Do Energy Prices Drive Outward FDI? Evidence from a Sample of Listed Firms.

• Author: Garsous, Gregoire

1. INTRODUCTION

   Foreign direct investment (FDI) has increased dramatically over the last four decades, both in developed and developing countries. In particular, developing countries have become a primary destination of FDI: in 2012, for the first time they received more than 50% of worldwide FDI. (1) A significant part of these investments has been made in the manufacturing sector. For example, in China, between 1997 and 2008, 70% of total FDI inflows were in the manufacturing sector for an amount of USD 388,679 million (Liu and Daly, 2011), making China the largest single FDI recipient already by 2002. Similarly in India, between 2000 and 2015, manufacturing FDI accounted for roughly 50% of total FDI inflows, amounting to USD 123,069 million (DIIP, 2015).

   FDI has potentially large benefits, for both host and investor countries. Such benefits can range from more efficient production patterns, technology and know-how transfers and economic development (see for example De Mello, 1998; Saggi, 2002; OECD, 2002 for surveys on this issue). However, reasons for deciding to invest abroad may also be controversial. For example, outward FDI can be motivated by comparatively lax environmental regulation in the destination country. This is in line with a pollution haven effect type of argument, which leads to concerns about the deterioration of competitiveness in environmentally stringent countries--in particular for pollution intensive manufacturing--and pollution leakage.

   Business associations, primarily in developed countries, often point out low energy costs and weak environmental regulation in developing countries as being partly responsible for these global FDI trends (Alliance for American Manufacturing, 2008, 2009). More generally, business leaders consider cheap energy as vital for manufacturing industries to compete on global markets. (2) High energy prices would arguably reduce industrial output and significantly reduce employment (Business for Britain, 2014). Consequently, such groups argue that the introduction of a (unilateral) carbon tax would have adverse effects on manufacturing industry activity (Morgan, 2012; Green, 2011; American Council for Capital Formation and National Association of Manufacturers, 2009; National Black Chamber of Commerce, 2015).

   In this paper, we shed light on these claims by estimating the effect of energy prices on firm-level outward FDI. Our estimations are based on an instrumental variable strategy that removes endogenous firms' choices of fuel substitution from observed energy prices. We use a sample of listed firms from 9 manufacturing sectors in 24 OECD countries over the period 1995-2008. Results suggest that only relative energy prices--i.e. the difference between domestic energy prices and prices in the FDI destination--rather than absolute energy prices are significantly correlated with firm level FDI. Second, we also find that firms heterogeneously respond to variations in energy prices. Only firms that faced increases in relative energy prices increased their international assets as opposed to those that experienced a decrease in relative energy prices, which did not respond to it. Considering only firms that experienced an increase in (relative) energy prices, we find that, on average, a 1% increase in energy prices is associated with an increase of 0.71 percent in firms' international assets. Compared to total assets, this increase appears to be small in magnitude. Further results suggest that a 1% increase in relative energy prices is associated with a 0.54 percent increase in firms' international-to-total-assets ratio on average.

   While listed firms represent only a fraction of total firms that are potentially involved in FDI, they are likely to be larger and more "internationalised" than other firms not included in our data set. Moreover, this data-driven approach is appealing in light of increasing evidence that movements in large firms help to explain major parts of aggregate fluctuations (see Gabaix, 2011, for evidence using US data). Our results are robust across various specifications.

   Conceptually, the question of whether outward FDI is driven by cross-country differences in energy prices is closely linked to the pollution haven effect. In theory, countries with relatively laxer environmental policies can gain a comparative advantage in pollution intensive industries (Pethig, 1976; Siebert, 1977; Yohe, 1979). Thus, stringent environmental regulations provide firms with incentives to relocate some stages of production to countries with laxer environmental policies--investing in "pollution havens" (Siebert et al., 1980; McGuire, 1982; Merrifield, 1988). Similarly, countries with lower energy prices can gain an absolute competitiveness advantage or a comparative advantage in energy intensive industries. Provided that capital is sufficiently mobile, a variation in the competitiveness advantage can be "internalised" by firms in countries with stringent environmental regulation through relocating their production overseas.

   From an empirical perspective, industrial energy prices can therefore be expected to have impacts similar to general emissions policies. Climate change mitigation policies through carbon pricing--carbon taxes, cap-and-trade mechanisms--are one obvious example but, in practice, even command and control instruments addressing climate or air pollution can ultimately increase energy prices. (3) Thus, estimating the effect of changes in energy prices on FDI improves our understanding of the impact of cross-country differences in climate policies on FDI and of the environmental efficacy of such policies. The main advantage of using energy prices in our analysis is that data are readily available and comparable for a large number of countries and a long time period.

   Our findings support the pollution haven effect even though the latter appears to be quantitatively small. In a simple simulation exercise, we illustrate that only a very high carbon tax would have a sizeable effect on FDI. Nonetheless, given that these effects are stronger in energy-dependent sectors and presumably geographically localised, we anticipate the political economy of carbon taxation to be complicated.

   The rest of the paper is organised as follows. Section 2 summarises the literature on the PHH debate. Section 3 explains our empirical strategy. Section 4 reports the main estimation results and discusses them. Section 5 provides a simulation of the effect of a carbon tax on FDI. Section 6 concludes.

2. EMPIRICAL LITERATURE ON POLLUTION HAVENS AND FOREIGN DIRECT INVESTMENT

   The debate on the pollution havens through FDI is based primarily on the empirical evidence from single-country effects, (4) mostly due to the dearth of relevant bilateral FDI data. (5) A large number of existing studies use US inward or outward FDI data (List and Co, 2000; List, 2001; Keller and Levinson, 2002; Xing and Kolstad, 2002; Eskeland and Harrison, 2003; Fredriksson et al., 2003; List et al., 2004; Cole and Elliott, 2005; Kellenberg, 2009; Hanna, 2010). Other studies use outward FDI from U.K. (Manderson and Kneller, 2012), Japan (Kirkpatrick and Shimamoto, 2008; Elliot and Shimamoto, 2008), France (Ben Kheder and Zugravu, 2012), Germany (Wagner and Timmins, 2009) and Korea (Chung, 2014). Another set of studies relied on inward FDI into China (Di, 2007; Dean et al., 2009), or Mexico (Waldkirch and Gopinath, 2008).

   Overall, results are mixed with some papers finding strong evidence that FDI decisions are significantly influenced by weaker environmental regulations (Hanna, 2010; Chung, 2014) or not affected at all (Eskeland and Harrison, 2003; Kirkpatrick and Shimamoto, 2008; Manderson and Kneller, 2012). Others find that the effect depends on industries' ability to relocate, proxied by capital intensity (Cole and Elliot, 2005; Kellenberg, 2009). The effects are also found to depend on the characteristics of the parent countries (Dean et al., 2009) or characteristics of the host countries (Ben Kheder and Zugravu, 2012). Conducting a meta-analysis over this literature, Rezza (2015) suggests that differences among results can be explained by different choices in the definition of FDI as a dependent variable or the use of different proxies for environmental stringency. The Table A1 in Appendix A1 summarises the main results of this literature.

   These studies suffer from at least two drawbacks that our paper attempts to address. First, given the single-country focus of this research, one can hardly generalize these results. Second, the tabulated overview of the related literature in Table A1 in Appendix A1 shows that these studies mostly relied on indexes derived from surveys, abatements expenditures or authorized pollution as proxies for environmental regulations, all of which are flawed with identification issues (Botta and Kozluk, 2014). In surveys for instance, respondents might perceive environmental regulation as more or less stringent depending on the business cycle. Abatement expenditures capture effects from non-environmental policies--such as technology and innovation policies--or firms' decisions (e.g. energy efficiency profit seeking investments that also improve environmental performance). Authorized pollution does not capture to what extent environmental regulations are enforced. That is, countries vary in the vigor of their law enforcement, which results in some discrepancy between the legal (de jure) and the actual (de facto) stringency of their environmental policies.

   The contribution of this paper is therefore twofold. First, we provide the first cross-country analysis at the firm level on the response of FDI to energy prices. Second, our estimates of FDI elasticity of energy prices allow for a better understanding of the effect of some important environmental policies, such as climate change mitigation policies. In practice, most upstream policy instruments addressing climate...