IN A RECENT ARTICLE [7] Jene Kwon and Richard Thornton (K-T) test the hypothesis that FHLB advances to Savings and Loan Associations (S&Ls) are an efficient method of intermediation, by which they mean that additional debt issued by the FHLB to finance advances is not acquired by actual or potential depositors at S&Ls. Their test of the hypothesis is based on a model of the demand for S&Ls deposits in which the desired stock of S&L deposits is a function of current income and the differentials between the rate offered on S&L deposits and the yields on alternative financial assets. The alternative assets used were FHLB securities, 3-month Treasury bills and AAA corporate bonds. Substituting this description of the desired stock of S&L deposits into a conventional stock adjustment model, and taking first differences, K-T estimated regressions in which the volume of net new savings in S&L deposits is a linear function of the change in income, the changes in yield differentials and the lagged volume of new new savings. They found that FHLB bond yield and Treasury bill yield coefficients had the expected negative signs but that only the FHLB bond yield coefficient was statistically significant when the Treasury bill yield as also included in the regression. On the basis of these results K-T concluded that FHLB bonds are substitutes for savings deposits, and that FHLB advances financed by debt issue are not efficient. They also argued that their results suggest a need for a reexamination of the current techniques of FHLB open market operations in an attempt to find improved methods of raising funds in the open market, methods to minimize the competitive effect of FHLB bond sales on thrift institution deposits.' This paper has elicited comments from Van Horne [10] and Grebler [4]. Van Horne's comment is directed primarily at econometric problems, among them the problems associated with lagged dependent variables, the possibility of simultaneous equations bias, and the use of first differences. Upon re-estimating the K-T equations using levels rather than first differences Van Horne found that the yield differential between Treasury bills and S&L deposits was statistically significant while the differential with FHLB bonds was not.2 Grebler took issue with the K-T