main-content

## Über dieses Buch

This book describes a series of laboratory experiments (with a total of 167 independent subjects) on forecasting behavior. In all experiments, the time series to be forecasted was generated by an abstract econometric model involving two or three artificial exogenous variables. This designprovides an optimal background for rational expectations and least-squares learning. As expected, these hypotheses do not explain observed forecasting behavior satisfactorily. Some phenomena related to this lack of rationality are studied: Concentration on changes rather than levels,underestimation of changes and overvaluation of volatile exogenous variables. Some learning behavior is observed. Finally, some aspects of individual forecasts such as prominence of "round" number, dispersion, etc.,are studied.

## Inhaltsverzeichnis

### 0. Literature Review

Abstract
Generally, studies on expectation formation/predictive behavior have the following pattern: There is a time series describing the development of a certain economic variable (endogenous variable). At the beginning of each period, the forecasters have to predict the next realization of this variable, using some information which includes the past history of the variable. Thus, the subjects already know something about the pattern of the time series when they have to give their first forecast.
Gunnar Brennscheidt

### 1. Rationality in Presence of Exogenous Variables

Abstract
Consider linear models of the form
$$p_t = \alpha_0 + r(t)'\alpha + u_t ,\,\,t = 1,...,T$$
(1.1)
or
$$p_t = \alpha _0 + r(t - 1)'\alpha + u_t ,\,t = 1,...,T$$
(1.2)
where pis the endogenous variable, r(t) is a vector of exogenous variables, αis a parameter vector of appropriate dimension, and uis stochastically independent white noise.
Gunnar Brennscheidt

### 2. Experimental Design and Notation

Abstract
The experiments took place at the computer terminals of the laboratory for experimental economics at the University of Bonn. 14 experiments were run, each experiment with about 12 subjects. In total, there were 167 subjects. The subjects were not allowed to participate in more than one experiment, and they were independent of one another.
Gunnar Brennscheidt

### 3. Aggregate Forecasts

Abstract
There are two concepts of aggregating individual forecasts: taking averagesand medians. For each experiment, the time series of average forecasts, f A , consists of the arithmetic means from individual forecasts1for each period:
$$f_{t,A} = \frac{1}{N}\sum\limits_{i = 1}^N {f_{t,i} \,,} \,t = 1,...,T$$
(3.1)
Analogously, the time series f M of median forecastsis obtained as
$$f_{t,M} = \text{Median}[f_{t,1} ,...,f_{t,N} ],\,t = 1,...T$$
(3.2)
For each experiment, one time series of average forecasts and one series of median forecasts result, i.e. 28 series from all 14 experiments. These series are analyzed in this chapter.
Gunnar Brennscheidt

### 4. Sum Results on Individual Forecasts

Abstract
This chapter presents some results which lend support to the statements on aggregate forecasts in Chapter 3: Comparison of observed forecasts to various prediction concepts, non-rational features, learning, and “improvements” of underestimating predictions (regression on single exogenous variables and psychological components).
Gunnar Brennscheidt

### 5. Features of Individual Forecasts

Abstract
This chapter analyzes some features of individual forecasts which cannot be discussed in context with aggregate forecasts. In most sections of this chapter, the analysis is subdivided according to three groups which we already know from Sections 3.6 and 4.4 on influence of single exogenous variables:
• Simple full-information experiments (Exp. 1–6 and 9),
• Seasonality experiments (Exp. 7 and 8),
• Incomplete-information experiments (Exp. 10–14).
Gunnar Brennscheidt

### 6. Conclusion

Abstract
This chapter briefly summarizes the experimental results presented in Ch. 3–5.
Gunnar Brennscheidt

### 7. Appendix

Without Abstract
Gunnar Brennscheidt

### 8. References

Without Abstract
Gunnar Brennscheidt

### Backmatter

Weitere Informationen