I once drove Milton Friedman from San Francisco to the Hoover Institute while he talked to my dad about his book Free To Choose. (Being dad's unofficial chauffeur for things he wanted keep secret - principally his visits to the hospital and meetings with, as yet, unsigned authors - was my penance for being the black sheep elder son.) Like any sensible author, Friedman spent most of the time talking about royalties, but I do remember a brief exchange on the subject of money. Since the subject of money supply was one of the book's themes, shouldn't there be a glossary or some definition of what it was? It could be put in the back, said dad. Friedman's response was to laugh. That would, he said, make the appendix ten times the size of the book.

C. Kin adds:

Central bank forecasting Published: January 30 2007 11:47 | Last updated: January 30 2007 20:57

Milton Friedman, in one of his final interviews, suggested that monetary policy should be run by a computer. Since the future is uncertain, any interest rate mechanism will make wrong decisions. But humans add another flaw. Monetary policy requires managing expectations of future inflation and interest rates. Policymakers must be able to communicate effectively. Ben Bernanke's initial difficulties and the Bank of England's current woes reflect this challenge. The problem is not policymakers' views but rather that few can understand what those views actually are.

Sweden's Riksbank, the world's oldest central bank, has just joined a small group of institutions with a no-nonsense solution: policymakers publish a forecast of where they expect to set interest rates in the future. This is not as radical as it sounds. If they are competent, they should have a view. And it forms another step towards transparency. Europe's central banks once made inflation forecasts on the assumption of constant interest rates - a pretty silly premise. Now the European Central Bank and BoE assume a more realistic market yield curve. But they expend an inordinate amount of energy hinting at how plausible they believe that curve is. Far better just to say, explicitly, what they think.

Why do most central bankers see forecasts as the last taboo? Partly, self interest - they would frequently be revealed as being wrong. Yet there are two credible objections. First, while financial markets are grown up enough to understand forecasts are uncertain, the general public might not be. Second, making explicit numerical rate forecasts by committee is difficult. Pioneering New Zealand, Norway and Sweden in effect have either a single dominant decision maker or small groups of bank insiders. The ECB's 19 rate-setters and the BoE's nine look unwieldy by comparison. Reformers can forget the Federal Reserve. It is not so long ago that a congressman told Alan Greenspan that he finally understood what the chairman had been saying. "I must have misspoken," was Mr Greenspan's famous reply.

Copyright The Financial Times Limited 2007

Rudolf Hauser adds:

The most important problem with central bank discretion is not the lack of transparency, although that certainly does not help, but rather, it's that central bankers might take actions that make the problems worse rather than better. The problem with a computer program is that it has to make some static assumption about some key variables that might in effect change over time. However, it is possible for a perceptive central banker to anticipate those changes and adjust central bank policy accordingly. It might be much less costly to adjust monetary policy than to force the market to adjust to a predetermined monetary policy.

Forgetting the many specific ways of implementation, there are basically two guides to monetary policy. One is to target high-powered money (the monetary base or in other words the sum of reserves held at the central bank and currency) or a definition of money selected that the Fed could reasonably control through the use of the monetary base adjustments. There are a number of problems here, including changes in the demand for money, the fact that other financial instruments aside from the chosen definition of money are quasi-money and could substitute for the chosen definition (which could itself be fully money - such as an IRA savings deposit), and effectively controlling that definition just by manipulating the monetary base. The other approach is to target the interest rate for high-powered money (the Fed funds rate) or some other interest rate relative to an equilibrium real interest rate. The main problem here is determining what that real equilibrium interest rate is, which includes being able to determine the market's inflation expectations included in the nominal interest rate. The advantage is that changes in the demand for money might not present a problem, and that real equilibrium rate might change over time with shifts in the savings/investment balance and changes in time preference of consumption. Discretionary central bank policy has often just focused on perceptions of the economic outlook and attempted to react to those forecasts by changing interest rates or other targets to moderate the impact of undesired directions in inflation or unemployment. Given that stable prices should be a central bank's key objective in its monetary policy, inflation targeting is a third alternative, with adjustments to interest rate, or money supply targets to be used more directly. Although standard assumption can be made for the demand for money (income velocity) and for the real equilibrium interest rate, setting up an automatic computer program for a policy aimed at inflation is more difficult. It need not be the case if one has a market forecast of future inflation based on an efficient market for inflation-indexed securities to plug into a program that then specifies what adjustments should me made to the more direct targets.

When one has an astute central banker such as Alan Greenspan, discretionary policy might work better. The more stable the environment, the more effective a computer program might be. I favored Friedman's computer approach of targeting M2 until the turbulent times of the 1990s when there appeared to have been significant changes in the demand for money. Until that time, monetary growth had been extremely stable under the Greenspan Fed. After that, monetary growth accelerated to accommodate increases in the demand for money related to various financial crises, etc. Relying on an arbitrary computer program might not have been as effective. In the past, discretionary policy often made economic performance much worse, as was particularly the case in the depression of the 1930s and inflation of the 1970s. Focus on recent performance often created a worse pattern of boom and bust. The Fed has learned something from those past mistakes, so it is possible that those mistakes are less likely to be repeated in the future. But one never knows, and reacting to new circumstances is not the same as avoiding exact repeats of past mistakes. Also, as new people come to the positions, those past lessons might not be as well remembered as they should be. At this point, I am reluctant to rely on a computer program, but I realize that relying on discretionary policies also has considerable risks. What should be the case is that a stable price environment with either no inflation or a predictable minimum rate of inflation should be the standard by which the Fed is judged.





Speak your mind


Resources & Links