Rules versus Discretion in Monetary Policy

Disagreements over the appropriate design of monetary policy have existed at least since the nineteenth-century controversy between the currency and banking schools in the United Kingdom.

Although the British Bank Act of 1844 adopted the views of the currency school, the actual operation of monetary policy during the pre-1914 gold standard era entailed significant elements of discretion in the setting of interest rates. This orthodoxy was analyzed in successive editions of Mill [1848] and in Jevons [1875]. In addition, it was gradually recognized that, in periods of financial crisis, central banks should operate as lenders of last resort (Bagehot [1873]).

However, the orthodoxy was also challenged, in view of the trend decline of the price level during the gold standard period. Alternative academic proposals for monetary policy emerged toward the end of the nineteenth century, none of which questioned the convertibility of banknotes into precious metals, such as gold or silver. Jevons [1875] favored supplementing gold with indexation to prices, as did Marshall [1871, 1887]. Both Marshall and Edgeworth [1895] advocated symmetallism over the gold standard. Fisher [1896, 1911] put forward a radical academic proposal for the stabilization of the price level, rather than the gold price, as the main rule for monetary policy. Another such proposal was the Wicksell [1898] interest rate rule for changes in the nominal interest rate in accordance with deviations of the price level from a target price level.

Renewed arguments concerning the conduct of monetary policy emerged in the United States, during the operation of the prewar classical gold standard, but especially during and after the Great Depression. Fisher [1936] backed up his earlier proposal for a constant price level objective with a rule for 100% reserves behind bank deposits, and so did Simons [1936]. Simons argued forcefully for rules as opposed to discretion, and favored a fixed money supply rule, although he worried about shocks to the velocity of money. The Simons proposal was later updated by Friedman [1960], who argued for a fixed rate of growth of the money supply as the optimal rule for monetary policy.

After World War II, the dominant Keynesian approach favored a more discretionary monetary policy as discussed in chapter 15. However, the instability of the Phillips curve eventually tilted the discussion again in favor of rules, especially following the important contributions of Friedman [1968] and Kydland and Prescott [1977].

Another argument in favor of rules is the virtual impossibility of perfect knowledge about key parameters of economic models, the difficulties of estimating and forecasting the state of the economy, and the need for robustness. Discretion requires much more knowledge about the workings of the economy than we actually possess, in contrast to abiding by simple rules, as forcefully suggested by Meltzer [1987]. Among others, Orphanides [2003a] has formally investigated the implications of imperfect knowledge for monetary policy.

Very little disagreement exists today about the view that monetary policy should be determined on the basis of rules. Yet there is still a very lively debate about the appropriate design of such rules.²

Should rules be contingent on the state of the economy, as, for example, suggested by the Wicksell [1898] rule? Or should they be noncontingent, like the Fisher [1896] price level stabilization rule, or the Simons [1936]–Friedman [1960] rule for the money supply? Contingent rules allow for some flexibility in addressing shocks that affect aggregate fluctuations, whereas noncontingent rules do not.

Another longstanding debate has concerned the appropriate instrument of monetary policy. Should central banks control the money supply or nominal interest rates? In recent years, the debate has tilted in favor of interest rate rather than money supply rules, in view of the practice long followed by most central banks that have to deal with the practical difficulties in controlling (or even defining) the appropriate measure of the money supply.³

The literature on interest rate rules in the past 20 years has focused on the analysis of one particular such rule, the Taylor [1993, 1999] rule. This rule has been shown to describe the monetary policy of the United States and other advanced economies relatively well since at least the early 1980s, if not earlier. The Taylor rule, which is a generalized version of the contingent rule proposed more than a century ago by Wicksell [1898], is usually analyzed in the context of new Keynesian models of aggregate fluctuations, such as the models discussed in chapters 16 and 17. It has been compared to optimal monetary policy, and proposals have been put forward for improvements in the choice of parameters and other characteristics of this particular rule.⁴

This chapter analyses the various issues concerning the appropriate design of monetary policy. To analyze monetary policy, let us use the periodic wage contracts model of chapter 17. This is an analytically tractable new Keynesian model in which nominal wages are set periodically by insiders in the labor market and are nonindexed. We shall use this particular model to discuss the question of rules versus discretion and the appropriate use of monetary policy instruments. We will also use it to analyze alternative rules and the design of optimal monetary policy.⁵

20.2