A History of the Federal Reserve, Volume 2 (3 page)

A major change in economic theory came with recognition of uncertainty and the role of information. This heightened attention to the role of expectations. Lucas (1972) developed earlier work on rational expectations.
⁸
Rational expectations raised a question about the meaning of discretion. In practice, many central banks responded by providing more and better information about current and future actions. Rational expectations implies that central banks depend on market responses and markets depend on central bank actions. Setting and achieving a target for inflation two or three years ahead is a recognized way of reducing uncertainty about future actions. Federal Reserve officials have not adopted a formal inflation target, but, for a time, they encouraged a belief that they try to hold inflation in the 1 to 2 percent range, and in 2007 they began to forecast inflation, output, and unemployment for three years ahead. In early 2008, however, they gave most weight to forecasts of possible recession and less weight to inflation.

These actions constitute a major change from the secrecy traditionally practiced by central banks. It recognized the developments in monetary theory about the role of information, the importance of anticipations, and the success achieved by foreign central banks that announced inflation targets. But United States governments have not adopted fixed rules and are unlikely to do so in the foreseeable future.

Central bankers continue to meet regularly to decide current actions. Prominent central bankers have explained why they do not commit to a fixed rule. The former chairman of the Federal Reserve, Alan Greenspan (2003), explained that a fixed rule could not take account of the many contingencies to which monetary officials might wish to respond. The contingencies are infinite and most are unforeseeable. Many of the contingencies arise from actual or potential financial failures. The monetary rules developed in the literature do not incorporate these contingencies. In the past, following Bagehot (1873 [1962]) the central bank or the government announced in advance that it would suspend the gold standard rule at such times and provide the increased reserves demanded. This became part of the monetary rule.

8. Brunner and Meltzer (1993) point out that rational expectations models usually assign considerable weight to information but zero weight to the cost of acquiring information.

Greenspan’s successor, Ben Bernanke (2004), recognized that the central bank can do a great deal to reduce uncertainty about its future actions, but “specifying a complete policy rule is infeasible” (ibid., 8). He accepted Greenspan’s reason for infeasibility. Mervyn King (2004), governor of the Bank of England, called for “constrained discretion.” “Suitably designed, monetary institutions can help to reduce the inefficiencies resulting from the time-consistency problem” (promising one thing but later doing another) (ibid., 1). Otmar Issing, former chief economist and board member of the European Central Bank, expressed a similar position on many occasions (Issing, 2003, for example). He regarded as impossible in practice the idea of following a fixed rule.

The chapters that follow show that the Federal Reserve changed its objectives and its target many times. Often it did not have a precise target. Even after Congress required the Federal Reserve to announce an annual monetary target, it did not adopt procedures to achieve the target and allowed excess money growth to remain by following the practice called “base drift.”

Table 1.1 from the 1980s shows the changing objectives pursued during 1985–88. The principal objective changed frequently, making it difficult for the public to plan. The Federal Open Market Committee (FOMC) did not announce the objectives at the time, and the statement of objectives was sufficiently vague that knowing the objectives would not help observers to anticipate policy actions. And because it chose four or five objectives, the public could only guess the relative importance of each or its influence on Federal Reserve actions.

By the 1990s, principal central banks followed King’s “constrained discretion.” Many used some version of Taylor’s (1993) rule as a guide, but they deviated when they chose to do so. Several adopted inflation targets and gave more information about proposed actions and objectives. None followed a precise rule.

Definition
of
Inflation

Economists use two definitions of inflation, and laymen use some others. Monetarists define inflation as a
sustained
rate of change in some broad, general price index. The more common definition includes all price increases. Popular usage includes some relative price increases such as wage, asset price, or energy price increases; an example is “wage inflation.”

Economic theory does not prescribe the choice of a stable price level over a stable sustained rate of price change. The former requires central bank policy to roll back or push up the price level following an event that
raises or lowers it. If this is successfully carried out, the public can expect an unchanged price level over time. It incurs a cost because price adjustment is costly, particularly if the price level increased following a large increase in the price of oil or in an excise tax on a subset of goods.

The monetarist position lets the price level become a random walk. Energy price, excise tax increases, currency depreciation, or reductions in productivity raise the price level; opposite movements reduce the price level. These changes up and down often are spread through time. They appear as changes in the rate of price change, but they are not sustained.

Sustained money growth in excess of output growth induces a sustained increase in the rate of price change. Milton Friedman’s often quoted statement that inflation is always a monetary phenomenon used the monetarist definition of inflation. It recognized implicitly that non-monetary price level changes are mainly relative price changes.

A central bank must choose whether to control the price level or the rate of price change. Each has different costs to society. Controlling the sustained rate of price change permits the price level to vary, probably as a random walk. Wealth owners have to accept price variability but can be more confident when planning lifetime asset allocation that inflation will be controlled. Controlling all changes in the rate of price change also incurs a cost. The monetary authority must force other prices to decline if oil (or other) prices rise and permit other prices to rise in the opposite case. Such changes induce allocative changes and temporary changes in output and employment. Experience under the classical gold standard suggests that these costs are not small.

In practice, some central banks ignore some transitory changes in the price level. The Federal Reserve targets the so-called core deflator for private consumption expenditures. This excludes changes in the prices of food and energy on grounds that these prices are volatile and that many of the changes are transitory. The public experiences the effects of food and energy prices and considers these changes as inflationary. In 2007 the Federal Reserve accepted responsibility for controlling these prices over the longer term.

The use of a core price index is an inexact way of separating transitory from persistent price level changes to get a better measure of sustained inflation. A superior alternative would use statistical estimation of the relative variance of the permanent and transitory components to estimate whether a given change is likely to persist. Muth (1960) suggested a procedure.

Persistent price changes—infl ations—occur if sustained money growth rises in excess of sustained output growth. The inflation rate changes, therefore, if money growth rises relative to output growth or if normal output growth changes relative to money growth. The latter change occurred in the mid-1990s in the United States. It produced a fall in the sustained rate of inflation.

Implementing a monetarist policy to control inflation requires commitment to the low or zero inflation rule. Implementation of the policy requires judgment about the permanent rates of change of money and output. Many central banks now use an inflation target that they try to meet over two or more years.

The
Role
of
Relative
Prices

The simple Keynesian model of the 1940 and 1950s had a single interest rate representing the bond market or, in practice, the Treasury bill or federal funds rate. In the IS-LM model of that period, money was a substitute
for bonds; money growth had little direct impact on output or employment. The real balance effect was small. Usually the price level remained fixed. Later a Phillips curve avoided fixed prices by making the rate of price change depend on some measure of the output gap.
⁹

Friedman’s (1956) essay on the demand for money broadened the interpretation of interest rates to include relative prices of assets and output. His analysis changed the explanations of the transmission of monetary impulses to include a wide range of substitutions between money and other objects. In place of the Keynesian transmission from money to Treasury bills found in textbooks and many versions of the Federal Reserve’s econometric models, monetarists claimed that changes in the quantity of money altered current and expected future prices on a wide variety of domestic assets and the exchange rate.

In classical monetary theory, monetary policy changed the quantity of real balances relative to the stocks of other assets and current consumption. Substitution occurred in many directions. An excess supply of real balances induces changes in asset prices and spending; a deficient supply does the opposite. A change in the price of existing capital relative to the price of current investment induces or discourages new production. Changes in real balances relative to current consumption expenditure encourage or discourage spending.

There is no possibility of a liquidity trap—a condition in which monetary changes are impotent. If the nominal rate on short-term bills falls to zero, this margin closes but other margins remain (Brunner and Meltzer, 1968). A central bank can always increase the quantity of real balances by buying long-term debt, foreign exchange, real assets, or claims to real assets until money holders find that they hold more real balances than desired. To reduce money holdings people spend on consumption or nonmoney assets, changing relative prices to restore portfolio balance.

In Federal Reserve history, deflation occurred several times. In some periods, such as 1938, the nominal short-term interest reached zero or slightly below. Each of these periods is highlighted in the text of the two volumes. Economic expansion followed monetary expansion. Other periods of deflation, including the early 1920s, when the real interest rate reached 20 percent or more, do not show failure of monetary policy. The principal examples used by proponents of a liquidity trap are usually the early 1930s in the United States or the late 1990s in Japan. In both cases, mon
etary policy was not expansive. Inept and inappropriate monetary policy in 1929–33 induced reductions in money growth, giving rise to anticipations of further deflation.
¹⁰

9. Early discussions of the Phillips curve did not recognize that original data came from a time when the gold standard anchored inflation and expectations.

The most comprehensive recent statement of modern macroeconomic theory, Woodford (2003), is an elegant, erudite development of the rational expectations model that currently dominates academic thinking. Like early Keynesian models, but for very different reasons, Woodford’s analysis has a single interest rate that is set by monetary policy. All other interest rates reflect the current short-rate and rational expectations of the duration, magnitude, and influence of current policies and events. Prices and output are determined by aggregate demand and supply. Since the single interest rate is fixed by policy action, money has no independent role. All relative prices fully reflect current rational expectations of future events. Spending in this and other models depends on the long-term real interest rate. The central bank controls the short rate. A strong assumption about the expectations theory or the term structure of interest rates assumes away the problem of determining long rates.

Many central bank economists use this model. No central banker uses it. There are many reasons for this difference in approach. Three are most important.

First, rational expectations models give importance to information and anticipations of future events. Decision makers use all available information when allocating resources. This is an important advance. However, few models recognize the cost of acquiring information and differences in this cost in different markets. Further, the meaning or interpretation assigned to observations depends on the particular model or framework used. Federal Reserve policy discussions show that major differences in interpretation and anticipations were common. Members lacked a common framework of analysis, so they often differed about the expected policy consequences of current information.
¹¹

10. At a zero interest rate, as in Japan in the 1990s, central bank purchases of treasury bills provide no stimulus. The two assets are nearly perfect substitutes at a zero price. Later, expansion followed purchases of longer-term securities.

11. A well-known example from the 1990s was the conclusion drawn by Chairman Greenspan that productivity growth had increased. Other members of the open market committee expected an increase in inflation and wanted to raise the interest rate. Several years passed before all agreed about increased productivity growth. This is one of many examples with differences in interpretation of common information. Blinder (2004, 39, 43) discusses differences of interpretation and opinion on the FOMC during his term as a member. He concluded that committee decisions are less extreme, and less volatile (ibid., 48). Some work on committee decision making in monetary policy suggests relevant differences between the single policymaker assumed in economic models and committee decisions (Chappell, McGregor, and Vermilyea, 2005).