Does India Use Development Finance to Compete With China? A Subnational Analysis

Asmus-Bluhm, Gerda; Eichenauer, Vera Z.; Fuchs, Andreas; Parks, Bradley Article — Published Version Does India Use Development Finance to Compete With China? A Subnational Analysis Journal of Conflict Resolution Provided in Cooperation with: Kiel Institute for the World Economy – Leibniz Center for Research on Global Economic Challenges Suggested Citation: Asmus-Bluhm, Gerda; Eichenauer, Vera Z.; Fuchs, Andreas; Parks, Bradley (2024) : Does India Use Development Finance to Compete With China? A Subnational Analysis, Journal of Conflict Resolution, ISSN 1552-8766, SAGE Publications, Thousand Oaks, Calif., Vol. 69, Iss. 2-3, pp. 406-433, https://doi.org/10.1177/00220027241228184 , https://journals.sagepub.com/doi/10.1177/00220027241228184 This Version is available at: https://hdl.handle.net/10419/319916 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc/4.0/ Article Journal of Conflict Resolution 2025, Vol. 69(2-3) 406–433 © The Author(s) 2024 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/00220027241228184 journals.sagepub.com/home/jcr Does India Use Development Finance to Compete With China? A Subnational Analysis Gerda Asmus-Bluhm 1,2 , Vera Z. Eichenauer 3,4,5 , Andreas Fuchs 2,4 , and Bradley Parks 6,7 Abstract China and India increasingly provide aid and credit to developing countries. This article explores whether India uses these financial instruments to compete for geopolitical and commercial influence with China. We build a new geocoded dataset of Indian government-financed projects in the Global South between 2007 and 2014 and combine it with data on Chinese government-financed projects. Our regression results for 2,333 provinces within 123 countries demonstrate that India’s Exim Bank is significantly more likely to locate a project in a given jurisdiction if China provided government financing there in the previous year. Since this effect is more pronounced in countries where India is more popular relative to China and where both lenders have a similar export structure, we interpret this as evidence of India competing with China. By contrast, we do not find evidence that China uses official aid or credit to compete with India through co-located projects. 1 Department of Economics, University of Hohenheim, Germany 2 Department of Economics, University of G¨ ottingen, Germany 3 KOF Swiss Economic Institute, ETH Zürich, Switzerland 4 Kiel Institute China Initiative, Kiel Institute for the World Economy, Germany 5 CESifo, Munich, Germany 6 AidData, Global Research Institute, William & Mary, Williamsburg, VA, USA 7 Center for Global Development, Washington DC, USA Corresponding Author: Vera Z. Eichenauer, KOF Swiss Economic Institute, Eidgen¨ ossische Technische Hochschule Zürich, Leonhardstrasse 21, Zurich 8092, Switzerland. Email: [email protected] Data Availability Statement included at the end of the article. Keywords development finance, foreign aid, official development assistance, official credits, SouthSouth cooperation, China, India, geostrategic competition, geospatial analysis Introduction China and India—the world’s two most populated countries—committed almost US$155 billion in aid and credit to 132 developing countries between 2007 and 2014 alone. 1 Casual observation suggests that China and India may be using their financial resources to compete for geostrategic and commercial advantages around the globe. Since 2013, China has engaged in an unprecedented effort to build a “Belt”of roads, rails, ports, and pipelines from China to Central Asia and Europe and a “Maritime Silk Road”consisting of deep-water ports along the littoral areas of the Indian Ocean (Brewster 2015;Mullen and Poplin 2015). India is one of the few countries in the Asia-Pacific region that formally opposes China’s Belt and Road Initiative (BRI) (Pant 2017). It also refuses to accept bilateral aid from China. According to Agrawal (2007, 7), Delhi competes with Beijing over “diplomatic influence, oil reserves, and markets for goods.”Journalistic reports and case studies also suggest that India is sensitive to any real or perceived attempts by China to encroach upon its existing spheres of influence or otherwise engage in expansionary efforts. 2 During our own interviews with decision makers in Delhi, one official from India’s Exim Bank indicated that countering Chinese influence is a consideration when the government decides on loan and export credit approvals. However, other interviewees claimed that India was “not benchmarking China at all”and such claims are a “media myth.” 3 The existing literature is largely silent on the issue of whether and how emerging powers use aid and credit as tools of strategic rivalry. We seek to close this evidence gap by conducting a rigorous analysis of whether and how India and China use development finance instruments to compete with each other around the globe. We first determine if the Indian government allocates aid and credit across developing countries and their subnational localities in response to the receipt of Chinese government financing. Specifically, we hypothesize that if a developing country receives a new financial commitment from the Chinese government, it becomes more likely that the Indian government provides funding to the same recipient country (Hypothesis 1) and same subnational locality (Hypothesis 2). We then evaluate whether India’s behavior is consistent with the notion that it is seeking to compete with China. 4 Those who assert that Delhi and Beijing use official financial instruments to compete with each other rarely articulate falsifiable hypotheses about government motivations. We follow Steinwand (2015, 451) by defining “competition”as the provision of government financing “to counteract the influence gained by other donors.”Therefore, if the Indian government responds to a new Chinese government-financed project by increasing its financial footprint in the same country or subnational locality, we consider this to be potential evidence of India seeking to compete with China. By Asmus-Bluhm et al. 407 increasing its presence in these jurisdictions, Delhi may be able to constrain or even undermine Beijing’sinfluence in relative terms—either at the country level or subnationally with respect to regional governments or local firms. However, to more directly test whether Delhi seeks to counteract actual influence gains achieved by China, we leverage public opinion data to determine if Delhi assigns special priority to jurisdictions where popular sentiment is more or less favorable towards India than China. We also differentiate between competition for commercial reasons (e.g., to secure export markets) and competition for geopolitical reasons (e.g., to strengthen ties with regional partner countries). Existing research on the competitive use of development finance focuses on rivalry between traditional providers of development finance (Davies and Klasen 2019;Fuchs, ¨ Ohler, and Nunnenkamp 2015;Mascarenhas and Sandler 2006) and rivalry between established and emerging powers (Bueno de Mesquita and Smith 2016;Hernandez 2017;Humphrey and Michaelowa 2019;Zeitz 2021). However, to the best of our knowledge, rigorous studies of emerging-power competition that cover a large set of developing countries are non-existent. One reason is the absence of comprehensive and reliable data. To better understand whether and how India and China use official financial instruments to compete with each other in developing countries, we construct a new dataset that contains information on 1,196 Indian government-financed projects in 4,308 locations within 93 countries from 2007 to 2014. 5 In sum, the monetary value of these projects amounts to US$14.58 billion (in constant 2014 values). The dataset captures official development assistance, concessional and non-concessional loans, export credits, and other state-sponsored financial flows from India’s two most important sources of development finance: the Ministry of External Affairs (MEA) and the Export-Import (Exim) Bank of India. We combine these data with geocoded Chinese official financing data from Bluhm et al. (2020), covering 3,485 projects in 6,184 subnational locations within 138 countries. We then compare the timing and location of Indian projects with those financed by China. Specifically, we run regressions on a sample covering 2,333 provinces within 123 countries and estimate the effect of Chinese government-financed projects on the allocation of Indian official financing using a linear probability model with countryprovinceand country-year-fixed effects. 6 Our results show that India’s Exim Bank is more likely to allocate a credit-financed project to a subnational locality if the Chinese government provided financing there in the previous year. We observe weaker effects at the national level. By contrast, development aid provided by India’s MEA does not follow China’s development activities in the average recipient country. We only find that the MEA allocates significantly more development aid in response to Chinese projects where it arguably matters most for geostrategic competition: in India’s neighborhood. Since our main finding is more robust for India’s Exim Bank, which primarily follows commercial objectives, we conclude that India is engaging primarily in commercial rather than geopolitical competition with China. Our main analysis focuses on whether the distribution of funding from the Chinese government influences the aid and credit allocation decisions of the Indian 408 Journal of Conflict Resolution 69(2-3) government—rather than the reverse relationship. India is the more likely follower given that Beijing oversees a substantially larger portfolio of overseas development projects (Asmus, Fuchs, and Müller 2020). 7 Nevertheless, we also test whether China competes with India, and consistent with our expectation, we find no evidence that Indian projects attract Chinese projects to the same localities. A major empirical challenge is disentangling competition from other factors that may lead to a positive association between Indian and Chinese government financing. Chief among these alternative explanations are selectivity and imitation (Davies and Klasen 2019). Selectivity refers to the possibility that China and India follow similar financial allocation criteria. For example, both countries might favor jurisdictions with lower levels of economic development or higher-quality policies and institutions. Imitation refers to the possibility that foreign donors and lenders—operating in unfamiliar settings with imperfect access to information—take cues from their peers with more local experience and tacit knowledge. We run several tests to disentangle competition from selectivity and imitation. First, we partially account for selectivity in our regression models by controlling for the various allocation determinants identified in the literature either directly or indirectly through the inclusion of fixed effects. Second, we run sectoral regressions to test whether “crowding-in”takes place within the same sectors. When China provides aid or credit to a particular sector within a particular subnational locality, we find no evidence that India is more likely to approve aid or credit for the same jurisdiction and sector in the following year. It is thus unlikely that India is simply responding (more slowly) to the same allocation criteria than China. In recognition of the fact that imitation most likely occurs within sectors but competition can occur across sectors, this sectoral finding is also inconsistent with the alternative explanation of imitation. Third, in order to test whether India is motivated by a desire for greater influence vis- ` a-vis China, we leverage data from the Gallup World Poll. We find that India’s Exim Bank is more likely to increase its financial footprint in jurisdictions where it is more popular relative to China. This is a pattern that is difficult to reconcile with any explanation other than competition with China. Since the crowding-in effect is also more pronounced in countries where both lenders have a similar export structure, it appears that both emerging economies compete over commercial influence. Therefore, the weight of the evidence suggests that India is competing with China rather than simply imitating it or following a similar set of allocation criteria. This article contributes to the literature in several ways. First and foremost, it is the first to analyze China-India competition in developing countries in a quantitative analysis. Second, to the best of our knowledge, we provide the first study of local competition using geocoded development finance data for two bilateral donors. Third, by including Exim Bank loans in our analysis, our study contributes to a relatively small body of evidence on South-South official financing flows to developing countries other than official development assistance (Bunte 2019;Horn, Reinhart, and Trebesch 2021; Kaya, Kilby, and Kay 2021;Werker, Ahmed, and Cohen 2009). Less concessional and more commercially-oriented projects make up a substantial proportion of the official Asmus-Bluhm et al. 409 financing that Southern providers of development finance commit to their peers each year, yet they largely represent a blind spot in the empirical literature. A New Dataset: Indian Development Finance at the Project Level In collaboration with AidData, a research lab at William & Mary, we collected projectlevel information on the two major Indian agencies that provide official financing to other developing countries: the MEA and the Exim Bank (see Appendix C for details on the data collection process). The first step of the data collection process was to retrieve project-level information from official government documents. The resulting database consists of 1,196 projects in 93 countries from 2007 to 2014. In sum, the monetary value of these projects amounts to US$14.58 billion (in constant 2014 values), of which US$9.56 billion were committed by the Exim Bank and US$5.02 billion originate from the MEA. In a second step, we allocated all projects to (sub)sectors according to the OECD-DAC definitions, ranging from “Education”to “Humanitarian Assistance.” In a third step, we geocoded all of the projects in this database. Our 1,196 projects are located at 4,308 project intervention sites at various administrative levels. Our dataset accounts for the varying levels of geographic precision and coverage of Indian projects. As illustrated in Figure 1 for the Nepalese case, we created line features on the map that follow the projected course of infrastructure projects such as roads and railways to identify all regions that were affected by infrastructure projects. For our empirical analysis below, we then extracted all first administrative units through which a line feature cuts. The spatial precision of 599 projects is sufficient so that we can assign 2,166 project locations to their respective province. Figure 2 presents these project locations in a world map. Aggregating the project-level data to the level of world regions, we observe that India prioritizes Africa and Asia. The African countries receiving most projects are Liberia, Ghana, Malawi, Mauritius, and Mozambique. Nepal, Afghanistan, Myanmar, Bhutan, and Sri Lanka top the list of India’s Asian recipients. We also observe that the MEA initiates more projects than the Exim Bank, while the Exim Bank commits larger amounts of official financing. In terms of financial values, 64 percent of India’sfinancing commitments in our dataset are from the Exim Bank, while only 36 percent come from the MEA. In contrast, the MEA dominates with 88 percent in terms of the number of projects and is represented in almost three times more subnational locations. This new dataset allows us to test whether our expectation (explained in more detail in Appendix B) that MEA aid is mainly driven by geopolitical interests and Exim Bank loans follow primarily commercial interests are reflected in the data. Figure 3 shows the results of a one-standard-deviation change in each of the explanatory variables on the logged monetary amount of Indian OOF and ODA, respectively. 8 In line with expectations, commercial variables enter significantly in OOF but not in ODA regressions. Specifically, larger Exim Bank funds (OOF) are given to countries to which India exports more, which is in line with the institution’s mandate. Moreover, smaller 410 Journal of Conflict Resolution 69(2-3) amounts of loans flow to more indebted countries, which likely follows from concerns about the ability of these countries to repay their debt. By contrast, MEA’s aid allocation (ODA) reflects its geopolitical interests. More precisely, India’s voting alignment in the UN General Assembly is significantly correlated with MEA aid. MEA aid is also more targeted towards countries in India’s neighborhood, which are arguably of larger strategic interest to India. Both political variables, UN voting and geographic distance, do not turn out to be significant predictors of Exim Bank loans. Based on these results and our theoretical considerations (Appendix B), we will interpret India’s MEA response to China’sofficial financing activities as evidence of geopolitical competition. Conversely, we will interpret India’s response to China’sofficial financing activities through India’s Exim Bank as evidence of commercial competition. Empirical Strategy We use the geocoded information about the provincial location of Indian and Chinese official projects to analyze whether and how India locates projects in the same Figure 1. Geocoded Indian and Chinese development projects in Nepal (2000/07–2014). Notes. This figure illustrates the locations of development projects in Nepal. The green triangles represent the locations of Indian development projects (2007–2014). The blue lines represent line features of Indian development projects, including the Terai Road project. The red circles represent the locations of Chinese development projects (2000–2014). Source: India: Authors’data, China: Bluhm et al. (2020). Asmus-Bluhm et al. 411 Figure 3. Marginal effects of potential determinants of India’s allocation of Exim Bank loans (OOF) and MEA aid (ODA) (standardized coefficients, OLS, 2007–2014). Notes. The figure shows the results of least-squares regressions of India’s country allocation of Exim Bank loans (in blue) and MEA aid (in red) over 8 years (2007–2014). The dots represent a one-standard-deviation change in the respective explanatory variable on the logged monetary amount of countrywide Indian OOF and ODA, respectively (together with 90 percent and 95 percent confidence intervals). All regressions include year dummies. Standard errors are clustered by country. The number of observations is 887 in 119 country clusters. The R-squared takes values of 9.4 percent (Exim Bank) and 32.8 percent (MEA), respectively. Source: Authors’calculations. Figure 2. World map of Indian and Chinese government-financed project locations (2000/07– 2014). Notes. Source: India: Authors’data, China: Bluhm et al. (2020). 412 Journal of Conflict Resolution 69(2-3) jurisdictions as China’sofficial financing activities in the 2007–2014 period. First, we test whether India is more likely to commit a development project to a province if China finances a new project in the same country in which the province is located (Hypothesis 1). Second, we test whether India is more likely to commit a development project to a province if China finances a new project in the same province (Hypothesis 2). To test our hypotheses about the allocation of Indian official finance across provinces, we use a linear probability model with country-provinceand year-fixed effects. Formally, we run the following regression model IndiaOFijt ¼αChinaOFjt1þβChinaOFijt1þc0 ijt1γþξij þψtþϵijt, (1) where IndiaOF ijt is a binary variable that takes the value one if India launches either a new Exim Bank loan or MEA aid project in province iof country jin year t; ChinaOF jt1 is a binary variable for a new Chinese project in the same country in the previous year t1; ChinaOF ijt1 is a binary variable for a new Chinese project in the same province in the previous year t1; and c 0 ijt1is the vector of lagged time-variant controls at the province level. While we use a binary dependent variable, IndiaOF ijt , and binary variables of interest, ChinaOF jt1 and ChinaOF ijt1 , in our baseline specification, we replace them by continuous variables that measure the logged monetary value of official finance commitments in robustness tests. The latter comes with the obvious advantage that it accounts for the size of projects. However, one important caveat is that 39 percent of Chinese projects lack information on their monetary value (Dreher et al. 2021,2022). Moreover, given that only 4.2 percent of province-years in our sample are “treated” with Indian and 7.7 percent with Chinese projects, the use of monetary amounts is more likely to be biased by outliers. Therefore, our preferred measure is the binary project variable. We include the following control variables that vary across provinces and time and are lagged by 1 year: the logarithm of average nighttime light to proxy the provincial level of development; the logarithm of average precipitation and the number of conflict fatalities to control for temporary shocks; the logged population level as a larger population might increase the probability of receiving a project; and a binary variable for projects extended by the World Bank to account for the regional dynamics of one of the most important providers of development finance. For the latter, we use loan projects extended by the World Bank’s International Bank for Reconstruction and Development (IBRD) in Exim Bank loan regressions and International Development Association (IDA) aid projects in MEA aid regressions. Country-province-fixed effects ξ ij account for time-invariant countryand province-specific characteristics (such as access to the sea, geographic distance to India, or a recipient’s historical ties with India). 9 Year-fixed effects ψ t absorb year-specific factors such as India’s budget for overseas official finance or changes in the Indian government. Standard errors are cluster-robust at the country level. Appendix Table C-4 provides detailed definitions Asmus-Bluhm et al. 413 Table 3. India’s MEA Aid and Chinese Official Finance (2007–14). Baseline Timing Placebo Baseline Timing Placebo (1) (2) (3) (4) (5) (6) ChinaOF ijt+1 0.009 (0.008) 0.001 (0.009) ChinaOF ijt 0.001 (0.007) 0.008 (0.009) 0.006 (0.007) 0.008 (0.008) ChinaOF ijt1 0.013 (0.013) 0.016 (0.014) 0.012 (0.011) 0.003 (0.008) 0.003 (0.009) 0.000 (0.009) ChinaOF ijt2 0.008 (0.009) 0.010 (0.011) 0.001 (0.008) 0.001 (0.009) ChinaOF ijt3 0.023 (0.014) 0.016 (0.010) 0.010 (0.009) 0.009 (0.011) ChinaOF jt+1 0.004 (0.003) ChinaOF jt 0.002 (0.004) 0.002 (0.004) ChinaOF jt1 0.005 (0.003) 0.005 (0.004) 0.007 (0.004)* ChinaOF jt2 0.002 (0.003) 0.003 (0.004) ChinaOF jt3 0.003 (0.004) 0.002 (0.004) Controls 33 3333 Country-province FE 33 3333 Year FE 33 3333 Country-year FE 33 3 Observations 18,673 18,673 16,338 18,657 18,657 16,324 Adjusted R-squared 0.454 0.457 0.451 0.607 0.607 0.600 Notes. The unit of observation is the province (ADM1 region). Dependent and explanatory variables are binary with 1 indicating if at least one Indian (respectively Chinese) project is committed to a province. All specifications control for the presence of World Bank IDA projects (t-1), log nighttime light (t-1), log precipitation (t-1), log population (t-1), and log conflict-related deaths (t-1). Columns 4–6 include country-year-fixed effects in addition to country-provinceand year-fixed effects. Robust standard errors clustered at the country level are presented in parentheses. Significant at: ***: p< .01; **: p<.05;*:p< .1. 420 Journal of Conflict Resolution 69(2-3) amounts and project counts at the 99.9 th percentile of the respective distribution (see Appendix Table E-3). These results suggest that India does not time the provision of its aid to strengthen its geopolitical ties with other countries at those moments when China engages more intensively with specific countries and provinces via ODA-financed projects. However, this finding may only hold true for the average recipient country. A different picture might emerge if we look at countries where India has particularly strong interests. Therefore, we now turn to an analysis of heterogeneous effects. Disentangling Competition from Alternative Explanations Sectoral Decomposition. As a first attempt to disentangle competition from selectivity and imitation, we analyze whether the crowding-in of Indian projects occurs in the same sectors or across sectors. Imitation should be visible as co-location of projects within the same sectors because foreign financiers often design and implement development projects in unfamiliar settings and with limited access to information, which gives them an incentive to follow cues from other donors and creditors with more local experience and tacit knowledge. Indeed, Davies and Klasen (2019, 244) note that a lack of Table 4. Sensitivity Analysis for Table 3. Projects in Response to No controls (log) $ amounts (log) Count Chinese OOF Chinese ODA Top recipients (1) (2) (3) (4) (5) (6) ChinaOF ijt1 0.002 (0.008) 0.010 (0.007) 0.006 (0.006) 0.011 (0.013) 0.000 (0.011) 0.002 (0.011) Controls 33333 Countryprovince FE 33 33 3 3 Year FE 33 33 3 3 Country-year FE 33 33 3 3 Observations 19,568 18,657 18,657 18,657 18,657 10,065 Adjusted R-squared 0.599 0.640 0.753 0.607 0.607 0.591 Notes. The unit of observation is the province (ADM1 region). Dependent and explanatory variables are binary with 1 indicating if at least one Indian MEA (respectively Chinese) project is committed to a province in columns 1 and 6, in logged US$ amounts in column 2, and in logged project counts in column 3, respectively. Columns 4 and 5 document how India’s MEA aid commitments react toward Chinese OOF and ODA projects, respectively. Column 6 includes the quartile of countries that receive most finance from MEA. Except for column 1, all specifications control for the presence of World Bank IDA projects (t-1), log nighttime light (t-1), log precipitation (t-1), log population (t-1), and log conflict-related deaths (t-1). Robust standard errors clustered at the country level are presented in parentheses. Significant at: ***: p< .01; **: p< .05; *: p< .1. Asmus-Bluhm et al. 421 information about the distribution of local needs and opportunities can prompt governments to “base their expectations in part on the choices made by other governments, leading to herding whereby one donor’s aid follows that of others due to the presumed information [that] their donations convey.”Likewise, if the co-location of projects was the mere outcome of selectivity, i.e., both donors following the same allocation criteria, one would expect co-location to occur in the same sectors. For example, a natural disaster would typically lead to Indian and Chinese aid in the same sector of humanitarian assistance if imitation or selectivity was at play. Competition, on the other hand, is equally likely to occur within or across sectors. If India wants to counteract the influence of China, it can seek to differentiate itself by pursuing projects in sectors where it has a comparative advantage vis-` a-vis China, or it can seek to design and implement projects in the same sectors but in more efficient, effective, or sustainable ways. Therefore, we will seek to determine if the relationship between the receipt of Chinese government financing and Indian government finance is primarily driven by within-sector co-location because this would provide strong grounds to question our interpretation of the results as competition. To test this, we regress Indian projects in a specific sector on Chinese projects in the same sector. Specifically, we look into the three broad development finance sectors defined by the OECD—Social Infrastructure & Services, Economic Infrastructure & Services, and Production Sectors—as well as the three largest narrow sectors in our Indian development finance dataset (in terms of project numbers), which are Energy Generation & Supply, Health, and Transport & Storage. Panel A of Table 5 documents coefficients for Exim Bank loans and panel B reports results for MEA aid. We find no evidence of India being more likely to provide an Exim Bank loan or MEA aid project to the same region and to the same sector as China in the previous year. All same-sector effects are smaller than the significant positive aggregate effects for Exim Bank loans in Table 1 (0.008). Overall, the co-location of Indian projects does not occur within the same sector. Our results are thus driven by projects committed in the same province but in different sectors. It appears unlikely that the crowding-in effect is mainly driven by imitation of specific activities or both donors following the same allocation criteria. Instead, this evidence is in line with our competition interpretation. However, with the sectoral decomposition, we cannot fully rule out that selectivity and imitation drive the positive association between India’s and China’s loan allocation. This is why we proceed with more direct tests of our competition interpretation. Public Opinion. As a more direct test of our competition interpretation of the crowdingin effect, we analyze whether India is particularly responsive to new Chinese development projects in jurisdictions where the gap between public opinion towards India and public opinion vis-` a-vis China is large. If Delhi is using its international development finance program to constrain or challenge Beijing’s growing influence, one might expect it to focus on any jurisdiction where China has established a local presence and could make a competitive gain at its expense. 15 However, given that 422 Journal of Conflict Resolution 69(2-3) Indian policymakers have scarce resources and must make risk-adjusted reward calculations, we expect Delhi to practice strategic entry deterrence, i.e., to increase spending in jurisdictions where it has a public opinion advantage over Beijing but its competitive (incumbent) advantage is under threat. One way to proxy for the relative levels of influence enjoyed by two (potentially competing) donor governments is public opinion about these donor governments in recipient countries. Most donor governments have policies and programs in place to win the “hearts and minds”of citizens in host countries (e.g., Blair, Marty, and Roessler 2022;Dietrich, Mahmud, and Winters 2018;Eichenauer, Fuchs, and Brückner 2021; Wellner et al., 2024). They understand that public perceptions can “filter up and influence elite policy to be more amenable to [their own] interests,”and that their strategic rivals are seeking to create a more favorable public opinion environment to promote their interests (Brazys and Dukalskis 2019, 567). Therefore, if one can consistently Table 5. Sectoral Decomposition. Social Economic Production Energy Health Transport (1) (2) (3) (4) (5) (6) Panel A: Indian Exim Bank loans ChinaOF ijt1 0.000 (0.000) 0.005 (0.004) 0.001 (0.003) 0.007 (0.007) 0.003 (0.002) 0.002 (0.005) Panel B: Indian MEA aid ChinaOF ijt1 0.003 (0.004) 0.010 (0.005)** 0.008 (0.005) 0.001 (0.004) 0.000 (0.005) 0.005 (0.003)* Controls 33 3333 Countryprovince FE 33 3333 Year FE 33 3333 Country-year FE 33 3333 Observations (A) 18,657 18,657 18,657 18,657 18,657 18,657 Observations (B) 18,657 18,657 18,657 18,657 18,657 18,657 Adjusted R-squared (A) 0.226 0.136 0.147 0.161 0.733 0.102 Adjusted R-squared (B) 0.495 0.565 0.171 0.633 0.585 0.797 Notes. The unit of observation is the province (ADM1 region). Dependent and explanatory variables are binary with 1 indicating if at least one Indian (respectively Chinese) project is committed to a province. The column labels indicate the sector within which the project has been allocated. Panel A reports estimates for Exim Bank loans, panel B reports estimates for MEA aid. All specifications control for the presence of World Bank IBRD (panel A) or IDA (panel B) projects (t-1), log nighttime light (t-1), log precipitation (t-1), log population (t-1), and log conflict-related deaths (t-1). All columns include country-year-fixed effects in addition to country-provinceand year-fixed effects. Robust standard errors clustered at the country level are presented in parentheses. Significant at: ***: p< .01; **: p< .05; *: p< .1. Asmus-Bluhm et al. 423 measure levels of public approval for two governments over space and time, one can effectively proxy for the relative gains and losses that one government is experiencing vis-` a-vis another government in specific jurisdictions. The Gallup World Poll provides such data (Gallup 2018). Each year, the survey is conducted in more than 160 countries worldwide. Gallup interviews at least 1,000 individuals in each country and weights them in a manner that ensures the final survey results are nationally representative. We use the Gallup World Poll question “Do you approve or disapprove of the job performance of the leadership of [country]?”, where [country] is either China or India. 16 This allows us to generate a measure of the distance in public approval rates between the two countries. More specifically, we augment our regression equation with an interaction between ChinaOF ijt1 and the difference between the approval rates of the Indian government and the Chinese government in the recipient country. In column 1 of Table 6,we compute the approval rates only for those respondents who provided an answer to the question. Column 2 includes respondents that refused to answer or replied with “don’t know”and treats these observations as absence of approval. According to both columns, the coefficients on the interaction term are positive and statistically significant at conventional levels. This implies that the increase in the probability of a new Indian Exim Bank loan in response to new Chinese development projects is more pronounced when popular opinion in the recipient country is relatively more favorable about India than about China. 17 As the insignificant coefficients on India’s and China’s approval and disapproval rates in Appendix Table E-4 show, this finding is driven by the difference in public sentiment towards India and China rather than by the absolute levels of public support for India in these countries. It also holds when we control for nationalist sentiment. 18 These interaction results are difficult to reconcile with any explanation other than donor competition between the two Asian powers. Replicating the analysis for the Indian MEA, we again find no evidence of a crowding-in effect after the commitment of new Chinese projects—not even in areas that hold a more positive view of India relative to China (columns 3 and 4 of Table 6). Commercial and Geopolitical Interests. Improved public opinion ultimately serves the goal of advancing a donor country’s interests. As a final test of our competition interpretation, we directly test whether commercial and geopolitical interests are associated with the crowding-in effect. Specifically, we consider commercial and geopolitical factors that have been suggested as “fueling”or intensifying competition. If competition is indeed the driver of the crowding-in effect, the effect should be more pronounced in the countries that matter most to Delhi. Conversely, it is unlikely that these factors matter if the crowding-in effect is driven by imitation or selectivity. Empirically, we separately add an interaction of one of two ‘competition-intensifying’ variables with our variable of interest to our baseline specification. First, with respect to commercial competition, we expect that India will be more sensitive to the receipt of Chinese aid and credit in countries to which China and India export similar goods. To test this expectation, we calculate an export similarity index 424 Journal of Conflict Resolution 69(2-3) Table 6. Interactions With Public Opinion, Commercial, and Geopolitical Variables. Public Opinion Commerce and Geopolitics Exim Bank Loans MEA Aid Exim Bank Loans MEA Aid (1) (2) (3) (4) (5) (6) (7) (8) ChinaOF ijt1 0.014 (0.008)* 0.013 (0.008) 0.017 (0.023) 0.020 (0.021) 0.021 (0.011)* 0.007 (0.004)* 0.016 (0.029) 0.000 (0.008) ChinaOF ijt1 × Approval distance 1 0.091 (0.044)** 0.151 (0.117) ChinaOF ijt1 × Approval distance 2 0.091 (0.053)* 0.191 (0.117) ChinaOF ijt1 × ESI 0.073 (0.033)** 0.046 (0.076) ChinaOF ijt1 × BIMSTEC 0.017 (0.020) 0.047 (0.022)** Controls 33333 333 Country FE 33333 333 Province FE 33333 333 Year FE 33333 333 Country-year FE 33333 333 Observations 5,560 5,560 5,560 5,560 18,577 18,657 18,577 18,657 Adjusted R-squared 0.238 0.238 0.613 0.613 0.266 0.266 0.610 0.608 Notes. The unit of observation is the province (ADM1 region). Dependent and explanatory variables are binary with 1 indicating if at least one Indian (respectively Chinese) project is committed to a province. Columns 1–4 explore interactions with the Approval Distance 1 or Approval Distance 2, respectively. The variable subtracts the approval rate for the Chinese government from the approval rate of the Indian government. Columns 1 and 2 report estimates for Exim Bank loans; columns 3 and 4 report estimates for MEA aid. Columns 5–8 explore interactions with the export similarity index (ESI) and an indicator for countries being part of the geopolitical alliance Bay of Bengal Initiative for Multi-Sectoral Technical and Economic Cooperation (BIMSTEC). Columns 5 and 6 report estimates for Exim Bank loans; columns 7 and 8 report estimates for MEA aid. All specifications control for the presence of World Bank IBRD (Exim Bank loans) or IDA (MEA aid) projects (t-1), log nighttime light (t-1), log precipitation (t-1), log population (t-1), and log conflict-related deaths (t-1). All columns include country-year-fixed effects in addition to country-provinceand year-fixed effects. Robust standard errors clustered at the country level are presented in parentheses. Significant at: ***: p< .01; **: p< .05; *: p< .1. Asmus-Bluhm et al. 425 (ESI). We follow the seminal contribution of Finger and Kreinin (1979) and proxy the similarity of the sectoral export structure between India and China in a given country as Pn s¼1MinðXIndia,j s;XChina,j sÞ, where Xrepresents the respective donor country’s exports to recipient country jin the product division sas a share of the donor’s total exports to recipient country j. 19 To mitigate concerns about reverse causality, we use trade values from 2007, i.e., the first year of our estimation sample. The resulting index ranges from zero to one, with higher values indicating more similar export structures. In our sample, India and China on average have the most similar export structure in Gambia (0.65), followed by Zambia (0.64), and the index indicates least room for commercial competition in Eritrea (0.03), followed by Somalia (0.09). Second, concerning geopolitical competition, we expect that the Delhi-based MEA is more responsive to Chinese activities in its neighborhood, as India’s geostrategic stakes are much higher in South Asia than elsewhere. If the effects in India’s neighborhood are stronger, this would support our competition interpretation of the crowding-in effect. We interact our variable of interest with a binary variable that takes a value of one if the recipient country is part of the multilateral organization of the region, Bay of Bengal Initiative for Multi-Sectoral Technical and Economic Cooperation (BIMSTEC). 20 We prefer this variable over a simple geography-based neighborhood dummy because of the long-standing conflict between India and neighboring Pakistan. As Shrivastava (2005, 973) notes, “[f]or India, membership of BIMSTEC implies closer ties with its eastern neighbours, offsetting the influence of China in the region, sidelining Pakistan, access to [the Association of Southeast Asian Nations], security, economic prosperity due to [the free trade agreement] and clout in regional and international affairs”. Table 6 presents the results. 21 As expected, we find that India’s Exim Bank is more likely to co-locate a project in response to a Chinese project in a given province if India and China have a similar export structure in the respective country (column 5). In quantitative terms, the likelihood of an Exim Bank response is 7.3 percentage points larger in a country where India and China have the same export structure compared to a country where India’s and China’s exports show no overlap. The likelihood of an Exim Bank loan increases in response to Chinese activities in a given province if the export similarity exceeds 0.4 (see also graphical visualization in Appendix Figure E-6). 22 This is further evidence for commercial competition of India with China. 23 By contrast, we find no evidence that India’s Exim Bank is more responsive to China’s engagement in its neighborhood than elsewhere (column 6). We repeat the analysis for India’s MEA, which is, as we explained and tested above, guided to a larger extent by geopolitical motives than the country’s Exim Bank. Column 7 shows that India’s MEA is not more (or less) likely to respond to a new Chinese project if it is located in a recipient country where both emerging economies have a similar export pattern. This is not surprising given that we have seen above that India’s MEA primarily follows geopolitical interests. By contrast, Chinese development projects lead to a crowding-in of Indian MEA aid in India’s neighborhood, the 426 Journal of Conflict Resolution 69(2-3) BIMSTEC countries (column 8). In quantitative terms, a new Chinese development project in the previous year increases the likelihood of an Indian MEA aid project commitment by 4.7 percentage points if the recipient country is part of this group of strategic importance to India. The effect is sizeable in light of the average likelihood of a new MEA aid project being set up in a province in a given year (3.7 percent). To sum up, while development aid provided by India’s MEA does not follow China’s development activities in the average recipient country, the MEA allocates its development aid to compete with China where it arguably matters most: in India’s neighborhood. Further Checks. A critical reader might be concerned that the observed co-location of Indian Exim Bank projects relative to Chinese projects is driven by a boost in local demand generated by Chinese projects rather than competition. 24 While this is unlikely to be the main driver of co-location in light of the above-documented links with India’s relative popularity and India-China export similarity, we cannot rule out such an effect. To test whether our observed pattern might indeed be driven by an economic boom created by Chinese projects, we repeat our baseline regression and add an interaction of our variable of interest IndiaOF with a 1-year lead of nighttime lights to proxy for local growth. If an economic boom was driving our results, we should observe a positive significant coefficient on the interaction term. Since this is not the case (see Appendix Table E-6), this further adds to our confidence in the competition explanation of the colocation of projects. Next, we estimate a spatial Durbin model (see Appendix E-5) to account for spatial dependence. More specifically, we allow for the possibility that a new Chinese project in neighboring provinces in the previous year affects the allocation of Indian projects. Reassuringly, our main finding is robust to accounting for potential spatial dependence. The insignificant coefficients on Wx suggest that there are no neighborhood effects of Chinese projects on India’s project allocation. In other words, there is no evidence that a new Chinese project in the neighboring provinces attracts a new Indian Exim Bank loan or, alternatively, a MEA project. Finally, competition between actors can be one-sided with only one party reacting to the other but it might also be reciprocal. In a final step, we, therefore, seek to evaluate whether China steps up its development activities when India launches new projects in countries or specific provinces. The non-findings reported in Appendix F suggest that while India competes with China through development finance, the opposite does not appear to be the case. This is in line with the characterization of Cheru and Obi (2011, 91) that India’s strategy vis-` a-vis China is one of “playing ‘catch up.’” Concluding Remarks Our regression results based on a joint sample covering 2,333 provinces within 123 countries confirm that India’s Exim Bank is more likely to allocate a creditfinanced project to a subnational locality if the Chinese government provided financing Asmus-Bluhm et al. 427 to the province in the previous year. We also observe co-location effects at the national level, albeit to a lesser extent. Since our effect is more pronounced in countries where India is more popular relative to China and where both lenders have a similar export structure, we interpret this as evidence of India competing with China. It is thus unlikely that India allocates projects (more slowly) to the same province as China simply because it follows the same allocation criteria as China. By contrast, development aid provided by India’s MEA does not follow China’s development activities in the average recipient country. We only find that the MEA allocates its development aid to compete with China where it arguably matters most: in India’s neighborhood. Since we find robust evidence of competition only for India’s Exim Bank, which primarily follows commercial objectives, we conclude that India is engaging primarily in commercial rather than geopolitical competition with China. Nevertheless, our finding about India using aid to compete with China in its neighborhood is also of high relevance since 36 percent of Indian aid remains in its own world region. Finally, analyzing China’s possible response to Indian development activities, we find no evidence that China competes with India in the same localities. At first sight, rivalry does not need to be detrimental from a development perspective. Rivalries can lead to more aid activities and this is to be welcomed if aid is effective. 25 They can also create policy space for developing countries, allowing them to choose the most competent partner country. Competition can also be beneficial if it leads donors to strive for the best development solutions and more effective projects and programs. However, research has shown that strategically motivated aid is less effective according to country growth rates and project evaluations (Dreher et al. 2013, 2018;Kilby and Dreher 2010). There are thus reasons to be concerned about adverse effects of the rivalries documented in this article. Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work would not have been possible without the invaluable support of reviewers, discussants and research assistants (see Appendix A for a full list). Gerda Asmus-Bluhm, Vera Eichenauer, and Andreas Fuchs are grateful for generous support from the German Research Foundation (DFG) in the framework of the projects “The Economics of Emerging Donors in Development Cooperation”(DR 640/5-1 and FU 997/1-1) and “Empirical Analyses of Emerging Donors in Development Cooperation”(DR 640/5-3 and FU 997/1-3). ORCID iDs Vera Z. Eichenauer https://orcid.org/0000-0002-0440-3136 Andreas Fuchs https://orcid.org/0000-0002-6309-5954 428 Journal of Conflict Resolution 69(2-3) Data Availability Statement The data that support the findings of this study are available in the supporting information of this article. We make the Indian Development Finance Dataset, which has been introduced in this article, publicly availabable at https://indiandevelopmentfinance.net. Supplemental Material Supplemental material for this article is available online. Notes 1. We introduce the Indian official financing data in the Empirical Strategy section. Data on Chinese official financing are from Dreher et al. (2021a),Dreher et al. (2022), and Bluhm et al. (2020). Dollar values are in constant 2014 US dollars. For comparison, the United States provided US$242 billion in official development assistance over the same time period (OECD 2020). 2. For example, Ramachandran (2015) notes that “Indian aid intensified in 2007 in response to China’s mounting interest in the Maldives.”Likewise, Bhogal (2016) asserts that “[i]n order to check China’s growing footprints in South Asia, India has expedited its own plans to establish links with Chabahar port in Iran via Afghanistan.” 3. Authors’interviews were carried out with Indian ministries, public institutions, think tanks, and non-governmental organizations on September 8–9, 2014 in Delhi. 4. Appendix B outlines our argument in detail. It provides an overview of the motivations that might guide the allocation of Indian and Chinese official finance and the conditions under which one would expect these emerging powers to compete with each other through the use of government financing instruments. 5. The dataset is publicly available at https://www.indiandevelopmentfinance.net. 6. In our study, we define ‘province’as the first subnational administrative (ADM1) region according to the GADM database (version 2.8). 7. During our own interviews with decision makers in Delhi (September 8-9, 2014), an expert on Indian aid noted that “India is much slower than China”when it comes to planning and implementation. The expert referred to the Indian-financed Afghani parliament building as a case in point. The project was initiated by the Indian government in 2007 and inaugurated in 2015. 8. We provide details on the variables used in Appendix C. 9. These fixed effects also account for a Chinese or Indian development presence prior to 2007, the beginning of our sample period. We, therefore, do not worry that a “stock”effect biases our results. 10. We acknowledge that we cannot control for unobserved province-specific variables that change over time. These could include province-year-specific need factors such as natural disasters that affect only parts of the country under analysis, events that increase a province’s international importance in a given year (e.g., international summit, trade fair), and the timevarying domestic political relevance of a province (for example driven by provincial or municipal elections). This prevents us from interpreting our estimates as causal estimates. Asmus-Bluhm et al. 429