The Announcement Effect: Evidence from Open Market Desk Data. (Session 1: The Reserves Market)

Abstract

I. INTRODUCTION The textbook view of monetary transmission mechanism rests on central bank's ability to manipulate overnight interest rate by controlling reserve supply, followed by a rational-expectations mechanism that ensures that movements in overnight rate reverberate into longer maturity rates. However, while few dispute fact that central bank controls overnight rate effectively, notion that it does so via a liquidity effect and nature of term structure relationships needs to be reexamined. Modern central banking is generally characterized by public announcements of an interest rate target, such as federal funds rate target in United States. In some cases, central banks (such as Bank of Australia and Bank of England) also disclose an inflation target, while in extreme cases, banks (such as Reserve Bank of New Zealand) disclose parameters of policy reaction function. These actions constitute a significant departure from traditional central banking. It is natural to question why central banks have abandoned their once-secretive behavior in favor of public disclosures of policy moves. Likely reasons include desire for better and more precise control of overnight rate, and, more important, enhanced communication of future policy moves--in essence, Holy Grail of controlling long rates by also manipulating expectations. This paper investigates these issues as they relate to U.S. Federal Reserve. In particular, we focus on how Federal Reserve's 1994 policy change--by which it began announcing target level for federal funds rate--had an impact on liquidity effect and manner in which central bank uses open market operations to control federal funds market. We also examine what effect this policy change may have had on behavior of term structure. Prior to Federal Reserve's Federal Open Market Committee (FOMC) meeting in February 1994, monetary policy objectives for federal funds rate and outcome of FOMC meeting itself had been confidential and had never been announced. (1) After policy change occurred, and inspired by similar developments in other central banks, Demiralp and Jorda (2000), Guthrie and Wright (2000), Taylor (2001), Thornton (2001), and Woodford (2000) began to investigate a central bank's ability to control overnight rate--not merely through traditional open market operations, but by effectively communicating desired level of overnight rate and standing ready to enforce that level. As Meulendyke (1998) observes, the [federal funds] rate has tended to move to new preferred level as soon as banks know intended rate. In this paper, we term this method of controlling overnight rate announcement effect (following Demiralp and Jorda [2000]); this effect differs from conventional liquidity effect in that volume of open market operations required to signal new target level is substantially smaller because of expectations. The strategy we pursue to investigate announcement effect consists of using two types of controls. The first is to analyze data with two primary subsamples: one predating and other postdating 1994 policy change. The second is to compare, within a subsample, pattern of open market operations surrounding days in which target was changed relative to rest of subsample. Most of time, open market operations conducted by Trading Desk of Federal Reserve Bank of New York (the Desk) are designed to accommodate variations in reserve needs that stem from a variety of factors, such as changes in currency holdings, float, and large Treasury balances; to manage currency in circulation; and to accommodate other variations in supply of reserves. Based on a particular type of variation (unexpectedly large Treasury balances), Hamilton (1997) calculates interest rate elasticity to an unanticipated shortfall in reserves. …

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 22 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.