Accounting for trade deficits

Highlights

• Total trade deficit can be decomposed into its components.

• Trade deficit of the U.S. is mostly due to its relative trade costs.

• Trade surplus of China is mostly due to its effective terms of trade.

• Alternative policy tools are required to balance trade at the country level.

Abstract

This paper proposes a decomposition for the total trade deficit of a country by using implications of a dynamic trade model. It is shown that the total trade deficit of a country can be decomposed into changes due to its effective terms of trade, its relative trade costs, and its macroeconomic developments with respect to its export partners. The implications for bilateral trade are estimated using both imports and exports data for 188 countries, and the decomposition of total trade deficit is achieved for each country. Empirical results show evidence for heterogeneity across countries regarding the decomposition of trade deficits, suggesting alternative policy tools to rebalance trade at the country level. A cross-country investigation further suggests that relative trade costs, followed by relative macroeconomic developments, have contributed the most to the heterogeneity of trade imbalances.

Introduction

Trade deficits (defined as the difference between total imports and total exports) have been experienced by more than 70% of the countries around the globe between 1979–2015.¹ Having a trade deficit is problematic, because it is simply financed by capital flows (from trade-surplus countries) of which sudden stop can be destabilizing not only at the country level but also globally (see Milesi-Ferretti and Razin, 2000, Blanchard and Milesi-Ferretti, 2009, Catão and Milesi-Ferretti (2014) or Caballero (2016)); moreover, a trade deficit can result in a dynamic Dutch disease (see Caballero and Lorenzoni (2014)). On the other hand, having a trade surplus is also problematic, because a trade surplus may reflect an underlying domestic distortion (see Blanchard and Milesi-Ferretti (2012)) or trade-surplus countries may become targets for protectionist measures by trading partners (see Carney (2017) or Obstfeld (2018)). Accordingly, having a balanced trade (or at least not having an excessive deficit/surplus) as investigated by Dekle et al. (2007) is desirable for any open economy, which requires the knowledge of the sources of trade deficit.

This paper investigates the sources of trade deficit by using an international trade approach. In particular, based on the implications of a dynamic trade model that incorporates implicitly additively separable nonhomothetic constant elasticity of substitution (CES) preferences as in studies such as by Hanoch (1975) or Comin et al. (2015), the trade deficit of any country is decomposed into the effects due to changes in effective terms of trade, relative trade costs, and relative macroeconomic developments. This is achieved in two steps. First, by using the implications of the dynamic trade model, bilateral imports and bilateral exports of 188 countries are estimated. As is standard in the international trade literature, these estimations result in fitted values representing bilateral trade costs, source-time fixed effects and destination-time fixed effects for both bilateral imports and bilateral exports in logs. Second, since the sum of logs is not equal to the log of sums due to Jensen’s inequality (i.e., one cannot take the sum of log bilateral trade deficits to obtain log total trade deficit), the fitted values obtained from these estimations are connected to the changes in total trade deficit of each country over time by using the Taylor series of bilateral trade expressions. This innovation results in a decomposition of the level changes in total trade deficit of a country into changes in its effective terms of trade (representing the difference between the weighted average of import prices and the weighted average of export prices), changes in relative trade costs of the country (representing the changes in the weighted average of import trade costs and the weighted average of export trade costs), and relative macroeconomic developments of the country (representing changes in both relative economic activity and relative saving decisions with respect to its export partners). Since cumulative changes over time in the level of total trade deficit of any country is equal to its level of total trade deficit for any given period, a final decomposition can be achieved for the level of total trade deficit for any country.

The empirical results suggest that each country has different patterns over time regarding the contribution of each component in the decomposition of trade deficits, although relative trade costs followed by relative macroeconomic developments have contributed the most to the magnitude (of the trade deficit) during the sample period, on average across countries. When countries are categorized as Organisation for Economic Co-operation and Development (OECD) versus non-OECD countries, the average OECD country has experienced a trade surplus that is mostly explained by effective terms of trade followed by relative macroeconomic developments, whereas the average non-OECD country has experienced a trade deficit that is mostly explained by relative trade costs followed by relative macroeconomic developments. When subsamples are considered, it is shown that the establishment of the World Trade Organization (WTO) coincides with higher trade deficits for non-OECD countries that are accounted for by relative trade costs and relative macroeconomic developments, although trade surplus of OECD countries and its components have been stable over time.

Regarding country-specific results, for example, the U.S. trade deficit is mostly explained by the positive contributions of relative trade costs followed by those of effective terms of trade. In contrast, the negative Chinese trade deficit (i.e., its trade surplus) is mostly explained by its negative effective terms of trade, despite high and positive contributions of its relative macroeconomic developments. Another interesting country is Japan of which negative trade deficit (i.e., its trade surplus) is mostly explained by its relatively negative macroeconomic developments, followed by its negative relative trade costs.

Since trade imbalance of a country is financed through changes in its net foreign asset position, the international macro literature has investigated the reasons behind current account imbalances from a macroeconomic perspective (e.g., see Gourinchas and Rey (2014) for an excellent survey of the literature). This literature has mostly focused on consumption/saving and investment decisions of economic agents (i.e., intertemporal approach) as in Obstfeld and Rogoff (1995), asymmetries between financial and economic development in advanced and emerging countries as in Caballero et al. (2008), cross-country differences in the ability to insure away idiosyncratic risk as in Mendoza et al. (2009) or Angeletos and Panousi (2011), interactions between financial frictions and international trade as in Antras and Caballero (2009), and the market value of claims and liabilities underlying a country’s net foreign position as in Lane and Milesi-Ferretti (2001).

This paper contributes to this literature by focusing on how trade-based variables such as source prices or international trade costs interact with macroeconomic developments to explain trade imbalances in a dynamic trade model. Although this trade-based approach is similar to studies such as by Dekle et al., 2007, Reyes-Heroles, 2016 or Alessandria and Choi (2018), different from them, this paper contributes by providing an estimation-based decomposition that is essential to understand the sources of total trade imbalances for 188 countries. On top of these studies, a cross-country investigation in this paper has further shown that the heterogeneity across countries regarding their total trade deficits is mostly connected to their relative trade costs, followed by relative macroeconomic developments.² When the same investigation is achieved for OECD versus non-OECD countries, the heterogeneity across the latter can be attributed more relative trade costs, while this attribution is less for the former.

The rest of the paper is organized as follows. The next section introduces a dynamic trade model that is connected to the decomposition of trade deficits. Section 3 discusses the estimation methodology and the data used. Section 4 depicts the country-specific results. Section 5 depicts summary of results for country groups and achieves a cross-country investigation. Section 6 concludes. Derivations are achieved in the Appendix A, whereas country-specific results are given in the Online Appendix.