Mr Greenspan is absolutely right that real-world monetary policy is about managing risks that are difficult to quantify, and so discretion is required. In defending his own practices, he implicitly criticised the monetary policy framework now commonly known as "inflation targeting" and the manner in which it is implemented in a number of countries. He is correct about the pitfalls in the practice of inflation targeting. But it is difficult to justify the unstated extension of his argument from flexibility in the execution of policy to flexibility in formulating policy objectives. To see why, we need to look at a little history.
Some 25 years ago, economists realised that a central bank controlled by politicians yields high inflation. The problem is that a monetary policy easing produces higher growth and employment in the short term, driving inflation up only several years later. Politicians are notoriously myopic, caring primarily about the next election.
Everyone agrees that this "inflation bias" is a problem and that the solution is to take monetary policy out of the hands of the politicians and hand it over to an independent central bank. The answer is to create a system with two basic elements. First, the politicians would dictate a set of goals and, second, the central bankers would publicly report their progress on meeting these goals. Explicit goals foster accountability, while disclosure requirements create transparency. Yes, central bankers are powerful - but we tell them what to do and then watch them do it. That means forcing them to explain their decisions in the context of their objectives. The explanations have to be plausible and accompanied by data to back them up.
The exact institutional arrangements establishing accountability and transparency differ from country to country. In some cases the government gives central bankers explicit numerical targets for inflation, while in others the central banks are left to define the exact goals. The British government sets a specific inflation target each year, while the treaty establishing the European Central Bank as well as the statutes governing the Fed simply refer to a goal of "price stability" and provide no technical definition.
There are similar differences in the timing and type of information made public by central banks. While every central bank today announces its policy actions almost immediately, the extent of the accompanying statements and the willingness to answer questions vary. The Fed's statements tend to be several sentences long and no one answers questions. By contrast, the ECB president holds a press conference to answer questions after reading a long statement on the bank's deliberations.
This brings us back to inflation targeting. In principle, it is no different from any other framework in which objectives are announced and policymakers' actions are explained to the public and the markets. The problem is in the practice.
There are two big pitfalls in the implementation of inflation targeting, both arising from formulaic implementation. One involves the technicalities of how interest rates are set. If you were to survey the academic literature that provides the conceptual underpinning for inflation targeting, you would find that most economists are discussing something called "inflation-forecast targeting". It is logical that policymakers should adjust interest rates in response to forecast deviations from the objective. That means having accurate forecasts as well as a quantitative estimate of the impact of interest rate changes on these forecasts. Getting these means building and using sophisticated models of a country's economy. This is an challenging task even for the smartest model-builders in the world. The Fed does an adequate job but employs hundreds of highly trained economists to do it.
Experience suggests that our models are at their worst when we need them the most. Since statistics is based on historical relationships, the technicians will be very bad at figuring out what to do about large disturbances of a type we have never seen before. Needless to say, such econometric models were practically useless in the wake of the Russian default and the bond market meltdown that followed in 1998.
My concern is that inflation-targeting central banks will employ economists to build models and then mechanically adjust policy in response to them. As one inflation-targeting central banker said to me once: "Forecasting is too important to be left to forecasters." Models are fine; but central bankers earn their pay on the days when the models fail. Here Mr Greenspan's warning about the dangers of following policy rules is well placed. We can only hope that the burgeoning ranks of the inflation-targeting central bankers do not learn this lesson the hard way.
The second pitfall is communication. Inflation-targeting central banks nearly all publish something called an inflation report, modelled on the one first published by the Bank of England in 1993. The idea of inflation reports is to convince financial markets and the public at large that the central bank is doing its job. But an inflation report is written for its local audience.
My concern is that, when it comes to communication, what works in one country may not work elsewhere. Cultural differences across countries dictate how central banks talk to their constituents. Neither western academics nor bureaucrats in international organisations can or should impose uniform codes of communication across the world. In the same way that the statistical models will not work all the time, communications strategies will not work in all places.
So we can agree with Mr Greenspan's warnings. Central banking is an art and it would be dangerous for any policy maker to adhere slavishly to mechanical rules. But no one has proposed that we replace informed decision makers with a bunch of econometric model-builders armed with ever-faster computers. Instead, critics have asked that Mr Greenspan and his colleagues give us a clearer sense of what their objectives really are. We are happy to leave them alone to do their jobs; we just want to understand some of the details of what that job is. Would it really do that much harm to the policymaking process if the Fed were to announce a medium-term numerical inflation objective?
The writer is Professor of Economics at Brandeis University, Massachusetts, and a Research Associate at the National Bureau of Economic Research.