Answers are grounded in this paper and cite their sources.
This paper discusses the shortcomings of the root-mean-sqare error as a measure of forecast accuracy. As an alternative, it is proposed that when forecasts over different horizons are being evaluated, the errors in different periods should be discounted. the discount rate might be positive when current information improved the accuracy of forecasts for the near future, and could be negative when the economic model includes long-run or equilibrium relationships which will eventually dominate short-run disruptions. An empirical example of annual forecasts for the UK is presented and the rankings of different forecasting organizations are shown to vary with the discount rate.
No takes yet. Share an insight, caveat, or question.
Kenneth Holden (1990) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: