Beyond the Norm

US federal agency macroeconomic forecasts are generally made in a manner consistent with the unbiasedness-forecast rationality (UFR) hypothesis. While this inference may be valid under normal policy conditions, cognitive biases can make ex ante policy analysis more difficult when macroeconomic circumstances deviate from the norm. I test whether policy forecasting under abnormal macroeconomic conditions generates predictions that exhibit cognitive biases (uncertainty-induced bias). Moreover, I investigate whether government agencies will either overstate (availability bias)or understate (mean anchoring bias) true conditions during both good and bad times. These three behavioral hypotheses are examined by performing simultaneous quantile regression analysis on a pooled dataset on US macroeconomic forecasts by the Office of Management and Budget (OMB), Congressional Budget Office (CBO), and Social Security Administration (SSA) for the 1976- 2002 period. The statistical evidence provides unequivocal support for the uncertainty bias hypothesis as well as strong support for the availability bias hypothesis. Specifically, deviations from frequently occurring (normal) policy conditions are positively related to cognitive bias involving ex ante policy decisions, and such biases tend to overstate abnormal policy conditions. These empirical findings highlight the power of cognitive psychology for explaining departures from classical rationality with respect to government agency forecasting under atypical macroeconomic conditions.