SINCE and nondurable goods are consumed quickly but durable goods last for some years, many economists adjust the concept of durable goods spending in the National Income Accounts (NIA) to allow for the element of saving in expenditures for durable goods.' Validly, a distinction is thus made between current consumption and outlays which are partly consumption and partly saving. Unfortunately, as commonly estimated with fixed life and straight-line depreciation assumptions, a of services definition of durable consumption may lead to misleading theoretical conclusions and affect the accuracy of economic forecasts.2 For theory, this treatment of durable stocks implies improvident consumption of durables during cyclical downturns and affects the case for money illusion. In forecasting, use of this concept in the consumption block of the FRB-MIT-Penn econometric model causes biases around cyclical turning points. In the first section of this article, it is shown that higher prices induce additional saving in the context of a consumption function with NIA definitions. The moving weight personal consumption deflator from the NIA and the fixed weight consumer price index are compared to confirm the positive effect of higher prices on saving and to suggest that inflation may force substitutions among customary purchases. Supporting data from consumer surveys are cited to illustrate that the price influenced changes in family expenditure patterns are associated with downgrading in the quality of purchases. The addition of several measures of price anticipations to our consumption equation suggests that the amount of inflation anticipated by households has a negative impact on saving but that this effect is less important in explaining variations in the saving rate than the positive saving associated with a rising consumer price index. In the second section, the saving rate consistent with the flow of definition of durables is contrasted with the NIA saving rate at times of heightened inflation and cyclical downturns by an analysis of the movements of the two time series on a graph. Some implications of the choice of net financial assets or total household net worth in consumption functions are examined. Finally, the predictions of the FRB-MITPenn econometric model are evaluated and the results are compared with a well-known consumption function and with the model presented in the first section. The flow of definitions of consumption, income, and wealth are found to generate overpredictions of consumption around cyclical turns.

gives opportunity-cost/marginal-product meanings to the factor prices. Unfortunately, this is not good enough. If one wants to compute social costs of a project in terms of the economy's long-run equilibrium, then the project's physical factor requirements should also be computed in this framework, allowing for the factor-substitution possibilities which exist in the long run and which are instrumental in assuring positive marginal products for all the primary resources. To conclude, while we agree with HK that in the presence of unemployed resources money expenditures on a project overestimate social costs, we do not believe (for the reasons given in this note) that the HK model allows computation of the correct adjustments.10