FINANCIAL SECTOR WEAKNESS AND THE M2 VELOCITY PUZZLE

Abstract

I. INTRODUCTION There has been a long-running debate over the usefulness of monetary aggregates as intermediate targets or information variables in the conduct of monetary policy. In the mid1970s the missing episode associated with M1 money demand was documented by Goldfeld (1976) and Judd and Scadding (1982). Passage of the Depository Institutions Deregulation and Monetary Control Act in 1980 and the Garn-St Germain Act in 1982 ushered in a period of financial deregulation, including interest-bearing checking accounts and money market deposit accounts. These innovations affected depositor behavior to such an extent as to make M1 unreliable as a target or indicator of policy. As a result, research efforts focused on M2 demand with initial evidence of stable short-run specifications for this aggregate, as in Small and Porter (1989). This issue remains timely, with the recent example of the European Central Bank debating whether to target a monetary aggregate or inflation in its implementation of monetary policy, as Svensson (1999) points out. In the United States, Feldstein and Stock (1994) argue that the Federal Reserve should use the M2 monetary aggregate as an intermediate target. On the other hand, a fair amount of work suggests that M2 is not reliable as either a target or an indicator of monetary policy. Friedman and Kuttner (1992) show that by the early 1990s the relationship between M2 and gross domestic producct (GDP) had weakened, and Estrella and Mishkin's (1997) work provides further support for this finding. Miyao (1996) presents evidence that a cointegrating M2 demand relationship broke down in the 1990s. As the link between M2 and GDP deteriorated, the forecasting ability of M2 money demand equations also suffered. The difficulties in forecasting M2 spurred a number of publications examining whether the deterioration in the M2 equation's forecasting ability was temporary, or whether more fundamental factors-such as flaws in the construction of the M2 aggregate, the opportunity cost, or both--were at work. Carlson and Parrott (1991) and Duca (1992) first argued that the existence of troubled thrifts and the length of time it took the Resolution Trust Corporation to resolve the thrifts' difficulties helped explain the weakness in M2. In particular, Duca found that the change in the volume of cumulated deposits at resolved thrift institutions accounted for a large part of the M2 weakness, although he suggested his findings be viewed cautiously because of the short time period of the analysis. In the same article, as well as in one subsequent to it, Duca (1995) also examined whether some of the weakness in M2 reflected substitution by households away from M2-type deposits and into bond and equity mutual funds. He found that adding a bond fund series to M2 reduced estimated M2 growth shortfalls. Orphanides et al. (1994), Carlson and Keen (1996), Feldstein and Stock (1996), and Mehra (1997) also present evidence that the unusual behavior of M2 was related to growth in stock and bond mutual funds. Koenig (1996a, 1996b) focused on the competitiveness problems of financial intermediaries in the face of tighter regulations and stricter capital standards. He proposed an alternative strategy for empirically modeling M2 by altering the opportunity cost measure to include a long-term Treasury bond rate. This addition substantially improved forecasts of M2 until 1995, when large prediction errors once again appear. Feinman and Porter (1992) and Mehra (1997) also attempted to explain recent M2 instability through alternative measures of the opportunity cost. Rather than refine the opportunity cost or redefine the M2 monetary aggregate, Carlson et al. (2000b) present consistent evidence of a cointegrating relationship through the 1990s between real M2, various scale measures, and an opportunity cost measure. This result is obtained after adjusting for periods of financial restructuring in the early 1990s, characterized by a sharp increase in bond and equity funds and a steep decline in small time deposits. …