Endogenous and Exogenous Investment in Macro-Economic Models

Abstract

M OST macro-economic models developed in recent years are expressed in terms of two principal types of variables: endogenous and exogenous.1 The latter may be defined as variables which influence the endogenous variables but which are not themselves affected by the endogenous variables.2 In other words the exogenous variables are considered to be outside the model. Endogenous variables, on the other hand, are determined by the model. The extent to which exogenous variables are used in a model depends to a considerable degree on the ultimate uses to which the system is to be put and on the ambitiousness and capacity of the model builder and the tools he has at hand. Conceivably a comprehensive system would have relatively few exogenous variables; perhaps only noneconomic phenomena such as temperature, rainfall, earthquakes, and other acts of God would be considered exogenous. This would indeed be an all-inclusive model. At the other extreme are systems having large blocks of economic activity represented as exogenous. For example in several Keynesian models all investment is regarded as exogenous.3 Other models, to varying degrees, fall between these two extremes. Examples may be found in the works of Samuelson,4 Hicks,5 Klein,6 and others. More will be said about these later. It would undoubtedly be desirable to have an all-inclusive model of the type mentioned in the preceding paragraph. However, the implications of such a system must be clearly recognized. What is involved is scarcely less than a complete theory of society, embracing the fields of economics, sociology, psychology, politics, and so forth. This is clearly a big order, and there have been relatively few major efforts to develop such a system, the Marxian theory probably being the most ambitious attempt. To construct this sort of model in some theoretical sense is difficult enough; but then to attempt to treat it statistically brings up further obstacles, especially in view of the limitations imposed by the analytical tools and basic data currently available.7 Apparently, whether we like it or not, we shall have to be satisfied for the time being with models which are considerably less ambitious than the grandiose scheme of things mentioned above. This means that fairly large areas in the system must be specified as exogenous. Thus the question is not so much whether there are to be large exogenous areas in macro-economic models, but rather whether these areas are chosen judiciously and so as not to give a markedly distorted view of the reality which the system is supposed to describe. The purpose of this paper is to discuss the endogenous-exogenous problem briefly with respect to one part of macro-economic models the investment part of the system. In the opinion of the present author this is an area which has not been handled satisfactorily in many of the models formulated in recent years. Basically the problem is that of determining which types of investment should be considered endogenous, which types should be considered ex*The author is indebted to F. S. Hoffman and R. N. McKean of the Rand Corporation for helpful comments, criticisms, and suggestions. The views expressed in this paper are personal and do not necessarily reflect those of the Rand Corporation. 'In stochastic models a third type of variable is found: the non-observable random variable or the disturbances in the system. We shall not be concerned with random variables at the moment. 2T. C. Koopmans, Statistical Inference in Dynamic Economic Models (New York, I950), p. 56. 'E.g., Lawrence R. Klein, Economic Fluctuations in the United States, 1921-194I (New York, I950), p. 7. 'Paul A. Samuelson, Interactions Between the Multiplier Analysis and the Principle of Acceleration, this REVIEW, XXI (May I939), 75-78; reprinted in Readings in Business Cycle Theory (Philadelphia, I944), pp. 26I-69. 'J. R. Hicks, A Contribution to the Theory of the Trade Cycle (Oxford, I950). 'Op. cit., pp. 84-II4. Klein's Model i is undoubtedly the most comprehensive dynamic system we have to date. To the knowledge of the present author, the only other model which even remotely compares with Klein's is the one presented by Tinbergen in I939. See J. Tinbergen, Statistical Testing of Business-Cycle Theories: A Method and Its Application to Investment Activity (Geneva, I939). 7This is not to suggest that a model, to be useful, must necessarily be set forth in econometric terms.