Abstract
The trade forecast for Taiwan conducted by the Chung-Hua Institution for Economic Research (CIER), the Directorate-General of Budget, Accounting and Statistics (DGBAS), and the Institute of Economics, Academia Sinica (lEAS) has received considerable attention from decision makers in the private and public sectors. We evaluate the forecasting performance of the three institutions in terms of the conventional criteria and the usefulness tests recently developed by Lin et al. (2011). More specifically, we analyze the samples for the annual and quarterly projections released by CIER, DGBAS and IEAS from 1996 to 2010. Our findings are as follows. First, the directional accuracy statistics show that the one-year-ahead annual projections are generally well produced. Second, Ashley's usefulness statistics indicate that the current-quarter forecasts released by those institutions perform the best. In addition, based on the tests for usefulness (Lin et al., 2011), the annual forecasts have also done a good job. Overall, the current year forecasts (prepared in the middle of the same year) produced by DGBAS and the next-year forecasts (prepared at the end of each year) produced by IEAS perform the best. Meanwhile, the current-year forecasts for the changes in trade between Taiwan and specific countries produced by CIER also provide useful information.