1. 2 and 3.4 for the de?nition of the shrinkage based tests Sign-N, Max-MSPE-Adjusted and MSPE-Adjusted. See section 4.1 for a description of the benchmark tests MSPE-Normal;MSPE-McCracken and CCS. 2. Rejections at 1% (***), 5% (**) and 10% (*) level of signi?cance
2. Next, we report its standard deviation and ?nally we report the value of the normalized statistic to be compared with critical values. For instance, in the case of Canada and Sign -N test the value of the statistic without normalization is 15. Its standard deviation is 6.44 and the ratio between both is 2.33. 5. The last column displays the di�erence between the MSPE of the null model and the MSPE of the alternative model using di�erent shrinkage factors. 6. Data range;U K For Switzerland;For each country and each statistic we report three numbers. First, we report the value of the statistic without normalization,1975
3. Stock return predictability and model uncertainty;D Avramov;Journal of Financial Economics,2002
4. Sign Tests for Dependent Observations and Bounds for Path-Dependent Options;D Brown;Yale ICF Working Paper,2005
5. The Econometrics of Financial Markets