowest dissimilarity in- PINNING DOWN THE dex, and therefore, the HOI is very close to the coverage. In “INFRASTRUCTURE GAP” contrast, countries with low levels of coverage (such as Ban- Over time, societies inherit man-made infrastructure stock gladesh) are associated with higher discount rates, and thus from previous generations. Yet different factors influence de- lower HOIs. However, if we take two similar access rates, mand and supply, and as countries grow these needs—both such as in the case of India (36 percent) and Nepal (37 per- the type of infrastructure and the quality of service provi- cent), we see a significant difference in how that access to sion—are likely to evolve. In this report, we assess the in- sanitation is distributed, with Nepal (0.14) being more even frastructure gap using a four-step process, as illustrated in than in India (0.24)—which results in India having a lower Figure 5. It shows (1) where a country is today; (2) where a HOI (0.27) than Nepal (0.32). Also, a country with higher country would like to be at a given point in time; (3) the dif- coverage, like Pakistan (44 percent), but featuring the same ference between the two points (i.e. the infrastructure gap- dissimilarity index as India (and thus higher than for Nepal), how far business-as-usual scenarios, shown by the dotted still ends up with a higher HOI than India but with a similar one to that of Nepal. In the case of access to improved wa- ter, rates are high enough to guarantee a low dissimilarity in- 14 These circumstances reflect previous inequality of opportunities studies and are in line with similar analyses that are part of the SAR dex. Electricity also follows the pattern of decreasing dissim- Regional Flagship Report on Inequality of Opportunities (forthcom- ilarity index as coverage is higher. ing in 2014). 10 Reducing Poverty by Closing South Asia’s Infrastructure Gap TABLE 4: BETTER COVERAGE TYPICALLY GOES WITH MORE EQUITABLE ACCESS AND THUS HIGHER HOIs Access to Infrastructure Services and Human Opportunity Index for household with children under 15 years old Country Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Year 2008 2006 2007 2007 2009 2011 2006 2010 Improved Sanitation Coverage — 45% 38% 36% 94% 37% 44% 90% Dissimilarity Index — 0.15 0.24 0.24 0.01 0.14 0.23 0.03 HOI — 0.38 0.29 0.27 0.93 0.32 0.34 0.87 Improved Water Coverage 46% 98% 96% 83% 86% 88% 92% 88% Dissimilarity Index 0.11 0.00 0.01 0.03 0.09 0.02 0.01 0.02 HOI 0.41 0.97 0.95 0.80 0.78 0.86 0.91 0.86 Electricity Coverage 17% 50% 72% 68% 100% 75% — 85% Dissimilarity Index 0.58 0.20 0.12 0.12 0.00 0.08 — 0.05 HOI 0.07 0.40 0.63 0.60 1.00 0.69 — 0.81 Source: Authors’ calculations based on NVRA 2008 for Afghanistan, MICS 2006 for Bangladesh, BLSS 2007 for Bhutan, DLHS-3 for India, DHS 2009 for Maldives, DHS 2011 for Nepal, DHS 2006 for Pakistan and HIES 2010 for Sri Lanka. All estimations represent the proportion of access in the sample with children below 15 years of age. Note: The column coverage is estimated over the predicted values of access to infrastructure for each household. Piped water is restricted – “Piped water in the premises.” For the Bangladesh improved water rate of access we did not use the assumption made by JMP where they discard 20 percent of protected wells due to arsenic contamination. For Bangladesh improved sanitation data, to make rates comparable, we include as JMP pit latrines without slabs in the category of improved sanitation (when is categorized as unimproved). For further information about these changes in the JMP methodology check JMP data by country (http://www.wssinfo.org/documents-links/documents/?tx_displaycontroller[type]=country_files). Due to the lack of information about landlines in Bhutan the definition of phone (mobile and landlines) is the same as mobile phones. Sri Lanka sewerage connection is not presented because it is not identifiable in the data. For Pakistan, there is no information about the tenancy of mobile phones in the household. The lack of information about access to electricity in Pakistan is caused by significant differences with the official data (World Energy Outlook) which motivates us to think the estimates in these countries are not comparable definitions of access to electricity. blue line, will take the country toward reaching its goal);15 FIGURE 5: A FRAMEWORK FOR ASSESSING (4) how far financial and policy options using existing re- INFRASTRUCTURE NEEDS sources (shown by the dotted red line) could take the coun- Coverage 2 Target try toward reaching its goal; and (5) the remaining financial 5 Financial Gap gap that will need to be bridged. Keep in mind that the im- 4 Financial and policy options portance of the financial gap will vary among countries, de- Infrastructure Gap using existing resources pending on how well better use of existing resources can 3 close the infrastructure gap. Level of investments (% of GDP) 1 At this point, most governments in SAR have some esti- mates of the investments required to reach certain targets, such as 24/7 electricity supply and the MDGs in water and Today Time sanitation, but those estimates are not consistent across the Source: Authors’ elaboration. 15 Given that (3) is the sum of (4) and (5), the four steps are (1), (2), (4), and (5). Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 11 TABLE 5: SAR’S TOTAL INVESTMENT TAB COULD REACH AROUND $2 TRILLION (Investment Requirements 2011–2020 (total, in billions of dollars 2010)   Bangladesh India Nepal Pakistan Sri Lanka SAR (5)a SAR (8)b   Low High Low High Low High Low High Low High Low High Low High Transport 36.0 45.0 340.0 595.0 3.7 5.5 17.2 21.5 10.8 18.0 408 685 411 691 Electricity 11.0 16.5 375.0 468.8 5.3 7.0 64.0 96.0 4.8 9.0 460 597 464 603 WSSc 12.0 18.0 95.0 162.0 1.7 2.6 9.3 14.0 0.6 1.8 119 198 120 200 Solid Waste 2.1 4.2 32.5 65.0 0.4 0.5 3.3 6.7 0.2 1.3 39 78 39 78 Telecom 5.0 5.0 150.0 225.0 0.4 0.6 12.4 12.4 2.0 2.5 170 246 171 248 Irrigation 7.7 11.6 140.0 210.0 1.6 2.3 9.7 14.6 2.5 3.1 161 242 163 244 Total 74 100 1,133 1,726 13 18 116 165 21 36 1,356 2,045 1,369 2,064 Source: Andres et al. (2013). Note: Estimations based on the technical models as well as the extrapolations for the other sectors where the models were not run. a SAR (5): Bangladesh, India, Nepal, Pakistan, and Sri Lanka. b SAR (8): Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan, and Sri Lanka. Based on an extrapolation from SAR (5) estimates. c WSS: Water Supply and Sanitation. region. That is why we developed different methodologies was electricity generation—which accounts for more than for different sectors. These models also allow us to calcu- a third (37 percent) of the total portfolio during this period. late and compare the costs of different sets of targets. Each In contrast, investment in transport (31 percent), irrigation model was applied to one country, which gives five sector- (15 percent), and water supply and sanitation (7 percent) country combinations: (i) power – Nepal; (ii) transport – were much more stable, although telecom (11 percent) has Sri Lanka; (iii) water and sanitation – India; (iv) solid waste been on a steady rise. management – Sri Lanka; and (v) irrigation – Afghanistan. As expected, these overall patterns have been largely driven So how much money will be needed to close the infrastruc- by India, which contributes the biggest share of total infra- ture gap? We estimate that SAR needs to invest US$ 1.7 to structure investment in South Asia during the 1973–2009 pe- US$ 2.5 trillion in infrastructure until 2020—equivalent to riod. In fact, infrastructure investment in India makes up on US$ 1.4 to US$ 2.1 trillion at 2010 prices (table 5).16 Going average 79 percent of total investment in the region. The sec- forward, a mix of investing in infrastructure stock and imple- ond largest contributor—Pakistan—has an average share of menting supportive reforms will enable the region to close its only 12 percent, and is followed by Bangladesh with 7.9 per- infrastructure gap. In GDP terms, if investments are spread cent, Nepal with 1.0 percent, and Bhutan with 0.2 percent. evenly over the years until 2020, SAR needs to invest be- tween 6.6 and 9.9 percent of 2010 GDP per year—an in- 16 See Andres et al. (2013) for the description on the methodology crease of up to 3 percentage points from the 6.9 percent of for computing these estimates. GDP invested 2009 (table 6). 17 While the main source of the data for these investments are multi-year development plans prepared by the National Planning Commissions, annual reports from ministries, state banks, and oth- What are the odds that SAR can put together enough funds er related government agencies have also been used in order to to meet these investment targets? An inspection of past in- form as much as a complete picture as possible. The data used is a mixture of estimated and actual expenditure, as not all plans state frastructure investment trends suggests this will be difficult actual expenditure from the previous plan or fiscal year. Further- to do.17 Certainly, the 2009 level of 6.9 percent is much high- more, these plans do not distinguish between Capital and Opera- er than the 1973 level of 4.7 percent, although there were tional Expenditures (CAPEX and OPEX). For public sector invest- ment, South Asia is defined as: Bangladesh, Bhutan, India, Nepal, many fluctuations around this trend from 1973 to 2011. The and Pakistan. Afghanistan, the Maldives, and Sri Lanka are not in- main driver of these fluctuations, especially in the 1980s, cluded in this definition due to data limitations. 12 Reducing Poverty by Closing South Asia’s Infrastructure Gap TABLE 6: CLOSING SAR’S INFRASTRUCTURE GAP WILL REQUIRE INVESTING A HIGHER SHARE OF GDP Investment Requirements 2011–2020 (% of GDP, per year)   Bangladesh India Nepal Pakistan Sri Lanka SAR (5)a   Low High Low High Low High Low High Low High Low High Transport 3.60 4.50 1.97 3.44 2.32 3.49 0.98 1.23 2.17 3.64 1.97 3.31 Electricity 1.10 1.65 2.17 2.71 3.34 4.46 3.66 5.49 0.97 1.82 2.22 2.89 WSSb 1.20 1.80 0.55 0.94 1.08 1.62 0.53 0.80 0.12 0.37 0.57 0.96 Solid Waste 0.21 0.42 0.19 0.38 0.24 0.30 0.19 0.38 0.04 0.27 0.19 0.38 Telecom 0.50 0.50 0.87 1.30 0.27 0.40 0.71 0.71 0.40 0.50 0.82 1.19 Irrigation 0.77 1.15 0.81 1.21 0.99 1.48 0.55 0.83 0.50 0.63 0.78 1.17 Total 7.38 10.02 6.55 9.98 8.24 11.75 6.63 9.44 4.21 7.23 6.55 9.89 Source: Andres et al. (2013). Note:These percentages are based on the investment requirements at 2010 prices. They are based on the technical models as well as extrapola- tions for the other sectors where the models were not run. a SAR (5): Bangladesh, India, Nepal, Pakistan, and Sri Lanka. b WSS: Water Supply and Sanitation. These differences in the shares of total infrastructure invest- happen in generation (60),20 mainly through build-operate- ments in the region are roughly in line with the relative size transfer arrangements. This is despite the fact that PPPs are of each economy. The average infrastructure investment as a the optimal organizational structure in transmission, not in percentage of GDP for the period 1973–2009 hovers around generation or distribution. 6 percent for India, Pakistan, and Bangladesh, and 5 percent for Nepal. Bhutan, with infrastructure investments represent- Even so, public provision is still the norm in South Asia. ing 14.6 percent of its GDP, has given significantly higher im- According to the World Bank Private Participation in In- portance to infrastructure development. frastructure Database,21 there are fewer than 1,000 active projects in the energy, telecom, transport, and water and Given that a significant share of the infrastructure invest- sanitation sectors under PPPs or fully owned by the pri- ment in the years ahead will have to come from the private vate sector.22 This number is low when compared with the sector, it is interesting to note that some sectors—such as energy and telecom—have drawn a lot more private investor 18 The core source for private sector investment is the World Bank interest than others (table 7).18, 19 On a country level, while Private Participation in Infrastructure Database (for a detailed ex- planation on the PPI methodology, refer to: http://ppi.worldbank. India has the largest presence by far, with 85 percent of re- org/resources/ppi_methodology.aspx). In this section, Private Par- gional private investment commitments, it follows Bhutan ticipation in Infrastructure (PPI) investment consists of all eight and Maldives, in terms of proportion of investment commit- countries in the region—Afghanistan, Bangladesh, Bhutan, India, the Maldives, Nepal, Pakistan, and Sri Lanka. ments with respect to GDP. 19 Irrigation is not included in the definition of infrastructure for this section. This is primarily because there is not enough data Of these investments, as Figure 6 shows, some sectors at- available but also because irrigation is primarily a public sector domain. tract full privatization more than others. In transport, the pri- 20 As of August 2013, there are 69 active projects in the power sec- vate sector tends to partner with the public sector through tor under lease contract, concession, or Build, Operate, Transfer (BOT), according to the World Bank Private Participation in Infra- Public-Private Partnerships (PPPs); while in telecoms, it structure Database tends to invest by itself (regulated privatization). When it 21 http://ppi.worldbank.org comes to energy, the private sector chooses to invest main- 22 The database considers 7 possible statuses for a project (i.e., cancelled, concluded, construction, distressed, merged, operation- ly by itself, but also through partnerships with the public al, under development). We consider a project is active, if it is not sector. Many of the PPPs in the power sector in South Asia cancelled or concluded. Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 13 TABLE 7: PRIVATE INVESTORS FAVOR ENERGY AND TELECOM (Total Private Sector Investment Commitments in Current US$ Mil (1990–2012) Water Supply % of total % of GDP   and Sanitation Transport Energya Telecom Total PPIb (2007–2012) Afghanistan 0 0 2 1,683 1,685 0.47 0.01 Bangladesh 0 0 3,285 6,855 10,140 2.83 1.14 Bhutan 0 0 201 18 219 0.06 2.65 India 470 81,098 135,703 89,054 306,325 85.49 2.57 Maldives 0 478 0 84 562 0.16 4.09 Nepal 0 0 997 135 1,132 0.32 0.66 Pakistan 0 2,555 13,416 17,090 33,061 9.23 1.59 Sri Lanka 0 740 1,438 3,003 5,181 1.45 1.04 SAR 470 84,871 155,042 117,922 358,305 100 2.34 Source: World Bank Private Participation in Infrastructure Database. Note: % of GDP was computed as a simple average across the specified period. For Afghanistan, % of GDP is from 2001-2011 due to a lack of data. a Energy entails a combination of electricity and natural gas. b PPI: Private Participation in Infrastructure. more than 400 power plants in the region; the extension outputs, and commonly debated market failures. The infra- of the electricity transmission network; the large number structure services listed often fall under the public sector, but of cities where electricity distribution, water, and sanitation in some cases may be a combination of public and private networks exist or are needed; the more than 400 seaports provision. For example, sanitation via off-site systems is typ- and airports; and the extension of the road network. ically provided by public utilities, but on-site sanitation such as septic tanks are generally private investments. The list is PRIORITIZING INFRASTRUCTURE not meant to be exhaustive and provides ranks from 1 to 3, INVESTMENT with 1 being the lowest relative weight, based on the existing Which investment projects should SAR countries tackle first, literature on infrastructure services and its impacts. or even second, third, and fourth? Not surprisingly, how to pri- oritize investment projects or portfolios is a common question How much financial resources should be allocated to infra- a government at any jurisdictional level asks. This question is structure development, within infrastructure sectors, and especially critical in developing countries, particularly in South other sectors (such as health, education, public safety, and Asia and Sub-Saharan Africa, where demand for investment is national defense)? This is another question asked by all de- huge and financial resources are limited. A few existing stud- veloping countries, but unfortunately, there is no rule to deter- ies attempted to address this question,23 but the methodolog- mine the investment allocations.24 It depends on a country’s ical framework they developed is narrow and can be applied priority, economic growth, and welfare objectives. Considering only to rank infrastructure investment projects. that infrastructure is both a means to facilitate this econom- ic growth and development, and a measure of the former, one For that reason, we developed a methodology to help priori- tize infrastructure needs in developing countries, particular- ly in South Asia. It consists of three main steps: (i) Identi- 23 See, for example Berechman and Paaswell (2005); Karydas and Gifun (2006). fying factors that affect infrastructure investment decisions, 24 Some existing literature attempted to address this question. (ii) Quantifying identified factors, and (iii) Ranking the infra- Based on information from a previous study (Fay and Yepes, structure projects. Table 8 presents some stylized rankings, 2003), Estache and Fay (2010) estimate that developing countries might need 6.5% of their GDP, in average, during 2005–2015 peri- which relate infrastructure services according to input intensi- od. Of which 2.3% would be needed to maintain the existing infra- ty of use, degree of spatial manifestation, typical development structure and 3.2% for new infrastructure projects. 14 Reducing Poverty by Closing South Asia’s Infrastructure Gap FIGURE 6: ELECTRICITY AND TELECOM FAVOR PRIVATIZATION WHILE TRANSPORT AND WSSa FAVOR PPPs 100% 90% 80% 70% Breakdown (%) 60% 50% 40% 30% 20% 10% 0% BGD BTN IND MDV NPL PAK LKA BGD BTN IND NPL PAK LKA IND MDS PAK LKA IND Electricity Telecom Transport WSS PPP Privatization Source: World Bank Private Participation in Infrastructure Database. Note: PPPs include lease contracts, concessions, and Greenfield projects under Build, Lease, Transfer (BLT) or Build, Operate, Transfer (BOT); while privatization includes Greenfield projects under Build, Own, Operate (BOO), merchant, or rental, and full divestitures. Partial divestitures are not considered due to the lack of information on the role of the private sector on the management of the facility. a WSS: Water Supply and Sanitation. could expect that a higher share of GDP (including funds re- economic growth. Given that most power sources are limit- ceived from bilateral and multilateral donors) would need to ed, there is a clear policy choice related to power allocation be allocated for infrastructure investment. This is the case, at for different usages. Instruments such as tariffs and other least, for developing countries, where there is greater scarci- incentives play a vital role in allocating this scarce resource. ty of man-made and human capital related to infrastructure. Moreover, policy makers should be cognizant that attempt- ing to apply the same standards across the board regardless That said, there is a false dichotomy between prioritizing of income may translate into no provision to the poor. large-scale infrastructure versus addressing the needs of the poor. At a very basic level, this dichotomy is false be- Ultimately, both types of investments are needed—those that cause many large-scale infrastructure investments may clearly target economic growth in the short run and those concurrently facilitate economic growth and increase the that attempt to reduce poverty in the short run. The right welfare of poorer populations. An example of this is a large combination as well as the level at which design and imple- transport project that may primarily target facilitating trade mentation take place is highly dependent on country level of raw materials, but at the same time it may also connect institutions, the policy makers’ objectives, and the econom- isolated poorer populations to better services. ic characteristics of the infrastructures under consideration. A more interesting debate is at which stage of development a particular infrastructure investment has a higher impact POLICY OPTIONS TO PROVIDE BETTER on economic growth versus welfare. For instance, a pow- INFRASTRUCTURE SERVICES er distribution project may have large welfare impacts giv- What exactly are the policy options that South Asian policy en that it enables education and health outcomes, which makers should focus on to improve the level and quality of may in turn translate into future economic growth as a more services provision for their diverse populations? educated, healthier labor force joins the labor market in the medium to long run. Yet, it may also facilitate growth in First, rehabilitate and maintain existing assets. South Asian manufacture today, which in turn may promote short-term governments should invest in rehabilitating and maintaining Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 15 16 TABLE 8: RANKING OF PRIORITIZATION OF INFRASTRUCTURE INVESTMENTS ON A SCALE OF 1 (LOWEST) TO 3 (HIGHEST) Development Challenges Spatial Inputs Intensityb Manifestation Outputs Externalities Natural Economic Local Climate Green Infrastructure Servicesa Capital Labor Land Resources Urban Rural Growth Welfare Agglomeration Pollution Change Growth Power Grid- Fossil Fuels 2 1 1 3 2 2 2 2 3 3 3 1 connected (Gas, Coal, etc) Generation Hydro 3 1 2 3 1 3 2 2 3 1 1 3 Wind 3 1 2 3 1 3 2 2 3 1 1 3 Geothermal 2 1 1 3 1 3 2 2 3 1 1 3 Biofuels 2 1 1 2 1 3 2 2 3 1 1 3 Off-grid Diesel 2 2 1 3 2 3 2 2 1 3 3 1 Generation Small Hydro 2 2 1 3 1 3 2 2 1 1 1 3 Wind 3 2 1 3 1 3 2 2 1 1 1 3 Solar 3 2 1 3 2 2 2 2 1 1 1 3 Reducing Poverty by Closing South Asia’s Infrastructure Gap Biofuels 2 2 1 2 1 3 2 2 1 1 1 3 Transmission Grid 2 1 2 1 3 1 2 2 3 1 2 2 Distribution Grid 2 1 1 1 3 1 2 2 3 1 2 2 Water Piped water into dwelling 2 2 1 3 3 1 1 3 3 1 1 3 Water well 1 3 1 3 1 2 1 3 1 1 1 2 Protected spring 1 3 1 3 1 3 1 3 2 1 1 2 Sanitation Piped sewer system 2 2 1 1 3 1 1 3 3 2 1 2 Septic tank 1 3 2 1 2 2 1 3 1 3 1 2 Solid Collection and processing 1 3 2 1 3 1 2 3 1 2 2 2 Waste Transport Roads Rural 2 2 2 1 1 3 3 2 3 2 2 2 Urban 2 2 2 1 3 1 3 2 3 3 3 1 Highway 2 2 2 1 2 2 3 2 3 2 2 2 Railways 3 2 2 1 2 2 3 2 3 1 1 2 Ports 3 1 1 1 3 1 3 1 3 1 1 2 Airports 3 2 1 1 3 1 3 1 3 3 3 1 Notes: a The provision modalities considered for each infrastructure service are the best available technologies (BAT) to provide the specific infrastructure service. The BAT for a specific infrastructure service is the best ranked technology according to a cost-benefit analysis. In the case of power generation, the BAT varies by location; hence the different options; b Inputs intensity is based on BAT to provide the specific infrastructure service and the BAT for building the infrastructure needed to provide the infrastructure service. infrastructure assets to deliver services efficiently and sus- Third, establish solid legal, policy, and regulatory frame- tainably, moving away from the “build, neglect, and rebuild” works. South Asian governments need to have solid le- mindset. Lack of adequate infrastructure maintenance is gal and policy frameworks; as well as transparent, well de- quite common across developing countries. In India, the signed, and implemented regulatory framework for both Working Group on Roads for the National Transport Devel- public and private operators; in order to attract private in- opment Policy Committee reports a 40–50 percent short- vestment in line with the best organizational form for each fall in the maintenance allocation for state highways and service. For example, governments across the region need major district roads. Under-spending on maintenance of to set the conditions for an even bigger role of the private infrastructure has direct and indirect costs. Without reg- sector in a service such as power generation, which is bet- ular maintenance, physical infrastructure can rapidly fall ter suited for liberalization than PPPs. Additionally, they shift into disrepair, requiring expensive reconstruction to bring it the public resources and efforts toward other services where back to adequate standards. For example, the cost of full the public sector has the comparative advantage (Box 1). reconstruction of roads that have been poorly maintained Fortunately, when the private sector investment in infra- is, on average, at least three times the cost of maintenance structure in SAR, it tends to choose the optimal organiza- (World Bank, 2005). Lack of adequate maintenance trig- tional forms. These frameworks provide clarity to the private gers a progressive deterioration of the quality of the infra- sector, increasing the attractiveness of private participation structure services, which hurts users (e.g., higher costs in infrastructure projects. They also allow the public sector because of imperfect and costly substitutes, worse social to clearly define responsibilities and manage the risks asso- outcomes in health and education) and development out- ciated with private sector participation. A stable yet dynam- comes. In India, the National Transport Development Poli- ic regulatory framework for infrastructure services is particu- cy Committee estimates that poor road maintenance costs larly critical for: i) Attracting and supporting desired levels of the country about Rs 350 billion annually. investment, ii) Ensuring service sustainability, iii) protecting customers, and iv) Guarding the public interest. While funds for new construction are sometimes easier to obtain and implement, those for maintenance are more dif- Fourth, decentralize service provision in an appropriate man- ficult as they need to be sustained on a regular basis. Differ- ner. SAR countries should be rethinking how much to decen- ent mechanisms can be implemented in different infrastruc- tralize (i.e., distribute the administrative powers or functions of ture sectors to improve maintenance. In the road sector, a central authority over a less concentrated area) as a means some governments have adopted or considered adopting of improving service delivery for the smallest units of soci- a “road fund” type of arrangement for supporting mainte- ety (households and individuals). As the World Bank’s 2004 nance. Under such arrangements, maintenance funds are World Development Report puts it: “Decentralization can be assured from a mandated tax on gasoline and diesel, and a powerful tool for moving decision making closer to those af- are deposited into an assured and independently operated fected by it. Doing so can strengthen the links and account- fund. A Board that includes the public sector, or the private ability between policymakers and citizens—local governments sector, or possibly both oversees this fund. are potentially more accountable to local demands. It can also strengthen them between policymakers and providers—local Second, reform service providers and ensure financial/opera- governments are potentially more able to monitor providers. tional sustainability. Service providers should be financially But local governments should not be romanticized. Like na- viable, able to plan and implement sound investment strat- tional governments they are vulnerable to capture—and this egies, and improve operational performance for the long might be easier for local elites on a local scale.” term. This requires: i) Reliable, steady, and adequate reve- nue streams to fund operations and investment; ii) Capaci- In practice, the experience with trying to decentralize infra- ty and independence without threat of political interference; structure service delivery is mixed—the biggest problem of- and iii) Appropriate incentives for becoming and remaining ten being a mismatch (often financial or fiscal in nature) be- more efficient. tween responsibilities in infrastructure service delivery and Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 17 BOX 1: OPTIMAL ROLES FOR THE PUBLIC AND PRIVATE SECTORS There is no single service provision approach that is better than the alternatives for all infrastructure services and under all degrees of insti- tutional development. In this report, we examine four possible organizational forms—with varying levels of public and private participation: (i) traditional provision, (ii) PPP, (iii) regulated privatization, and (iv) liberalization (deregulated privatization). Following Engel et al. (2009), we assume that private firms build, operate, and maintain the infrastructure under all of these forms, and hence, the benefits of each form stem from the incentive structure—not the degree of private participation. Depending on the features of each infrastructure service, one of the four organizational forms brings the highest social welfare. When market liberalization is optimal Power generation is an example of an infrastructure service that is produced under constant or decreasing returns to scale, and for which user fees can be charged. In cases like this, the optimal organizational form is market liberalization (that is, privatization plus price deregulation). Competition together with private ownership induces firms to select optimal life-cycle cost saving investments and provide the optimal service quality, thereby solving the trade-off between productive efficiency and quality considerations. When traditional provision is optimal In the case of flood control, which is a non-excludable service, it is not possible for the government to set service standards that are enforceable. Furthermore, quality considerations dominate productive efficiency, making traditional provision the optimal organizational form. If the costs of quality reduction were not as important as the benefits of reducing life-cycle cost, then a PPP would be preferred over traditional provision. Similarly, if it were not possible to charge fees for the use of a service, but service standards could be designed and enforced, then a PPP would be the optimal organizational form. When PPPs are optimal This occurs when the service is produced under increasing returns to scale (i.e., natural monopoly) or there are technical aspects that create barriers to entry (e.g., the scarcity of radio spectrum for wireless communications), conditions that would rule out market liberalization. So in a case like power transmission —which is a natural monopoly, and where expansion requires significant network planning—PPPs would dominate over regulated privatization. PPPs have the advantage of leaving the government with the authority to decide on future expansions. The same applies for most transport services. When regulated privatization is optimal This occurs when competition is not feasible (e.g., because of increasing returns to scale or technical and/or legal entry barriers), user fees can be collected, the government can design and enforce service standards, and planning is best done at the firm level. Hence, regulated privatiza- tion is optimal for power distribution, and ICT services (fixed and mobile). In the latter case, network externalities are important, creating the need to regulate interconnection charges. Regulated privatization is also optimal for sanitation and water services, particularly at the distribution level in the latter case. An issue of planning and coordination in the use of a natural resource that is beyond the project level arises in water production or catchment, which makes PPPs the optimal organizational form in water production. the ability to execute such responsibilities. But when de- the access of the poor—keeping in mind the following five centralization succeeds, it is thanks to: (i) A fully democrat- principles. ic, transparent, and inclusive (of the beneficiaries) local de- cision process; (ii) The cost of local decisions fully borne by Access is fundamental, but usage determines impact. That is local government; and (iii) No spillover of benefits to other why policy makers should complement access to infrastruc- jurisdictions. ture with policies to incentivize the use of services, or make potential benefits more obvious or attainable. One way to do this is to focus on subsidizing (implicitly or explicitly and with POLICY OPTIONS TO HELP sunset clauses) the infrastructures that provide the greatest THE POOREST GAIN BETTER public benefit (public good) in contrast to those that provide INFRASTRUCTURE large private benefits. This should be true across infrastruc- At the same time as South Asian governments are mov- ture sectors as well as within sectors. Another way is to focus ing to improve the overall level and quality of infrastruc- on improving women’s access to services, as the improve- ture services, they must take deliberate steps to improve ment in household outcomes can be larger when women 18 Reducing Poverty by Closing South Asia’s Infrastructure Gap benefit fully from access. Still another way is to focus on en- infrastructure), while allowing for wealthier populations to hancing quality and maintenance, which are major issues in shoulder most of the burden of improving coverage for all. South Asia, as there are an average 42 power shortages a Given the equality achieved in improved water in South Asia, month and 21 water shortages a month. one would be tempted to conclude that this objective is pres- ent as the service expands. But this conclusion might con- Ability-to-pay for access to infrastructure services cannot ceal rent-seeking behavior, where the wealthier capture pro- be the only instrument to determine provision. Infrastructure portionally larger amounts of rents that otherwise could be services have strong market failure characteristics, under- used for expansion and quality improvement for all and not scoring the need for adequate regulation. Some infrastruc- just a few. The literature also argues that infrastructure ser- tures are still close to natural monopolies (such as pipe wa- vice expansion is closely linked to rent seeking, since rich- ter and off-site sanitation services). Many are associated er districts are better able to lobby the government for infra- with strong externalities (negative and positive) and public structure provision (Cadot et al., 1999). good (and bad) characteristics, as in the case of a lack of pollution treatment or a lack of access to cooking gas. Infor- Although subsidies may improve affordability among under- mation issues also abound. Moreover, since infrastructure privileged groups, they can also have the effect of increasing may act as a spur to growth, relying on the ability-to-pay cri- income inequality. Subsidies tend to be captured by those terion might undercut efforts to reduce poverty. who have political connections, which, at least among un- connected households, tend to be the more middle class Some infrastructure programs are too costly to be sustainably households. Wodon and Ajwad (2002) found that in Boliv- implemented without cost-recovery mechanisms that allow ia and Paraguay, the marginal benefit of improved access to them to be self-supporting. The trade-off between providing a service tended to be two to three times higher among the access to infrastructure services and fully charging for these upper two quartiles. Thus, while all income quartiles ben- services is seldom an easy one to equate. It involves un- efited from decentralization, the richer 50 percent benefit- derstanding the economic characteristics of particular in- ed more than the poorer 50 percent, a net effect that would frastructure sectors and the technology available for provi- tend to increase income inequality. Estache (2005) points sion under different physical, political, and socio-economic out that in Latin America as much as 60 to 80 percent of conditions. Take the case of piped water provision, which cross-subsidy schemes “were aimed at households well is a private good. It has important market failures associat- above the poverty threshold, while as much as 80 percent ed with it, but essentially individual households have clear of poor households failed to benefit.” Thus, it is not surpris- incentives to pay for a superior service compared to other ing that even as absolute levels of connection increase, re- forms of getting water in an urban environment. Yet, piped gressivity in access to infrastructure may still prevail. water is seldom charged to attain full cost recovery and of- ten relies on direct or indirect subsidies that burden public budgets. Nonetheless, the expansion of piped water provi- MENU OF INSTRUMENTS sion is often part of political manifestos during election cam- These principles, in turn, would point to the following menu paigns. Now take the case of flood control, which is a pub- of instruments: lic good. Direct cost recovery mechanisms (like tariffs) are difficult to design, but the lack of adequate flood control in Subsidies for connection rather than service consumption. a locality for example can cause substantial costs to house- To avoid some of the drawbacks of subsidies, policy mak- holds via the loss of private assets and lives. Budgetary allo- ers can adopt measures that reduce the cost of providing cation for flood control is often inadequate and the service network services or improve the ability of poor households is underprovided. to pay for service at a given cost (Komives et al., 2005). These would be available only to unconnected households, The likely aim of the policy maker is to attain a certain de- reducing or eliminating the price customers have to pay gree of balance in infrastructure access (especially basic to connect to the system. Alternatively, policy makers can Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 19 subsidize lower service levels that the better off find less at- user groups can contribute to planning and operations; tractive, such as social connections. NGOs can help with monitoring and evaluation, promoting social accountability and raising awareness; and the private Targeted interventions. Usually these instruments are cen- sector can get involved with investment and delivery (An- tered narrowly on a certain district or group that is perceived dres and Naithani, 2013). These alternative mechanisms, as underserved. This approach has the advantage of few- which are context-specific, are becoming part of the policy er spillovers—that is, there is less likelihood that the inter- toolkit as they are tested and mainstreamed. vention ends up benefitting those who were not intended to be its beneficiaries. In Mexico, this is now being tried with a program that provides conditional cash transfers for CONCLUSIONS the poorest segments of the population, named Oportuni- If South Asia hopes to meet its development goals and not dades (“Opportunities”). Under this program, energy subsi- risk slowing down—or even halting—growth, poverty alle- dies were channeled using the same targeting mechanisms viation, and shared prosperity—it is essential to make clos- so these funds are reaching the poorest population in the ing its huge infrastructure gap a priority. Even though SAR’s country. A note of caution here, however, is that because economic growth follows that of EAP, its access to infrastruc- these interventions are operating only within impoverished ture rates (sanitation, electricity, telecom, and transport) are and underserved areas, they may face issues like inade- closer to that of SSA—the one exception being water, where quate staffing, funding, technical capacity, and lack of polit- SAR is comparable to EAP and LAC. According to business- ical will (Menéndez, 1991). es in South Asia, infrastructure is a major or severe hin- drance to their growth, and electricity is the largest problem. Institutional groups. There are also a number of options to Transport is also an obstacle for regional and international design programs to reduce elite capturing and increase the trade. The good news is that policy makers do not have to power of impoverished groups to allocate resources toward choose between growth and welfare, as there is enormous their priorities. These include: potential for them to be mutually supportive. ƒƒ Institutional re-centering. Organizations can be creat- The cost to close this gap by 2020 will be an estimated US$ ed whose primary concern is to reduce poverty through 1.7 to US$ 2.5 trillion. If investments are spread evenly over providing infrastructure. For instance, Bolivia’s Emer- these years, SAR needs to invest between 6.6 and 9.9 per- gency Social Fund was a temporary organization that cent of 2010 GDP per year—an increase of three percent- was created to finance infrastructure projects in under- age points over the current 6.9 percent invested by SAR served communities. countries in 2009, up from 4.7 percent in 1973. This in- ƒƒ Community participation. Incorporating transparent crease was driven mainly by the region’s large investment in mechanisms for underserved people to easily provide electricity generation. input into the design and decision-making process behind infrastructure projects could pote low them The challenge of increasing access to infrastructure servic- to compete with the more informal mechanisms that es across South Asia is compounded by the inequality in the richer populations use to influence decision-making distribution of existing access for households and individuals. (Menéndez, 1991).25 That is why providing some level of access is a start—even if those services are not of the highest quality. At the same time Innovative mechanisms. Service delivery mechanisms need policy makers should take into account which types of servic- to evolve to respond to the challenges of coverage, afford- es best fit each population’s needs (such as septic tanks for ability, use, and sustainability. This is particularly important given that poor households tend to pay more for services 25 The way incentives are designed play an important role in mitigat- when they have to obtain them through non-network so- ing rent seeking. Community Driven Development Projects are par- lutions. For example, community-based organizations and ticularly concern with elite capture even within poor communities. 20 Reducing Poverty by Closing South Asia’s Infrastructure Gap a mountainous region but sewerage lines for a more acces- frameworks in the decision making process of establishing sible urban area). Our study sheds light on who has access priorities is highly desirable, they are no substitute for con- in terms of space (a current framework) and in terms of time sensus building and political negotiations. (future generations). Countries with higher per capita income (like the Maldives and Sri Lanka) enjoy better access to in- SAR also needs to rethink the infrastructure service paradigm frastructure services both spatially (geographically within the to bring in the private sector and decentralize administrative country) and income. This is despite the fact that conflict ar- powers functions. The sheer size of the gap and the macro- eas are clearly worse off. Among sectors in SAR countries, economic situation in South Asia dictate that the region taps some, such as water, tend to be more equally distributed than other funding sources. However, this situation should also others (such as sanitation, energy, and phones). One surpris- be seen as an opportunity to rethink and improve how in- ing discovery is the widespread use of firewood for cooking, frastructure services are delivered. One way to do this is by especially among the poor. Moreover, within SAR countries, broadening service provision to give the private sector a big- some states and districts have better access than others. In ger role—whether through PPPs or regulated privatization addition, leading regions within a country typically have better and market liberalization. Another way is by giving greater access but lagging regions do not necessarily have worse ac- administrative powers and functions to lower levels of gov- cess. However, if a poorer country or a poorer state can have ernments, although the degree to which such decentraliza- better access to a given infrastructure service than in a richer tion is desirable will depend on the nature of the investment, country or a richer state, then there is hope that policy mak- the reason it is being provided, how it is being financed, and ers can adopt measures that will improve access in a manner where it is located. that increases shared prosperity. Policy choices should be aimed at increased shared pros- There is no simple explanation for these inequalities, al- perity. Key principles to keep in mind are: (i) Access is fun- though certainly geography matters, policy intent matters, damental but usage determines impact; (ii) Ability-to-pay and some household characteristics matter. At the country for access cannot be the only instrument to determine pro- level, household characteristics like location and education vision; (iii) Some infrastructure programs are too costly to are the main explanatory factors. Location seems obvious, be sustainably implemented without cost-recovery mecha- but education does not—unless it is linked to income pov- nisms that allow them to be self-supporting; (iv) Market fail- erty and remoteness of household location (even among ru- ures need to be corrected to avoid rent-seeking behavior; ral areas). At the state level, education actually starts to be- and (v) Subsidies should be designed in a way that they come a bigger factor than at the country level. While the do not exacerbate income inequality. The menu of possible contribution of gender of the head of the household seems instruments includes subsidizing connections (rather than negligible at the country level, there are a number of states service consumption), adopting targeted interventions, cre- (and districts within them) where access is clearly biased to- ating organizations dedicated to reducing poverty through wards male-lead households. providing infrastructure, and asking NGOs to help with mon- itoring and evaluation, promoting social accountability, and Given the size of the gap and limited fiscal and financial re- raising awareness. sources, it is essential for SAR to prioritize infrastructure in- vestments. We propose a generic methodological framework If shared prosperity is one of the ultimate goals of policy mak- for doing just that, building on the existing literature. It is ers, it is important to get accurate infrastructure data. The ex- not desirable to have a single methodology, providing a sin- isting data allowed us to create a baseline for tracking prog- gle ranking of infrastructure investments, because of the ress in closing the infrastructure gap and to answer some complexities of infrastructure investments. Rather, a mul- questions about who has access to infrastructure services tidisciplinary approach should be taken. Decision makers now and how that is likely to affect future generations—and will also need to account for factors that are often not eas- thus the equality of opportunities across the region. This ily measured. While having techniques that enable logical work can be expanded and improved by considering more Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 21 circumstances, along with exploring alternative indicators of Department of Census and Statistics. 2009–10. “Household In- access, use, and quality of infrastructure services. However, come Expenditure Survey (HIES).” Ministry of Finance and in order to do that, better data is needed. Without it, private Planning, Colombo. and public investments may miss their targets of leveling the Engel, E., R. Fischer, and A. Galetovic. 2009. “Public-Private playing field and end up increasing inequality of infrastruc- Partnerships: When and How.” Centro de Economía Apli- ture service provision. cada, Universidad de Chile, HYPERLINK “http://ideas. repec.org/s/edj/ceauch.html” Documentos de Trabajo 257. REFERENCES Estache, A. 2005. “On Latin America’s Infrastructure Experi- ence: Policy Gaps and the Poor.” ”, in J. Nellis and N. Ajwad, M. I., and Q. Wodon. 2002. “Who Benefits from In- Birdsall (2005), Reality Check: The Distributional Impact creased Access to Public Services at the Local Level? A of Privatization in Developing Countries, Center for Global Marginal Benefit Incidence Analysis for Education and Ba- Development. 281–296. sic Infrastructure.” In Washington DC: World Bank Econo- Estache, A. and M. Fay. 2010. “Current Debates on Infrastruc- mists’ Forum (Vol. 2, pp. 155–175). ture Policy.” Globalization and Growth: Implications for a Andres, L. and S. Naithani. 2013 “Mechanisms and Approach- Post-Crisis World. M. Spence and D. Leipziger Eds. World es in Basic Service Delivery for Access and Affordability.” Bank, Washington, DC. Washington DC: World Bank. Fay, M. and T. Yepes. 2003. “Investing in Infrastructure: What Andres, L., D. Biller, and M. Herrera-Dappe. 2013. “Infrastruc- is needed from 2000–2010”. Policy Research Working Pa- ture Gap in South Asia: Infrastructure Needs, Prioritization, per 3102, World Bank, Washington, DC. and Financing.” Washington DC : World Bank. Institute for Population Sciences. 2007–08. “District Level Household Survey (DLHS-3).” Ministry of Health and Fam- Bangladesh Bureau of Statistics and UNICEF. 2007. Multiple ily and Welfare, Delhi. Indicator Cluster Survey (MICS) 2006. Dhaka, Bangla- desh: Ministry of Planning and UNICEF. Karydas, D.M. and J.F. Gifun. 2006. “A method for the efficient prioritization of infrastructure renewal projects,” Reliability Berechman, J. and R. E. Paaswell. 2005. “Evaluation, prioriti- Engineering & System Safety, Vol. 91, No. 1, pp. 84–99. zation and selection of transportation investment projects in New York City,” Transportation, Vol. 32, pp. 223–249 Kojima, M. 2011. “The Role of Liquefied Petroleum Gas in Re- ducing Energy Poverty.” Extractive Industries for Develop- Biller, D., L. Andres, and M. Herrera Dappe. 2013. “Inequality ment Series No. 25, World Bank, Washington, D.C. of Access to Infrastructure Services in South Asia .” Wash- ington DC: World Bank. Kojima, M., R. Bacon, and X. Zhou. 2011. “Who Uses Bottled Gas? Evidence from Households in Developing Countries,” Cadot, O., L. H. Röller, and A. Stephan. 1999. “A political econ- Policy Research Working Paper No. 5731, World Bank, omy model of infrastructure allocation: an empirical as- Washington, DC. sessment,” (No. FS IV 99–15). Discussion papers//WZB, Wissenschaftszentrum Berlin für Sozialforschung, Forsc- Komives, K., J. Halpern, V., Foster, Q. Wodon, and R. Abdullah. hungsschwerpunkt Marktprozeß und Unternehmensent- 2005. “Water, Electricity, and the Poor: Who Benefits from wicklung. Utility Subsidies?, World Bank, Directions in Development, Washington, DC. Debroy,B. and L. Bhandari. 2003. “District-level Deprivation in National Institute of Population Studies (NIPS) [Pakistan], and the New Millennium.” Rajiv Ghandi Institute for Contem- Macro International Inc. 2008. Pakistan Demographic and porary Studies and Indicus Analytics. Health Survey (DHS) 2006–07. Islamabad, Pakistan: Na- Debroy, Bibek. District-level deprivation in the new millennium. tional Institute of Population Studies and Macro Interna- Delhi: Konark Publishers, 2003. tional Inc. 22 Reducing Poverty by Closing South Asia’s Infrastructure Gap National Statistics Bureau. 2011. “Bhutan Living Standards Vulnerability Analysis Unit. 2007–08. “National Risk Vulnera- Survey (BLSS) 2007.” Royal Government of Bhutan, bility Assessment (NRVA).” Ministry of Rehabilitation and Thimpu, Bhutan. Rural Development and Central Statistics Organization, Government of Afghanistan, Kabul. Menendez, A. 1991. Access to basic infrastructure by the ur- ban poor (No. 28). World Bank Publications. World Bank. 2004. “World Development Report 2004: Mak- ing Services Work for Poor People.” Washington: World Ministry of Health and Family (MOHF) [Maldives] and ICF Bank. Macro. 2010. Maldives Demographic and Health Survey (DHS) 2009. Calverton, Maryland: MOHF and ICF Macro. World Bank Public-Private Infrastructure Advisory Facility (PPIAF). 2013. “Public-Private Infrastructure Advisory Ministry of Health and Population (MOHP) [Nepal], New ERA, Facility Database.” World Bank, Washington DC. http:// and ICF International Inc. 2012. Nepal Demographic and ppi.worldbank.org/index.aspx Health Survey (DHS) 2011. Kathmandu, Nepal: Ministry of Health and Population, New ERA, and ICF Internation- World Bank. 2005. “Why Road Maintenance Is Important and al, Calverton, Maryland. How to Get it Done.” Transport Note No. 4. Organization for Economic Co-operation and Development World Bank. 2013. “World Development Indicators.” http:// (OECD)/ International Energy Agency (IEA). 2010. “World data.worldbank.org/country Energy Outlook 2010.” Paris: France. World Health Organization (WHO)/United Nations Chil- Paes de Barros, R., F. H. Ferreira, J. R. Molinas Vega, and J. dren’s Fund (UNICEF) Joint Monitoring Pro- Saavedra Chanduvi. 2009. “Measuring Inequality of Op- gramme (JMP) for Water and Sanitation. 2013. http:// portunities in Latin America and the Caribbean.” Wash- www.wssinfo.org/documents-links/documents/?tx_ ington DC: Palgrave MacMillan and The World Bank displaycontroller[type]=country_files Luis Andrés, Dan Biller, and Matías Herrera Dappe December 2013 23