Experimental feature

Listen to this article

Experimental feature

With Ben Bernanke, the untested chairman at the Federal Reserve, and other central bankers pondering how far to raise interest rates, anxiety in asset markets runs high. Many imagine that relief might be found in an automated monetary policy. A monetary “rule”, advocates say, would serve to remove the unnecessary uncertainty added by the Fed’s possession and exercise of “discretion”, leaving only the inherent uncertainty about future market outcomes.

“Inflation targeting”, as its champ­ions know, would require the central bank to follow a rule so it could prove to legislators that it aimed for the inflation rate target. Mr Bernanke himself belongs to a generation of economists whose schooling has cultivated a taste for rules – fixed, hence predictable, and mechanical, hence transparent.

But would a monetary rule make sense in the spontaneous and generally innovative economies of, say, the US and the UK? It is a good question.

The rule brilliantly argued for by Milton Friedman in the 1960s was a money supply rule. A fail-safe feature it had was that if the rule somehow made money too tight, product and labour markets would come to the rescue by scaling back price and wage increases, thus sparing the nation a recession or swiftly reversing one. What sunk that sort of rule was a collapse of the structural relationship between money and inflation.

The rule advocated now, however, would set the short rate of interest. In the standard textbook rule, created by John Taylor, the Stanford University economist, in the 1980s, the real short rate (the money interest rate less the inflation rate) is set higher the greater is inflation, and lower the greater is unemployment. If perchance the inflation rate is at its “target” and unemployment is at the “natural unemployment rate”, the real interest rate is to be set equal to the “natural interest rate” – the real rate businesses could afford to pay if unemployment were
at the natural unemployment rate.
Mr Taylor took the natural rates to be constants.

Commitment to an interest rate rule would be dangerous because the product and labour markets could not rescue the economy from the consequences of an error in the rule. The natural interest rate is complicated to estimate.* Should the natural interest rate be below the level the rule took it to be, no fall in prices and wages could restore unemployment to its natural rate: their fall would pull down the money supply with them, leaving no net restorative effect.

That was Keynes’s message in the 1930s, when the Bank of England’s tacit policy was to keep rates around what it supposed the natural interest rate level to be, with little knowledge of where that level was.

The possibility exists, then, that the particular rule selected would leave the natural interest rate well below the real interest rate required by the rule in some periods, thus risking a slump, and well above it in other periods, thus risking rising inflation. A more dire risk is that the rule selected proves to be pitched too high (or too low) all the time, owing to a large overestimate (or underestimate) of the range in which the natural rate will lie over the future.

Are these likely possibilities? Research of mine and others in the 1990s strongly suggests that the natural interest rate and the natural unemployment rate have taken long swings and experienced sharp shifts to new territory.**

Defenders of interest rate rules reply that a central bank can deal with moving natural interest and natural unemployment rates. It can draw on an econometric model to estimate how exactly the natural interest rate varies with its determinants – wealth, the stocks of business assets, overseas interest rates and incomes, and business prospects for the future. With that estimated structural relationship in hand, the bank can calculate what the natural interest rate is implied to be this month or next, using the current observed values of its determinants. This formula for the variable natural interest rate and the formula for the natural unemployment rate are thus embedded in the rule. The formulae are to be constant to meet the rule advocates’ desire for commitment to a fixed structure. The bank may not fiddle with them.

Such an extended rule might improve on the Taylor rule. Yet it raises questions in practical minds. Might not the bank’s econometric model prove to be way off the mark? Might not the economy’s structure change? Does that matter?

In recent articles and a forthcoming book on exchange rates and risk, Roman Frydman and Michael Goldberg have got to the bottom of what is wrong with fixed rules: in entrepreneurial economies, people possess only “imperfect knowledge” about the economy’s structure. And the structural relationships themselves are “unfolding” as learning and creativity occur. It follows that commitment to a fixed rule would add new sources of uncertainty – possibly worse than the sources removed by barring “discretion”.

At a time when seismic shifts may be coming in the global economy, we may need from our central bankers not a rule but all the flexibility and inventiveness they have got.

* H. T. Hoon, E. S. Phelps and G. Zoega, “The structuralist perspective”,
W. Semmler (ed.), Monetary Policy and Unemployment (Routledge, 2005). **Phelps, “Structural Slumps” (Harvard, 1994); Phelps and Zoega, “Structural booms”, Economic Policy (2001)

The writer is director of the Center on Capitalism and Society, Columbia University.

Get alerts on World when a new story is published

Copyright The Financial Times Limited 2019. All rights reserved.

Comments have not been enabled for this article.

Follow the topics in this article