The following observations are relevant: (a) Since the sum of the squares of the residuals can be expressed as a quad- ratic form in the true errorslB the ratio R is homogeneous of degree zero in the true error terms; hence the ratio is independent of 0; under the null hypothesis and a, if the alternative hypothesis is true. (b) The ratio R is independent of the regression coefficients 0. (c) We normally cannot use the standard test for the homogeneity of several varianccs since we generally do not have repeated observations for given x’s.’ In addition, the number of observations is frequently too small for an analogous multiple division of a given sample. (d) The power of this test will clearly depend upon the value of k, the number of omitted observations; for very large values of k the power will be small but it is not obvious that the power increases monotonically as k tends to O.* (e) The power of the test will clearly depend on the nature of the sample of values for the variable which is the deflator. Thus, if the variance of zm is small relative to the mean of z, the power can be expected to be small and conversely.