A positive approach of Prediction and Explanation in Economics

Summing up the argument of the preceding sections, we find that the difficulties encountered in economics flow from the fact that hypotheses cannot be strictly disconfirmed. As long as it is logically impossible to disconfirm a hypothesis, the hypothesis remains in doubt no matter how much confirmatory evidence is accumulated in its support. This does not mean that such a hypothesis may not be used nor that it may not command some degree of confidence. The point is not that intended empirical laws may not be used nor command confidence; it is rather that no evidence need be recognized as disconfirming them.

Our purpose in this section is to recommend certain approaches through which we may develop hypotheses that can be disconfirmed. While we would not argue that there is a unique methodology, we do maintain that our recommendation can lead to the discovery of general laws and hence to models with predictive and explanatory power. Essentially, we argue that economics should be viewed as part of the study of human behavior and that the empirical laws of economics should be based on propositions of behavior that can be disconfirmed independently from the economic situations. In a basic sense our position is consistent with J. N. Keynes’ assertion that economics presupposes psychology. However, we will elaborate our position and show the operational significance for research.

Perhaps the issue can best be joined by referring again to the FriedmanSavage billiard player. Friedman proposes a theory based on the assumption that the billiard player makes his shots “as if he knew the complicated mathematical formulas that would give the optimum directions of travel, could estimate accurately by eye the angles, etc., describing the location of the balls, could make lightning calculations from the formulas, and could then make the balls travel in the direction indicated by the formulas.” He then goes on to argue that quite obviously the confidence in such a hypothesis is not based on any actual belief that a billiard player would have such abilities, but rather from the belief that “unless in some way or other they were capable of reaching essentially the same results they would not in act be expert billiard players.” Further, Friedman, at least by implication, excludes the possibility of utilizing empirical methods to develop a more detailed theory of behavior in this case. He argues that if one asked the billiard player how he decides where to hit the ball, the billiard player might say that he “just figures it out” but then might also rub a rabbit foot to make sure. Friedman is here taking the position that it is impossible to find out from the decision maker himself the process by which the decision is made. This means that it is impossible to gain any insight into the actual processes of decision making in the two primary micro-units of economics, the firm and the household. If such an inference were correct, economists would be forever condemned to making counterfactual hypotheses of the billiard-player type.

In fact, however, such a limited view of our potential knowledge of decision-making behavior is not justified. A variety of decision-making behavior has been studied by a number of methods — including interviewing, observation, and analysis of  the records of past decisions (see Chapters 7 and 10). Models that predict and explain decisions have been developed in a number of areas. ¹⁶ The trend of the evidence leads to the view that any decision can be successfully subjected to scientific analysis.If we look at the billiard player as an example, the approach we suggest would be to study his decision-making process, build a model of the process, and test the model by making predictions and comparing the predictions with the actual result. In this way we can not only join our knowledge with that of other disciplines studying similar behavior, but we will gain explanatory value for our models as well as predictive ability.One of the methodological breakthroughs that  makes complicated models of a decision-making behavior feasible is the digital computer (see Appendix B). The computer model written in computer language makes it possible to build “positive” models of processes that were previously too complicated to be viewed as legitimate units of study.Thus, one program of research that would lead out of the difficulties described in earlier sections involves the following suggestions:

1. There should be a concentration of empirical study of the microdecision-making units — firms and households
2. The empirical study should be directed toward determining the actual decision- making processes involved, which requires the use of such techniques as interviews, observations, and analyses of records.
3. The models formed as a result of such study should be decisionprocess models. They should show the complete map of alternatives under the variety of conditions of the real world. Thus we will be able to eliminate the usual ceteris paribus condition.
4. The language of the computer should probably be used because of the complex and dynamic character of such models.

In short, we are striving for a science of economics that is based on behavioral propositions relating to the micro-units. These propositions would relate to the decision-making behavior of the micro-units and would be disconfirmable either by direct behavioral tests or by inference from economic tests. The propositions would no longer refer to phenomena observable only as unique events but would refer to repetitive situations.

Source: Skyttner Lars (2006), General Systems Theory: Problems, Perspectives, Practice, Wspc, 2nd Edition.