An important aspect of the globalisation process (Rodrik 2011) is foreign direct investment (FDI), which underlies the fragmentation of supply chains (Baldwin 2016) and the emergence of global companies (Navaretti et al. 2004). Time and effort have been devoted to developing a satisfactory understanding of the rationale and the mechanisms through which the benefits for the host of FDI take root, leading to theoretical support for a positive impact from FDI (for example, through increases in total factor productivity and accelerating capital accumulation). This led to the expectation that the gains from FDI in terms of economic growth and business performance would be easily identified, allowing the debate to be focused on the scale of these benefits. However, consistent positive empirical effects have proved to be elusive. Rather, recent research at both the overall and firm level has converged on the view that the effect of FDI on the host economy is conditional. Hence in one stream, the overall effect of FDI is argued to depend upon whether recipient countries have attained minimum levels of human capital (Borensztein et al. 1998), financial development (Alfaro et al. 2004), and institutional quality (De Santis and Lührmann 2009). In the other, the firm-to-firm effects of FDI operating through horizontal and vertical spillovers, are considered to be conditional upon the form of inter-firm linkages, with vertical dominating horizontal and backward forward linkages (Javorcik 2004).
This column takes contemporary stock of the evidence, treating overall and firm-to-firm effects jointly (Bruno et al. 2017). We reassess the conclusions of the literature based on a comprehensive systematic quantitative review of these two bodies of empirical evidence within a comparable empirical model.
The empirical literature documents the relationship between FDI and host economy growth, investment, and productivity. The impact of FDI has typically been identified at the micro level by looking at firm-to-firm evidence, usually restricted to a single country, on spillover effects on firm productivity and performance from foreign to domestic competitors within an industry (horizontal spillovers) or through supply chains (vertical spillovers). These studies typically have little to say about the overall impact of FDI. In contrast, macro-level analyses have a cross-country perspective and therefore need to overcome more cogent potential methodological problems in terms of, for instance, endogeneity and omitted variable bias. Although drawing on different theoretical underpinnings and data analyses, both streams of the literature, when compared and contrasted, have strong findings.
We devote equal attention to the two bodies of evidence (overall and firm-to-firm; or micro and macro) because we think that taken jointly, they can further improve our understanding of FDI. The latter throws light on private returns and localised firm-to-firm effects, while the former can reveal important features of social returns and the net effects of FDI inflows. Our analysis draws on meta-regression techniques, an empirical methodology for summarising and distilling the lessons from a given body of econometric evidence (Anderson and Kichkha 2017). For this exercise, we construct a new hand-collected dataset covering 565 micro (or firm-to-firm) and 551 macro (or overall) level estimates of the effects of FDI on performance, from 104 and 71 empirical primary studies, respectively. We also construct variables covering 13 features of these econometric estimations with respect to, among other things, various features of sampling and econometric methodology. Our study has five main areas of originality:
- We put forward a new methodological framework in which we jointly analyse the ‘overall’ and ‘firm-to-firm’ FDI effects;
- We create a unique new data set that can implement our proposed methodological innovations;
- We base the analysis upon a substantially larger number of papers than previous FDI meta-analyses;
- We exploit a wider set of moderator variables and controls; and
- We employ a sophisticated empirical model allowing the studies to be a random sample from the universe of all possible studies and hence assuming that there are real differences (which we test for) between all studies in the magnitude of the effects.
One of our main findings concerns the distribution of the FDI impacts between overall and firm-to-firm effects. We find that across our sample of studies, 44% of the firm to firm effects are positive and statistically significant; 44% are insignificant; and 12% are negative and significant (Figure 1). On the other hand, 50% of the overall effect estimates are positive and statistically significant, 39% are insignificant and 11% are negative and significant (Figure 2). In these studies, the overall spillover effects are at least six times larger than the firm to firm ones. This suggests that by focusing on external effects, say through imitation, within an industry or up and down a supply chain, the micro studies miss important broader spillover effects across the entire host economy.
The second set of results tries to identify the reasons for the observed variation in the estimated effects of FDI. Bruno et al. (2017) show that the choice of statistical method and specification are more important factors in macro studies than in micro, because the latter often uses more sophisticated empirical models. There is evidence that empirical specifications controlling for panel unobserved heterogeneity report significantly smaller effects of FDI and the same applies to those studies that take into account the interaction of FDI with, for example, R&D expenditures, trade openness, human capital, and financial openness (in macro studies). Macro studies have attempted to overcome methodological challenges of endogeneity and omitted variable bias often via interaction terms so as to identify potential channels.
Figure 1 Distribution of firm-to-firm effects across studies
Figure 2 Distribution of overall effects across studies
Figure 3 plots partial correlation coefficients (a measure of the impact of FDI on growth) on the vertical axis against macro effects (in red) which can be seen to be larger than micro effects (in blue), though the difference is more muted for, in this case, panel data estimates.
Figure 3 Impact of FDI on growth, micro and macro effects
The third main result is that we find that FDI spillover effects are heterogeneous through time and across countries, even when controlling fully for statistical method and specification. Can one find any systematic patterns for time and country heterogeneity? Interestingly, we do for the latter, but not for the former. We cannot identify any over time pattern. Yet we are able to identify a wedge between the average macro and micro effect across countries. The average wedge is 6 to 1 on the whole sample. However, there is variability among countries – for example, if we look at a given year in the middle of the sample (e.g. 1995), the micro effect in China, in Ghana, and in Ukraine is respectively about 7%, 12%, and 60% smaller than the macro effect.
The main lesson from the literature – namely, that the spillover effects of FDI are conditional on countries having reached certain thresholds with respect to human capital and financial/institutional development – is disjoint with the finding that the effects are larger for countries below such critical thresholds (Cohen and Levinthal 1990, Acemoglu et al. 2006). We posit that the gap between private and social returns, albeit largely missing in most of the current academic and policy discussions, may provide an explanation. Private returns to FDI are higher in low income countries, but because of institutional deficiencies, infrastructure problems, pervasive rent-seeking and/or generalised lack of competition, the benefits from these investment projects are highly localised. This wedge between the overall and micro effects is important for the dynamics and distribution of the benefits from globalisation but has received, as our analysis suggests, insufficient attention in the literature so far.
Acemoglu, D, P Aghion and F Zilibotti (2006), “Distance to frontier, selection, and economic growth”, Journal of the European Economic Association 4(1).
Alfaro, L, A Chanda, S Kalemli-Ozcan and S Sayek (2010), “Does foreign direct investment promote growth? Exploring the role of financial markets on linkages”, Journal of Development Economics 91(2): 242–256.
Anderson, R and A Kichkha (2017), “Replication, meta-analysis, and research synthesis in economics”, American Economic Review 107(5): 56–59.
Baldwin, R (2016), The Great Convergence, Cambridge, MA: Harvard University Press.
Borensztein, E, J Gregorio and J Lee (1998), “How does foreign direct investment affect economic growth?”, Journal of International Economics 45(1): 115-135.
Bruno, R, N Campos and S Estrin (2017), “The benefits from foreign direct investment in a cross-country context: A meta-analysis”, CEPR Discussion Paper No. 11959.
Cohen, M W and D L Levinthal (1990), “Absorptive capacity: A new perspective on learning and innovation”, Administrative Science Quarterly, 35: 128-152.
De Santis, R A and M Lührmann (2009), “On the determinants of net international portfolio flows: A global perspective”, Journal of International Money and Finance, 28: 880-901.
Javorcik, B S (2004), “Does foreign direct investment increase the productivity of domestic firms? In search of spillovers through backward linkages”, American Economic Review 94(3): 605-627.
Navaretti, G, A Venables and F Barry (2004), Multinational Firms in the World Economy, Princeton University Press.
Rodrik, D (2011), The Globalization Paradox, Oxford: Oxford University Press.