ABSTRACT This study measures the power of accounting information to predict unexpected changes in earnings one year ahead. Ou (1990) and Ou & Penman (1989) have already carried out such experiment in the United States. Our study shows that Ou's (1990) methodology succeeds in predicting future earnings changes when applied to a substantially different context, namely that of Hong Kong. The Hong Kong environment is especially appropriate for observing the impact of more limited reporting practices and different economic conditions on the validity of empirical findings such as those of Ou's (1990).