Energy Abundance, Trade and Specialization.

• Author: Gerlagh, Reyer

1. INTRODUCTION

   Do energy endowments affect industry location and trade flows? For energy-intensive manufacturing sectors such as metals and chemicals, access to low-cost energy is of crucial importance. Though oil, gas and coal are traded globally, national prices have not converged, since transport costs create a wedge between prices. Hence in the absence of policy intervention, we expect energy to be available more cheaply in energy-abundant countries. Moreover several OPEC members subsidize fossil fuel consumption below the world market price, (1) and industries in the coal-abundant U.S. face much lower coal and electricity prices than their European counterparts. Yet, taxes and subsidies could also counter the effect of local energy availability and whether on average energy endowments lead to lower prices is an empirical question.

   In this paper, we test whether energy-intensive industries are overrepresented in and export more from energy-abundant countries. Using a panel data set of 14 OECD countries, we find support for both hypotheses. Energy-abundant countries have 7 to 10 percent higher employment and 13 to 17 percent higher net exports per value added in energy-intensive sectors vis-a-vis otherwise comparable countries. Conversely, energy-scarce countries have 7 to 10 percent higher employment and 13 to 17 percent higher net exports in energy-non-intensive sectors than otherwise comparable countries.

   The rest of this paper is organized as follows. Section 2 reviews the theoretical and empirical literature and issues on determinants of industry location. Section 3 outlines the methodology. Section 4 describes the data. Section 5 presents the results, and section 6 concludes. The online appendix contains a description of data and a series of robustness checks.

2. LITERATURE REVIEW

   The Heckscher-Ohlin model has established the relation between factor endowments and industry location and trade flows as one of the pillars of trade theory. (2) If energy, like final goods in the H-O model, is traded without friction in competitive markets, energy endowments have no effect on industry location. Indeed, most empirical work inspired by the H-O model does not include energy as a production factor (cf. Bowen et al. 1987, Trefler 1995). When energy trade is costly or countries use strategic trade policy, energy prices are lower in energy-abundant countries though. Romalis (2004) argues that factor price differences are an important mediator of the effect of factor endowments on industry location. We can thus expect energy abundance also to significantly affect industry location and trade flows.

   Existing evidence on the location of energy-intensive industries is limited and focuses on the U.S. Assuming capital-energy complementarity, Hillman and Bullard (1978) find that the U.S. has a comparative advantage in labour-intensive sectors, and conversely a disadvantage in energy-intensive sectors. Ellison and Glaeser (1999) report that energy-intensive industries are overrepre-sented in U.S. states with low energy prices, and Gustavsson et al. (1999) find that energy-intensive sectors are more competitive in OECD countries with low energy prices. Hillman and Bullard (ibid) calibrate a modified version of the Heckscher-Ohlin model to aggregate U.S. import and export data, rather than performing an econometric estimation. Ellison and Glaeser (ibid) and Gustavsson et al. (ibid) interact energy prices with energy intensity but leave the direction of causality unsettled; they do not address the concern that energy prices may be endogenous to the industry structure. Our approach is more closely related to Michielsen (2013) who addresses this concern by directly estimating the effect of coal and gas resources on sector location, showing fuel resources are a major determinant of U.S. industry location. Grether et al (2013) apply a similar methodology to Chinese provinces over the period 1999-2009. They find that larger energy endowments are significantly correlated with larger production of energy-intensive sectors. Disaggregating across energy carriers shows that coal, the most costly energy carrier to transport, exhibits the strongest effect.

   Other studies with an international dimension use computable general equilibrium models to numerically simulate the movements of energy-intensive sectors in response to internationally differentiated energy prices (Felder and Rutherford 1993, Babiker 2005). The implementation of the Kyoto protocol has resulted in some recent empirical studies on 'carbon leakage' where the focus is on climate policy and energy abundance is not directly addressed (Aichele and Felbermayr 2012, 2013).

   This paper contributes to the literature by investigating the effect of energy abundance on industrial activity and trade flows in several countries over an extended period of time. To the best of our knowledge, this is the first paper investigating the effect of energy abundance on trade flows and industry location in a panel of countries. Following the literature investigating the importance of factor endowments and to assure the robustness of our results, we use both sectoral production and sectoral trade flows as dependent variable (see for example Romalis 2004 using trade data and Midelfart-Knarvik et al. 2000 for production data as dependent variable).

3. METHODOLOGY

   Our methodology is closely grounded in theory and has been applied in several previous empirical studies (Midelfart-Knarvik et al., 2000; Crafts and Mulatu, 2005; Mulatu et al., 2010; Michielsen 2013). All other things being equal, all industries would be interested to locate in e.g. capital-abundant countries. However, desirable country characteristics increase location costs for the average firm in general equilibrium. Midelfart-Knarvik et al. (2000) and Romalis (2004) develop models to show that in equilibrium, the sectors that end up in capital-abundant countries are those that benefit most from capital abundance, i.e. capital-intensive sectors. Non-capital-intensive sectors move somewhere else, and therefore tend to locate in capital-scarce countries.

   The model in Midelfart-Knarvik et al (2000) combines both comparative advantage and geographical elements. Countries differ in factor endowments and face transport costs when trading. Sectors use inputs to produce differentiated goods, which allows for positive trade flows and intraindustry trade even in the presence of transport costs. Linearizing the equilibrium relationships allows them to derive the econometric equation directly from their theoretical model.

   We test whether the predictions from these models also hold for energy endowments. Our first hypothesis is hence that energy-abundant countries have larger energy-intensive sectors than countries that are scarcely endowed with energy. The latter countries specialize in non-energy-intensive sectors. We measure industrial activity by employment. This...