Test for normality in the econometric disequilibrium markets model (Q788454): Difference between revisions

The econometric model for a market in disequilibrium is in the form of demand \((D_ t)\) and supply \((S_ t)\) equations \(D_ t=x^ T_{1t}\beta_ 1+v_{it}\), \(S_ t=x^ T_{2t}\beta_ 2+v_{2t}\), \(Q_ t=\min(D_ t,S_ t)\), where the actual quantity \(Q_ t\) transacted is the minimum of demand and supply. For such disequilibrium models it is known that if the disturbances \(v_{1t}\), \(v_{2t}\) are not normal but misspecified to be so, then the maximum likelihood estimates are not only asymptotically inefficient but also may be inconsistent. Hence a test procedure for normality is developed on the basis of the efficient score test of \textit{C. R. Rao} [see ''Linear statistical inference and its applications. 2nd ed. (1973; Zbl 0256.62002)]. This test is based on some measures of cumulants of third and fourth orders of the disturbances constructed from the samples. The test is then empirically applied to test the bivariate normality hypothesis in the housing starts disequilibrium model, where the normality assumption was strongly rejected.