JBNST - Vol. 237/1 - 2017

Existing theory suggests that three “order effects” may emerge in multi-stage tournaments with information feedback. First, participants adjust effort across stages, which could advantage the leading participant who faces a larger “effective prize” after an initial victory (leading-effect). Second, leading participants might engage in sabotage activities to protect their lead thereby decreasing the rivals’ output. Finally, participants lagging behind may increase risk at the final stage as they have “nothing to lose” (risk-taking). The expected order effects based on existing theory cannot be supported empirically in a natural experiment setting, where professional teams compete in a two-stage tournament with asymmetric initial conditions and clear incentives.

The Great Recession in Germany was noticed only seven months after its onset. This study examines whether the available data could have helped to predict or identify the crisis in real time. After assessing the accuracy of previous recession forecasts, we examine that of forecasts published from April to December 2008 by twelve major national and international forecast institutions and confront them with real-time data from official statistics, major surveys, and indicators. While annual forecasts for 2008 were unusually accurate due to errors of semi-annual forecasts offsetting each other, forecasters failed to observe the onset of the recession in Q2 2008, although from May onward an increasing amount of data indicated that the economy was in recession or was likely about to enter one. Though the data were neither ambiguous nor misleading, forecasters recognised the onset of the recession as late as mid-November, but also failed to warn of a coming recession. The most convincing explanations for these failures to recognize the crisis in time appear to have been the ‘truth effect’ and forecasters’ ‘low priors about the likelihood of a recession’.