Paving Their Own Road? Local Chinese and World Bank Aid and Foreign Direct Investment in Africa

Abstract

China’s emergence as a major donor and economic partner to sub-Saharan Africa has prompted questions if the aims and impacts of China’s efforts in the region are driven by self-serving commercial motives. While there are strong reasons to think that the foreign aid may make a location more attractive to investment, generally, by improving the infrastructural, institutional, or human capital environment and/or by serving as a signal of a location’s risk, there is also reason to suspect that Chinese aid may be “paving the way” for investment by Chinese firms. To investigate, this paper uses spatial and temporal variations in localized, geo-referenced data to find a strong overall support that local aid boosts local foreign direct investment (FDI). However, we also find some evidence that Chinese aid boosts its own FDI compared to FDI from third-party countries; but this differential effect is also visible with the Chinese FDI’s response to the World Bank aid as well as in the relationship between both aid and FDI from the USA, suggesting that the Chinese FDI may simply be following any aid and that the relationship between its aid and FDI is not exceptional among bilateral donors.

Introduction

Since the turn of the century, Chinese involvement in sub-Saharan Africa (SSA) has expanded markedly and the country is now one of the leading sources of foreign direct investment (FDI). However, the aims of China in the region, and the benefits to the countries therein, remain unclear. While many hail the potential of China as a “new” development partner that can bring fresh ideas and resources to countries that have struggled to develop, others are more sceptical and think that China’s efforts are mainly directed by China’s foreign strategic and economic goals, which have become significantly more ambitious under the premiership of Xi Jinping.¹ Overall, the empirical evidence is nuanced. While some work suggests that Chinese involvement is effective in boosting growth,² other work suggests that Chinese development efforts may undermine local governance,³ hamper traditional donors’ conditionality efforts,⁴ or reduce local support for both the Chinese development model and domestic political leaders.⁵ In this paper, we focus on the relationship between China’s overseas development assistance and Chinese FDI. Understanding the extent to which Chinese development assistance serves its own commercial interests, compared to the broader development needs of recipient countries, is crucial in answering larger questions about the way in which China is positioning itself as an emerging superpower.⁶ In recent research, China’s increasing influence is illustrated through its creation of the Asian Infrastructure Investment Bank and the World Bank’s adoption of a similar focus on infrastructure-intensive lending.⁷

At a general level, FDI is crucial to those countries whose levels of income and domestic savings are low given that FDI can spur economic growth and development by increasing firm productivity and capability and integrating domestic firms into global markets. Thus, FDI is a very important source of external financing to countries in the developing world. Yet, multinational firms perceive SSA as being inherently risky for business due to its political instability and lack of credibility of reform.⁸ Although it is difficult for SSA countries to attract much FDI despite their efforts,⁹ the source, magnitude, and timing of FDI in SSA vary greatly.

Meanwhile, SSA is also host to a vast quantity of official development assistance (ODA), which also varies in terms of time and location and is ostensibly purposed, among other aims, with improving the investment environment. Again, at a general level, this variation brings up two important questions: (1) To what extent can the siting of ODA influence the firms’ decisions to invest in SSA, especially given high levels of political risk? and (2) is aid from specific donors used to facilitate FDI flows from their own firms? To answer these questions, we develop a theory of the political economy of foreign aid and investment, focusing on subnational locations of aid and FDI using the geo-referenced data. Our theoretical framework suggests that foreign aid, from any source and especially at the local level, helps to explain patterns of FDI inflows into SSA. We argue that the aid provides investors with new information about investment in a given region. Aid boosts economic infrastructure, increasing the expected productivity and, consequently, anticipated profits of investments. Furthermore, aid signals political or economic stability for a given region, which lowers the investment risk. Thus, foreign aid induces FDI in the same region.

Turning to the relationship between bilateral ODA and FDI from the same-source country, our theoretical framework suggests that donors may provide an informational advantage to their home firms and/or help smooth other transaction costs when investing abroad. These relationships may be especially prevalent between the Chinese government and Chinese firms, be they public or privately owned. We argue that Chinese FDI is more likely to be directed by strategic state policy planning than FDI from other source countries because Chinese firms are likely to coordinate location decisions, either indirectly or directly, with the Chinese government. As a result, we expect that Chinese aid will have a more pronounced effect on Chinese FDI than FDI from other source countries. As a foil, we compare the impact of aid from the World Bank on FDI from several source countries/regions. We make this comparison since, in the absence of a single principal, World Bank aid should be less likely to have differential effects on FDI from different source countries.

To evaluate these claims, our paper combines several geo-coded foreign aid datasets in Africa from the AidData project with several thousand geo-referenced FDI project locations from the Financial Times fDi Markets database. The geo-spatial revolution in development data has facilitated the rise of new research agendas which consider the political economy of subnational and local aid allocation and impact. Recent studies have used geo-referenced data to consider questions of aid’s local impact on topics including growth, welfare, the environment, and governance.¹⁰ Other work has considered the political motivations behind sub-national allocation of aid.¹¹ Sub-national investigations allow both for more fine-grained empirical analysis and for an evaluation of subnational discrepancies or inequalities. In our case, these geo-referenced data allow for the estimation strategies that leverage information on project timing and location to evaluate an aid treatment effect. Accordingly, this paper uses a difference-in-difference-like identification strategy that compares location-years with active aid projects to those locations that do not yet have an active aid project, but subsequently will and those sites’ local neighbours who never receive aid.

To the best of our knowledge, this is the first subnational analysis of the relationship between aid and FDI in SSA over time. The results suggest that, in general, aid is substantially effective in attracting local FDI. Location-years with active aid projects are more likely to receive any FDI project compared to location-years without current aid projects, but who will have aid projects in the future, and neighbouring locations with no aid projects at any time. When looking at aid and FDI from individual-source actors, the results are more nuanced. Chinese aid attracts FDI from all sources, but the increase compared to the baseline is noticeably larger for Chinese FDI compared to other sources. However, this result is not markedly different from active aid from the World Bank which also attracts FDI projects from all sources, although again the impact on Chinese FDI still outstrips that on other sources, although the difference is smaller. While some of these effects may derive from the fact that China had a lower base-rate of FDI in SSA, the differential effect of Chinese aid on Chinese FDI may be suggestive of active coordination between Chinese aid and Chinese FDI efforts. However, when comparing with a more limited sample of countries, we find a similar substantive impact of USA’s aid on its own FDI, suggesting that China may not be unique in this behaviour, but instead may simply be acting in line with the traditional “self-interested” donors. However, given that siting of Chinese aid and FDI are likely in many instances to be driven by coordination stemming from the same internal policy, we present these findings as evidence of spatial–temporal correlated behavior, rather than a causal treatment effect. This finding contributes to the literature on foreign aid and FDI and the broader literatures on the heterogeneous behaviour of investors and on China’s aims and ambitions both in Africa and globally.

Aid and FDI

A substantial amount of literature has suggested that aid may facilitate FDI,¹² although others have noted that aid may also simultaneously serve to crowd out FDI.¹³ At an abstract level, foreign aid can boost economic infrastructure,¹⁴ serve in a signalling function, especially in post-conflict countries,¹⁵ or facilitate human capital and social cohesion¹⁶ which in turn attracts FDI. While most studies have suggested that the relationship between aid and FDI is positive, some have suggested this only applies to some countries¹⁷ while other work has found a negative relationship between the two.¹⁸ Although a large body of work has examined the relationship between Official Development Assistance (ODA) and FDI at the recipient country level, research on the impact of aid on FDI has not considered the subnational impacts of aid.¹⁹ Even in the literature that considers the broader determinants of FDI, there are relatively few papers that consider the politics of subnational siting of FDI.²⁰

In this paper, we first make a general argument that foreign aid will increase investment across Africa, especially at a local level, before turning to bilateral donor, and ultimately, China-specific effects. Generally, we suggest two broad mechanisms by which aid might lead to changes in investors’ perceived risks and profits of investment. The two channels through which aid may impact the likelihood of investment are by building public goods (functional effect) and by signalling information about the investment environment (signalling effect), both of which we expect to be most salient at a local level.

Infrastructure and Productivity

The functional argument asserts that aid may serve a specific economic purpose by increasing the expected productivity of new FDI. This may occur particularly with aid that is classified as “Aid for Trade” (AfT). AfT is a broad conceptualization and not only includes categories of productive infrastructure—including transportation, energy, communications, and utilities infrastructure—but also can include, often industry-specific, technical training, or research and development.²¹ This type of aid may help improve the regulatory and infrastructural business environments, increasing attractiveness to FDI.²² Donaubauer et al. show that this functional effect can also be indirect in that infrastructure aid may increase a recipient countries’ infrastructure endowment.²³ This functional argument becomes even more plausible when considering local proximity. For most physical infrastructure, it is reasonable to assume that the firms need to be sufficiently near to take advantage of the amenities in a manner that will increase productivity. New roads, electric transmission lines, or water supplies will only be attractive for firms if they can directly access that infrastructure. Even infrastructure such as airports or seaports, or international road or rail links, will likely only serve those firms within a defined geographic catchment area.

The literature has also argued that functional aid may consist of that which improves the human capital in a given area. Most notably, Donaubauer et al. argue and evidence that education aid attracts FDI in Latin America. However, the causal link in this argument again is stronger when considering local effects.²⁴ While there may be some human factor mobility, one can assume that improved education is most likely to upskill the local labour pool. Health aid may work in a similar fashion, as it has been shown that healthier workers are more productive.²⁵ Again, health projects are likely only to disproportionately impact those in geographic proximity to the project. Projects that promote health and education may then increase the productivity of the local labour pool, making that locality more attractive to foreign investors. These functional arguments are unlikely to be donor specific—aid from any bilateral or multilateral donor should induce the effects described above.

Risk and Signalling

Since all FDI are irreversible to some degree, multinationals face the “obsolescing bargain” problem,²⁶ which arises when firms lose the initial bargaining advantage after investment as the balance of bargaining power shifts to host governments. Firms are vulnerable to outright expropriation and arbitrary policy changes in host countries. As a result, it is difficult for countries to make credible commitments to their business-friendly policies and attract FDI. This is an important problem for many developing countries that depend on FDI as a major source of external financing. Aid can mitigate adverse effects such as expropriation risk and can also provide new information to potential investors about the investment climate.²⁷

Locality also strengthens the signalling argument of aid attracting foreign investors. Garriga and Phillips examine aid as a signal in post-conflict situations, suggesting that aid can serve a role in low-information environments, signalling to investors that a country has sufficient political or economic stability for investment.²⁸ As many conflicts are subnational, aid may signal to investors that a region previously affected by conflict or instability is now politically and economically safe for investment.

Donaubauer et al. also explain how aid may serve as a signal to investors to anticipate an increase in (domestic) (economic) infrastructure.²⁹ Thus, the aid itself may not (only) be doing the heavy lifting in increasing productivity but may also send a signal on the infrastructure priorities and focus of a host government. Again, this signal is likely to be local. If, for instance, an aid project supports the construction of a bridge then it may be a reasonable assumption that the host government will also invest in upgrading the local thoroughfares and access roads that utilize that bridge. States often invest heavily in promoting infrastructure in specific areas, most notably by fostering special economic or export zones (SEZs).³⁰ Aid may be a leading indicator of further state investment in these areas. Empirically, however, China’s development efforts in Africa appear to be the only ones that have been explicitly focused around these types of special economic areas.³¹ Importantly, some authors note that some types of aid, such as assistance from International Monetary Fund (IMF) programs, may instead act as a critical signal, sending information about the ongoing or impending crisis.³² However, with this latter caveat in mind, the aforementioned discussion generally suggests a positive relationship between aid and FDI. Once again, the signalling mechanisms described earlier should be germane for any bilateral or multilateral donor. Accordingly, our first hypothesis is: 

Hypothesis 1: Local aid will increase local foreign direct investment.

Bilateral Aid and FDI

While the cross-country literature is broadly supportive of this positive link between aid and FDI, the existing literature is relatively agnostic regarding source-country heterogeneity in the aid–FDI relationship.³³ However, there is substantial reason to suspect country heterogeneity in the political economy of aid allocation. Considerations of foreign economic policy motivation date to at least McKinlay and Little³⁴ and McKinlay,³⁵ sustaining a prolonged debate on whether foreign aid is given to suit “donors’ interests”, “recipients’ needs”, or both.³⁶ More recently, Bermeo has argued that aid may be “targeted” for development purposes in countries most likely to engender spillovers to the donor.³⁷ Extending the logic of this literature, it is eminently plausible to think that aid from some donors may be used to facilitate FDI from that country in a self-interested, strategic, and fashion. Indeed, the use of aid to “pave the way” for subsequent FDI from the same originating country would be highly suggestive of an imperialist critique that has undermined discourses such as the “new scramble for Africa.”³⁸

These bilateral-specific effects work primarily through the information/signalling channels. Aid from a particular host country signals a country’s broader engagement with a partner country. As described at greater lengths by Morgan and Zheng, these may induce donor-specific FDI in two ways.³⁹ First, a donor-aid presence can reduce the information costs for that donor’s firms. This may occur via the creation of what they deem “social capital.” This can be direct—wherein individuals formerly involved with aid projects may impart knowledge of a locality directly to the (potentially) investing firm—or it can be indirect—wherein the legacy of the aid project leaves behind social capital that can be utilized to create an investment.

Second, Morgan and Zheng argue how bilateral aid can reduce institutional risk for investors.⁴⁰ Specifically, they argue that aid relationships can give donors the ability to influence the host government’s receipt of FDI from the bilateral donor. The bilateral donors may be able to use aid—either via its good will or through direct leverage—as a means by which to induce favourable host policy towards FDI from their country. Bilateral donors with strong ties in a host country can also reduce ex post risks (like expropriation), either via softer policy influence or through a direct leveraging of aid to protect the investments of firms from their countries. New or increased aid amounts might also suggest that the future economic relations between the countries may intensify, say through a closer trading relationship, which could positively impact the return on the investment. While these institutional risk/reward arguments are not necessarily location specific, when combined with the informational/signalling arguments they provide a strong rationale for thinking that FDI from a given country may be more likely to follow aid from that same country at a local level.

Thus, while aid facilitates the provision of public goods which are not completely excludable, we think that bilateral aid may provide a disproportionate boost to FDI from the same donor country via informational advantages on the institutional quality, market opportunities, and risk in a locality. This information is both private and costly, and non-donor firms will find it more difficult to obtain. Moreover, bilateral aid efforts may be directly coordinated with firms via internal state coordination, and thus giving those firms a “first mover” advantage in locating FDI to an area. Firms from the donor country can begin planning investment decisions alongside aid planning decisions which may not be immediately known to non-donor firms. Notably, what this implies is that the local siting of aid and FDI may not be exogenous, but instead is the result of this coordination. Thus, our empirical tests and results below should be viewed as testing for evidence of this coordination mechanism, rather than something more akin to an “average treatment effect” of aid.

Paving the Road “With Chinese Characteristics”?

China has substantially increased its global net outflows of FDI from $4.6 billion in 2000 to a zenith of $216.4 billion in 2016,⁴¹ with the vast bulk of that directed to the developing world.⁴² However, as noted by Carmody and Wainwright,⁴³ there has been a conspicuous fall in investment to the continent as it appears the Belt and Road Initiative (BRI) has scaled back significantly since 2016. In Africa, China has become one of the leading sources of FDI since China’s Going Global Strategy along with the Forum on China–Africa Cooperation (FOCAC) in 2000, with an FDI stock of nearly $50 billion by 2019. In the past years, China has overtaken the USA as the largest foreign direct investor in Africa.⁴⁴ Much literature on Chinese FDI explores why Chinese firms choose to invest in Africa despite its high political risk. Indeed, work examining Chinese outward FDI suggests that it is more likely to be directed to weak governance states than FDI from other sources.⁴⁵ Scholars have suggested the imperfect property and capital market,⁴⁶ coordination between state and business,⁴⁷ or provincial features of China⁴⁸ as key determinants of Chinese FDI inflows into such risky markets. Other studies have suggested that outward FDI stems from strong interest in securing access to natural resources, and that this has also been a key determinant of Chinese aid to developing countries,⁴⁹ though the results have been mixed.⁵⁰

While the arguments in the earlier section may apply to any “self-interested” bilateral donor, there is reason to suspect these arguments might be particularly salient when considering the Chinese aid and the Chinese FDI. Indeed, at the country level there is some evidence that Chinese aid flows increase Chinese FDI flows to the same country.⁵¹ As a state-directed economy, Chinese firms operating abroad are tightly regulated by Beijing, even when they are in private ownership, as all outward investment is ostensibly registered with the Chinese Ministry of Commerce (MOFCOM).⁵² As such, these firms are generally either directly controlled (as SOEs) or indirectly influenced (private enterprises) by the state.⁵³

Due to close coordination between state and business, as well as direct control by the state, it seems likely that Chinese aid could be used as a handmaiden to Chinese FDI. Indeed, aid and FDI were explicitly linked (along with trade) as part of the PRC’s “going out” strategy which has largely directed the PRC’s economic statecraft since the 1990s and has since been rebranded as the BRI.⁵⁴ Morgan and Zheng’s historical study of Chinese aid and contemporary investment, in particular, specifically highlights how Chinese aid built “social capital”, or familiarity and knowledge of an area, which was then used to inform commercial investment in the same area.⁵⁵ This coordination between aid and commercial efforts is likely to be especially salient given the PRC’s economic statecraft grand strategies. Thus, Chinese FDI, especially in developing countries, is heavily influenced by national and/or provincial governments.⁵⁶ Although some research has found that the PRC may cede control of subnational siting of aid to local political elites, this doesn’t necessarily undermine the arguments we’ve made earlier.⁵⁷ Instead, what it does suggest is that, if there is a relationship between Chinese aid and FDI, it is FDI following aid, rather than aid going to locations that have been a priori identified as attractive for (Chinese) FDI.

In contrast to aid from bilateral donors, aid from a multilateral donor with no single principal, like the World Bank, should provide no such informational or coordination advantage to FDI from any particular source country. World Bank project planning is transparent and likely to be accessible to all parties equally, while any World Bank knowledge on location-specific institutional quality, market opportunities, or risk is also unlikely to be private to FDI from any particular source. Moreover, the World Bank is unlikely to use its donor leverage to influence host-country FDI policy to be favourable, ex ante or ex post, to FDI from any specific source country. Accordingly, while we expect local aid to boost local FDI generally, we expect that bilateral aid will disproportionally boost bilateral FDI but World Bank aid will bestow no country-specific effect. We expect the bilateral effect to be particularly salient when considering Chinese aid and FDI given the close relationship between states and firms in that country. As such, our second and third hypotheses states that:  

Hypothesis 2: Bilateral (Chinese) Aid will disproportionally increase Bilateral (Chinese) FDI compared to FDI from other sources.

  

Hypothesis 3: World Bank aid will not lead to a disproportionally relative increase in FDI from any particular source country vis-à-vis any other.

Data and Methods

We limit our analysis to SSA host countries to examine the unique relationship between aid and FDI in the poorest global region which is not only recognized as a risky location to conduct business but has also seen a recent uptick in Chinese investment. The outcome phenomena of interest are greenfield and expansion FDI projects drawn from the Financial Times fDI markets database, which has been used in a small but growing number of recent studies in economic and political science.⁵⁸ These data include timing information, source country, and geographic data for 9864 greenfield and expansion announced FDI projects in 56 African countries from 2003 to 2017. These data are meant to capture the population of announced FDI projects into Africa. Of these, 5953 of the original project records contain geographic destination information at the city level; 3979 of which are in the sub-Saharan region, and accordingly, the analyses used in these projects. While the data does include information on project size, both in terms of investment amount and job creation, the bulk of this is estimated and potentially biased. Accordingly, this paper relies on project events as these data are verified and cross-referenced in the original FDI Markets methodology.⁵⁹ With that caveat, the projects with “non-specified” city-level geographic information appear roughly equivalent to those with corresponding city-level information. In both instances, the proportion of greenfield projects is just over 90%, with expansion and co-location projects making up the remainder. The “non-specified” projects may be slightly larger with a median-estimated value of $16.6 million and jobs created of 86 compared to an estimated $11 million and 26 jobs created for the median project with geographic information. This may be due to the fact that the geographically “non-specified” projects, while accounting for less than 40% of the total projects, are disproportionately in sectors that tend to have larger investments like coal, oil, and natural gas (63%), Alternative/renewable energy (69%) and metals (57%). While the lack of geographic information for these projects potentially bias our results, we do not suspect that this introduces any systemic bias into our results, but future research which tracks down more detailed geographic information on these projects, or disaggregates the analysis by FDI sector, would be a useful replication of our analysis.

The main explanatory variables come from AidData’s geo-coded datasets: the World Bank geo-coded research release 1.4.2 (AidData 2017) which covers World Bank projects across the globe and similar project-level, geo-coded, and data on Chinese development efforts⁶⁰ which also has pan-African coverage. Crucially, given that China eschews traditional classifications of concessional finance and the expectation and evidence that the lines between Chinese commercial and concessional finance are blurred, we only use projects from these data that have been classified as “ODA-like.” These are projects identified by AidData as being concessional along the lines of DAC ODA classifications. In contrast, other types of Chinese “development” finance (“Other Official Flows (OOF) in the AidData parlance) includes projects that might well be consider more akin to FDI, like telecommunications projects, mines, or factories. Across SSA, we identify 17 217 unique aid locations (15 315 World Bank; 1902 China) from 1995 to 2014. Descriptive statistics can be found in the Supplementary Appendix. Figure 1 displays the spatial locations, in navy, of the World Bank (triangle) and Chinese (circle) ODA projects, while the FDI projects are displayed as squares, coloured by source country (China = Tan, USA = Magenta, Europe = Green, Africa = Red, and “Other”=Blue).⁶¹

Fig. 1.

FDI and World Bank and Chinese Aid in SSA

This paper employs a spatial-temporal identification strategy to evaluate the hypotheses. Starting with a panel of 5-min grid cells across SSA, we geo-locate both aid and FDI within a capture radius of the centroid of these cells. This gives a panel of spatial units from 2003 to 2017 that contain aid, FDI, both or neither. With these data construction, we employ a difference-in-difference-like approach similar to that used in Knutsen et al.’s study of mining and local corruption.⁶² To set-up the analysis, the paper takes advantage of the fact that both the aid and FDI project records indicate the timing of projects. The outcome measure is a new FDI project in the capture radius in a site-year. To evaluate our hypotheses, we not only evaluate FDI from all sources combined, but also identify FDI by country/region source including FDI from China, the USA, Europe (EEA and UK), Africa, and all other countries. The primary analysis relies on a linear probability model. For these models, the outcome variable is a binary indicator that equals “1” if the site had any new FDI project (from a given source) in a given year and “0” otherwise.

In terms of the treatment, three groups of sites are considered. First, some site-years are within the capture radius of active aid project, i.e. one that has an aid commitment in or prior to that year.⁶³ Second, there are site-years that capture a future aid project at the site, i.e. a project that has not yet begun in the panel-year but subsequently will—coined as “inactive” in Knutsen et al. (2017). Finally, there are sites which are outside the capture radius of any aid project (active or inactive). However, as discussed in a similar approach in Christensen,⁶⁴ rather than comparing the active and inactive sites to all other sites, it may be more appropriate to compare these sites only to their geographic neighbours. Similar approaches have been employed in recent studies, for example, Wang et al. and McCauley et al.’s study on the impact of Chinese FDI in Africa.⁶⁵ This comparison is based on the fact that location-specific unobservable factors are likely to be more similar at neighbouring sites rather than the sites far afield. Additionally, many grid cells are likely to be in locations that are theoretically unlikely to ever attract aid or FDI (deserts, mountain peaks, lakes, etc.). Accordingly, keeping those sites as comparators might not only bias the treatment effect but also artificially reduce standard errors. As such, we match sites that receive aid at any time to their eight closest neighbours (their contiguous grid-cell neighbours) and only keep these units as comparators.

Paraphrasing Knutsen et al.’s explanation,⁶⁶ interpreting the active coefficient alone in the setup above would assume that aid siting is uncorrelated with FDI placement before the aid project becomes active. However, this is a strong assumption since unobserved characteristics may make a particular site attractive to both aid and FDI. Accordingly, including inactive (future) sites allows for comparing sites before an aid project becomes active with sites after an aid project becomes active, and not only areas near and far from aid projects. As such, test results are provided for the difference between active and inactive (β_1—β₂) which gives a difference-in-difference-like measure that accounts for any latent, time-invariant, features that influence both aid and FDI siting. Thus, our sample for any given donor model includes all grid cells that host a donor’s aid project during our sample timeframe and the eight neighbouring cells of each of those “project” cells. As such, our sample size varies depending on what aid is being considered, with models with more donor project locations having a larger sample of grid cells.

In setting active site-years, we assume that once a site becomes active it stays active for the duration of our sample period.⁶⁷ This assumption is driven by two arguments. First, aid projects may take some (variable) amount of time to be implemented and completed. Second, we expect that once aid has “improved”, a location (either through the provision of new infrastructure or the improvement in the quality of human capital or institutional environment), that improvement is durable, at least for the duration of our sample period which is no more than 15 years. Accordingly, we expect that “active” aid increases the attractiveness of a site to FDI at any time after the aid has been implemented.

The spatial identification assumes that aid will attract FDI within a given cut-off distance. As discussed in Knutsen et al.,⁶⁸ one has to make an assumption about the geographic reach of the capture radius for the treatment. This is ultimately an empirical question that includes a trade-off between the precision of the geo-location in the data, noise, and the size of the treated unit. The analysis employs precision code “2” or better in the AidData, which is equivalent to “‘near’, in the ‘area’ of, or up to 25 km away from an exact location”.⁶⁹ Accordingly, we use a 25-km capture radius in our primary analyses. However, we are also cognizant of the Modifiable Area Unit Problem (MAUP).⁷⁰ The MAUP can introduce biases into the spatial analysis as arbitrary borders may introduce bias. In essence, the concern is exemplified by a situation where an outcome of interest and its explanatory factor may lie very close but on either side of an (arbitrary) border. While the phenomenon might be spatially contiguous, they would not be coded as being in the same areal unit. Accordingly, to try and mitigate this problem, in the robustness checks we run our primary models across a range of capture radii for the treatment.

We use country-year fixed effects with Conley standard errors to account for dependence based on spatial proximity.⁷¹ The country-year fixed effects capture any unobserved country-level, temporal, phenomena which might otherwise influence FDI location such as the establishment of a new trade agreement,⁷² changes in institutional or infrastructural environment,⁷³ participation in an IMF program,⁷⁴ regime change/stability/expropriation, risk/institutional, quality/electoral rules,⁷⁵ or the presence/absence of conflict.⁷⁶

The baseline reduced form specification is:

$${Y_{it}} = \,{\beta _1}*active + \,{\beta _2}*inactive + {\alpha _{ct}} + {\varepsilon _{it}}$$

where the FDI project outcome measure at site i in time t, is regressed on active and inactive. As discussed, (α_ct) are the country-year fixed effects, and |${\varepsilon _{it}}$| are the Conley standard errors. However, it is also important to note that the data are temporally truncated, as the fDi Markets data coverage only begins in 2003 and, as such, the existing stock of FDI prior to 2003 is unknown. Thus, in the robustness check below, we consider a model where countries who were “active” in the first 3 years (2003–2005) are omitted to ensure that every site in that model was initially “untreated”. We do not include any other exogenous controls in the models as almost any site-level, observable, control that we might imagine—poverty, population, nighttime light, governance quality, etc.—is likely to also be influenced by aid and incorporating these controls would introduce post-treatment bias,⁷⁷ even if we could find data at this level of granularity.

Results

To substantively evaluate our hypotheses, we transform the results from the linear probability models, available in full in Supplementary Tables A.2–A.4 in the Appendix. The difference-in-difference measure in those models is interpreted as the absolute increase in the probability that a site-year receives a FDI project from a given source. As receiving FDI is a relatively low-probability event, the magnitude of the coefficients is quite small. The probability that any site-year in the combined sample receives an FDI project is just less than 2%. However, more problematically, the baseline probability differs depending on the source of the FDI, thus the coefficients and differences are not directly comparable across models. Accordingly, a better comparison is the difference measure as a percentage increase on the mean sample probability of the “untreated” sites (both the “inactive” and the neighbour sites).⁷⁸

Accordingly, we construct a ratio measure that gives the effect size (of the difference-in-difference measure) as a percentage of the probability of an untreated or neighbour site receiving FDI from a given source. We use the estimates, their errors, and the distribution statistics from FDI at the untreated and neighbour sites for propagation of errors to construct the confidence intervals around these ratio measures.⁷⁹ The confidence intervals are generally larger for the models with comparatively fewer aid and FDI projects (China and US compared to the models which aggregate all EU, African, Other, or all FDI).

We present these results graphically in Figure 2 through 4. Before turning to the China/World Bank comparisons, we make a few general observations. First, the general hypothesis (1) that aid attracts FDI is strongly supported as the ratio measures of all models are positive and substantively large. Overall, the increase of the difference between active and inactive sites on the probability of a site-year receiving FDI is between roughly 780% the probability of inactive and neighbour sites receiving FDI. Substantively, This result holds both when considering all FDI, and FDI from specific subgroups of actors. Second, as shown in the regression tables in the Appendix, and likewise holding for all FDI and FDI from specific subgroups, there is strong evidence that there are indeed selection effects, as the positive and significant finding on the inactive coefficients in most models indicates that site-years where there was not yet received aid, but would be in later years, were also more likely to have FDI compared to neighbouring sites that would never receive aid. This finding stresses the importance of accounting for these selection effects in the “active-inactive” approach. However, the magnitude of the difference between “active” and “inactive” is large and strongly significant in all models.

Fig. 2.

Difference in Receiving FDI as Percentage of Sample Mean (Chinese and WB Aid)

Note: 90% (thick) and 95% (thin) confidence intervals.

When considering aid from both China and the World Bank (Figure 2), we see that the impact on Chinese FDI is quite substantial. The increase of the difference between active and inactive sites equals roughly 2670% of the baseline Chinese FDI probability in the sample of inactive and neighbour sites, while the increase of this difference is between roughly 770% and 820% for Europe, the USA, and “Other” FDI sources. Notably, the increase from the difference between active and inactive sites for FDI from other African countries, at roughly 160% of the inactive and neighbour sample mean, is also markedly higher than the USA, European and “other” sourced FDI, although this impact is still significantly less than the impact for Chinese-sourced FDI.

Turning to the results of World Bank and Chinese aid, individually, in Figures 3 and 4, respectively, we see that aid from each source again appears to have a disproportional impact on Chinese FDI. As shown in Figure 3, while World Bank aid has an elevated impact on Chinese (1190%) and African (985%) sourced FDI compared to USA (645%), EU (560%), and “Other” (710%) FDI. However, the absolute magnitude is smaller than in Figure 4 which shows the impact of Chinese aid. In this figure, the absolute impact on Chinese FDI (177%) outstrips that of the USA (780%), EU (650%), Africa (1035%), and “Other” (475%), and the confidence interval only interlaps with the impact on FDI from Africa. These findings are consistent with our expectation that Chinese aid has the largest impact on Chinese FDI (Hypothesis 2). However, Figure 3 shows that WB aid has a similar, and in fact larger, differential effect on Chinese FDI. This means that Hypothesis 3, that World Bank aid does not favour FDI from any particular source country, is not supported, but, instead, that Chinese FDI takes the largest advantage of World Bank aid. This latter result prompts a few considerations. First, some of the difference in the impact on Chinese FDI is likely driven by the base effects of a lower initial probability of Chinese FDI. The massive ramp up of Chinese FDI from a low initial level means that percentages effects are likely to be larger—for Chinese or World Bank aid. China may not have received any particular private information benefits from World Bank aid, but the fact that it was simply looking to invest meant that it was also happy to take advantage of “public good” area improvements driven by the World Bank. Thus, while Chinese aid does appear to disproportionally “pave the way” for Chinese FDI, it may simply be that Chinese FDI was best positioned during this period in SSA to take advantage of any aid.

Fig. 3.

Difference in Receiving FDI as Percentage of Sample Mean (WB Aid)

Note: 90% (thick) and 95% (thin) confidence intervals.

Fig. 4.

Difference in Receiving FDI as Percentage of Sample Mean (CN Aid)

Note: 90% (thick) and 95% (thin) confidence intervals.

Extensions and Robustness Checks

In this section, additional (features of the) data are utilized to extend the analysis and check on the robustness of the baseline results, the model which includes aid from both China and the World Bank and FDI from all countries. In the first robustness check, we address the MAUP by running model 1 from Supplementary Table A.2 at all treatment capture radii from 5 to 75 km, where each areal unit is determined by a radius emanating from the grid cell centroid.⁸⁰ We would expect that the results will be noisier at smaller capture radii, given the paucity of treated units, while increasing in precision at larger capture radii. The results are presented graphically below in Figure 5. As seen there, the DiD estimate changes smoothly with differently sized treatment capture areas, increasing as the capture distance increases. While one might normally expect these estimates to decay towards 0 as the capture distance increases, with the use of the active–inactive strategy, as capture distances increases, there are fewer and fewer “inactive” comparator units (as increased capture distance makes a unit more likely have an “active” aid project). At the same time, the increased capture distance also makes it more likely that “active” units will have FDI projects in their radius. However, the smoothness of the change suggests that the main results are not unduly biased by, or a random artifact of, the MAUP and our choice of a 25 km capture distance.

Fig. 5.

Difference-in-Difference, WB and China Aid, All FDI, at Different Treatment Radii

Note: 90% and 95% confidence intervals from clustered standard errors.

The full table of our further robustness checks is available in Supplementary Table A.5 in the Appendix. First, we investigate if our results are robust to using a less precise precision code from AidData. Here, we use all projects at precision code “3” or better, which is equivalent to an “Administrative 2” Unit (i.e. a county or district). We also check if our results are robust to the exclusion of any sites that were “active” in the first 3 years as we do not have FDI data prior to 2003 and, as such, we do not know the FDI behaviour at “inactive” sites prior to that year. Using a 3-year “burn in” ensures that each site was without aid for at least 3 years before being deemed “active”. Next, whereas in our main specifications we assume that once a site becomes “active” it remains “active”, we test models where sites only stay “active” for 5 years following the arrival of aid. While our primary assumption is that aid improvements are “durable”, restricting the active period is a more conservative approach that also better tests the causal temporal logic of our argument. In a further robustness check, we conduct a placebo test using the neighbouring cells where we randomly assign active (and inactive) status by site. The difference between active and inactive in these results is insignificant. We also evaluate models that includes two-way country and year fixed effects, despite recent critiques of this approach,⁸¹ and an even more conservative approach where we only compare sites that received aid, using the temporal aspect of the data to compare times before and after the site received aid. As a final check, we delay the implementation of projects by 3 years to account for the fact that we use commitment years rather than implementation years. While some committed projects may never be implemented, this check at least partially accounts for the fact that project commitments take time to translate into implemented projects. Our main conclusions remain robust to each of these checks.

However, as China may not be unique in its donor behaviour, as an extension we evaluate the relationship between aid and FDI from the USA. Unfortunately, detailed geo-referenced data for US aid is only available for seven countries in SSA, again from AidData’s geo-referencing of “Aid Information Management System” (AIMS) datasets in Burundi, Democratic Republic of the Congo, Malawi, Nigeria, Senegal, Sierra Leone, and Uganda.⁸² As shown in Supplementary Table A.6 and Supplementary Figure A.1, using this sample, we see less impact from US aid on inward FDI. While most results are positive, they are only (just) significant in that model that considers FDI from the “other” source countries. While the largest percent change impact is on FDI from the USA, this result is not significant. Interestingly, the point estimates on FDI from the EU and China are negative, although again these are not significantly different from zero. We also consider Chinese aid only in these countries in Supplementary Table A.7 and Supplementary Figure A.2. Here, once again, Chinese impact has a positive and large impact on FDI from all source countries. These impacts on Chinese FDI are again larger than that on other source countries, with the exception of the USA whose point estimate is substantially larger than all other sources, including China. However, this estimate is not statistically significant as there are comparatively few co-located Chinese aid and US FDI projects in these AIMS countries. While we hesitate to read too much into these results given the small, non-random, sample of aims countries, they are suggestive that Chinese aid may be more “effective” at building the foundations for future localized inward investment from any source, at least compared to aid from the USA. This result would mesh with other recent findings that Chinese aid is indeed effective at promoting local economic growth.⁸³

Conclusions

This paper sought to clarify both the extent to which aid projects influence where firms decide to locate their foreign direct investment and the extent to which aid donors, China, in particular, might use aid as a precursor for commercial interests from their own country. We believe that China presents a most-likely case for this kind of evaluation due to the close and documented ties between the Chinese government and its firms which are often directly state-owned or indirectly influenced by the state. While aid should mostly provide public goods, which should make a location more attractive to investment from any environment, we think it is highly likely that the Chinese state coordinates its aid efforts with their commercial interests, which may enable Chinese firms either to access private information about a location and/or give them a temporal “first mover” advantage. We compared Chinese aid with aid from the major multilateral development donor, the World Bank, whose aid should not confer any country-specific FDI benefits but, instead, build public goods which may locate more attractive to investment from all sources.

Using spatial-temporal data on both aid and FDI, we find strong evidence in support of the overall contention that local aid has a large and significant impact on a location receiving FDI in the future. This finding is robust to several different approaches to the data and estimation. When considering only aid from China, we see a noticeably disproportionate impact on Chinese FDI. That said, we also find evidence that World Bank aid disproportionately benefits Chinese FDI. While some of this impact may be attributable to base effects of lower overall levels of Chinese investment, the discrepancy is suggestive that Chinese, if not World Bank, aid is “paving the way” for Chinese investment. However, given the high likelihood that Chinese aid and FDI and driven by the same internally coordinated policies, we present these results as evidence of a spatial-temporal correlation rather than an aid “treatment effect.” When evaluating the same relationship for the USA, albeit on a smaller set of countries, we find less evidence that US aid boosts FDI from any source. Thus, while Chinese aid may give a slight advantage to FDI from its own firms, it also appears to be building local environments that foster inward investment from a range of sources. This “effectiveness” result is consistent with other recent work finding a positive impact of Chinese aid on local economic development.⁸⁴

While many papers have evaluated the aid–FDI relationship at the cross-national level, there are few that test the relationship at a local level. Thus, we believe our findings represent an important step forward. First, as elaborated earlier, the theoretical mechanisms about aid serving a signalling or functional role to attract FDI by reducing informational or production costs, respectively, are largely dependent on proximity between the flows. Evaluating aggregated, cross-country flows of aid and FDI means that aid and FDI project sites in the data could be hundreds of miles from each other, undermining the theoretical linkage. Thus, the subnational and local-level analysis improves our understanding of the efficacy of aid.

Second, evaluating local aid and FDI provides a better chance to observe the politics of aid allocation. Scholars have long debated about the extent to which aid is provided to meet the needs of recipient countries compared to serving the self-interest of the donor actors. China, in particular, has been accused of using its development assistance to further its own commercial interests in SSA. By evaluating the proximity of Chinese aid projects to FDI projects from that same actor, the paper evaluated the extent to which strategic economic aims may be influencing donor development behaviour. Indeed, many question for whose benefit China (or any donor) is active in Africa. However, our results suggest that while Chinese aid may pave the way for FDI from their own firms, it also provides a local boost for FDI from other sources and Chinese FDI also follows World Bank aid. These results may temper concerns that China’s aid is exclusively for the benefit of Chinese firms. Moreover, to the extent there is a differential impact, we qualitatively find similar results for US aid, suggesting that, at worst, the consequences and impacts of Chinese aid, at least economically, may simply mirror those of other great powers.⁸⁵

Supplementary data

Supplementary data is available at The Chinese Journal of International Politics online.

Conflict of interest statement.

None declared.

Funding

This research was supported by European Commission, Horizon 2020 Framework Programme [No. 693609], “Reconsidering European Contributions to Global Justice,” and Irish Research Council Laureate Programme Grant [No. IRCLA/2017/92] (TRADE ME).

Footnotes

1

Chris Alden and Daniel Large, eds., New Directions in Africa-China Studies (London and New York, NY: Routledge, 2019).

2

Axel Dreher, et al., “Aid, China, and Growth: Evidence from a New Global Development Finance Dataset,” American Economic Journal: Economic Policy, Vol. 13, No. 2 (2021), pp. 135–74; Huang Zhenqian and Cao Xun, “The Lure of Technocracy? Chinese Aid and Local Preferences for Development Leadership in Africa,” Foreign Policy Analysis, Vol. 19, No. 3 (2023), https://doi.org/10.1093/fpa/orad010.

3

Samuel Brazys, et al., “Bad Neighbors? How Co-Located Chinese and World Bank Development Projects Impact Local Corruption in Tanzania,” Review of International Organizations, Vol. 12, No. 2 (2017), pp. 227–53; Ann-Sofie Isaksson and Andreas Kotsadam, “Chinese Aid and Local Corruption,” Journal of Public Economics, Vol. 159 (2018), pp. 146–59; Ping Szu-Ning, et al., “The Effects of China’s Development Projects on Political Accountability,” British Journal of Political Science, Vol. 52, No. 1 (2022), pp. 65–84.

4

Diego Hernandez, “Are ‘New’ Donors Challenging World Bank Conditionality?” World Development, Vol. 96 (2017), pp. 529–49; Mitchell Watkins, “Undermining Conditionality? The Effect of Chinese Development Assistance on Compliance with World Bank Project Agreements,” Review of International Organizations, Vol. 17, No. 4 (2022), pp. 667–90.

5

Xiaonan Wang, et al., “Foreign Direct Investment, Unmet Expectations, and the Prospects of Political Leaders: Evidence from Chinese Investment in Africa,” Journal of Politics, Vol. 84, No. 3 (2022), pp. 1403–19; John F. McCauley, Margaret M. Pearson and Xiaonan Wang, “Does Chinese FDI in Africa Inspire Support for a China Model of Development?” World Development, Vol. 150, (2022), p. 105738. https://doi.org/10.1016/j.worlddev.2021.105738.

6

Robert A. Blair, Robert Marty and Philip Roessler, “Foreign Aid and Soft Power: Great Power Competition in Africa in the Early Twenty-First Century,” British Journal of Political Science, Vol. 52, No. 3 (2022), pp. 1355–76; Jue Wang and Michael Sampson, “China’s Multi-Front Institutional Strategies in International Development Finance,” Chinese Journal of International Politics, Vol. 15, No. 4 (2022), pp. 374–94.

7

Qian Jing, et al., “The Impact of China’s AIIB on the World Bank,” International Organization, Vol. 77, No. 1 (2023), pp. 217–37; Alexandra O. Zeitz, “Emulate or Differentiate? Chinese Development Finance, Competition, and World Bank Infrastructure Funding,” Review of International Organizations, Vol. 16, No. 2 (2021), pp. 265–92.

8

Elizabeth Asiedu, “Foreign Direct Investment in Africa: The Role of Natural Resources, Market Size, Government Policy, Institutions and Political Instability,” World Economy, Vol. 29, No. 1 (2006), pp. 63–77; Elizabeth Asiedu, “On the Determinants of Foreign Direct Investment to Developing Countries: Is Africa Different?” World Development, Vol. 30, No. 1 (2002), pp. 107–19.

9

John C. Anyanwu, “Why Does Foreign Direct Investment Go Where It Goes? New Evidence from African Countries,” Annals of Economics and Finance, Vol. 13, No. 2 (2012), pp. 425–62.

10

Axel Dreher and Steffen Lohmann, “Aid and Growth at the Regional Level,” Oxford Review of Economic Policy, Vol. 31, No. 3–4 (2015), pp. 420–46; Jürgen Bitzer and Erkan Gören, “Foreign Aid and Subnational Development: A Grid Cell Analysis,” AidData Working Paper #55, Williamsburg, VA: AidData at William & Mary (2018), https://www.aiddata.org/publications/foreign-aid-and-subnational-development-a-grid-cell-analysis; Bruno Martorano, et al., “Chinese Development Assistance and Household Welfare in Sub-Saharan Africa,” World Development, Vol. 129 (2020), https://doi.org/10.1016/j.worlddev.2020.104909; Robert A. Blair and Philip Roessler, “Foreign Aid and State Legitimacy: Evidence on Chinese and US Aid to Africa from Surveys, Survey Experiments, and Behavioral Games,” World Politics, Vol. 73, No. 2 (2021), pp. 315–57.

11

Ryan C. Briggs, “Does Foreign Aid Target the Poorest?” International Organization, Vol.71, No. 1 (2017), pp. 187–206.

12

Philipp Harms and Matthias Lutz, “Aid, Governance and Private Foreign Investment: Some Puzzling Findings for the 1990s,” Economic Journal, Vol.116, No. 513 (2006), pp. 773–90; Anyanwu, “Why Does Foreign Direct Investment Go Where It Goes?” pp. 425–62; Kafayat Amusa, et al., “The Nexus Between Foreign Direct Investment and Foreign Aid: An Analysis of Sub-Saharan African Countries,” African Finance Journal, Vol.18, No. 2 (2016), pp. 45–68.

13

Pablo Selaya and Eva Rytter Sunesen, “Does Foreign Aid Increase Foreign Direct Investment?” World Development, Vol. 40, No. 11 (2012), pp. 2155–76.

14

Julian Donaubauer, Birgit Meyer and Peter Nunnenkamp, “Aid, Infrastructure, and FDI: Assessing the Transmission Channel with a New Index of Infrastructure,” World Development, Vol. 78 (2016), pp. 230–45.

15

Ana Carolina Garriga and Brian J. Phillips, “Foreign Aid as a Signal to Investors: Predicting FDI in Post-Conflict Countries,” Journal of Conflict Resolution, Vol. 58, No. 2 (2014), pp. 280–306.

16

Julian Donaubauer, et al., “Does Aid for Education Attract Foreign Investors? An Empirical Analysis for Latin America,” European Journal of Development Research, Vol. 26 (2014), pp. 597–613; Emmanuel A. Cleeve, et al., “Human Capital and FDI Inflow: An Assessment of the African Case,” World Development, Vol. 74 (2015), pp. 1–14; Tony Addison and Mina Baliamoune-Lutz, “The Effects of Aid on Foreign Direct Investment in Africa and Other Regions,” Social Sciences Association (2016), pp. 1–25.

17

Hidemi Kimura and Yasuyuki Todo, “Is Foreign Aid a Vanguard of Foreign Direct Investment? A Gravity-Equation Approach,” World Development, Vol. 38, No. 4 (2010), pp. 482–97; Annageldy Arazmuradov, “Can Development Aid Help Promote Foreign Direct Investment? Evidence from Central Asia,” Economic Affairs, Vol. 35, No. 1 (2015), pp. 123–36.

18

Julian Donaubauer, “Does Foreign Aid Really Attract Foreign Investors? New Evidence from Panel Cointegration,” Applied Economics Letters, Vol. 21, No. 15 (2014). pp. 1094–8.

19

The notable exception is Blaise’s study on Japanese inflows to China. See Severine Blaise, “On the Link Between Japanese ODA and FDI in China: A Microeconomic Evaluation Using Conditional Logit Analysis,” Applied Economics, Vol. 37, No. 1 (2005). pp. 51–5.

20

Samford and Gomez’s study of FDI in Mexico is a notable exception. See Steven Samford and Priscila Ortega Gómez, “Subnational Politics and Foreign Direct Investment in Mexico,” Review of International Political Economy, Vol. 21, No. 2 (2014). pp. 467–96.

21

Samuel Rueckert Brazys, “Evidencing Donor Heterogeneity in Aid for Trade,” Review of International Political Economy, Vol. 20, No. 4 (2013), pp. 947–78.

22

Hyun-Hoon Lee and John Ries, “Aid for Trade and Greenfield Investment,” World Development, Vol. 84 (2016), pp. 206–18.

23

Donaubauer, et al., “Aid, Infrastructure, and FDI,” pp. 230–45.

24

Donaubauer, et al., “Does Aid for Education Attract Foreign Investors?” pp. 597–613.

25

Joshua Graff Zivin and Matthew Neidell, “The Impact of Pollution on Worker Productivity,” American Economic Review, Vol. 102, No. 7 (2012), pp. 3652–73; Robert E. Baldwin and Burton A. Weisbrod, “Disease and Labor Productivity,” Economic Development and Cultural Change, Vol. 22, No. 3 (1974), pp. 414–35.

26

Raymond Vernon, Sovereignty at Bay:The Multinational Spread of U.S. Enterprises (London: Pelican, 1971).

27

Elizabeth Asiedu, et al., “Does Foreign Aid Mitigate the Adverse Effect of Expropriation Risk on Foreign Direct Investment?” Journal of International Economics, Vol. 78, No. 2 (2009), pp. 268–75.

28

See Garriga and Phillips, “Foreign Aid as a Signal to Investors,” pp. 280–306.

29

See Donaubauer, et al., “Aid, Infrastructure, and FDI,” pp. 230–45.

30

Andrzej Cieślik and Michael Ryan, “Location Determinants of Japanese Multinationals in Poland: Do Special Economic Zones Really Matter for Investment Decisions?” Journal of Economic Integration, Vol. 20, No 3, (2005), pp. 475–96.

31

Deborah Bräutigam and Tang Xiaoyang, “African Shenzhen: China’s Special Economic Zones in Africa,” Journal of Modern African Studies, Vol. 49, No.1 (2011), pp. 27–54; John Page and Abebe Shimeles, “Aid, Employment and Poverty Reduction in Africa,” African Development Review, Vol. 27, No. S1 (2015), pp. 17–30.

32

Michael Breen and Patrick J.W. Egan, “The Catalytic Effect of IMF Lending: Evidence from Sectoral FDI data,” International Interactions, Vol. 45, No. 3 (2019), pp. 447–73.

33

Kimura and Todo who find that Japanese aid only attracts Japanese FDI being an important exception. See Kimura and Todo, “Is Foreign Aid a Vanguard of Foreign Direct Investment?” pp. 482–97.

34

Robert D. McKinlay and Richard Little, “A Foreign Policy Model of US Bilateral Aid Allocation,” World Politics, Vol. 30, No. 1 (1977), pp. 58–86.

35

Robert D. McKinlay, “The Aid Relationship: A Foreign Policy Model and Interpretation of the Distributions of Official Bilateral Economic Aid of the United States, the United Kingdom, France, and Germany, 1960–1970,” Comparative Political Studies, Vol. 11, No. 4 (1979), pp. 411–64.

36

Alberto Alesina and David Dollar, “Who Gives Foreign Aid to Whom and Why?” Journal of Economic Growth, Vol. 5 (2000), pp. 33–63; Jean-Claude Berthélemy and Ariane Tichit, “Bilateral Donors’ Aid Allocation Decisions—A Three-Dimensional Panel Analysis,” International Review of Economics & Finance, Vol.13, No. 3 (2004), pp. 253–74; Brazys, “Evidencing Donor Heterogeneity in Aid for Trade,” pp. 947–78.

37

See Sarah Blodgett Bermeo, “Aid Allocation and Targeted Development in an Increasingly Connected World,” International Organization, Vol. 71, No. 4 (2017), pp. 735–66.

38

Alison J. Ayers, “Beyond Myths, Lies and Stereotypes: The Political Economy of a ‘New Scramble for Africa’,” New Political Economy, Vol. 18, No. 2 (2013), pp. 227–57.

39

Pippa Morgan and Zheng Yu, “Tracing the Legacy: China’s Historical Aid and Contemporary Investment in Africa,” International Studies Quarterly, Vol. 63, No. 3 (2019), p. 561.

40

Ibid.

41

World Bank, “Current USD,” https://data.worldbank.org/indicator/BM.KLT.DINV.CD.WD?locations=CN.

42

Lo Dic, “Towards a Conception of the Systemic Impact of China on Late Development,” Third World Quarterly, Vol. 41, No. 5 (2020), pp. 860–80.

43

Pádraig Carmody and Joel Wainwright, “Contradiction and Restructuring in the Belt and Road Initiative: Reflections on China’s Pause in the ‘Go world’,” Third World Quarterly, Vol. 43, No. 12 (2022), pp. 2830–51.

44

“Data: Chinese Investment in Africa,” China Africa Research Initiative, http://www.sais-cari.org/chinese-investment-in-africa; Shirley Ze Yu, “Why Substantial Chinese FDI is Flowing into Africa,” Africa at LSE, 2 April 2021, https://blogs.lse.ac.uk/africaatlse/2021/04/02/why-substantial-chinese-fdi-is-flowing-into-africa-foreign-direct-investment/.

45

Chen Wenjie, et al., “Why is China Investing in Africa? Evidence from the Firm Level,” World Bank Economic Review, Vol. 32, No. 3 (2018), pp. 610–32.

46

Peter J. Buckley, et al., “The Institutional Influence on the Location Strategies of Multinational Enterprises from Emerging Economies: Evidence from China’s Cross-Border Mergers and Acquisitions,” Management and Organization Review, Vol. 12, No. 3 (2016), pp. 425–48; Chen, et al., “Why is China Investing in Africa?” pp. 610–632; Ruey-Jer Bryan Jean, et al., “Ethnic Ties, Location Choice, and Firm Performance in Foreign Direct Investment: A Study of Taiwanese Business Groups FDI in China,” International Business Review, Vol. 20, No. 6 (2011), pp. 627–35.

47

Peter J. Buckley, et al., “The Determinants of Chinese Outward Foreign Direct Investment,” Journal of International Business Studies, Vol. 38, (2007), pp. 499–518.

48

Pippa Morgan, “‘Many Chinas?’ Provincial Internationalization and Chinese Foreign Direct Investment in Africa,” Oxford Development Studies, Vol. 49, No. 4 (2021), pp. 351–67; Chen Chunlai, “Determinants and Motives of Outward Foreign Direct Investment from China’s Provincial Firms,” Transnational Corporations, Vol. 23, No. 1 (2015), pp. 1–28.

49

Giles Mohan and Marcus Power, “New African Choices? The Politics of Chinese Engagement,” Review of African Political Economy, Vol. 35, No. 115 (2008), pp. 23–42; Jean Claude Berthélemy, “China’s Engagement and Aid Effectiveness in Africa,” African Development Bank Working Paper, No. 129 (2011).

50

Cullen Hendrix and Marcus Noland, Confronting the Curse: The Economics and Geopolitics of Natural Resource Governance (Columbia University Press, New York, 2014); Axel Dreher and Andreas Fuchs, “Rogue Aid? An Empirical Analysis of China’s Aid Allocation,” Canadian Journal of Economics/Revue canadienne d’économique, Vol. 48, No. 3 (2015), pp. 988–1023.

51

Su Huaqiang, et al., “An Empirical Analysis on the Relationship between Chinese Outward Foreign Aid and Outward Foreign Direct Investment from China,” Journal of China Studies, Vol. 20, No. 3 (2017), pp. 47–68; Morgan and Zheng, “Tracing the Legacy,” pp. 558–73; Liu Ailan, et al., “Does Informal Economy Undermine the Effects of China’s Aid on its Outward Foreign Direct Investment?” International Review of Economics & Finance, Vol. 75 (2021), pp. 315–29.

52

Morgan, “‘Many Chinas?’ Provincial Internationalization and Chinese Foreign Direct Investment in Africa,” pp. 351–67.

53

Buckley, et al., “The Determinants of Chinese Outward Foreign Direct Investment,” pp. 499–518; Peter J. Buckley, et al., “An Investigation of Recent Trends in Chinese Outward Direct Investment and Some Implications for Theory,” Centre for International Business University of Leeds Working Paper (2006).

54

Yang Yi Edward and Liang Wei, “Introduction to China’s Economic Statecraft: Rising Influences, Mixed Results,” Journal of Chinese Political Science, Vol. 24 (2019), pp. 381–5.

55

See Morgan and Zheng, “Tracing the Legacy,” pp. 558–73.

56

Ibid.; Morgan, “‘Many Chinas?’ Provincial Internationalization and Chinese Foreign Direct Investment in Africa,” pp. 351–67.

57

Axel Dreher, et al., “African Leaders and the Geography of China’s Foreign Assistance,” Journal of Development Economics, Vol. 140 (2019), pp. 44–71.

58

Salvador Gil-Pareja, et al., “The Effect of the Great Recession on Foreign Direct Investment: Global Empirical Evidence with a Gravity Approach,” Applied Economics Letters, Vol. 20, No. 13 (2013), pp. 1244–8; Erica Owen, “Foreign Direct Investment and Elections: the Impact of Greenfield FDI on Incumbent Party Reelection in Brazil,” Comparative Political Studies, Vol. 52, No. 4 (2019), pp. 613–45; Johanne Døhlie Saltnes, et al., “EU Aid for Trade: Mitigating Global Trade Injustices?” Third World Quarterly, Vol. 41, No. 12 (2020), pp. 1992–2010; Samuel Brazys and Andreas Kotsadam, “Sunshine or Curse? Foreign Direct Investment, the OECD Anti-Bribery Convention, and Individual Corruption Experiences in Africa,” International Studies Quarterly, Vol. 64, No. 4 (2020), pp. 956–67; Wang, et al., “Foreign Direct Investment, Unmet Expectations, and the Prospects of Political Leaders,” pp. 1403–19; McCauley, et al., “Does Chinese FDI in Africa Inspire Support for a China Model of Development?”.

59

The FDI markets data are gathered from media sources, industry organizations, investment promotion agencies, market research companies, and from the Financial Times’ own newswires and sources. The dataset is cross-referenced to multiple sources with preference for direct company sources. The World Bank, UNCTAD and over 100 national governments use the information as primary source data for investment trends.

60

Austin M. Strange, et al., “Tracking Underreported Financial Flows: China’s Development Finance and the Aid–Conflict Nexus Revisited,” Journal of Conflict Resolution, Vol. 61, No. 5 (2017), pp. 935–63.

61

Where “Europe” FDI projects include those from the EU-27, the EEA, Switzerland and the UK.

62

See Carl Henrik Knutsen, et al., “Mining and Local Corruption in Africa,” American Journal of Political Science, Vol. 61, No. 2 (2017), pp. 320–34.

63

In the first instance, we use transaction start year or commitment year as this data is available for the vast majority of projects. However, when this information is missing and planned start, transaction end or planned end is available, we use that information. While this likely introduces some measurement error, we think this will be less than bias introduced by omitting these observations. In the robustness checks, we evaluate a model where we delay assigning “active” status for two years to account for a delay in implementation.

64

Darin Christensen, “Concession Stands: How Mining Investments Incite Protest in Africa,” International Organization, Vol. 73, No. 1 (2019), pp. 65–101.

65

Wang, et al., “Foreign Direct Investment, Unmet Expectations, and the Prospects of Political Leaders: Evidence from Chinese Investment in Africa,” pp. 1403–19; McCauley, et al., “Does Chinese FDI in Africa Inspire Support for a China Model of Development?”.

66

Knutsen, et al., “Mining and Local Corruption in Africa,” pp. 327–8.

67

As such, as shown in Supplementary Table A.1 in the Appendix, the number of “active” grid cell-years is substantially larger than the number of “inactive” grid cell-years.

68

Knutsen, et al., “Mining and Local Corruption in Africa,” pp. 320–34.

69

See the AidData geocoding methodology at http://docs.aiddata.org/ad4/files/geocoding-methodology-updated-2017-06.pdf, accessed on 13-02-2019.

70

A. Stewart Fotheringham and David W.S. Wong, “The Modifiable Areal Unit Problem in Multivariate Statistical Analysis,” Environment and Planning A, Vol. 23, No. 7 (1991), pp.1025–44.

71

Timothy G. Conley, “GMM Estimation with Cross Sectional Dependence,” Journal of Econometrics, Vol. 92, No. 1 (1999), pp. 1–45.

72

Alberto Osnago, et al., “Do Deep Trade Agreements Boost Vertical FDI?” The World Bank Economic Review, Vol. 30, No. Supplement_1 (2017), pp. S119–25.

73

Nathan M. Jensen, “Fiscal Policy and the Firm: Do Low Corporate Tax Rates Attract Multinational Corporations?” Comparative Political Studies, Vol. 45, No. 8 (2012), pp. 1004–26.

74

Breen and Egan, “The Catalytic Effect of IMF Lending,” pp. 447–73.

75

Li Quan, “Democracy, Autocracy, and Expropriation of Foreign Direct Investment,” Comparative Political Studies, Vol. 42, No. 8 (2009), pp. 1098–127; Oliver Morrissey and Manop Udomkerdmongkol, “Governance, Private Investment and Foreign Direct Investment in Developing Countries,” World Development, Vol. 40, No. 3 (2012), pp. 437–45.

76

Nigel Driffield, Chris Jones, and Jo Crotty, “International Business Research and Risky Investments, An Analysis of FDI in Conflict Zones,” International Business Review, Vol. 22, No.1 (2013), pp. 140–55.

77

Jacob M. Montgomery, et al., “How Conditioning on Posttreatment Variables Can Ruin Your Experiment and What to Do about It,” American Journal of Political Science, Vol. 62, No. 3 (2018), pp. 760–75.

78

In order to calculate simulated standard errors for this measure, we first simulate a number of active and inactive draws for each model equal to the number of observations from the model. From these simulations, we derive a distribution of the difference measure. We then simulate the same number of draws of the non-active sample mean (inactive and neighbor sites). We then use a Taylor expansion using the moments from the difference and sample mean distributions to approximate the variance of the ratio measure.

79

Following the formulas for error propagation of sums and ratios. See Georg Fantner, “A Brief Introduction to Error Analysis and Propagation,” February 2013, https://www.epfl.ch/labs/lben/wp-content/uploads/2018/07/Error-Propagation_2013.pdf.

80

For computational reasons, these results use clustered (at the grid cell level), rather than Conley, standard errors. Running each model with Conley standard errors requires approximately 90 hours on the personal computing equipment to which the researchers had access. Compared to the main models above, the Conley standard errors are roughly 2 to 2.5 times as large as the clustered standard errors.

81

Kosuke Imai and In Song Kim, “On the Use of Two-Way Fixed Effects Regression Models for Causal Inference with Panel Data,” Political Analysis, Vol. 29, No. 3 (2021), pp. 405–15.

82

Christian Peratsakis, et al., “Geocoded Activity-Level Data from the Government of Malawi’s Aid Management Platform,” Washington DC AidData and the Robert S. Strauss Center for International Security and Law (2012); AidData, “DRC-AIMS_GeocodedResearchRelease_Level1_v1.3.1 geocoded dataset(2016a),” Williamsburg, VA and Washington, DC: AidData, http://aiddata.org/research-datasets; AidData, “NigeriaAIMS_GeocodedResearchRelease_Level1_v1.3.1 geocoded dataset(2016b),” Williamsburg, VA and Washington, DC: AidData. http://aiddata.org/research-datasets; AidData, “UgandaAIMS_GeocodedResearchRelease_Level1_v1.4.1 geocoded dataset(2016c),” Williamsburg, VA and Washington, DC: AidData. http://aiddata.org/research-datasets; AidData, “SenegalAIMS_GeocodedResearchRelease_Level1_v1.5.1 geocoded dataset(2016d),” Williamsburg, VA and Washington, DC: AidData. http://aiddata.org/research-datasets; AidData, “BurundiAIMS_GeocodedResearchRelease_Level1_v1.0 geocoded dataset(2017a),” Williamsburg, VA and Washington, DC: AidData. http://aiddata.org/research-datasets; AidData, “SierraLeoneAIMS_GeocodedResearchRelease_Level1_v1.0 geocoded dataset(2017b),” Williamsburg, VA and Washington, DC: AidData. http://aiddata.org/research-datasets.

83

Dreher, et al., “Aid, China, and Growth,” pp. 135–74.

84

Ibid.

85

Samuel Brazys and Krishna Chaitanya Vadlamannati, “Aid Curse with Chinese Characteristics? Chinese Development Flows and Economic Reforms,” Public Choice, Vol. 188, No. 3–4 (2021), pp. 407–30.