The Disequilibrium Model in a Controlled Economy: Comment
评论David Howard对Barro-Grossman非均衡模型在苏联背景下的检验，指出其方法缺陷，特别是用零售销售额衡量商品供应量不合理，导致储蓄与消费的负相关关系被误读。
In a recent paper in this Review, David Howard concludes that his test of the predictive ability of the Barro-Grossman disequilibrium model is successful within the Soviet context. The Barro-Grossman (B-G) model (1971, 1974) focuses on the responses of saving and labor supply to conditions of excess demand, defined as the case where at the prevailing price, the demand for consumer goods exceeds their supply. The B-G model predicts that under conditions of excess demand, referred to also as repressed inflation, an increase (decrease) in the quantity of goods available leads to a decrease (increase) in saving and an increase (decrease) in labor supply, with its consequent multiplier effect on output. Howard tailors the B-G model to the Soviet economy by including an uncontrolled market, specifically the collective farm market, and by eliminating the role of profits as an argument in the effective saving and labor supply functions. While the B-G model supplies an important theoretical framework for predicting responses to changes in the constrained availability of goods, the model is difficult to test empirically. Moreover the methodology used by Howard is flawed in several ways. An adequate measure of the supply of goods available on the constrained market (defined by Howard as B?, goods available on controlled market) is needed to test the B-G model. However, Howard's choice for B?, a composite of goods sold on the state and cooperative retail markets, is not such a measure. The significance of introducing a variable to measure availabilities is to capture the notion that what is purchased under conditions of excess demand will not reflect desired purchases, but rather actual purchases, thereby revealing a point of market disequilibrium. Despite the argument that sales seem a good proxy for availabilities because quantities are so limited that whatever is available will be purchased, the use of sales to represent availabilities is unjustified for two reasons. The first reason focuses on the relationship between two of the most important variables in Howard's analysis: B?, state and cooperative retail sales, and sd, the change in savings bank deposits, which can simply be termed saving in the Soviet context where the purchase of interest-earning assets (or even lottery bonds) and the other usual alternatives to savings bank deposits are negligible. Certainly the change in state and cooperative retail sales (AB') and the change in saving (As') must be highly negatively correlated. This can be easily seen in the context of the Howard model by constructing the ratios ABO/A(wLs) and ASdl/A(wLs) where w is the wage rate, Ls is the average nonprivate civilian employment, and the product wLs is the wage bill. Assuming the wage bill is approximately equal to income, these ratios are approximately the marginal propensity to consume and the marginal propensity to save, which must sum to unity, neglecting the fraction of the change in total income spent on the uncontrolled market and assuming other forms of saving to be zero. Noting this obviously negative relationship between ASd and ABIR, the fact that Howard's estimations yield coefficients for aSd/aBO that he claims are both correct and significant with respect to one-tailed t-tests (see his Table 2: OSd OBR = -.424641 with an absolute t-statistic of 2.292 and his Table 3: dSd/0RO = -.42054 with an absolute t-statistic of 2.158) is uninformative except when viewed from a somewhat different perspective; that is, the fact that the coefficients are only so weakly signif*Assistant professor of economics, Graduate School of Business Administration, New York University. This work was funded in part by a New York University faculty research grant.