Currency movement and the forecasting power of interest rates.

• Author: Supanvanij, Janikan
• Position: Report

1. INTRODUCTION

   Currency trading market has been growing rapidly in the last decade. Exchange rates in floating currency system are commonly related to demand and supply. In addition, macroeconomic factors have shown as important factors in asset pricing. Many researchers in finance area recently have turned to macroeconomic factors in order to explain the pattern of the stock returns. Some explain the link between stock returns and domestic macroeconomic activity (Balvers, Cosimano and McDonald, 1990; Chen, Roll, and Ross, 1986; Fama, 1990). Some examine the relationship between stock returns and both domestic and international macroeconomic activity (Canova and De Nicolo, 1995; Patelis 1997). Mukherjee and Naka (1995) find evidence to support the cointegration between return and six macroeconomic variables in Japanese market, including exchange rate, money supply, inflation, real economic activity, long-term government bond rate, and call money rate.

   Exchange-rate determination theories, like purchasing power parity (PPP) and Interest Rate Parity (IRP), associate the currency movement with the expectation of future economic events. PPP states that exchange rates reflect the relative purchasing power of currencies in the long run; whereas IRP explains the relationship between exchange rates and interest rates.

   Taguchi (2007) studies the post-crisis exchange rate management in East Asian Countries and finds that inflation rate is one of the significant factors that determine a reference rate in the exchange rate management. Falk and Wang (2003) test the long-run purchasing power parity hypothesis on industrial countries' monthly data when exchange rates and inflation rates are assumed to be heavy-tailed stochastic processes. They note that their results are marginally less supportive of PPP. The study of Wu, Lee, and Wang (2011) show that nominal exchange-rate adjustments dominate in the reversion toward PPP regardless of a nominal exchange-rate shock or a price shock.

   This paper looks at the time series behavior of exchange gain/loss due to economic condition changes. It examines the long-run relationship between currency movements and economic factors, including interest rates, maturity spread, gross domestic product, and inflation differences between domestic and foreign countries. The sample consists of the exchange rate movements of Australia, Canada, Great Britain and Japan during 1994-2010. The long-run predictive power of economic factors on exchange rates is analyzed in this study.

   Johansen's cointegration test and Variance decomposition method are also employed to investigate the cointegration between currency movements and the independent variables. Johansen's cointegration method is useful because it can explain dynamic co-movements among variables examined better than the vector autoregressive (VAR) model. The cointegrated time series occur when the linear combination of at least two non-stationary series is stationary, which indicates a long-run equilibrium relation. The variance decomposition method is also employed to decompose the variation in an endogeneous variable into component shocks to variables in the vector autoregressive model.

   The following section presents data and model specification. Section III presents results and discussion. Section IV concludes.

2. DATA AND MODEL SPECIFICATION

   According to fundamental analysis, exchange rate is based on expected inflation rate. So investors' perception regarding the inflation rate is important since it determines the exchange rate. The three-month Treasury bill rate is used to measure the effect of short-term interest rates while the thirty-year Treasury bond rate is used to capture the effect of long-term interest rates. Based on the pure expectation hypothesis, it can be viewed as weighted averages of expected future short-term interest rates. However, they are not used in the same model to avoid the possible multicollinearity problem. In the short-run model, the maturity spread, which is the difference between the three-month Treasury bill rate and the thirty-year Treasury bond rate, is used as an independent variable to measure the risk premium and avoid the multicollinearity problem between short-term and long-term interest rates. In many finance studies, researchers also suggest using the term spread in the models to avoid the multicollinearity problem (Chen, 1991; Fama and French, 1988, 1989; Fama and Bliss, 1987; Campbell, 1987; Keim and Stambaugh, 1986; Fama, 1984; Shiller, 1979; Fama and Schwert, 1977)

   To guarantee comparability of all series, the quarterly data are used instead of monthly data. The information on exchange rates and interest rates are available on a continuous basis and have no informational lag; whereas monthly macroeconomic data, such as consumer price index and gross domestic product, have informational lag since they are available one or two months after the period covered by the data. Estrella and Mishkin (1998) suggest that the observations actually available as of the end of a quarter are assigned to that quarter.

   The data consists of quarterly exchange rate movements of Australian dollar, Canadian dollar, British Pounds and Japanese yen during the sample period of 1994-2010. The daily exchange rates are obtained from the Pacific Exchange Rate Service. The data on Treasury bill rates, gross domestic product and inflation rates are obtained from...