01562naa a2200193 a 450000100080000000500110000800800410001910000170006024500390007725000100011626000090012630000140013552010720014970000160122170000210123770000180125870000150127677300770129116476471996-06-26       1988    bl ---       0--  u    #d1 aJUDGE, G. G.  aSets of linear statistical models.  a2. ed  c1988  ap.443-496  aIn most of the earlier chapters we concentrated on estimation and hypothesis testing for the parameter vector B in the single equation model y=XB + e. Different estimation methods and tests were considered, depeding on what further assumptions were made about the disturbance convariance matrix E[ee]. In this Chapter we turn to a situation where there is more than one equation to estimate. For example, interest might center on demand equations for a number of commodities, investment functions a number of firms, or consumption functions for subsets of the population. Suppose that time-series data suitable for estimation of a number of demand equations are available. The disturbances in these different equations at a given time are likely to reflect some common unmeasurable or omitted factors, and hence could be correlated. The same is also likely to be true when time-series observations are used to estimate different investment functions for different firms, different consumption functions for different subsets of the population, or many other examples.1 aHILL, R. C.1 aGRIFFITHS, W. E.1 aLUTKEPOHL, H.1 aLEE, T.-C.  tIntroduction to the theory and Practice of Econometrics. New York: 1988.