Simulation-Based Tests of PTM

Abstract

Pricing-to-market (PTM) theory stipulates that exporting monopolistic firms set different prices for their goods depending on the destination markets. In other words, firms adjust their destination-specific markup in order to conform to differing demand characteristics in each country and to account for bilateral exchange rate fluctuations that could potentially affect the demand for their goods.

Empirical support for PTM comes mostly from ordinary least squares and instrumental variable (IV) estimations of partial equilibrium models. Unfortunately, this evidence may be spurious. Indeed, it was recently shown that standard inference methods in instrumental regressions (including exogeneity tests and Wald tests on regression coefficients) are highly unreliable, particularly when the instruments at hand are uninformative.

This study revisits the PTM evidence on various Japanese manufacturing product groups and industries, taking the above concerns explicitly into account. Exogeneity tests, as well as bounds and bootstrap-type likelihood ratio tests, are therefore carried out, based on OLS and limited-information maximum likelihood estimates of the well-known PTM model developed by Marston (1990). Unlike the standard IV-based procedures, these tests are known to achieve level control even when instruments are weak.

Results reveal significant right-hand-side endogeneity in the examined PTM equations. Furthermore, whereas statistical tests which ignore this problem (i.e. which replicate Marston’s testing strategy) overwhelmingly reject the no PTM null, the tests which properly correct for endogeneity fail to reject this null for several industries. This suggests that available partial-equilibrium-based evidence in favor of PTM may not be as statistically conclusive as was previously believed.

The first author thanks the Fonds F.C.A.R. of the government of Québec for financial support. Both authors thank Richard Marston and Thomas Klitgaard for kindly providing the data sets used in this paper. We also thank Christopher Nicol and two anonymous referees for helpful comments. The views in this paper are our own and should not be attributed to the Bank of Canada.