Volume 10, Issue Supplement

Economic experiments interact with economic theories in various ways. First of all they are used to test economic theories. However, they can neither confirm nor falsify them in a strict sense. They rather inform us about the range of applicability, the robustness and the predictive power of a theory. Furthermore, economic experiments discover and isolate phenomena and challenge economic theorists to explain them. Finally, many economic experiments are “material” models. They are used to analyze and predict how changes in the environment affect economic outcomes. However, they cannot offer an explanation for what we observe. This has to be provided by economic theory.

Economic engineering is the science of designing real-world institutions and mechanisms that align individual incentives and behavior with the underlying goals. Institutions matter because they affect incentives, and decision makers respond to incentives. Yet, they do not always do so rationally. Experimental economics complements economic theory by observing the performance of mechanisms in the context of actual decision processes faced by real people. It also answers questions that cannot be answered by theory and field data, tests hypotheses and identifies causalities suggested by theory and field observations, collects facts and phenomena that may stimulate behavioral theories of market design, eases cross-disciplinary cooperation, and communicates economic research to market participants, managers and other real-world decision makers. This article presents selected examples to illustrate how experimental economics may interplay with the more traditional economic toolbox to promote economic engineering both in research and in practice.

In this article we discuss the complementarity of laboratory and field data. Experiments offer highly controlled environments that allow precise testing and causal inferences. Survey and field data on the other hand provide information on large and representative samples of people interacting in their natural environment.We discuss several concrete examples how to combine lab and field data and how to exploit potential complementarities. One example describes an experiment, which is run with a representative sample to guarantee control and representativeness. The second example is based on the idea to experimentally validate survey instruments to ensure behavioral validity of instruments that can be used in existing panel data sets. The third example describes the possibility to use the lab to identify causal effects, which are tested in large data sets. Topics discussed in this article comprise the relation of cognitive skills (IQ) and risk and time preferences, determinants, prevalence and economic consequences of risk attitudes, selection into incentive schemes and the impact of unfair pay on stress.

How should we react to experiments that refute pure selfishness? Are deviations in action space only important if they imply considerable losses? For (un)bounded rationality it is unimportant which payoff declines lead to more or less adequate decision considerations. But also for unreflected actions a bad outcome need not necessarily arise if for example institutional frameworks exclude or restrict the exploitation of naïve agents. The comparison of rational and unreflected decisions is not reduced to action and payoff spaces, but extended to cognition space. The conceptual comparison is illustrated by results from laboratory experiments.