Nonlinear, Nonparametric, Nonessential Exchange Rate Estimation

Abstract

A wide variety of empirical exchange rate mo.dels have been estimated over the years. But, despite the considerable energies that have been devoted to this work, the economics profession has remarkably little to show for itself. There is little evidence that conclusively links the bilateral exchange rates of typical OECD countries to fundamental macroeconomic determinants of exchange rates, such as money, output, relative prices, or interest differentials. Coefficient estimates are notoriously unstable and frequently mis-signed (compared with theoretical predictions); exchange rate equations do not fit particularly well, and forecast no better than the simplest naive alternatives. Recently, a new class of exchange rate models was introduced by Paul Krugman (1988). These models provide a potential reason for the poor performance of traditional exchange rate models, because they are nonlinear. If the exchange rate actually depends in a nonlinear way on exogenous macroeconomic fundamentals, linear exchange rate models may work poorly, even though the exchange rate is closely linked to fundamentals. In this paper we provide a brief sketch of some of this work, as well as some preliminary evidence on the actual performance of these nonlinear models. In our empirical analysis, we use a nonparametric estimator that can handle a wide variety of nonlinear phenomena. We examine fixed exchange rate regimes, where nonlinearities should be quite easy to detect. However, we do not find strong empirical support for the hypothesis that the incorporation of nonlinear effects significantly improves models of exchange rate determination. In Section I, we briefly review the theoretical literature on nonlinear target zone exchange rate models, linking this work to the tests for intrinsic bubbles (we draw heavily on recent papers by Kenneth Froot and Maurice Obstfeld, 1989a,b). Our methodology and data are discussed in Section II; Section III contains new empirical tests for nonlinearities in exchange rate models.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 151 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.