Academic journal article Journal of Money, Credit & Banking

Federal Funds Rate Prediction

Academic journal article Journal of Money, Credit & Banking

Federal Funds Rate Prediction

Article excerpt

THE IMPORTANCE of the effective federal funds (FF) rate in U.S. financial markets is unquestionable. The Federal Reserve (Fed) implements monetary policy by targeting the effective FF rate. The ability of market participants to predict the FF rate is important to modern analyses of monetary policy in that other interest rates are believed to be linked to the FF rate by the market's expectation of monetary policy actions that directly affect the FF rate. It is therefore not surprising that a vast body of research has studied the behavior of the FF rate and proposed empirical models designed to explain it. One strand of this literature focuses on the FF rate using data at monthly or quarterly frequency to establish the extent to which arguments of interest to the Fed--such as inflation and the output gap--are sufficient to explain the variation in the FF rate (e.g., Taylor, 1993, 1999, Clarida, Gali, and Gertler, 1998, 2000, and the references therein). A related literature investigates the impact of monetary policy shocks on key macroeconomic aggregates, again using low frequency data, identifying shocks to monetary policy using the FF rate in structural vector autoregressions (e.g., Christiano, Eichenbaum, and Evans, 1999, and the references therein). Other studies locus on the high-frequency behavior of the FF rate, using data at the daily frequency. This frequency is appealing because each day the Trading Desk of the Federal Reserve Bank of New York (hereafter Desk) conducts open market operations designed to move the FF rate in the desired direction (e.g., Hamilton, 1996, Roberds, Runkle, and Whiteman, 1996, Balduzzi, Bertola, and Foresi, 1997, Taylor, 2001). (1)

The literature has suggested that several variables have predictive power for explaining FF rate movements: FF futures rates (Krueger and Kuttner, 1996), the FF rate target (Rudebusch, 1995, Taylor, 2001), and other interest rates linked to the FF rate via no-arbitrage conditions or the term structure of interest rates (e.g., Enders and Granger, 1998, Hansen and Seo, 2002, Sarno and Thornton, 2003, Clarida et al., 2004). A number of models, both univariate and multivariate, linear and nonlinear, have been proposed to capture the unknown process that drives FF rate movements. To date, however, there appears to be no consensus on what variables and models best characterize the behavior of the FF rate at the daily frequency. This paper attempts to fill this gap in the literature by examining a variety of univariate and multivariate, linear and nonlinear empirical models of the FF rate, largely taken directly from or inspired by previous research in this context. We estimate these models using daily data for the period from January 1, 1990 through December 31, 1996 and generate forecasts over the remaining four years of data. We also examine the potential to improve on the individual or 'primitive' models by using combinations of forecasts (see, inter alia, Diebold, 2001, Stock and Watson, 1999b, 2003, Swanson and Zeng, 2001).

To anticipate our results, we find that, in general, most of the models and predictor variables considered produce satisfactory one-day-ahead forecasts of the FF rate. However, the best forecasting model is a very parsimonious univariate model where the one-day-ahead funds rate is forecast using the current difference between the funds rate and its target. This model can be thought of as a simple variant of the Desk's reaction function proposed by Taylor (2001). Combining the forecasts from various models provides generally modest improvements on the Desk's reaction function. We argue that these results have a natural interpretation and that they are in line with the growing empirical evidence suggesting that the Fed's policy is well described as a forward-looking interest rate rule. (2)

The remainder of the paper is set out as follows. In Section 1, we describe the empirical models of the daily FF rate considered in the paper. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.