The response of manufacturers' inventory holdings to changes in the volume of sales and the backlog of unfilled orders is examined on a quarterly basis for the period 1948-55 within a buffer-stock flexible accelerator framework. The hypothesis that manufacturers successfully hedge against increases in the price of purchased materials, enlarging their stocks in advance of actual price increases, is rejected. By introducing explicitly the impact of prediction errors it is possible to infer that manufacturers tend to underestimate actual changes in sales volume, but by a surprisingly small amount. An analysis of discrepancies between desired and actual inventory holdings reveals that manufacturers tolerated sizable deficiencies in stocks throughout the Korean conflict.

A multi-sector buffer-stock inventory model is developed in an attempt to resolve the problem of aggregation involved in deriving implications for the stability of the economy from a consideration of inventory practices of individual firms. It is demonstrated that stability depends upon a multitude of parameters, some of which are suppressed in aggregative model construction. The economy is necessarily unstable when perfect, if myopic expectations are assumed. With naive expectations stability becomes a definite possibility, particularly if firms attempt only a delayed adjustment of inventories to the equilibrium level. Although the empirical evidence marshaled in order to illustrate the application of the theorems does not prove sufficiently accurate to permit precise conclusions, it is apparent that the conditions for stability may well be satisfied for reasonable values of the system's parameters. Tax schemes which have been suggested as means of stabilizing fluctuations in inventory investment are appraised in the concluding section.