International financial market integration: Empirical analysis with data from forward and futures currency markets.

摘要

Chapter 1 of this dissertation uses a rich data set to assess the degree of integration of 25 countries in world financial markets. Deviations from covered interest parity, uncovered interest parity, and real interest parity are examined. It is concluded that covered interest parity is, empirically, the most sensible definition of capital mobility. Deviations from uncovered interest parity are often relatively large among those countries with financial markets expected to be most open. Real interest parity is not a measure of capital market integration because it requires ex ante purchasing power parity. However, the correlation between real interest rates of the 25 economies provides information about linkages between policy makers in different countries. Strong linkages are found among the countries of the European Monetary System, while linkages are weaker among the rest of the countries.; Chapters 2 and 3 add new data sources to extend the research on the unbiasedness of the forward currency price in its prediction of actual depreciation. Chapter 2 presents tests of forward discount bias for the 25 country sample. Using the forward rate and lagged inflation as regressors, unbiasedness is rejected for 20 countries. In most cases where inflation provides significant information predicting the forward discount errors, the forward discount seems to be underestimating the effect of inflation on exchange rates. The magnitude of the coefficients on the forward discount is discussed, showing that it is difficult to interpret the rejections solely as evidence of a time-varying risk premium.; In Chapter 3 currency futures prices are used to test unbiasedness at 63 different horizons from 3 to 89 days. Futures rate unbiasedness is rejected for at least 17 horizons in each of the five currencies tested, with rejections occurring more frequently at long horizons. In pooled regressions the null is rejected at nearly every horizon greater than 6 days. Very short horizons again provide less evidence of futures rate bias. Since theoretical models suggest the risk premium should be independent of the horizon, the failure to reject at shorter horizons suggests either a changing process as contract expiration nears or systematic expectational errors at longer horizons.