WPS5728 Policy Research Working Paper 5728 Aid to the Services Sector Does It Affect Manufacturing Exports? Esteban Ferro Alberto Portugal-Perez John S. Wilson The World Bank Development Research Group Trade and Integration Team July 2011 Policy Research Working Paper 5728 Abstract This paper evaluates the impact of foreign aid to (i) the variation of aid flows to service sectors, and (ii) five service sectors (transportation, information and the variation of service-intensities across industrial communications technologies, energy, banking/financial sectors and countries using input-output data. The services, and business services) on exports of downstream authors find a positive effect of aid to services, in general, manufacturing sectors in developing countries. To address on downstream manufacturing exports of developing the reverse causality between aid and exports, the analysis countries across regions and income-level groups. relies on an original identification strategy that exploits This paper is a product of the Trade and Integration Team, Development Research Group. It is part of the Global Trade and Financial Architecture Project, under the Transparency and Competitiveness Trust Fund. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at eferro@worldbank.org, aportugalperez@worldbank.org, and jswilson@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Aid to the Services Sector: Does It Affect Manufacturing Exports?‡ Esteban Ferro*, Alberto Portugal-Perez§, and John S. Wilson† The World Bank Keywords: Foreign Aid; Aid for Trade; Services; Manufacturing; International Trade JEL: F13; F14; F35; L16; O19 * Consultant, eferro@worldbank.org., §Economist, aportugalperez@worldbank.org., †Lead Economist, jswilson@worldbank.org. at DECRG - Trade and International Integration, The World Bank Group. 1818 H St. NW, Mailstop: MC3-303, Washington DC, 20433. ‡ The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Without implicating them, we would like to thank Olivier Cadot, Ana Margarida Fernandes, Gene Grossman, Aart Kraay, Daniel Lederman, Catherine L. Mann, Aaditya Mattoo, Ariell Reshef, and participants of seminars at the World Bank and at the World Trade Organization for helpful ideas and comments throughout the preparation of this paper. This research is part of the Trust Fund on Transparency and Competitiveness and Global Trade and Financial Architecture Project at the World Bank. 1. Introduction The global recession has placed aid budgets of donor countries under significant strains. In addition, aid effectiveness as it relates to trade is at the forefront of development policy discussions. There is also increased attention to project planning and focusing aid for trade in areas of maximum impact and return on investment. Moreover, there is very limited empirical evidence on the impact of aid to service sectors. Aid for trade has rapidly gained importance in trade and development circles as well as in the donor community. Launched at the Hong Kong SAR, China WTO Ministerial Conference in December 2005, the aid for trade initiative aims at helping developing countries, particularly least developed countries, overcome their supply-side constraints, so as to expand trade and benefit from global integration. Despite enjoying preferential market access and facing lower tariffs, several developing countries —especially low-income countries— have seen their share of world exports diminish over the past years. Clearly, the reduction of tariff and non-tariff barriers faced by developing countries is an essential condition for their export growth, but not a sufficient one. These countries face supply- side constraints that severely limit their ability to reap the benefits from global trade integration. Evidence on the effectiveness of aid for trade in improving trade-related performance in developing countries is still limited. An important limitation prevalent in any econometric assessment is the potential reverse causality between aid and trade. Indeed, aid for trade is expected to have an impact on exports, however, aid may also be determined by export performance. If, for instance, better performing countries are rewarded with more aid, estimates of aid for trade coefficients would be biased upward. In this paper, we propose an identification strategy that helps to overcome the reverse causality of the trade-aid relationship. Using input-output data, we exploit the differential 2 service-intensities of manufacturing sectors to evaluate the impact of aid in five service sectors (transport, communications, energy, banking/financial services, and business services) on exports of downstream manufacturing sectors for 132 countries between 2002 and 2008. Our estimates, in general, show a positive effect of aid to services on downstream manufacturing exports of developing countries. Aid to transport, energy, and banking sectors have consistently a significant and positive impact on downstream manufacturing exports. The effectiveness of aid to transportation in terms of exports growth diminishes for country groups with higher income, whereas the effectiveness of aid to energy and business services increases with the income of the group. The rest of the paper is organized as follows. Section 2 provides a literature review on aid and aid for trade. Section 3 explains our identification strategy. Section 4 describes the data and data resources used in our empirical analysis. Section 5 presents our results. Finally, the last section concludes and discusses potential avenues for further research. 2. Literature Review Our review is divided in two parts. We start with a brief account of the large literature on aid and growth, where the reverse-causality of aid-growth is prevalent. The second part defines briefly aid for trade and reviews the literature that is more specific to it. 2.1. Aid and Growth The literature on the impact of aid on the level of development of recipient countries is very large and keeps growing, and we provide a small and selective review in this paper. 3 Unfortunately, the results of the studies have been mixed.1 There is no robust evidence of either a positive or negative correlation between foreign aid inflows and the economic growth of poor countries. Most studies that find a positive effect between aid and growth also find this holds only under specific conditions such as quality of institutions, geographic location, and social elites of a country whereas other studies find that the type and timing of aid is what matters most.2 On the other hand, Rajan and Subramanian (2011) argue that the costs emanating from foreign aid offset its benefits. The authors find that aid has a Dutch disease effect on the terms of trade of recipient countries resulting in a negative impact on traded goods and on growth.3 The existing literature also stresses the reverse causality between aid and growth that could lead to misleading results. A few studies attempt to address this problem using instrumental variables. Rajan and Subramanian (2008) estimate a bilateral aid specification including variables traditionally used in the gravity model of bilateral trade (i.e., common language, colony relationships, among others) to generate a fitted aid measure (first stage) that is used as an instrument for aid in a GDP growth equation (second stage). They find little robust evidence of a positive (or negative) relationship between aid and GDP growth. On the other hand, Bruckner (2011) adopts a different two-stage strategy. In the first stage, he estimates an equation explaining aid using rainfall, commodity price shocks, GDP growth, and other controls. In the second stage, the fitted residuals are used as instruments for aid in a per capita GDP 1 See for example: Burnside and Dollar (2000); Collier and Dollar (2002); Easterly (2003); Easterly, Levine, and Roodman (2003); Clemens, Radelet, and Bhavnani (2005); Bourguignon and Sundberg (2007); Rajan and Subramanian (2008). 2 See: Burnside and Dollar (2000); Dalgaard et al. (2004); Angeles and Neanidis (2009); Clemens et al. (2004). 3 Some other studies of aid and exchange rates include: Younger (1992); Arellano et al. (2005); Berg et al. (2005); Prati and Tressel (2006). 4 growth specification. Bruckner finds that aid has a significant and positive effect on real per capita GDP growth. Finally, Bourguignon and Sundberg (2007) point to the wide heterogeneity of aid motives and the limitations of the tools of analysis. They suggest that the complex causality chain linking external aid to final outcomes has been handled mostly as a kind of ―black box‖, and progress on estimating rigorously aid effectiveness requires opening that box. 2.2. Aid for Trade According to the WTO, Aid for Trade aims to help developing countries, particularly least-developed countries, build-up the trade-related skills and infrastructure that are needed to implement and benefit from WTO agreements and to expand their trade.4 Aid for trade is an integral part of regular official development assistance (ODA). Donors have, in fact, been providing substantial amounts of aid to trade-related programs for many years. Moreover, the scope of aid for trade has expanded considerably. During the 1986-1994 Uruguay Round of trade negotiations, trade related assistance was mainly aimed at technical support to help developing countries negotiate and implement trade agreements. Subsequently, the scope expanded to include building supply-side capacities, for instance in private sector development and trade- related infrastructure. Now the agenda also includes trade-related structural adjustment programs and other trade-related needs. Aid for trade, as defined by the OECD, has nearly tripled between 2001 and 2008 as shown in Figure 1. Yet, as Figure 2 displays, the share of total aid that targets trade related projects has substantially decreased over the years only to marginally increase after 2006. Since 4 http://www.wto.org/english/tratop_e/devel_e/a4t_e/aid4trade_e.htm 5 2006 there has been an important surge in aid for trade in both absolute and relative terms; however, evidence of its effectiveness is still scant. A limited number of studies focus on the impact of aid for trade.5 Brenton and Uexkull (2009) analyze the effectiveness of export development programs. Using a difference in difference approach, they aim at isolating the impact of the policy interventions and draw four main conclusions. First, most export development programs have coincided with or predated stronger export performance. Second, such programs appear to be more effective where there is already significant export activity. Third, there is some concern about the ―additionality‖ of the programs as support may be channeled to sectors that would have prospered anyway. Finally, conclusions strongly depend on what one postulates would have happened in the absence of the policy intervention, so the definition of a credible counterfactual is of utmost importance for the evaluation of technical assistance for exports. Helble, Mann, and Wilson (2009) make one of the first attempts to analyze how foreign aid spent on trade facilitation increases trade flows in developing countries. The authors use a gravity model of bilateral trade and find that the bulk of the relationship between aid and trade appears to come from a narrow set of aid flows directed toward trade policy and regulatory reform, rather than broader aid-for-trade categories directed toward sectoral trade development or infrastructure development. Other studies on the effect of aid on trade have found similar positive results to those found by Helble, Mann, and Wilson (2009), including Cali and te Velde (2010), the latter being the closest study to our own. Indeed, Cali and te Velde (2010) evaluate whether aid for trade has improved export performance. They find that aid for trade facilitation, and to some extent aid for trade policy and 5 For and extended review of the literature on the relationship of aid and trade, see Suwa-Eisenmann and Verdier (2007). 6 regulations, helps reduce the cost of trading (both in terms of exports and imports). In addition, their results suggest that aid to economic infrastructure increases exports, whereas aid to productive capacity appears to have no significant impact on exports. The authors correctly point out that aid for trade is possibly endogenous to exports; particularly aid to productive capacity. For instance, if better performing sectors tend to receive more aid for trade than other sectors, it would generate an upward bias in the aid coefficient. To address this endogeneity, the authors instrument aid for trade with the degree of respect for civil and political liberties measured by the Freedom House Index and with an index proposed by Gartzke (2009) that measures the ―affinity of nations‖6. The authors argue that many donors choose recipients based upon development and democratic measures like these which make these indicators correlated with aid but not with exports. However, the authors acknowledge that their instruments are not appropriate for sectoral analysis, as they only vary across country-year and not across country-sector-year. We propose an alternative identification strategy that allows us to analyze the impact of sectoral aid. To address reverse causality, we exploit the links between the service sector and the manufacturing sector relying on input-output tables. Although input-output data has not been used in the analysis of aid effectiveness so far, a number of studies, particularly in the FDI literature, make use of it. Given the difficulty of finding consistent input-output matrices across countries most studies rely on input-output data from the United States to describe the technological possibilities of firms in a given economy. Acemoglu, Johnson, and Mitton (2009) investigate how contracting costs and financial development determine the extent of vertical integration across countries. Alfaro and Charlton (2009) use new firm level data at the 4-digit sector level and U.S. input-output tables to classify firms between horizontal and vertical 6 In preliminary research we also attempted to use other instrumental variables for aid for trade such as immunization rates, gender health access. However, statistical tests suggested that they are not appropriate instruments for aggregate aid for trade. 7 subsidiaries. The authors find that at the two-digit industry level, there are considerably more horizontal (subsidiaries in the same industry as their parents) than vertical (subsidiaries that supply their parents with inputs) FDI. However, disaggregating to the four- digit level reveals that many of the foreign subsidiaries in the same two-digit industry as their parents are, in fact, located in sectors that produce highly specialized inputs to their parents’ production. Thus, contrary to the conventional wisdom, the authors find that the number of vertical multinational subsidiaries is larger than commonly thought. Among the few studies using input-output data for countries other than the U.S, Hummels, Ishii, and Yi (2001) define a measure of vertical specialization that captures a country’s role in the fragmentation of production into multiple stages in multiple locations. The authors use input-output tables from ten OECD nations and four other countries to measure a country’s vertical specialization as its exports weighted by the share of imported inputs in its total output. Trefler and Zhu (2010) use 20 input-output tables from the Global Trade Analysis Project (GTAP) to reassess Vanek’s (1968) factor content of trade predictions in 41 countries. 3. Tackling Reverse Causality: An Identification Strategy Aid flows are expected to have an impact on exports, but a country’s exports may also affect the aid the country receives. It is plausible that donors target industries in recipient countries where exports are expanding or declining. As mentioned above, Brenton and Uexkull (2009) find that exports have increased owing to the effect of donor-funded export development programs in a number of countries. However, although the programs have preceded stronger export performance, causality cannot be clearly determined. Factors like the initial size of the 8 export sector, or selection bias (i.e., technical assistance may target products with already promising prospects) appear to be the real reasons behind the targeted commodities. To address this issue, we propose an original identification strategy. Instead of directly focusing on how aid that targets a specific manufacturing sector impacts its own exports, we analyze how aid that targets service sectors—such as banking, energy, and business services— has an impact on exports of downstream manufacturing using these services. OECD donors report disbursed flows of yearly aid by service sector to each recipient. Table 1 reports total aid disbursed in 2008 into the different categories of aid for trade and displays the link that we use in our identification strategy between service categories and manufacturing sectors. Our identification strategy could be considered an extension to that used by Rajan and Zingales (1998). The main idea of their work is that industries differ in terms of their dependence from external finance because of industry specific technological reasons. Therefore, when a country's financial system develops those sectors that rely more heavily on external finance will benefit and grow disproportionately faster. In our case we exploit the dependence of manufacturing industries not only on financial services but also on transportation, ICT, energy, and business services. As Rajan and Zingales (1998) we expect those manufacturing industries that use transportation more heavily to benefit the most from aid to transportation services, industries that use energy more intensively to benefit the most from aid to energy services, and so on. If aid has a positive impact it should improve the quality of the services provided and/or reduce the costs to downstream users of these services. Thus, industries that use these services more intensively should be able to produce more (i.e., cheaper inputs) and ultimately export more as well. 9 To implement our identification strategy, we need both the amount of aid received by each service sector in a country as well as information on how intensively a manufacturing sector uses upstream services sectors. Information on service intensity can be computed from input- output tables which provide information on the inter-industry relations of an economy. We use the total input requirements which are estimates of the inputs—including services—for each industry that are directly and indirectly required to deliver a dollar of the industry output to final users. Thus, exports are determined by: (1) where: Xijt is exports of sector j in country i in year t; aidikt, is the amount of aid addressed to the service sector k of recipient country i in year t; k_intensityij is the intensity with which the manufacturing sector j uses upstream service sector k in country i. which can be roughly interpreted as the cents of service k used in the making of a dollar of product j. We also include country-sector fixed effects, , to control for invariable characteristics specific to a manufacturing sector in a given country, such as constant taxes and subsidies. Country-year effects, , control for shocks specific to a country in a given year, such as inflation, exchange rates, political or economic shocks, and climate shocks such as natural disasters. Sector-year effects, , control for shocks specific to a sector worldwide in a given year, such as any supply or demand shock having an impact on world market prices. 4. Data Description Aid flow data were compiled from the OECD’s Creditor Reporting System (CRS); exports, specifically mirrored imports, were taken from UNCTAD’s COMTRADE database; output was taken from UNIDO’s INDSTAT 4; and input-output tables were compiled from 10 GTAP7 and Argentina’s INDEC. Since each of these sources uses different classification schemes, we first merged all databases together through the concordances described in Table 2, which also describes the five input service sectors and the nine manufacturing sectors used in our analysis. The final sample consists of 132 developing countries over the period 2002-2008. The OECD CRS database includes detailed information about the donor and recipient country, purpose of the aid disbursed, and the amount disbursed. We use this information to classify data on aid flows targeting each service sector as well as aid directly addressed to each manufacturing sector. Figure 2 illustrates total aid disbursements to service sectors between 2002 and 2008. Our sample only starts in 2002, due to the quality of data on aid disbursements. Indeed, OECD affirms that the annual coverage of the CRS data on disbursed aid is below 60% before 2002, whereas it is over 90% since 2002 and reached nearly 100% starting with 2007 flows.7 The key ingredient in our analysis is the link between inputs and outputs across sectors in an economy, as described in the input-output matrix. Extracting balanced input-output tables from the GTAP dataset requires several steps: the essential reference is McDonald and Thierfelder (2004). We first extracted all the social accounting matrix (SAM) components that are part of an input-output table, and then calculate the total input requirements for each sector.8 Out of the 132 countries in the sample, only 60 have country specific input-output tables in GTAP7; the remaining 72 countries are matched to 18 regions for which regional input-output tables are available. 7 http://www.oecd.org/document/50/0,3746,en_2649_34447_14987506_1_1_1_1,00.html 8 Input-output data extracted from the GTAP data set are the result of the GTAP fitting procedure. Original data are not provided in the GTAP data set. 11 Any change in a service sector will affect directly downstream manufactures using it, but also indirectly through other inputs using this service9. We are interested in the total effect that can be defined as the sum of direct and indirect effects. It can be derived from the total requirement matrix that is computed using a simple multiplier, (I-A)-1, where I is the identity matrix and A is a matrix of direct input coefficients. Figure 3 depicts the average service intensities of each manufacturing sector across countries in our sample. Table 3 presents summary statistics for the total input requirement of all five service sectors. 5. Results This section describing results is divided in three subsections. The first subsection presents the baseline estimates and a number of robustness checks. The second one displays results for sub-samples of countries grouped by income and regional breakdowns. The third one analyzes the impact of aid on total output rather than on exports. 5.1. Baseline Estimates Table 4 reports estimates of equation (1) using ordinary least squares on a panel covering the period 2002-2008. The baseline estimates in column 1 indicate that a 10 percent increase in aid to transportation, ICT, energy, and banking services is associated a 2 percent, 0.3 percent, 6.8 percent, and 4.7 percent increase of manufacturing exports, respectively. Aid to the business service sector is positive but not statistically significant. We carry out a number of robustness checks on the baseline estimates to address concerns about data limitations. As mentioned in Section 3, only 60 out of the 132 countries in the sample have country-specific input-output tables in GTAP. To confirm that our results are 9 For a detailed discussion on input-output analysis see: Miller and Blair (2009). 12 not driven by the 72 countries sharing 18 regional input-output tables, we estimate the baseline specification on the sample of 60 countries that have country-specific input-output tables; column 2 displays the results. With less than half of the total number of observations, aid to transportation, energy, and banking services maintains the positive significant relation with manufacturing exports whereas aid to ICT and business services is no longer positive; however it is not statistically significant. A different way to verify the sensitivity of the input-output data in our estimates is to use the same input-output table for all countries. We replace services intensities computed from GTAP input-output tables with intensities obtained from the total input requirement table from Argentina.10 The estimates are shown in column 3 of Table 4. The main results hold: aid to transportation, energy, and banking services is positively associated with higher levels of manufacturing exports. However, in this case, aid to business services becomes positive and significant whereas the coefficient on aid to ICT services remains statistically insignificant. As mentioned in the data description, CRS aid disbursements coverage has been improving over time. As a way to check for measurement error in aid disbursement data we estimate the baseline regression using only the last three years of available data. Column 4 of Table 4 displays the results. Aid to all service sectors displays a positive relation with manufacturing exports; however, aid to banking is not statistically significant. Notice that the magnitude of the coefficient on business services increases substantially. This could be due to the concentration of more resources into this type of services in the last few years as shown in Figure 2. As a final robustness check, we estimate equation (1) using only the sample of countries that have export data for every manufacturing sector and every year. This sample consists of 117 10 http://www.indec.gov.ar/ 13 countries, omitting 15 mostly smaller island countries such as Palau, Samoa, and Vanuatu. The results are shown in column 5 of Table 4. The main results still hold: aid to the transportation, energy, banking, and business service sectors reveals a positive and significant relation with exports whereas aid to ICT services again is not statistically significant. 5.2. Estimates by Income and Regional Samples We estimate the differentiated impact of aid effectiveness by region and income level of recipient countries. A priori, aid effectiveness may differ in each region and income group, as each subgroup of countries produce and export a different bundle of goods given their initial endowments, quality of services, and production technology. Attempting to measure the effectiveness aid to services in each region and income group can provide a starting point for the analysis of individual country policy recommendations. Table 5 displays the results of estimating equation (1) for each income group of countries, as defined by the World Bank.11 Akin to baseline estimates, aid to transportation, ICT, energy, and banking in low-income countries has a positive and significant effect on manufacturing exports. Only aid to business services records a negative impact on exports in low-income countries. In lower-middle income countries, aid to ICT services is associated with lower levels of manufacturing exports whereas in upper-middle income countries aid to transportation services seems to have a negative impact. The effect of aid to transportation on exports growth is lower the higher the income of the group of countries. As aid to transportation mainly consists in investments on physical infrastructure, this result is consistent with Portugal-Perez & Wilson’s (2010) finding that the 11 Economies are divided according to 2009 GNI per capita, calculated using the World Bank Atlas method. The groups are: low income, $995 or less; lower middle income, $996 - $3,945; upper middle income, $3,946 - $12,195; and high income, $12,196 or more. 14 marginal effect of physical infrastructure improvement on exports appears to be decreasing in per capita income. Aid to energy and business services follows the opposite progression across income groups, as their impact on exports increases with the income of the group. Regarding the progression of aid to banking and business services across income groups, aid to banking has a positive impact in low-income groups and aid to business services has a negative impact. By opposition, the impact of aid to banking services is negative for upper- middle countries, whereas the impact of aid to business services is positive. Lower-middle income countries experience a positive impact of both aid to banking and services. These results suggests that, as upper-middle income countries might be more financially developed, aid to banking services does not have the positive impact it does in low income countries. Furthermore, once firms have access to financial services, other services such as business development and technical advisory seem to become a priority; consequently, aid to business services in middle income countries has a positive effect on manufacturing exports. Estimates on sub-samples of countries by regions, presented in Table 6, also confirm these results. For example, for the regions with the highest percentage of upper-middle income countries—Latin America and the Caribbean (LAC) and for Middle East and North Africa (MNA)—aid to business services is positively associated with higher manufacturing exports whereas aid to banking services displays a negative relation with manufacturing exports. On the other hand, in South Asia where there are only low income and lower-middle income countries, aid to banking is positively associated with manufacturing exports whereas aid to business services is negatively associated. 15 5.3. Aid and Output If aid has a positive impact it should improve the quality of the services provided and/or reduce the costs to downstream users of these services. Thus, industries that use these services more intensively should be able to produce more (i.e., cheaper inputs) and ultimately export more as well. We use the same methodology described in Section 3 to confront this idea with actual output data. Table 7 shows the results for a selected number of countries that have output data available. The dependent variable in column 1 is the natural logarithm of individual industry output whereas the dependent variable in column 2 is, as in previous results, the natural logarithm of exports for each industry. We can see that the effect of aid to services is very similar for manufacturing output as it is for exports except that the effect, in absolute terms, is slightly greater for output. These results make sense as any improvement in services will benefit all firms—exporting and non-exporting—thus the effect of aid is slightly greater for output than for exports. 5. Conclusions Assessing the effectiveness of trade-related projects is a challenging task. Rigorous methods of impact evaluation used more extensively in education, cash transfers, and health programs, are less easily implementable in trade projects. In the absence of micro-level impact evaluation, macro-level evaluations can provide a general assessment on whether aid has had the expected impact on a specific variable. Certainly, macro-level exercises are not free from data and limitations and the presence of endogeneity, and results have to be taken carefully for policy recommendations. 16 We propose an original identification strategy to measure the impact of aid on exports exploiting the input-output linkages in an economy to analyze how aid to service sectors affects manufacturing exports. We believe that our estimating strategy is econometrically sound and the results suggest areas for further exploration, as it relates to discussions on aid for trade. A similar analysis can be extended to estimate the spillover effects of aid on different input sectors on downstream exports. We find that aid to the transportation and energy sectors is the most effective when the objective is to increase the exports of a recipient country. These results are robust to a number of different specifications. The effectiveness of aid to transportation in terms of export growth diminishes for country groups with higher income, whereas the effectiveness of aid to energy and business services increases with the income of the group. Finally, we believe that the results presented can help inform future research and policy discussions on an area of aid for trade largely unexplored, the link between aid and the services sector. A more detailed analysis involving rigorous impact evaluation of specific projects covering costs and benefits for a developing country can provide guidance on how to increase aid efficiency. 17 References Acemoglu, D., Johnson, S. and Mitton T. (2009). Determinants of vertical integration: financial development and contracting costs. Journal of Finance, 64 (3), pp. 1251–90. Alfaro, L. and Charlton, A. (2009). Intra-industry foreign direct investment. American Economic Review, 99 (5), pp. 2096-2119. Angeles, L., and Neanidis, K., (2009). Aid effectiveness: the role of the local elite. Journal of Development Economics, 90 (1), pp. 120-134. Arellano, C., Bulir, A., Lane, T. and Lipschitz, L. (2005). The dynamic implications of foreign aid and its variability. IMF Working Paper 05/119. International Monetary Fund,Washington. Berg, A., Hussain, M., Aiyar, S., Roache, S. and Mahone, A. (2005). The Macroeconomics of Managing Increased Aid Inflows: Experiences of Low-Income Countries and Policy Implications. International Monetary Fund, Washington. Bourguignon, F. and Sundberg, M. (2007). Aid effectiveness: opening the black box. American Economic Review, 97 (2), pp. 316-321. Brenton P., and von UexKull, E. (2009). Product specific technical assistance for exports – has it been effective? The Journal of International Trade and Economic Development, 18 (2), pp. 235- 254. Bruckner, M. (2011). On the simultaneity problem in the aid and growth debate. University of Adelaide School of Economics Research Paper No. 2011-01. Burnside, C., and Dollar, D. (2000). Aid, policies, and growth. American Economic Review, 90 (4), pp. 847-868 Calì, Massimiliano & te Velde, Dirk Willem, 2011. "Does Aid for Trade Really Improve Trade Performance?," World Development, vol. 39(5), pages 725-740. Clemens, M., Radelet, S. and Bhavnani, R. (2004). Counting chickens when they hatch: the short-term effect of aid on growth. Center for Global Development Working Paper No. 44. Collier, P. and Dollar, D. (2002). Aid allocation and poverty reduction. European Economic Review, 46 (8), pp. 1475-1500. Dalgaard, C., Hansen, H. and Tarp F. (2004). On the empirics of foreign aid and growth. Economic Journal, 114 (496), pp. 191-216. Easterly, W. (2003). Can foreign aid buy growth? Journal of Economic Perspectives, 17 (3), pp. 23-48. 18 Easterly, W., Levine, R. and Roodman, D. (2003). New data, new doubts: revisiting aid, policies, and growth.‖ Center for Global Development Working Paper No. 26. Ellison, G., Glaeser, E, and Kerr, W. (2010). What causes industry agglomeration? Evidence from coagglomeration patterns. American Economic Review, 100 (3), pp. 1195-1213. Freedom House (2009). Freedom in the world country ratings (online database). . Gamberoni, E. and Newfarmer, R. (2009). Aid for trade: matching supply and demand. World Bank Policy Research Working Paper No. 4991. Gartzke, E. (2009). The Affinity of Nations Index, 1946–2002, version 4.0 and related documents. . Helble, M., Mann, C.L., and Wilson, J.S. (2009). Aid for trade facilitation. World Bank Policy Research Working Paper No. 5064. Hummels, D., Ishii, J., and Yi, K. (2001). The nature and growth of vertical specialization in world trade. Journal of International Economics, 54 (1), pp. 75–96. Miller, R., and Blair, P. (2009). Input-Output Analysis Foundations and Extensions. Cambridge University Press: New York. Mitton, T. (2008). Institutions and concentration. Journal of Development Economics, 86 (2), pp. 367-394. Moretti E. (2004). Education, spillovers, and produtctivity. American Economic Review, 94 (3), pp. 656-690. Narayanan, B., and Walmsley,T. Eds. 2008. Global Trade, Assistance, and Production: The GTAP 7 Data Base, Center for Global Trade Analysis, Purdue University. Available online at: http://www.gtap.agecon.purdue.edu/databases/v7/v7_doco.asp Portugal-Perez, A, and J. S. Wilson (2010) ―Export Performance and Trade Facilitation Reform: Hard and Soft Infrastructure‖, World Bank Policy Research Working Paper No 5261, World Bank Prati, A., and Tressel, T. (2006). Aid volatility and Dutch disease. Is there a role for macroeconomic policies? mimeo, Washington: International Monetary Fund. Radelet, S., Clemens, M. and Bhavnani, R. (2006). Aid and growth: the current debate and some new evidence, in P. Isard, L. Lipschitz, A. Mourmouras, and B. Yontcheva (eds) The Macroeconomic Management of Foreign Aid: Opportunities and Pitfalls, Washington, DC: International Monetary Fund, pp. 43–60. 19 Rajan, R., and Subramanian, A. (2008). Aid and growth: what does the cross-country evidence really show? The Review of Economics and Statistics, 90 (4), pp. 643-665. Rajan, R., and Subramanian, A. (2011). Aid, Dutch disease, and manufacturing growth. Journal of Development Economics, 94 (1), pp. 106-118. Rajan, R., and Zingales, L. (1998). Financial dependence and growth. American Economic Review, 88 (3), pp. 559–586. Suwa-Eisenmann, A., and Verdier, T. (2007). Aid and trade. Oxford Review of Economic Policy, 23 (3), pp. 481-507. Trefler, D., and Zhu, S. (2010). The structure of factor content predictions. Journal of International Economics, 82 (2), pp. 195-207. Younger, S. (1992). Aid and Dutch disease: macroeconomic management when everybody loves you. World Development, 20 (11), 1587–1597. 20 TABLES AND FIGURES Figure 1 – Aid for Trade (USD millions) 10,000 15,000 20,000 25,000 5,000 0 90 93 94 95 98 99 03 04 05 08 91 92 96 97 02 00 01 06 07 19 19 19 19 19 19 20 20 20 20 19 19 19 19 20 20 20 20 20 AFT - capacity AFT - infrastructure AFT - regulation Figure 2 –Total AFT vs. Share of AFT in Total Aid 25000 .6 .5 20000 (sum) usd_disbursement share_AFT .4 15000 .3 10000 .2 5000 .1 1990 1995 2000 2005 2010 year Total AFT AFT/Total Aid 21 Figure 3 – Aid to Service Sectors (USD millions) 8,000 6,000 4,000 2,000 0 2002 2003 2004 2005 2006 2007 2008 Transportation ICT Energy Banking Business Figure 4 – Average Service Intensity by Manufacturing Sector .2 .15 .1 .05 0 s t. d. ls ls ry s d. s al qm al ile a ra ne In In et et ic xt e E hi o d m M M in Te gr oo ac rt he M us ic A po W M s C llic rro Ba n as a - fe et Tr on -M N on N Trasnport ICT Energy Banking Business 22 Table 1 – Total 2008 AFT by categories 2008 code/ sector name Disbursements (USD millions) (%) Infrastructure 13,112 51% 210 Transport & Storage 7,494 29% 220 Communications 461 2% 230 Energy 5,157 20% Production Capacity 11,741 45% 240 Banking & Financial Services 2,892 11% 250 Business & Other Services 1,943 8% 311 Agriculture 4,668 18% 312 Forestry 534 2% 313 Fishing 341 1% 321 Manufacturing 1,362 5% Agro-industries 104.5 0.40% Wood industries 3.9 0.01% Textiles 10.3 0.04% Chemicals 119.5 0.46% Non metallic products 0.8 0.00% Basic Metals 11.4 0.04% Non-ferrous metals 0.3 0.00% Machinery 41.8 0.16% Transport equipment 21.1 0.08% Energy manufacturing 1.4 0.01% Ind. policy, development, R&D 1,047 4.04% 322 Mineral Resources & Mining 241 1% Trade Policies and Regulations 795 3% Total 25,888 100% 23 Table 2 - Databases' Concordances CRS COMTRADE INDSTAT GTAP (aid) (exports) (output) (input-output) Sector Industry purpose code SITC ISIC Rev. 3 GTAP AGG agriculture agriculture 311 11 pdr, wht, gro, v_f, osd, c_b, pfb, ocr, ctl, oap, rmk, wol forestry forestry 312 12 frs fishing fishing 313 13 fsh mining mining 322 21, 22, 23, 29 coa, oil, gas, omn construction construction 323 cns manufacturing agro-industries 32161 31 15, 16 cmt, omt, vol, mil, pcr, sgr, ofd, b_t forest industries 32162 33, 34 20, 21, 22 lum, ppp textile 32163 32 17, 18, 19 tex, wap, lea chemicals 32164, 32165, 32168 35 23, 24, 25 p_c, crp non-metallic mineral 32166 36 26 nmm basic metal 32169 371 271 i_s non-ferrous metal 32170 372 272 nfm engineering 32171 382, 383, 385 29,30,31,32,33 ele, ome transport equipment 32172 384 34, 35 mvh, otn other manufacturing Other under sector 321 381, 39 28, 36, 37, 273 fmp, omf services transportation 210 otp, wtp, atp ICT 220 cmn energy 230 ely, gdt, wtr banking 240 ofi, isr business 250 obs other inputs trd, ros, osg, dwe 24 Table 3 – Summary Statistics Variable Obs. Mean Std. Dev. Min Max ln_trade0 8243 16.768 3.388 3.714 27.253 ln_intaid_trans 8243 10.234 4.970 -10.353 17.892 ln_intaid_ict 8243 6.182 4.386 -9.370 13.375 ln_intaid_energy 8243 8.784 6.037 -7.796 18.750 ln_intaid_bank 8243 6.950 5.960 -11.926 16.513 ln_intaid_bus 8243 7.742 5.252 -6.755 15.222 Transport int. 8243 0.026 0.019 0.00003 0.196 ICT int. 8243 0.006 0.004 0.00001 0.042 Energy int. 8243 0.048 0.059 0.00000 0.404 Financial int. 8243 0.015 0.012 0.00001 0.086 Business int. 8243 0.018 0.015 0.00002 0.086 Transport disb. 8243 33.900 77.885 0 813.754 ICT disb. 8243 1.871 5.283 0 73.600 Energy disb. 8243 17.570 45.730 0 446.367 Financial disb. 8243 8.910 27.555 0 309.197 Business disb. 8243 6.525 24.402 0 477.660 Source: Own calculations for baseline sample. 25 Table 4 – Baseline Estimations (1) (2) (3) (4) (5) Baseline I-O only ARG I-O 06-08 Complete Transportation 0.196*** 0.493*** 0.175*** 0.311*** 0.187*** [0.017] [0.028] [0.017] [0.023] [0.016] ICT 0.033* -0.011 -0.02 0.401*** -0.021 [0.017] [0.013] [0.017] [0.030] [0.016] Energy 0.684*** 0.442*** 0.612*** 0.035* 0.651*** [0.020] [0.017] [0.018] [0.018] [0.018] Banking 0.468*** 0.467*** 0.541*** 0.036 0.576*** [0.026] [0.024] [0.020] [0.044] [0.020] Business 0.007 -0.018 0.030*** 0.405*** 0.031*** [0.014] [0.020] [0.011] [0.045] [0.011] Observations 8243 3339 8243 3552 7371 R-squared 0.99 0.99 0.99 0.99 0.99 Robust standard errors in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. Dependent variable is ln(exports). All regressions control for country-sector, country-year, and sector-year effects. Table 5– Estimates by Income Groups (1) (2) (3) LOW MIDLW MIDUP Transportation 0.763*** 0.696*** -0.355*** [0.038] [0.039] [0.035] ICT 0.061*** -0.395*** 1.035*** [0.020] [0.032] [0.053] Energy 0.142*** 0.283*** 1.092*** [0.018] [0.025] [0.027] Banking 0.222*** 0.309*** -1.460*** [0.040] [0.024] [0.025] Business -0.189*** 0.346*** 1.443*** [0.020] [0.030] [0.043] Observations 3265 3195 1783 R-squared 0.99 0.99 0.99 Robust standard errors in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. Dependent variable is ln(exports). All regressions control for country-sector, country-year, and sector-year effects. 26 Table 6 – Estimates by Geographic Regions (1) (2) (3) (4) (5) (6) EAP ECA LAC MNA SAS SSA Transportation 0.576*** 1.420*** 1.020*** 0.846*** 0.802*** 0.025 [0.050] [0.022] [0.015] [0.026] [0.050] [0.045] ICT -0.283*** -0.007 0.457*** 0.324*** 0.118*** 0.483*** [0.056] [0.032] [0.017] [0.041] [0.022] [0.031] Energy 0.437*** -0.074* -1.276*** 0.177*** 0.153*** 0.637*** [0.032] [0.041] [0.013] [0.009] [0.048] [0.031] Banking 0.388*** 0.060*** -0.364*** -0.272*** 0.137** 0.223*** [0.053] [0.019] [0.015] [0.020] [0.054] [0.047] Business 0.412*** 0.001 1.663*** 1.030*** -0.146*** 0.090* [0.031] [0.029] [0.015] [0.016] [0.020] [0.047] Observations 1237 972 1827 756 504 2947 R-squared 0.99 0.99 0.99 0.99 0.99 0.99 Robust standard errors in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. Dependent variable is ln(exports). All regressions control for country-sector, country-year, and sector-year effects. Regional abbreviations are as follows: East Asia Pacific (EAP); East Europe and Central Asia (ECA); Latin America and the Caribbean (LAC); Middle East and North Africa (MNA); South Asia (SAS); and South Saharan Africa (SSA). Table 7 – Output vs. Exports (1) (2) output exports Transportation 0.389*** 0.323*** [0.007] [0.015] ICT -0.369*** -0.324*** [0.005] [0.015] Energy 0.079*** 0.064*** [0.003] [0.006] Banking 0.384*** 0.342*** [0.004] [0.008] Business 0.327*** 0.307*** [0.008] [0.016] Observations 2070 2069 R-squared 0.99 0.99 Robust standard errors in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. Dependent variable in column 1 is ln(ouput) in column 2 is ln(exports). All regressions control for country-sector, country-year, and sector-year effects. 27 Appendix. List of countries Afghanistan Eritrea Mongolia Turkey Albania Ethiopia(excludes Eritrea) Montenegro Turkmenistan Algeria Fiji Morocco Uganda Angola Gabon Mozambique Ukraine Argentina Gambia, The Namibia Uruguay Armenia Georgia Nepal Uzbekistan Azerbaijan Ghana Nicaragua Vanuatu Bangladesh Grenada Niger Venezuela Belarus Guatemala Nigeria Vietnam Belize Guinea Oman Yemen Benin Guinea-Bissau Pakistan Zambia Bhutan Guyana Palau Zimbabwe Bolivia Haiti Panama Botswana Honduras Papua New Guinea Brazil India Paraguay Burkina Faso Indonesia Peru Burundi Iran, Islamic Rep. Philippines Cambodia Jamaica Rwanda Cameroon Jordan Samoa Cape Verde Kazakhstan Sao Tome and Principe Central African Rep. Kenya Senegal Chad Kiribati Seychelles Chile Korea, Dem. Rep. Sierra Leone China Kyrgyz Republic Solomon Islands Colombia Lao PDR Somalia Comoros Lebanon South Africa Congo, Dem. Rep. Lesotho Sri Lanka Congo, Rep. Liberia St. Kitts and Nevis Costa Rica Libya St. Lucia Cote d'Ivoire Macedonia, FYR St. Vincent Grenadines Croatia Madagascar Sudan Cuba Malawi Suriname Djibouti Malaysia Swaziland Dominica Maldives Syrian Arab Rep. Dominican Rep. Mali Tajikistan East Timor Mauritania Tanzania Ecuador Mauritius Thailand Egypt, Arab Rep. Mexico Togo El Salvador Micronesia, Fed. Sts. Tonga Equatorial Guinea Moldova Tunisia 28