Exchange rate pass-through: evidence from aggregate Japanese exports.

• Author: Parsley, David C.

1. Introduction

   The pass-through of exchange rate changes to prices has been the subject of both theoretical, and empirical investigations.(1) Much of the recent interest in pass-through was stimulated by the unprecedented movements in the U.S. dollar during the 1980s and by the potential for these movements to have lasting consequences on the domestic economy.

   Still unresolved in these studies is whether pass-through to the U.S. market has declined over the past decade, and if so, why? Possible explanations include e.g., a growing dependence on imports, changes in the structure of competition in the U.S. market, or even "unfair" practices by foreign firms. This issue is crucial to evaluating the long-term effects of recent movements in U.S. dollar exchange rates. This paper seeks to shed light on these issues indirectly by focusing on the stability of pass-through in aggregate Japanese exports. Japan is an obvious case study both because of its increasing presence in world trade and because of the continuing concern, expressed frequently in the popular press, over the large and persistent bilateral trade imbalances between Japan and her major trading partners, especially the U.S.

   Two general criticisms of empirical studies in this area are the poor quality of the data analyzed and the lack of formal justification for the empirical modeling strategies employed. These problems are addressed in this study in two ways. First, the data set used represents a significant improvement over those employed in other aggregate analyses. This data, from the Bank of Japan, is unique because it is available on a monthly basis, and because export prices (as opposed to unit values) are reported in both domestic and contractual currencies. Thus, the exchange rate measure this study uses corresponds exactly to the export bundle; that is, we do not use the "trade-weighted" exchange rate index commonly employed. In addition to aggregate export price data, this data set also contains six sector specific price indexes, which permits a comparison of aggregate results with those at the sector, or industry level.

   Second, the problem of dynamic specification is addressed by estimating each equation in error correction form. This methodology has been advocated as a means of avoiding illusory inference when the exact lag structures are unknown by Davidson et al. |4~, and, Hendry, Pagan, and Sargan |10~. Inferences about shifting pass-through relationships may be unwarranted due to this potential misspecification of standard pass-through equations. This is a specific manifestation of the problem noted by Hendry |11, 219~, "if all true structural equations in a system remained unaltered but the behavior of some exogenous variables changed, then all mis-specified econometric approximations to the equations of that system could manifest 'shifts' (i.e., apparent structural breaks)."

   While much existing research has focused on exchange rate pass-through to U.S. import prices, the empirical investigation reported here focuses on aggregate pass-through across all Japanese export markets. This has practical as well as pedagogic benefits. First Japanese exports are an important and growing fraction of world trade. This increased presence itself begs analysis of the Japanese experience. More importantly, the detail of the data set used in this study enables a more thorough investigation of the stability/instability issue than is possible using other aggregate data.

   Below, the issue of stability is examined by formally testing the equality of pass-through estimates obtained from different sub-samples of the data at both the aggregate, and at the industry levels. This approach focuses attention on industry differences as determinants of (and thus changes in) pass-through. The approach taken here follows existing literature in estimating a log-linear, reduced-form real export price equation.

   The remainder of this paper is organized as follows. Section II summarizes the theoretical considerations relevant to the econometric specification adopted. The data are described in the third section, and section IV presents the results from estimating the aggregate and industry level equations. A final section concludes.

   II a. The Determinants of Pass-Through

   In the simplest static optimization, pass-through does not depend on rivals' strategic behavior; instead, the degree to which prices in the importing country change following an exchange rate change depends upon the relative proportion of domestic (exporting) and foreign (importing) country suppliers. This can be demonstrated by considering the pricing problem for an exporter with prices set in foreign currency. In the case where the product is produced solely by perfectly competitive exporting firms, it is easy to see that pass-through will be complete. In other words, a one-percent change in the exchange rate will be matched by a one-percent change in price. This condition follows since in perfect competition: P = SC*, where S = foreign currency/domestic currency, P = the foreign currency price of the good, and C* = the marginal cost in domestic currency. Taking logs and differentiating we see that pass-through (|Phi~ = d lnP/d lnS) is equal to one because all exporting firms experience the same shock. For the case of aggregate Japanese exports, this case does not seem relevant because the global output base is highly diversified.

   Moving away from the perfect competition case, the value pass-through takes fundamentally depends on the number of foreign and domestic suppliers as well as their response to the assumed exogenous exchange rate changes. In the other extreme case, i.e., an exporting monopolist, it is possible to show that pass-through will be less than one in the non-constant elasticity case, and less than one-half with the simplifying assumption of a linear demand curve in the export market. To see this, recall the standard monopoly pricing formula:

   SP = C*|Epsilon~/(|Epsilon~ - 1), (1)

   ...