Using a new dataset, Investing Across Borders 2010, on selected regulations that govern foreign direct investment (FDI) in 87 countries, and direct investment they receive from 30 OECD source countries, this paper explores how much FDI-specic policies and institutions inuence FDI inows when rms are heterogeneous in terms of productivity. The analysis overhauls conventional estimation techniques by directly addressing biases resulting from country and rm selection. The paper nds evidence of

more »

... antial heterogeneity bias in the eects of distance and related policy barriers on the inow of FDI. Controlling for rm heterogeneity and country selection, as well as key determinants of FDI like market size, a statistically signicant relationship is found between FDI regulations and the value of inward direct investment. However, when the quality of logistics infrastructure is accounted for, the salience of FDI regulations diminishes. For a clear source of identication when correcting for selection biases, the paper uses another new indicator of the number of days and procedures required to establish a wholly foreign-owned subsidiary. for helpful comments. I also thank Biswo Poudel and Uttam Sharma for feedback on earlier drafts. Please send comments to 1 Over the past three decades, FDI has been associated with several successful growth transitions, notably in Asia. In 1980, total inward ow of FDI globally was just over US$50 billion; in 2007, this peaked by nearly a factor of 40, at US$1.98 trillion. Between 2005 and 2007, the developing countries received an average annual FDI inow of around US$410 billion which outstripped total remittance inow of US$217 billion and ocial development assistance of around US$100 billion. 1 While the global recession after 2007 broke the upward trend, FDI ows are likely to recover in the future, with perhaps a more pronounced share in favor of greater South-South ows, and in services relative to manufacturing. 2 In this paper, I ask to what extent country-specic policies and institutions that are directly related to FDI inuence aggregate FDI inows after controlling for market size, quality of infrastructure, income per person, and factors that aect transaction costs such as distance, and colonial or linguistic ties between countries. 3 To do so, I adopt two new approaches. First, I use new sets of indicators that specically measure, i) a country's openness to foreign investment, ii) quality of institutions related to investment disputes, and iii) time and procedures required to set up a wholly foreign-owned subsidiary. These new indicators are notable not only for being the most comprehensive to date in terms of coverage of topics and countries, 4 but they also obviate the need to rely on proxy indicators for FDI openness or institutions, as has been common in the literature. Second, the analysis incorporates rm heterogeneity without the use of rm-level data, and corrects for sample selection, following an analogous paper on trade ows by Helpman, Melitz, and Rubinstein (2008) . This involves an econometric methodology that corrects for two major biases prevalent in standard gravity models. 5 The rst one arises when limiting the sample to only those countries that actually have an investment relationship with each other and excluding those that do not (problem of country selection). The second bias arises when rms are not dierentiated by their ability to meet xed costs of investing abroad (problem of rm heterogeneity). Together, the paper purges a theoretically derived gravity model of the two biases, and then asks whether indices of the quality of FDI-related institutions, and openness to FDI matter for FDI inows. The paper proceeds as follows. Section 2 scans the new literature on the determinants of FDI ows from general and partial equilibrium perspectives, especially the policy and institutional determinants of FDI on which the paper builds. Section 3 derives a gravity model for FDI from micro-theoretical foundations. Section 4 describes how the gravity equation is implemented empirically. Section 5 introduces the new sources of data used. Section 6 shows the main results by comparing the benchmark estimates with those obtained after correcting for biases resulting from sample selection and rm heterogeneity. Section 7 uses alternative dependent 1 FDI involves either building of new facilities for production of goods and services (greeneld), or acquiring of existing ones (mergers and acquisitions). It does not include portfolio investment. 2 In 2006, services accounted for 62 percent of estimated world inward FDI stock, up from 49 percent in 1990 (UNCTAD 2008). 3 The interpretation of physical distance ought not be literal. It could proxy for other forms of friction such as lack of information and unfamiliarity. 4 See www.investingacrossborders.org 5 Gravity models predict bilateral ows such as trade, investment, and migration to depend positively on economic pull such as GDP of both home (exporter) and host (importer) countries, and negatively on frictions such as distance and policy barriers. 2 and explanatory variables to check for robustness of the main results including endogeneity of key regressors. Section 8 concludes. Related Literature The new trade models in the 1980s based on monopolistic competition assumed similar size and productivity across symmetric rms within an industry, and predicted that all rms export to all countries, only more or less depending on trade costs (transport and taris). 6 The literature on rm heterogeneity, begun notably by Bernard and Jensen (1995) , points out that rms that co-exist in an industry nevertheless could dier widely in terms of productivity. Because exporting abroad entails sunk costs, only the more productive are in a position to meet those costs. Anecdotally, it is not dicult to see that rms that export and withstand international competition are probably more productive than those that only serve the domestic market. However, until Melitz (2003) , among others, showed how diering heterogeneity among rms aects their propensity to participate in world trade, the fact that rms self-select into export market was not robustly established. Today, we know that only a small share of rms export, and those that do tend to be more productive and larger (see survey by Greenaway and Kneller 2007). This framework applies to FDI as well. Helpman, Melitz and Yeaple (2004) show, as summarized in Appendix 1 and in Figure 1 , that when prots (π) are a function of varying productivity (θ) and diering xed costs (f ) that rms face, there is a natural sorting of rms. Fixed cost of exporters (such as setting up of distribution networks, advertising, and complying with product standards) is assumed to be less than that incurred by those undertaking FDI which involves establishing and operating an entire new plant. Exporters have higher marginal costs, and because FDI does not involve transport cost, the prot schedule of FDI-oriented rms is steeper ensuring an intersection with the atter curve of the exporter. Self-selection of rms ensues as a result of heterogeneous productivity. The most productive undertake FDI (I), the next tier of productive rms serve the foreign market through export (X), and the least productive serve the domestic market (D). 7 This is supported by data. Helpman, Melitz and Yeaple (2004) show for US manufacturing rms across 52 sectors and 38 countries, multinational enterprises had 15 percent more labor productivity than exporters in 1994. Exporters were 39 percent more productive than non-exporters. Girma et al. (2004) , Girma et al. (2005) , and Arnold and Hussinger (2005) all nd signicant productivity dierences between rms that invest abroad and those that do not. Chen and Moore (2010) nd in the case of French multinational rms that those with low productivity are less likely to invest in host countries with a small market size, high production costs, or low trade costs. 6 Trade costs aect only the intensive margin, the volume of export per rm, not the decision of whether to export in the rst place. 7 The implication of this model is that the weakest multinational rm is more productive than the strongest exporting rm. Home and foreign countries are assumed to be identical and the only motivation for FDI is horizontal. Head and Ries (2003) show that when rms undertake vertical FDI to take advantage of low factor costs, the ordering of productivity as predicted by Helpman et al. (2004) can reverse.