Implications of Observed Properties of Daily Exchange Rate Movements
本文综述了汇率变动并非独立同分布的文献，指出均值和方差随时间变化，条件异方差是数据非线性的最佳刻画，并探讨了对国际经济学、金融学及计量统计分析的启示。
This paper has reviewed the current literature which shows that exchange rate movements are not independent and identically distributed. Instead means and variances appear to change over time. Conditional heteroskedasticity is perhaps the best characterization of the nonlinearity in the data. There are a number of implications. In the area of international economics, short term models for exchange rate must be developed to account for this nonlinearity. In the area of international finance, mean-variance portfolio selection rules and option pricing models are not applicable. In the area of econometric and statistical analysis, maximum likelihood estimation, which assumes that exchange rate changes are IID, can lead to misspecified models. Regressions of exchange rate changes should allow for heteroskedasticity in the residuals. Event studies must use statistical procedures that take into account heteroskedasticity, perhaps by boostrapping the empirical distribution of any test statistic.