DOI
10.1080/07350015.2012.634340
Document Type
Article
Publication Date
1-2012
Publisher Statement
Copyright © 2012 American Statistical Association. Article first published online: 22 FEB 2012. DOI: 10.1080/07350015.2012.634340.
The definitive version is available at: http://amstat.tandfonline.com/doi/pdf/10.1080/07350015.2012.634340
Full citation:
Croushore, Dean. "Comment." Journal of Business & Economic Statistics 30, no. 1 (February 22, 2012): 17-20. doi:10.1080/07350015.2012.634340.
Recommended Citation
Croushore, Dean D., "Comment" (2012). Economics Faculty Publications. 22.
https://scholarship.richmond.edu/economics-faculty-publications/22
Comments
In the forecasting literature, researchers often seek to determine stylized facts, such as: are forecasts rational? But forecasts can be characterized in many dimensions and answering the question about whether forecasts are rational may require a multidimensional answer. I think about forecasts in three dimensions: (1) horizon, (2) sub-sample, and (3) vintage.
In their paper, “Forecast Rationality Tests Based on Multiple-Horizon Bounds,” Patton and Timmermann handle two of the three dimensions of forecast rationality tests: they look across alternative forecast horizons and they develop tests for which choosing an “actual” is not needed. They don’t, however, look at the sensitivity of their results to alternative sub-samples.