Identifying the Liquidity Effect at the Daily Frequency

Abstract

Listen

Translate article

It is widely believed that the Fed controls the overnight federal funds rate, and thereby other short-term interest rates, through open market operations.1 The Fed sells government securities, reducing the supply of reserves relative to demand, when it wants to raise the funds rate. Similarly, it purchases securities when it wants the funds rate to fall. The change in the short-term interest rates brought about by the exogenous policy actions of the monetary authority is referred to as the liquidity effect. By many accounts—public and professional—the Fed’s use of this procedure is responsible for most, if not all, lower frequency variation in the federal funds rate and, thereby, other short-term interest rates at least since the late 1980s. Despite the widespread belief in the Fed’s ability to control short-term interest rates through open market operations, attempts to estimate the liquidity effect—using a variety of approaches—generally have been unsuccessful.2 Many analysts attribute this lack of success to the inability of economists to isolate the exogenous policy actions of the Fed. Hamilton (1997) notes that, most often, the Fed adjusts its policy instrument in response to new information about current or future values of output, inflation, the exchange rate, or other variables. Because of this, Hamilton argues that “the correlation between such ‘policy innovations’ and the future level of output of necessity mixes together the effect of policy on output with the effect of output on policy” (p. 80). This problem tends to be more severe when the monetary aggregates used to estimate the liquidity effect are only loosely associated with the policy actions of the Fed or when the data are averaged over a period of a month or more. Hamilton suggests that “a more convincing measure of the liquidity effect” can be obtained by estimating the “instantaneous consequences of an open market purchase. For daily data, Federal Reserve accounting and Trading Desk procedures suggest some quite plausible identifying assumptions that can be used to measure the instantaneous liquidity effect” (pp. 80-81). Specifically, Hamilton estimates the response of the funds rate to reserve supply shocks that are analogous to those that the Fed could create through open market operations. Hamilton finds that the funds rate responds significantly to reserve supply shocks, but only on the last two days of the maintenance period, and that the response is economically significant only on settlement Wednesdays. I model the reserve market based on the Fed’s operating procedure and show why the liquidity effect cannot be identified using Hamilton’s methodology. Consistent with this analysis, I show that Hamilton’s settlement-day result is fragile and due to a few settlement Wednesdays when there were unusually large changes in the funds rate. In addition, I show that there is no liquidity effect using his methodology for sample periods before and after his. I then suggest an alternative approach to estimating the liquidity effect at the daily frequency. Specifically, I argue that the liquidity effect can be identified by estimating the change in nonborrowed reserves associated with changes in the Fed’s target for the federal funds rate. If the Fed controls the federal funds rate by changing the funds rate target, reserves should change when the target is changed. Estimates of this model using data prior to 1994 suggest that, while the Fed undertakes open market operations consistent with changing the funds rate when the funds rate target is changed, the size of the actions is relatively small compared with stochastic fluctuations in reserves. One possible interpretation