Search

IN THE JUNE 1971 ISSUE, Kwon and Thornton attempt to demonstrate that Federal Home Loan Bank obligations "sold in the open market" compete with savings and loan associations for the same funds and "thus partially deprive them of their deposits," which they consider to be contradictory to the original purpose of the FHLB System.' Their findings, relating to the 1965-1969 period, rest mainly on a regression analysis of a stock adjustment model for selected financial assets using quarterly data. The model explains net new savings at FSLIC-insured S & L associations principally as a function of yield differentials between S & L deposits and other financial assets. Among the latter, FHLB securities emerge as stronger competitors for S & L deposits than do either Treasury Bills or AAA Corporate bonds. That savings flows at S & L associations as well as other deposit institutions were adversely affected by the competition of high-yield credit market instruments during some periods of the 5 years investigated, notably 1966 and 1969, is not a matter of dispute. But the specific results obtained by Kwon and Thornton, which indicate an especially close nexus between S & L deposits and FHLB obligations, warrant close re-examination. First, it should be made clear that the regression analysis merely infers yieldinduced shifts of funds from savings deposits to selected debt securities. It does not show the destination or the magnitude of funds diverted from the savings deposit market. Indeed, the authors make no such claim. It would have been helpful, however, if their cautionary statements on technical aspects of the analysis, such as simultaneous equation bias and intercorrelation of interest rates, had been matched with cautionary statements on substantive interpretation. Second, the analysis is marred by faulty specification of net new savings at S & L associations as the dependent variable. There is no reason in theory or empirical observation to assume that the behavior of S & L depositors varies significantly from that of household savers2 in mutual savings banks or commercial banks when they face alternative investment opportunities at yields substantially exceeding deposit rates. In all three types of institutions, the bulk of funds comes from relatively large accounts whose holders are presumably sensitive to the rate of return, as the authors show in their Table 1 for S & L depositors. Hence, to the extent that FHLB security issues were in competition with savings deposits, the appropriate dependent variable

THE OBJECTIVES of this thesis are threefold. First, an analysis of the Government of Canada bond market is undertaken with a view to ascertaining which of the extant theories of the term structure of interest rates is most representative of reality. Second, theoretically optimal investment strategies for investors constrained to this market are developed and third, what is theoretically optimal is compared with the portfolios which the Canadian chartered banks are actually holding. With respect to the first objective, the thesis is unique in a number of important respects. First, the unbiased expectations, liquidity preference and market segmentation theories of the term structure are restructured in terms of a holding period return concept in a more complete fashion than heretofore. Second, uncertainty, both in terms of a variance about an expectation and the more theoretically complete covariance with the market, is explicitly introduced into the analysis and the theories of the term structure, then viewed in the context of payments for risk undertaken. Third, investor horizons of 3 months through 5 years are employed in hopes of isolating certain market segments, and fourth, a procedure is developed for at least partially compensating for unanticipated interest rate movement, a problem which has given classical yield type term structure researchers a problem for some time. The analysis itself proceeds in the following fashion. Government of Canada bond data for the period January 1, 1949 through January 1, 1969 are utilized to generate expected holding period returns on instruments of maturity 3 months through 20 years in 3 month intervals for investor horizons of 3, 6, 12, 24 and 60 months, respectively. The results are then compared with the restructured models of the term structure and with various risk parameters. The conclusion is that there is strong evidence to support the existence of a multiple horizon market with a short horizon (3 months or shorter) and probably a series of longer horizons. The Canada market can not on the basis of this research be construed as segmented in the very rigid Culbertson sense, however, but rather should be viewed in the preferred habitat sense articulated by Modigliani and Sutch. The second objective is to develop an optimal portfolio strategy for a short horizon investor constrained to the government bond market, a market now construed as possessing multiple habitats only one of which is short. The third objective is to compare this strategy with that actually undertaken by the chartered banks, whom, it is argued, are the best institutional example of a short horizon investor. The optimal strategy is argued to be a well diversified portfolio of short bonds and, in comparing this with what the banks in aggregate actually held, a discrepancy is observed. The banks seem to be holding an excess of the longer instruments. A number of explanations for this are then advanced; in particular, a desire on the part of the government that the banks maintain the long end of the market, uncertainty with respect to bank holdings in the 3 to 5 year range and the bank's own view ofthemselves as superior interest rate forecasters.