Outlier-robust evaluation of fixed-event macroeconomic survey expectations

Evaluating macroeconomic forecasts for their unbiasedness and efficiency is essential for policymakers, economists, and investors. The degree to which these stakeholders incorporate expectations into their decision-making processes depends heavily on how these forecasts have been formed. Existing methodologies do not explicitly address critical dimensions, such as the variability of bias across target events and forecast horizons, the forecast errors’ heteroscedasticity, and the potential state-dependence in bias. More importantly, they encounter difficulties during high-uncertainty periods, which can lead to inaccurate inference due to the presence of outliers. Apart from generalising the unbiasedness tests, this study contributes to the literature on both strong and weak efficiency by incorporating these aspects. Finally, the proposed methods are applied to the expectations of a crucial survey of the US economy, namely, the Survey of Primary Dealers (SPD). The findings from this application indicated that interested parties should investigate unbiasedness and efficiency in an outlier-robust way, while also allowing for greater flexibility in the methods regarding the variables and periods examined.