Jansen et al. (2016), who conduct a systematic comparison of the short-term forecasting abilities of 12 statistical models and professional forecasters, found forecasts by professional forecasters tend to perform better during periods of crises. However, there are few studies that use professional forecasters’ information. One of the few studies is Legerstee and Franses (2015), which examines the predictive power of disagreement among forecasters.
In this study, we combine professional forecasters information with single-equation approaches, such as mixed-data sampling (MIDAS) regressions and bridge equations. We use not only cross-sectional disagreement among forecasters, but also skewness and kurtosis measures in the ESP forecast survey, which is the first monthly survey of macroeconomic forecasts conducted by professional forecasters in Japan. We compare our results with those of Bragoli (2017), who was the first to propose a nowcasting framework for the country of Japan.
During the evaluation period (from 2005: Q1 to 2015: Q4), our results are similar to the forecast performance in Bragoli (2017), in spite of the different economic events that have characterized the Japanese economy during the period considered: low growth in the first period, the Great Recession in 2009, and the Great Earthquake in 2011.