Sunk cost models of new trade theory have demonstrated that large, unanticipated swings in currency values can have permanent effects on trade flows by altering market structure in imperfectly competitive markets. The author tests an implication of these models using panel data on export unit values from Germany and Japan to a number of foreign markets, including the United States. He finds weak evidence in support of the thesis that U.S. product markets became more competitive as a result of the large dollar appreciation of the 1980s, although the results vary by industry and comparison country.

The paper measures the degree of price discrimination across export destinations that is associated with exchange-rate changes using U.S., U.K., German, and Japanese industry-level data. Given the industries sampled, more price discrimination across destinations is observed in the U.K., German, and Japanese data. For industries that match across source countries, however, behavior is very similar across source countries. Furthermore, destination-specific price adjustment on exports to the United States from Germany and Japan is similar to price adjustment observed on shipments to other destinations. Most variation in the data appears to be related to industry.

The insensitivity of the U.S. trade balance to the sharp depreciation of the dollar in the past three years has revived interest in the relationship between exchange rates and the trade balance. A central issue in most analyses of the relation between the current account and the exchange rate concerns the price adjustment process.' The simple integrated, competitive market model predicts that local currency prices should change in proportion to the nominal exchange rate for a country too small to influence world prices. The relationship between local currency import prices and exchange rates has been referred to as the relationship in the empirical literature in international economics. If the proportional relationship between import prices and exchange rates holds, pass-through is said to be complete. Accounts of recent U.S. experience cite the failure of dollar prices of imported goods to rise in proportion to exchange rates (i.e., incomplete pass-through) as an important factor in explaining the persistence of the trade deficit. Unfortunately, observations on pass-through alone provide limited insight into the behavior of markets, since incomplete pass-through is consistent with at least two fundamentally different paradigms. One is the standard competitive model of trade in which the law of one price holds, but exchange rate fluctuations are associated with large changes in import demand due to other factors. For example, if dollar appreciation is correlated with increases in world demand and industry marginal cost is increasing, then pass-through will be less than complete. The other is an imperfectly competitive model in which exporters are capable of price discriminating across destination markets, a phenomenon. Paul Krugman (1987) has labeled pricing to market. In this model incomplete pass-through is typically associated with fluctuations in the markup of price over marginal cost on exports.2 These fluctuations in markups are believed to be countryspecific; they do not reflect the behavior of prices to other export destinations. Both models are plausible explanations of recent U.S. experience. Unfortunately, it is impossible to distinguish between these competing models using only information on import prices and exchange rates for a single country. In order to distinguish between the competing models, an empirical analysis of goods prices and exchange rates must be capable of measuring either marginal cost or the markup over marginal cost. Either of these tasks poses formidable empirical problems. Much of the empirical literature in industrial organization has been concerned with precisely these issues, since markups are an important measure of the competitiveness of industries.