1. 36 In Table 3, the statistics denoted "MSE-f" and "MSEt" test for equal mean squared error of the restricted and unrestricted forecast series. The latter is in the spirit of the regression-based test for equal mean squared error (MSE) proposed by;Clark Tests Of;incorporate and further develop the forecast accuracy tests of Granger and Newbold (1976) and Diebold and Mariano,1976
2. As noted above, the null hypothesis is that of "equal accuracy" for the two MSE tests and "forecast encompassing" for the ENC tests, and, as discussed in Clark and McCracken (2001), these tests are all one-sided, with the only rejection regions of interest under the alternative hypothesis residing in the right-hand tail. The results of the tests in Tables 3 and 4 strongly reject both the null hypotheses of equal forecast accuracy and;draws on Diebold and Mariano (1995), while the ENC-f test owes its origins to Clark and McCracken,1998
3. Quarterly Real GDP Estimates for China and Asean4 with a Forecast Evaluation;T References Abeysinghe;Journal of Forecasting,2004
4. Factor Forecasts for the UK;M Artis;Journal of Forecasting,2001