Modelling Individual Expectation Formation

The practice of forecasting is dominated by judgemental techniques. Due to this practical relevance the abilities of individuals to predict time series were explored in numerous experimental studies. Researchers want to find out which circumstances influence or improve the human forecasting performance and when statistical models are superior. Despite this increasing interest in describing expectation formation there are hardly any mathematical models explaining forecasting behaviour. In the literature there are explanation models that are based on unrealistic assumptions or which assume overly simple behaviour patterns. The bounds & likelihood (b&l) heuristic, BECKER ET AL. (in press), is a simple but very effective procedure that models collective forecasts of individuals very well. It was demonstrated that the model works in a variety of situations, for instance when the subjects are provided with a multitude of time series (leading indicators) for the forecast of a time series. For individual behaviour, two explanatory approaches were developed in the preceding FWF project: The similarity analysis and the scheme theory. The following questions are of particular interest: - Can the b&l-heuristic be extended to more realistic information environments in which coincidental indicators instead of leading indicators are available? - Can the scheme theory be extended to explain forecasting behaviour involving several sources of information and more complex time series as it has been successfully done for the b&l-heuristic? - How do subjects behave when they are repeatedly confronted with similar or even identical forecasting situations? Does their intrasubjective performance differ significantly? Is forecasting behaviour stable enough to be explained by one model? Our objective is to find answers to these questions by conducting new experiments and linking the results to our database of prior experiments.

The practice of forecasting is dominated by judgemental techniques. Due to this practical relevance the abilities of individuals to predict time series were explored in numerous experimental studies. Researchers want to find out which circumstances influence or improve the human forecasting performance and when statistical models are superior. Despite this increasing interest in describing expectation formation there are hardly any mathematical models explaining forecasting behaviour. In the literature there are explanation models that are based on unrealistic assumptions or which assume overly simple behaviour patterns. The bounds & likelihood (b&l) heuristic, BECKER ET AL. (in press), is a simple but very effective procedure that models collective forecasts of individuals very well. It was demonstrated that the model works in a variety of situations, for instance when the subjects are provided with a multitude of time series (leading indicators) for the forecast of a time series. For individual behaviour, two explanatory approaches were developed in the preceding FWF project: The similarity analysis and the scheme theory. The following questions are of particular interest: Can the b&l-heuristic be extended to more realistic information environments in which coincidental indicators instead of leading indicators are available? Can the scheme theory be extended to explain forecasting behaviour involving several sources of information and more complex time series as it has been successfully done for the b&l-heuristic? How do subjects behave when they are repeatedly confronted with similar or even identical forecasting situations? Does their intrasubjective performance differ significantly? Is forecasting behaviour stable enough to be explained by one model? Our objective is to find answers to these questions by conducting new experiments and linking the results to our database of prior experiments.