Money is a collection of liquid assets that is generally accepted as a medium of exchange and for repayment of debt. In that role, it serves to economize on the use of scarce resources devoted to exchange, expands resources for production, facilitates trade, promotes specialization, and contributes to a society's welfare (Thornton, 2000). This theoretical definition serves two purposes: It encompasses new forms of money that may arise as a result of financial innovations related to technological change and institutional developments. It also distinguishes money from other assets by emphasizing its general acceptability as a medium of exchange. While all assets serve as a store of wealth, only a few are accepted as a means of payment for goods and services.

While this definition provides a clear picture of what money is, it does not specify exactly what assets should be included in its measurement. There are several liquid assets, such as coins, paper currency, checkable-type deposits, and traveler's checks, which clearly act as a medium of exchange and definitely belong in its measurement. However, several other assets may also serve as a medium of exchange but are not as liquid as currency and checkable-type deposits. For example, money market deposit accounts have check-writing features subject to certain restrictions, and savings accounts can be converted into a medium of exchange with a negligible cost. To what extent such assets should be included in money's measurement is not clear.

As an alternative, economists have proposed defining and measuring money using an empirical approach. This approach emphasizes the role of money as an intermediate target for monetary policy. As Mishkin (1997) points out, an effective intermediate target should have three features: It must be measurable, be controllable by the central bank, and have a predictable and stable relation with ultimate goals. Thus, an asset should be included in the measurement of money if it satisfies the above requirements. Evidence is mixed on which measures of money have a high predictive power. A measure that predicts well in one period might not perform well at other times, and a measure that predicts one goal might not be a good predictor of others.

The Federal Reserve System (the Fed) has incorporated both the theoretical approach and the empirical approach in constructing its measures of the money supply for the United States. The results are four measures of monetary aggregates—M1, M2, M3, and L—that are constructed using simple summations of some liquid assets. M1 is the narrowest measure, corresponding closely to the theoretical definition of money. It consists of six liquid assets—coins, dollar bills, traveler's checks, demand deposits, other checkable deposits, and NOW accounts held at commercial banks and at thrift institutions. These assets are clearly money because they are used directly as a medium of exchange. The M2 aggregate adds to M1 two groups of assets: (1) other assets that have check-writing features, such as money market deposit accounts and money market mutual funds shares, and (2) other extremely liquid assets, such as savings deposits, small-denomination time deposits, overnight repurchase agreements, and overnight Eurodollars. Similarly, the M3 aggregate adds to M2 somewhat less liquid assets, such as large-denomination time deposits, institutional money market funds, term repurchase agreements, and term Eurodollars. Finally, L is a broad measure of highly liquid assets. It consists of M3 plus several highly liquid securities, such as savings bonds, short-term Treasury securities, bankers' acceptances, and commercial paper.

A potential problem with the simple summation procedure, which underlies the construction of the monetary aggregates, is the assumption that all individual components are perfect substitutes. As Barnett, Fisher, and Serletis (1992) point out, this procedure is useful for constructing accounting measures of monetary wealth but does not provide reliable measures of monetary services. As a solution, Friedman and Schwartz(1970) have proposed weighting individual components by their degree of "moneyness," with the weights varying from zero to unity. Another more rigorous solution proposed by Barnett and colleagues (1992) is based on the application of aggregation and index number theory. Evidence along this line of research (Chrystal and McDonald, 1994) suggests that these measures of monetary aggregate are superior to the traditional measures in their predictive contents.

Knowledge of the money-supply process and information about its behavior are important for two interrelated reasons. First, changes in money growth may have significant effects on the economy's performance. Its short-run variations may affect employment, output, and other real economic variables, while its long-run trend determines the course of inflation and other nominal variables. Second, money supply serves as an important intermediate target for the conduct of monetary policy. As a result, changes in money growth may be instrumental in attaining economic growth, price stability, and other economic goals.

Three groups of economic agents play an important role in the process of money-supply determination. The first and most important is the Fed, which sets the supply of the monetary base and imposes certain constraints on the set of admissible assets held by banks and on the banks' supply of their liabilities. Next is the public, which determines the optimum amounts of currency holdings, the supply of financial claims to banks, and the allocation of the claims between transaction and nontransaction accounts. The last is banks, which absorb the financial claims offered by the public, set the supply conditions for their liabilities, and allocate their assets between earning assets and reserves subject to the constraints imposed by the Fed. The interactions among the three groups are shaped by market conditions and jointly determine the stock of money, bank credit, and interest rates (Brunner, 1989).

The level of money stock is the product of two components: the monetary multiplier and the monetary base. The monetary base is the quantity of government-produced money. It consists of currency held by the public and total reserves held by banks. Currency is the total of coins and dollar bills of all denominations. Reserves are the sum of banks' vault cash and their reserve deposits at the Fed. They are the non-interest-bearing components of bank assets, consisting of required reserves on deposit liabilities established by the Fed and additional reserves that banks deem necessary for liquidity purposes.

The Fed exercises its tight control over the monetary base through open-market operations and extension of discount loans. Open-market operations, which are the Fed's authority to trade in government securities, are the most important instrument of monetary policy and the primary source of changes in the monetary base. An open-market purchase expands the monetary base, whereas an open-market sale works in the opposite direction. The Fed's control of the discount loans results from its authority to set the discount rate and limit the level of discount loans through its administration of the discount window.

The money multiplier reflects the joint behavior of the public, banks, and the Fed. The public's decisions about their desired holdings of currency and nontransaction deposits relative to transaction deposits are one set of factors that influence the multiplier. Banks liquidity concerns, and thus their desire to hold excess reserves relative to their deposit liabilities, are another set of factors. The Fed's authority to change the required reserve ratios on bank deposits constitutes the third set of factors. Given the rather infrequent changes in the reserve-requirement ratios, the multiplier primarily reflects the behavior of the public and private banks as well as market and institutional conditions.

For example, a decision by the public to increase its currency holdings relative to transaction deposits results in a switch from a component of money supply that undergoes multiple expansion to one that does not. Thus, the size of the multiplier declines. Similarly, a decision by banks to increase their holdings of excess reserves relative to transaction deposits reduces bank loans, causing a decline in deposits, the multiplier, and the money supply. Finally, a decision by the Fed to raise the reserve-requirement ratio on bank deposits results in a reserve deficiency in the banking system, forcing banks to reduce their loans, deposit liabilities, the money supply, and the multiplier.

Over the 1980-1999 period, the M1 and M2 aggregates grew at average annual rates of 5.5 and5.7 percent, respectively. However, the growth rates were not stable. They varied between the low of 3.5 percent and the high of 16.9 percent for M1, and between the low of 0.4 percent and the high of 11.4 percent for M2. What factors contributed to the long-run growth and short-run fluctuations in the money supply? During the same time period, the monetary base grew at an average annual rate of 7.5 percent, due primarily to open market operations. Thus changes in the monetary base and open-market operations are the primary source of long-run movements in the money supply. For shorter time periods, however, changes in the money multiplier may also have contributed to the fluctuations in the money supply.