Federal Open Market Operations and Variations in the Reserve Base

Abstract

IN A RECENT ISSUE of this Journal Professors Bonomo and Schotta presented evidence on the nature of open market operations.' Bonomo and Schotta use regression analysis to determine the fraction of changes in factors affecting free reserves and total reserves which have been offset by defensive open market operations.2 However, Bonomo and Schotta misspecify the importance of the regression constant, and as a result of this misspecification, the conclusions they reach are inappropriate. The estimating procedure employed in their study makes use of the following equations (6 and 7):3 AR=al + P AR* + (o and AFR = a2 + 32 AFR* + o2 where ARweek-to-week change in total reserves, AFR = week-to-week change in free reserves, AR* week-to-week change in factors affecting total reserves other than open market operations, AFR* week-to-week change in factors affecting free reserves other than open market operations, a,, a2 estimates, allegedly, of net of other factors influencing open market opera-A A tions, i.e., dynamic objectives; and Pt,2 2 =estimates of (1 + k1) and (1 + k2). Bonomo and Schotta assume that the equation constant may be treated as representing the dynamic element in open market operations. Their results indicate that the constants, al and &2, are not significantly different from zero at the 1 per cent level;4 therefore, the conclusion to be drawn from their tests is that the System has not conducted dynamic open market operations with regard to total reserves since 1949 nor with regard to free reserves since 1945! This conclusion, that ". . . the dynamic element, . . ., is zero for all but 5 of the individual annual periods" (P. 662), results from the fact that Bonomo and Schotta have equated dynamic operations with a monotonic time trend variable-the equation constant in a first difference regression.5 There are several plausable reasons why Bonomo and Schotta's regression constants are not statistically significant. First, it may be due to the strong seasonality of the unadjusted weekly data. Secondly, it may 1.

Reply

“…I n the preceding comment [2], Duggar disagrees with several aspects of our recent article [1]. We shall argue that, in the main, Duggar's criticisms arise from a totally different, and, we think, erroneous conception of the goal of the analysis in our article.…”