← Back to Search
A Note On Regime Classification In Disequilibrium Models
Published 1980 · Economics
Reduce the time it takes to create your bibliography by a factor of 10 by using the world’s favourite reference manager
Time to take this seriously.
Recent applications of the disequilibrium econometric models of Fair-Jaffee (1972), (Amemiya (1974) and Maddala-Nelson (1974) give further details on maximum likelihood estimation), have reported estimate probabilities that each sample observation is from one regime or the other. Laffont and Garcia (1974) in their analysis of the market for business loans and Portes and Winter (1980) in their model of the consumption sector in centrally planned economies report for each sample point the probability that the point was from one of the regimes. Rosen and Quandt (1978) also try to identify periods of excess demand in a disequilibrium model of the US labour market. The purpose of this note is to point out that the classification method used to date, conditional on the estimated parameters, is not efficient and that the probabilities obtained by this method are not the appropriate probabilities to report. Looking at the disequilibrium model as giving a joint distribution of the dependent variable q and a binary variable D indicating the regime (see below), shows that reported probabilities have been based on the marginal distribution of D, which is appropriate for prediction rather than the distribution conditional on q, which is appropriate for classification. The latter takes a particularly simple form.