I. Introduction, 20. — II. A linear decision rule, 22; decision rule for static uncertainty, 23; decision rule for dynamic uncertainty, 25; plans, 28. — III. Simulation tests of the dynamic decision rule under uncertainty, 29; the policy of inaction, 30; decision rule stabilization with poor forecasts, 32; decision rule stabilization with perfect forecasts, 33; performance comparisons, 34; errors in estimating the economic relations and the welfare function, 38. — IV. Economic growth, 39. — V. Stabilization of an inherently unstable economy, 42. — VI. In conclusion, 43.

The work on the Phillips curve has been predominantly empirical, but policy intervention designed to decrease both inflation and unemployment requires a better theoretical understanding of the determinants of this relation. This paper attempts to sketch out the basic labor market relationships that appear to the author to account largely for the Phillips relation2 and then to consider the kinds of policy measures that this analysis suggests are relevant for moving the Phillips curve. The basic issue involved in the stability or instability of the Phillips curve when the inflation rate is constant is explored with a simple model in the appendix. The emphasis in this paper is on the atomistic operation of the labor market because it appears sufficient to generate the Phillips relation, but we do not deny that union bargaining and price dynamics may also be involved to some degree. This paper concentrates on conceptual issues. Continuing research will be devoted to the statistical measurement of relationships.

The current inability of economists to prescribe policies for achieving full employment without inflation can be traced to the unresolved schism between macro and micro analysis. There are calls for structural reforms that will impact at the micro level, but we still don't know enough about the microdynamics of the economy to specify effective programs. We need to identify the critical parameters at the micro level which account for frictional and structural unemployment and the bias toward inflation. This paper addresses one part of this problem by proposing an approach to explaining the dynamic and static structure of wages and unemployment in a segmented market, drawing policy implications and contrasting them with present programs.

The comment by John Barron constitutes a very interesting extension of earlier work rather than a correction of it. He introduces the additional dimension of searching firms to find vacancies. The earlier work had concentrated on the searching of vacancies to find job offers. Clearly both are relevant and, not surprisingly, he finds that when different assumptions are made, different conclusions follow. He introduces the concept of mean (firm) search time T to find a particular firm with a vacancy in a given set of firms. Assuming random search, this time depends on the number of firms in the set. Thus, he finds that when a labor market is divided into N equal compartments, mean search time in each compartment is reduced by a factor of (1/N). The earlier work to which he refers implicitly assumed that the vacancies in a compartment could be located fairly readily. The hiring firms could be easily located but time consuming search was required to determine the particular vacancies which would produce job placements for particular workers. Thus, the relevant search time measure is the mean (vacancy) search time to locate a placement from the set of vacancies in the