Knowledge that Transforms

Forecasters cannot yet rely on indicators alone, for the reason that (as in periods like 1952 and 1962) they may give misleading signals; in expansions the indicators often turn down too many months before the true peak; in recessions, they turn up only a month or two before the upturn -by the time the data become available for smoothing, their main (important!) purpose is to tell us we have indeed already turned the corner. Economists look forward with interest to further improvement in these techniques...

R ECENT labor market analysis has been conducted by two rather distinct groups of economists. Each seems to have neglected the contribution of the other. Solow, Minasian, and Ferguson have explored extensively the impact of regional wage differentials on factor proportions.' Others such as Dunlop, Ross, Kerr, and Thompson have explored wage decision-making in various labor markets with little treatment of the employment effect.2 Empirical evidence will be presented here to support the arguments advanced by each group of labor market analysts. In addition, a possibility of combining the contribution of these groups into a single framework within which to study the impact and behavior of wage differentials will be suggested.

IN order to evaluate alternative stabilization policies, it is necessary to-ascertain the properties of the system into which these are to be introduced. Since this is very difficult in economics, the usual practice is to postulate simple systems which are amenable to analysis by general methods, and then to consider the impact of various stabilization on these systems. Such an approach led Baumol, for example, to conclude that policies automatic or not which appear to be properly designed may very well turn out to aggravate fluctuations [2, p. 21]. This somewhat pessimistic conclusion was suggested by an analysis of the transient response of a deterministic linear system. The question naturally arises whether results similar to those derived by Baumol also hold for stochastic systems. The purpose of this paper is to extend the tools and results derived by Baumol to a linear stochastic system. Since attention will be focussed on the stochastic response of the system, this paper can be viewed as an elaboration of some of the problems discussed by Friedman [5] in connection with stabilization policy. In the next section, some of the properties of stochastic linear systems and linear stabilization are described. Methods which may be used to evaluate alternative stabilization are considered in section III. The paper concludes with several comments on the implications of minimum-variance stabilization policies.