مقاله سال ۲۰۱۶

Given the available information, conventional theory suggests that the optimal forecast, in the sense of minimized expected mean squared forecast errors (MSFE), must be the ‘‘rational expectation’’ of the target variable. The optimal forecast is unbiased because the induced forecast error has a zero mean, and efficient because the error is unpredictable and orthogonal to the components of the given information set when the forecast is made.

There are two conventional types of regression models proposed in the literature for investigating whether aforecast is optimal. One focuses on the relationship between realized and forecasted values directly (e.g., Mincer & Zarnowitz, 1969), while the other turns to the properties of the forecast errors (e.g., Holden & Peel, 1990 and Nordhaus, 1987). Note that the parameters of these two types of models are assumed (implicitly) to be constant over time, and thus the implied bias (non-zero mean of the forecast error) of a forecast is time-invariant; see e.g. Artis and Marcellino (2001), Gavin and Mandal (2003), Holden and Peel (1990), Loungani(2001), Mincer and Zarnowitz (1969), and Nordhaus (1987).

However, many studies analyzing survey data have shown that the forecast performance can be correlated strongly with the underlying economic conditions; see for example Chauvet and Guo (2003), Döpke (2001), Fildes and Stekler (2002), Grunberg and Modigliani (1954), Messina, Sinclair, and Stekler (2015), Sinclair, Joutz, and Stekler(2010), Sinclair, Stekler, and Carnow (2015), and Swanson and van Dijk (2006). This empirical evidence reveals that the bias of a forecast could be time-varying, and may depend on the underlying economic state and its duration.