Why inflation targeting is not for the long-run

By: Wolfgang Munchau

Since the late 1980s, increasing numbers of central banks have adopted the strategy of inflation targeting. The idea was to provide an anchor for expectations of future inflation and interest rates – both by means of the target itself and by means of an inflation forecast. While this strategy has worked well for some central banks – notably the Bank of England – it did not work out as well for others.

The Swedish Riksbank was one of the early adopters of inflation targeting. Yet, in Sweden, the strategy has come unstuck – though in a most peculiar way. Like many other central banks, the Riksbank defined price stability as an annual increase in consumer price inflation of 2 per cent, with a one percentage point margin on either side. Yet between January 2004 and February 2006 annual consumer price index inflation persistently undershot the lower band of the target, hitting a low point of minus 0.4 per cent in February 2004. Under its own definition, the Riksbank produced price instability. Of course, this description does not make logical sense. How can you have price instability when inflation is almost exactly zero?

In theory, inflation targeting is a relatively flexible monetary policy framework that gives central banks discretion to pursue particular strategies within the confines of the upper and lower tolerance levels. Ben Bernanke, now chairman of the Federal Reserve, once described inflation targeting as “constrained discretion”. The bands act as the constraint. Going outside the bands would have required the Riksbank to take swift policy action to get inflation back towards 2 per cent – or at least above the lower band of 1 per cent. If it had taken its own strategy seriously, the Riksbank should have opted to pursue a Japanese or Swiss-style policy of zero nominal interest rates. But Swedish policy rates only came down to a low of 1.5 per cent in June 2005, at which level they stayed for only seven months.

Instead of cutting rates further to meet the target, the Riksbank subsequently raised interest rates. In a memorandum published two weeks ago, it “clarified” its policy – a euphemism for a change of strategy. It said it remained committed to its CPI inflation target and to the band, but made two further important qualifications. First, it will look at various measures of core inflation on the grounds that “there is no single measure of inflation that consistently indicates the appropriate stance of monetary policy”.

Second, the Riksbank will also take a close look at asset prices and other financial variables.

On the face of it, this is a perfectly sensible strategy for a central bank to adopt. The Riksbank had good reasons to deviate temporarily from its strict inflation target. At a time when inflation was close to zero, the country experienced high economic growth and a housing boom. The Riksbank concluded that a Dutch- style boom-bust development in the housing market could destabilise the economy and have negative effects on price stability. For that reason, the decision not to cut policy interest rates below 1.5 per cent was probably correct.

But while the Riksbank’s monetary policy was sensible, it was not consistent with its inflation target framework. At the very least, the Riksbank should have got rid of the band, while remaining committed  to the central target. The strategy  as it is formulated now is simply  not credible.

Is the inconsistency of the Swedish central bank a sign of the beginning of the end of inflation targeting, or – at least – the end of the strictly banded version? It may well be. The Swedish experience demonstrates that inflation targeting can work extremely well as a self-fulfilling prophecy for some time. But once this system faces stress – such as a persistent undershooting and overshooting of the target range – central banks are confronted with the choice of either acting against their better judgment to preserve the integrity of the strategy, or to risk their credibility. Most central banks would choose the latter option, while trying to pretend otherwise. The demise of monetarism in the 1980s played out to a very similar script.

The purpose of targets is to provide a nominal anchor for inflationary expectations. It is probably a good idea for central bankers to say what level of inflation they are comfortable with and what they consider excessively high or low. The Federal Reserve and the Bundesbank (and now the European Central Bank) managed to do this for years without adopting a formal inflation target. An explicit or implicit inflation target of
2 per cent seems to have emerged as most central banks’ preferred choice. Targets should also be symmetrical. Positive or negative deviations should be treated with equal concern.

But a fixed band is probably a bad idea unless you are determined to defend it. Since this is evidently not the case, inflation targeting is a violation of Occam’s razor: the principle that theories – and models – should use as few explanatory variables as possible. Inflation targeting is an overspecified strategy because of the fixed band, not the inflation target itself. Take the overspecification away and you end up, approximately, with a monetary policy strategy similar to that of the Fed or the ECB. Strict inflation targeting will ultimately prove just as impractical as other rule-based systems before it.

Copyright The Financial Times Limited 2006