| Part 1 derives and tests a bilateral balance of payments (BOP) flow model of bilateral currency exchange rate determination, incorporating triangular arbitrage ties among exchange rates.;The BOP flow model of exchange rates is based on the BOP equilibrium condition, which is solved for the exchange rate. Most BOP flow models are multilateral and hence are used for effective exchange rate determination. Unlike the multilateral BOP, the bilateral BOP has to take into account such possible complications as when one of the two countries in the bilateral relationship uses the other country's currency in transactions with third countries. In the few cases of bilateral BOP flow models, such things are assumed not to occur. In this study, however, extra bilateral BOP constituent accounts (labelled "bilateral currency leakage and injection") are added in order to incorporate them, so that the bilateral BOP is always equal to zero by definition, rather than by assumption.;Movements in exchange rates due to macroeconomic fundamentals alone may possibly result in divergences in cross-exchange rates from direct exchange rates. Such a divergence creates a triangular exchange rate arbitrage opportunity. In turn, triangular arbitrage moves and ties together exchange rates until actual cross-exchange rates equal actual direct exchange rates again. Triangular arbitrage ties are a much broader and richer concept than cross-exchange rates alone, and have not been used, nor even described, in previous exchange rate models.;The valuation effect of a nominal exchange rate (NER) movement on domestic currency-measured, foreign currency-denominated BOP flows also plays a significant role in the model, relating all of the BOP constituent accounts to the exchange rate. For example, if a U.S. company has a subsidiary in Japan, and if the dollar depreciates against the yen, then the value of the profits denominated in yen is unchanged, but measured in dollars, it rises.;No empirically-tested, nonhybrid bilateral BOP flow model of bilateral exchange rate determination is known to exist in the literature. The model significantly outperforms a simple random walk, with or without drift, in 3-month out-of-sample forecasting for three prominent NERs. No other NER model does so. The triangular arbitrage ties provide more predictive power for the model than the BOP ties. The purchasing power parity (PPP) condition, expressed in rate of change, also provides a surprisingly large degree of the predictive power of the model.;Part 2 first examines theoretical long-run exchange rate determination under the BOP flow model. Empirically, the model is tested using cointegration analysis. The results indicate that nominal or real exchange rates are not necessarily mean-reverting, even in the long-run, contrary to other empirical findings which uphold the PPP model in the long-run.;Then part 2 works with an intertemporally-optimizing BOP flow model of the real exchange rate. Numerous relationships of the venerable but non-optimizing absortion-elasticities approach to the trade balance, such as trade adjustment lags, are replicated in the model. As a result, the model provides a rigorous theoretical underpinning for such relationships, thereby helping to justify their use in part 1. |
