This analysis was pioneered by Tinbergen (1952).

"As the discussion earUer in this section suggests, it is not clear that this is what policymakers should in fact be trying to achieve.

Meulendyke (1990) describes the specifics of the Federal Reserves operating procedures.

a change in policy would have. This natiually raises the issue of how these lags and uncertainties should affect policy.

The traditional analysis of policymaking under uncertainty distinguishes among objectives, instruments, intermediate targets, and indicators of pol-icy." The objectives are the ultimate goals of policy, such as inflation and unemployment. The instruments are the variables that policymakers can control directly, such as open-market operations, reserve requirements, tax rates, and government purchases.

Indicators and intermediate targets fall between the instruments and the objectives. Indicators are variables that provide information about the current or future behavior of the objectives. Some examples are orders for new goods, prices of raw materials, and measures of money and lending. As policymakers obtain new information about the likely behavior of the objectives by observing the indicators, they may adjust the settings of the instruments. Intermediate targets, in contrast, are variables that poUcymakers choose to focus on in place of the ultimate objectives. The most famous candidate target is the money stock. Many economists have argued that it is better to instruct policymakers to try to keep the growth rate of a measure of the money stock (such as Ml or M2) as close as possible to some steady, low rate (such as 3% per year) rather than to try to maximize some broader objective function (see, for example, Friedman, 1960).

To see how instruments, indicators, targets, and objectives are used in practice, consider the following stylized description of U.S. monetary policy in recent years. The main ultimate objectives of policy are the behavior of unemployment (or real output) and inflation. Policymakers appear to want inflation to be around 2% or 396 per year and to avoid large swings in un-employment.2 Thus, for example, when inflation is above the 2-3% range, policymakers have sought to reduce it gradually. Other objectives, such as keeping exchange rates and interest rates moderately stable, also appear to get some weight in policymakers objective function.

Over the short term (say, day-to-day and week-to-week), the key intermediate target of policy is the Federal funds rate. The Federal Reserve conducts its daily open-market operations to try to keep the funds rate close to its current target level. Although on a day-to-day basis there are noticeable departures of the funds rate from the target, on a weekly or longer-term basis the Federal Reserve usually hits the target quite accurately.

Over the slightly longer term (say, month-to-month), the Federal Reserve does not focus on any single intermediate target. Instead, it adjusts the target level of the funds rate in response to many variables that can provide information about the future paths of real activity and inflation.

2For the United States today, estimates of the natural rate range from under 6% to almost 7%. Obviously where in this range the true ftgure hes has important implications for policy.

Finally, over the medium term (say, quarter-to-quarter), there usually is clearer information about how real output and unemployment are likely to behave than about inflation. Thus over this horizon, real output and unemployment are not only among the main objectives of policy, but are also the main indicators or intermediate targets. When inflation is above the desired range, for example, pohcymakers typically aim to keep unemployment moderately above the natural rate; when inflation is in the desired range, they usually try to maintain unemployment roughly at the natural rate. In either situation, policymakers adjust the target as they obtain more information about the behavior of inflation.

The Traditional Argument for Rules

A natural question about indicators and intermediate targets is why policymakers would ever adopt an intermediate target. It seems that policymakers should take aU relevant information into account in their efforts to achieve their ultimate goals. A particular indicator, such as a measure of the money stock, may turn out to be particularly informative; but even then, it appears that there is a cost and no benefit to targeting that variable.

One possible answer involves the dynamic-inconsistency issue, discussed in Sections 9.4 and 9.5: adopting a binding rule about the behavior of an intermediate target can overcome the dynamic-inconsistency problem, and can therefore lead to lower average inflation. But support for money-stock rules and other intermediate targets long predates concern about dynamic inconsistency. Moreover, many proposed ways of adopting intermediate targets do not involve binding commitments, and thus would not overcome the problem.

The basis for the traditional argument for instructing pohcymakers to target some intermediate variable is twofold. Consider for concreteness a money-stock target. The first, and less important, part of the argument for targeting the money stock is that the relation between the money stock and the ultimate objectives of pohcy is strong enough, and the uncertainty about the effects of departures of the money stock from a path of steady growth large enough, that the potential for improvement over a money-stock rule is small. And since the rule would not be completely binding, in the event of a calamitous breakdown it could be abandoned.

The second, and more important, part of the argument is that instructing pohcymakers to try to achieve the ultimate goals of pohcy to the best of their ability may lead to systematic errors in pohcy. Those potential errors ha\ e several sources.

"The possibility of the Federal Reserve pursuing objectives other than social welfare (either because of its own preferences or because of political pressures) suggests that fluctuations can arise from political forces rather than exogenous disturbances. For examples of theories of such political business cycles, see Nordhaus (1975); Alesina and Sachs (1988); Rogoff and Sibert (1988); and Harrington (1993).

"Karamouzis and Lombra (1989) present one piece of evidence of a tendency for over-optimism among policymakers: during the 1970s, the Federal Open Market Committee tended to adopt combinations of interest-rate and money-growth targets that were systematically off the frontier (in the direction of lower money growth and lower interest rates) of possibilities presented by the staff as being feasible.

First, policymakers are subject to political pressures. Policymakers outside the Federal Reserve, and the public, may place too much weight on the short-run cost of lower unemployment relative to the long-run cost of higher inflation. This could arise from a higher discount rate than is appropriate, or from a failure to understand how the economy operates. Some evidence for this view is provided by the fact that during periods (such as 1979-1982) when the Federal Reserve has pursued policies that involved very high interest rates, it has not explicitly acknowledged that it was doing so. Instead, policymakers have characterized policy as focusing on some intermediate target (such as nonborrowed reserves in 1979-1982) and as not being directly concerned with interest rates.

Second, monetary poUcymakers may have objectives other than maximizing social welfare, and providing them with only vague instructions about how to conduct poUcy may increase their ability to pursue those objectives. For example, they may wish to improve the Presidents chances of being reelected, or to increase seignorage revenues."

Finally, policymakers may genuinely try to maximize social welfare but may nonetheless make systematic errors. Individuals are often overconfident in their judgments (of the state of the economy, or of the likely effects of policy, for example). In addition, they may be reluctant to admit that, given the lags and uncertainties in the effects of policies, the best reaction to a problem may be to do Uttle or nothing. As a result, policy may systematically overreact, easing too much in recessions and thereby causing the subsequent expansions to be too strong, and tightening too much in expansions and thereby causing recessions (see, for example, Friedman, 1960). Simflarly, given the suffering associated with unemployment, poUcymakers may have a tendency to read the evidence about the natural rate optimistically. This can generate an inflationary bias in policy. And, as with the tendency to overreact, it can generate fluctuations. Policymakers may first, out of concern about tmemployment and in hopes that the natural rate is low, push unemployment below the natural rate; then, when signs of rising inflation become clear, they may tighten and cause a recession.

This discussion suggests several potential sources of inflation other than dynamic inconsistency: political pressures on policymakers, policTnakers pursuit of objectives other than social welfare, and overoptimism about the