Sweep Programs: The Fall of M1 and Rebirth of the Medium of Exchange
通过1994-2000年货币需求的动态模拟，研究扫荡计划对M1的影响，发现扫荡计划导致M1被高估，并构建了包含扫荡余额的交易媒介指标M1S。
Sweep Programs:The Fall of M1 and Rebirth of the Medium of Exchange Donald H. Dutkowsky (bio) and Barry Z. Cynamon (bio) Abstract This paper investigates the effect of sweep programs on M1 using dynamic simulations of money demand over 1994-2000. The postsample period constitutes when sweep programs have been in effect. All models generate predictions systematically above reported M1. Using data on newly initiated programs, test findings indicate that sweeps account for the overprediction within the conventional money demand model with a long-term interest rate. We construct a medium of exchange measure, M1S, equal to M1 plus estimated holdings of sweep balances. M1S velocity compares favorably with that of broader aggregates. Evidence indicates cointegration within M1S money demand. From the time the Federal Reserve adopted monetary targeting, M1has played a central role in conducting policy. Long before then, the narrow aggregate's construction has emphasized money's medium of exchange function. This feature helped make M1 an attractive candidate for an intermediate target from 1970 through the mid-1980s. Along the way, the Fed redefined the measure in 1980 to include NOW and ATS Accounts along with other checkable deposits (OCDs). Even with these financial innovations, M1 retains its fundamental characteristic of including only assets that serve as a medium of exchange without transactions restrictions. In recent years, though, M1 has dropped precipitously out of favor. The FOMC stopped reporting target ranges for annual M1 growth in 1987 (see, for example, Pakko 1995), even though they continued with M2 and M3. Over the years, money in general has been de-emphasized by the Federal Reserve in the conduct of policy. As [End Page 263] stated in the Monetary Policy Report to the Congress for July 20, 2000 (which appears in the August 1993 issue of the Federal Reserve Bulletin), the FOMC stopped reporting annual growth objectives for money in 2000 and 2001 (Note 2, p. 540). But the text goes on to say that the FOMC still believes that money has value for gauging economic and financial conditions. The Report contains a detailed discussion regarding the recent behavior of M2 and M3. The observed movement of M1 over the past decade points to further instability. In the mid-1990s, the aggregate abruptly ceased its general upward trend. The level of nominal M1 decreased throughout 1995-97 and then rose at a much slower rate than before. M1 in the first quarter of 2000 is approximately equal to its value for the same quarter of 1993. Sweep programs have generally been cited as the major factor behind this observed shift in M1 (for example, Anderson 2000). As Anderson (1997) describes, since January 1994 hundreds of banks and other depository institutions have initiated sweep programs to avoid statutory reserve requirements on transactions deposits. In a sweep program, banks analyze customer use of checkable deposits such as demand deposits, ATS and NOW Accounts, and OCDs. Then they "sweep" forecasted excess funds into money market deposit accounts (MMDAs), which have zero statutory reserve requirements. Unlike checkable deposits, MMDAs are included in M2 but not in M1. The reallocation reduces M1, with the broader aggregates remaining unaffected. Anderson (2000) states that although only a small number of sweep programs were initiated during 1994, they lowered OCDs by about $9.8 billion. Holdings of sweep balances have accelerated since then. Anderson (2000) concludes that while their final extent is uncertain, sweep programs may have decreased checkable deposits by 50% or more. This paper empirically investigates the effect of sweep programs on M1. Our method consists of postsample simulation of M1 money demand. The demand for money has a long history in macroeconomic research and policy, reflected in a sizable body of literature. The systematic overprediction of the demand for real M1 found by Goldfeld (1976) set off a flurry of research attempting to resolve the problem, as surveyed by Judd and Scadding (1982). After a period of dormancy, this line of research was rejuvenated with a number of studies in the 1990s that apply the econometrics of nonstationarity. Researchers continue to empirically investigate the demand for money, as evidenced by a number of recent studies, including Carlson et al. (2000), Duca (2000), and...