Statistical approximation to ex ante variables

Abstract

In the previous chapters we questioned the conventional assumption that observed transactions generally belong to demand and supply schedules. Other authors have expressed similar doubts in recent years,¹ but the number of empirical studies dealing with disequilibrium transactions explicitly is still relatively small. All the studies of others we know of² are based on the minimum—trade assumption, implying that observations belong either to the demand or the supply schedule. In this view, transactions are completely determined by demand factors in some periods, and completely by supply factors in others. Time—series regressions of transactions on these factors have to switch from one set of factors to the other from time to time. Usually, the statistical information about the adequate regime is deficient, and some criterion has to be developed for the pooling of observations. The most elementary pooling criterion is the maximum likelihood to be attained by alternative classifications of observations. Somewhat more advanced techniques take other information into account. The use of the direction of the price change as an indicator of excess demand or excess supply in the pioneering study of Fair and Jaffee (1972) on the mortgage market is an example of the use of extra information. The principle of the exploitation of additional information seems to deserve support, but the integration of this information in the description of market—phenomena is deficient in the present case. The integrated treatment of price changes in Rosen and Quandt (1977) is more satisfactory.