Foreign currency translation adjustments as predictors of earning changes

Bernard (1993) contends little progress has been made in identifying nonearnings accounting numbers which can predict future changes in income. As a test of the value relevance of foreign currency translation adjustments, this study links year-over-year changes in earnings per share to changes in the value of the cumulative translation adjustment account. The empirical tests are conducted on a sample of 204 U.S. multinational firms for the time period 1991-1996. The sample is derived from firms with subsidiaries in Mexico and Germany. In order to account for the heterogeneity of exchange rate exposure, the sample firms come from a wide range of industrial groupings, have varying sizes and different levels of capital intensity. A panel data set and weighted-least-squares (WLS) are employed to obtain parameter estimates. A major finding of this research is that the lagged value of per-share foreign currency translation adjustments can be used to predict year-over-year changes in earnings per share. The results of the current study also indicate that predictive ability is enhanced when firm specific interaction terms, which control for size, industry, location of direct foreign investment and capital intensity, are included in the estimating equation.