by Allan Hynes
The implications of economic theory for many questions depend on how people form their expectations. Nowhere is this more important than in theories of the business cycle, and I will discuss rational expectations as it applies to one part of these theories the role of monetary policy in determining changes in real economic activity. Modern analysis of this problem brought rational expectations into the mainstream of economics, and the story thus provides a useful context.
Readers may remember the following question from their undergraduate courses: what are the consequences of an increase in the nominal quantity of money. David Hume gave an answer in 1752. In the long run, all nominal prices increase by the same proportion as the money supply, and real output and relative prices are unchanged. In the short run nominal prices rise by less, and real economic activity improves. Trade quickens, to use an old phrase, because people interpret some of the changes in demand as arising from real sources and respond accordingly. As people learn, these short run responses dissipate, and long run neutrality holds.
In the late 1960s and early 1970s, macroeconomic theorists in the U.S. once again debated whether a central bank can control output and employment. The problem re-emerged because monetary economists, led by Milton Friedman, were challenging Keynesian orthodoxy. And, in the 1970s the American economy was experiencing high rates of inflation accompanied by higher than normal rates of unemployment, a situation not easily encompassed by Keynesian theory of the time.
In his presidential address to the American Economics Association (1968), Professor Friedman outlined a model in which workers and firms have good information about their local circumstances but poor information about current monetary policy. An unexpected increase in the money supply leads to an increase in real economic activity that dissipates as people learn the real structure of the economy is unchanged. The logic is symmetrical for an unanticipated reduction in the money supply.
From Hume to Friedman, explanations of short run real effects of changes in the money supply depended on people having incomplete information in the short run but also having the ability to learn from experience. This problem is not unique to monetary theory, and after World War II, as formal modeling became the standard approach when analyzing theory and policy, an operational way to deal with expectations was required. The common strategy was to assume people revise their expectations of a variable by a fixed fraction of the difference between the current and the expected value. This implies the expected value of the price level or rate of inflation, for example, is a weighted average of current and past values. This model is easy to apply; but an important problem is that adaptive expectations assumes the only knowledge people have about a variable is contained in observations of current and past values of the variable. Moreover, the assumption that people form their expectations in this way implies there is always a policy strategy the central bank can use to "fool" the private sector and thereby control real economic activity.
It became clear that adaptive expectations is not a satisfactory explanation. And in 1961 John Muth proposed that in dynamic models expectations should be consistent with the equilibrium structure of the economy. When forming expectations, people should exhibit the same degree of rationality as with other decisions and should not ignore relevant information. He illustrated his point with a model of an agricultural product. Demand in the current period depends negatively on the current price, and supply depends positively on the equilibrium price farmers , last period, expected to rule this period. There is uncertainty because "weather" implies supply outcomes cannot be perfectly forecast when commitments are made at planting. The equilibrium price and quantity (those values consistent with demand being equal to supply) in each period equal those consistent with the equilibrium implied by certainty plus an unpredictable random amount equal to the forecast error. Because useful information is not ignored, the expected value of this error is zero. Prices and quantities fluctuate randomly around their long run equilibrium values. There will be forecast errors, but these do not contain new information about the structure of the market and do not call for systematic revisions of expectations. Economists call a model with this property a stochastic equilibrium model.
Muth's contribution is one of the most cited papers in macroeconomics in the last forty years. But despite its revolutionary thrust, the idea languished for a decade. Then, in 1972 Robert Lucas published a model of a monetary business cycle in which expectations are assumed to be rational in the sense proposed by Muth. Lucas's objective was to analyze the implications of changes in the nominal quantity of money when expectations do not have the limitations of adaptive expectations.
When expectations are rational, the characterization of monetary policy needs attention. A central bank has systematic goals, and in an open democratic society it is reasonable to assume the goals are widely understood. Nevertheless, it is unlikely the target for the money supply can be hit with complete accuracy, and the actual money supply has a systematic (predictable) component and an unsystematic (unpredictable) random component. Because all available information is used, the expected value of the latter is zero. In the construction of models, it is assumed systematic goals are translatable into central bank policy functions.
Lucas's model, like Hume's and Friedman's, assumes the economy is competitive, and firms and households observe there own markets in the current period. Given rational expectations, they know the central bank's policy function but do not observe the current money supply and its unsystematic random component. The conditional expectation of the money supply is then given by the systematic policy function. There are real demand shocks that are specific to individual markets. In the current period the latter are confounded with the effects of the unobserved money supply shock which is spread across all markets. In equilibrium output and the nominal price level fluctuate randomly around their full employment values as the nominal money supply fluctuates randomly around the long run value defined by the monetary policy function. The business cycle produced by the model is the variation of output and employment in a stochastic equilibrium as defined in Muth's original example from agriculture.
The central bank's policy function and the private sector's knowledge of this function are central features of models with rational expectations. There are therefore important questions about the specification of this function. Advocates of activist monetary policy argue for a discretionary policy function whereby the central bank chooses the current money supply based on indices of the state of the economy. Given rational expectations, people understand systematic policies and expect their implied outcomes. Discretionary money supply changes will not reduce output variability because unexpected shocks to the money supply cause output to vary. This policy invariance theorem implies the choice of the systematic policy function should be based on long run considerations. One option is a policy function that aims to provide nominal price stability. Whatever the choice, the monetary authority should state clearly its long run objectives and conduct policy in a manner that minimizes the unsystematic variation in the money supply. Monetary policy should not add more background uncertainty to the economy than necessary.
Monetary theories of fluctuations under rational expectations model the business cycle as an equilibrium process. Problems of transitions, while important, are not addressed. Consider an example. In wartime, central banks often bow to fiscal demands and print money at higher than normal rates, using inflation as part of the fiscal package. At the close of hostilities, there is the problem of returning to peacetime policies with lower inflation. If a central bank could commit to the new policy without cost, the transition would be easy. Unfortunately, policy choices are seldom this simple. Circumstances will have changed, and there will be uncertainty about the new policy function. Announcing a desire to reduce inflation (a new policy function) will have little effect on expectations unless accompanied by fiscal policies that reduce expenditures or raise other taxes. It is by actions that a central bank and treasury coordinate their policies and signal true commitment to reduced monetary expansion and inflation.
Initial efforts to test monetary theories of the business cycle under rational expectations appeared promising. However, these tests require an independent partition of the money supply changes into expected and unexpected components. And this necessitates independent specifications of policy functions. Unfortunately we do not have a positive theory of central bank behavior, and policy functions must be jointly estimated with the models. This makes convincing partitions impossible. These, and other problems, have led to a decline in this research program and a rise in interest in the real business theories discussed by Professor Restucia.
This said, the general approach to problems of theory and policy implied by rational expectations has had wide spread and enduring influence. The core of the rational expectations hypothesis is that peoples's expectations, and thus behavior, depend on their understanding of the economy, including the goals and actions of policy makers. Peoples's induced responses to policies therefore often result in outcomes that are different than those originally envisioned by the policy makers. Rational expectations direct attention to policy functions, policy makers abilities to commit to these functions, and peoples expectations relating to these features of the environment. Each of these is illustrated in our post-war adjustment example, but the same issues arise in the theories of taxation, debt management and foreign exchange rates to name only a few of the areas that have been touched in important ways. It is not an exaggeration to claim that rational expectatins analysis represented a revolutionary shift in macroeconomic analysis.