Inflation targeting comes to Washington DC

Introduction: changes in the conduct of monetary policy at the Fed

It has taken a while, just under two years since Ben Bernanke took over from Alan Greenspan as Chairman of the Fed, but the deed now is done: the Fed has moved to de-facto inflation targeting. It will continue to be an inflation targeting that dare not speak its name. The Fed has introduced inflation targeting inside the twin Trojan horses of improved communications and greater transparency. An indeed, these proposals are likely to improve the clarity of the Fed’s communications to the market and the public at large and to enhance its transparency. But there is more that that involved. I discern a movement away from the Fed’s symmetric dual mandate to a greater emphasis on price stability as the primary objective of monetary policy. This reform will not take the Fed the whole way towards the lexicographic or hierarchical inflation targeting of the ECB and the Bank of England, whose primary objectives are price stability and without prejudice to, or subject to, the price stability target being met, output, employment and all things bright and beautiful. It does, however, represent a significant step in that direction.

The Fed’s modus operandi under Greenspan could be described as formally symmetric but in fact biased towards low unemployment, extremely flexible inflation targeting without a firm, let alone a numerical, inflation target. The existence of the ‘Greenspan put’, referring to the asymmetric reaction of the Fed’s policy rate to asset price increases and asset price declines (and specifically to increases and declines in equity prices) remains a hotly disputed issue. There can be no doubt, however, about another asymmetry in the reaction function of the Greenspan Fed. With unemployment at or near the best guestimate of the natural rate, when faced with the choice between a rate cut that would reduce the likelihood of an increase in the unemployment rate at the expense of a higher risk of excessive inflation, or tighter monetary policy that would increase the likelihood of higher unemployment but would lower the risk of excessive inflation, the Greenspan Fed would opt for lower unemployment. 

This is no longer true for the Bernanke Fed. This may in part reflect differences in the interpretation of the Fed’s mandate between the two Chairman, or differences in their view of the transmission mechanism of the Federal Funds target rate to inflation and unemployment. It may also reflect differences among the two Chairman in their willingness and/or ability to impose their own views on the majority of the voting members of the FOMC. I have the impression that the Regional Federal Reserve Bank Presidents have become more vocal, assertive, and influential than they were under Greenspan. While among the Regional Fed Presidents there is a range of views and objectives – there is at least one distinguished modern Keynesian among them, Janet Yellen of the Fed of San Francisco – the Regional Fed Presidents tend to give greater weight to maintaining price stability than to maintaining high levels of employment and output. While the time series of observations on the Bernanke Fed is still to short for meaningful statistical analysis, I believe we will be able to identify in due course this break in Fed behaviour in the direction of putting greater weight on price stability. 

The new ‘communication and transparency’ framework provides most of the ingredients for inflation targeting. As explained below, at the very least, the new three-year horizon for the forecasts will provide the equivalent of a numerical point inflation target. In my view it will do more than that. A de-facto numerical inflation target could just turn the Fed into a standard flexible inflation targeter, trading off deviations of inflation from its target against the output gap or the deviation of the actual from the natural rate of unemployment. It could also be just the first step on the road to something closer to lexicographic or hierarchical inflation targeting, which will be approximated more and more closely as the relative weight given to inflation in the objective function of the Fed increases over time. 

Governor Bernanke denies that the substance of US monetary policy making has been changed in any way: “The steps being taken by the Federal Reserve, I must emphasize, are intended only to improve our communication with the public; the conduct of policy itself will not change.” (Bernanke (2007)). I don’t know whether this is an example of what that well-known moral philosopher Mandy Rice-Davies referred to as “Well, he would, wouldn’t he”, or whether the statement is sincere.[1] In either case, I don’t believe it. 

The Fed will no doubt continue to pay lip service to what is deemed (both by the Fed itself and by Congress) to be its official, legal dual mandate – maximum employment and stable prices. Congress will continue to insist on parity for these two objectives, and Fed officials and Governors will nod and agree. As the least independent of the leading central banks, the Fed cannot afford to thumb its nose at Congress by openly admitting to having adopted inflation targeting, be it ever so flexible a variety of inflation targeting. Even under Greenspan, however, maximum employment was typically interpreted by the Fed as the highest level of employment that does not threaten price stability; the changes announced today move the Fed closer to making price stability the primary objective of monetary policy, as it is in the Euro area, in the UK and in Japan. Actions speak louder than words, and I expect that the weight given in future Fed rate-setting deliberations to the achievement of sustained low inflation will not only be greater than it was during the Greenspan years (such has already been the case in the past year or so), but will continue to grow in relative importance. 

Incidentally, the fact that the official monetary policy objectives actually specify a triple, not a dual mandate appears to have been forgotten: “The Board of Governors of the Federal Reserve System and the Federal Open Market Committee shall maintain long run growth of the monetary and credit aggregates commensurate with the economy’s long run potential to increase production, so as to promote effectively the goals of maximum employment, stable prices, and moderate long-term interest rates.”[2] This is probably because (a) nobody knows what moderate long-term interest rates are, and (b) nobody knows why the Fed should care about them and target them. 

The move of the Fed towards de facto inflation targeting came in the form of an announcement of changes in the Fed’s forecasting and communication procedures, in a speech by Chairman Ben Bernanke (see Bernanke (2007)) at the Cato Institute in Washington DC on November 14, 2007. From the point of view of inflation targeting, there were three material changes[3]:

(i) A longer forecast horizon – three years rather than two for real GDP growth, the unemployment rate and inflation.

(ii) Forecasts for both headline PCE inflation (referred to by Bernanke as ‘overall’ PCE inflation) and core PCE inflation

      (iii) More information about the dispersion of individual forecasts 

The longer horizon matters, because, even allowing for long, variable and uncertain lags in the effects of monetary policy, over a three year horizon a monetary authority like the Fed should expect to hit its inflation target, if it has one. Second, the Fed forecasts are made on the assumption of ‘appropriate monetary policy’, that is, not on the basis of a constant Federal Funds target rate or on the assumption that the future path of the Federal Funds target rate is that implied by the market yield curve. This reinforces the presumption that at a three year horizon, if not earlier, the forecast for inflation should equal the inflation target. 

As regards the forecasts for the unemployment rate and the growth rate of real GDP, the influence of monetary policy has all but dissipated after three years. The unemployment forecast at the three-year horizon can therefore be interpreted as an estimate of the natural rate of unemployment in three years’ time. Likewise, the growth rate of real GDP in three years’ time can be interpreted as the estimate of the growth rate of potential output in three years time. 

Unlike the Bank of England, where the inflation forecast is, in principle, the forecast of the entire Monetary Policy Committee, and unlike the ECB, were the inflation forecast is a staff forecast rather than the forecast of the Governing Council, the Fed’s forecasting procedure involves each of the participants in the FOMC meeting — the Federal Reserve Board members and all the Reserve Bank presidents (not just those voting) — providing their individual forecasts. In the future there will be forecasts for growth, unemployment, core PCE inflation and headline PCE inflation. These forecasts will naturally differ, and in the future I expect us to see the entire joint distribution of the 19 forecasts (seven Board members and 12 Regional Fed Presidents) for each of the four variables. 

If inflation were the only objective, and if rate setting were a majority decision, the median voter theorem would apply and the effective or operational inflation target could be backed out of the distribution of forecasts as the median inflation forecast (that is, the one that has half of the forecasts below it and half above — number 9, if the forecasts were ranked).[4] With a dual objective, and with a non-majoritarian decision making procedure (the Chairman carries much greater weight than the rest of the voting members – there is an unwritten rule that no more than two votes are cast against the Chairman’s preferred action – and there is a strong desire to achieve a consensus if at all possible), the simple median voter result will not hold. Nevertheless, knowing the mean, the median and the modal forecast for inflation will provide the markets and other observers with essential information about the inflation target. We will also know the dispersion of the forecasts (that is, the extent of uncertainty and/or disagreement among the forecasters) and the skew, that is, the ‘balance of risks’. All these are key ingredients in standard inflation targeting. 

One element of conventional inflation targeting (both lexicographic and flexible) that we don’t have and are unlikely to get in the near future, is an announced numerical inflation target (typically a point target, although some countries, including New Zealand, have a target range). It would be helpful to have such a publicly stated numerical inflation target, to set alongside the target implied by the ‘central tendency’ of the two-year and three-year inflation forecasts, but for the moment, this would, politically, be a bridge too far. The Congress would no doubt insist on a dual target for employment, unemployment or output growth, and the Fed will not want to risk getting itself into a position where it is expected to target something that it will, in all likelihood, be unable to deliver.

Before he assumed the throne on 20th street and Constitution Avenue, Ben Bernanke made statements to the effect that 1.5 percent per annum was the centre of his comfort zone for core PCE inflation. This comfort zone was generally thought to range from 1.0 to 2.0 percent (see Bernanke (2005)).[5] Since becoming Chairman, Ben Bernanke has not made any statement that could be interpreted as implying a numerical inflation target. Recently, Frederic Mishkin, one of the Governors, has indicated that his comfort zone for core PCE inflation was more likely to be centred on 2.0 percent (Mishkin (2007a)). For sake of argument, let’s split the difference and assume that the centre the inflation comfort zone for the core PCE deflator is 1.75 percent. This is likely to reflect a slightly higher inflation target than the ECB’s below but close to 2.0 percent for the Harmonised Index of Consumer Prices (HICP) and the Bank of England’s 2.0 percent target for that same HIPC, which it calls the CPI. This is because the US HICP index (available only since 1997) behaves almost exactly like the US CPI index, and because the US CPI index has systematically increased faster than the US PCE index, as is clear from Figure 1.



During the 32-year period 1965-2007, the headline CPI increased 24 percent more than the headline PCE index, an average annual difference in inflation rates of around 0.67 percent.  the core CPI increased 28 percent more than the core PCE index, an average annual inflation rate difference of 0.77 percent. A core PCE target of 1.75 percent therefore would correspond to a core CPI target of 2.5 percent. If the US CPI is indeed not just similar in design and to the US HIPC but also to the European HICP (and the UK’s CPI), then the implicit Fed inflation target is about half a percentage point above the UK inflation target, and just over half a percent above the Euro area inflation target. 

As is clear from Figure 2, there has been a very slight increase in the ratio of the headline CPI index to the core CPI index since 1965, about 2.7 percent over almost 32 years, and an even smaller increase in the ratio of the headline PCE index to the core PCE index, about 2.3 percent over almost 32 years.


There have, however, been big long-term swings in the headline to core ratios over the period. The large increase during the first oil price shock, in 1973-74 and the peak in 1980 following the second oil price shock are clear from the date, as is the 20-year decline since then and the reversal of this decline because of the growth of the BRICs since about 2000. 

And the target is: headline inflation

Americans eat and drink (the first of these not infrequently to excess). They live in houses that are bone-chillingly air-conditioned in the summer and over-heated during the winter. They drive cars with lamentable energy inefficiency. So, yes, food and energy are part of the American consumption bundle, which makes low and stable headline inflation the proper objective of monetary policy. Because of the lags in the operation of monetary policy, inflation targeting means inflation forecast targeting. Anything that helps predict or infer headline inflation is therefore welcome and should be used in the headline inflation forecasting and targeting process. 

Core inflation – the inflation rate of a bundle of consumer goods and services that excludes food and energy.- may well, at certain times and under certain conditions, be a useful predictor of future headline inflation (see e.g. Mishkin (2007b) for a statement of this view). At other times it is likely to be a predictably dreadful predictor of inflation. This is the case since the beginning of the decade, as pointed out by Charlie Bean (2006) and myself (Buiter (2007a,b)) because globalisation has brought us a steady, secular increase in the relative price of non-core goods to core goods, as is clear from Figure 2.

It is true by definition that, if a forecaster is not predicting any future change in the relative price of core goods and services to non-core goods, then forecasting core inflation is equivalent to forecasting non-core inflation or headline inflation (and vice-versa). It may also well be factually correct that the Fed has, historically, not forecast any changes in the relative price of core and non-core goods. That would hardly have been consistent with optimal forecasting. There is a marked positive association between the global business cycle and the relative price of non-core goods, and intelligent forecasters should at times lean out of the window to make sure that there are no new developments under way (of a kind not captured in past statistics) that may be relevant to the relative price of core and non-core goods and services. Globalisation, with China, India and the other BRICs entering the global markets as suppliers of core goods and services (manufactured goods, back-office functions and call centres) and as demanders of non-core goods (food, energy and other commodities) is one such development. 

Both globalisation and the global business cycle have contributed to the strength of commodity prices since the end of the slowdown that followed the late-2000 tech bubble crash, as shown by the increase in the headline price indices relative to the core price indices in Figure 2. 

Because non-core goods prices tend to be flexible while the prices of core goods and services are subject to nominal rigidities caused by long-term incomplete nominal contracts, shocks to the relative global demands for and supplies of core and non-core goods will have short-run effects on headline inflation that would not be present if all prices were either subject to similar nominal rigidities or behaved as freely flexible ‘auction-style’ prices. Increases in the relative price of non-core goods to core goods and services therefore tend to be associated with higher headline inflation, as shown empirically in Figures 3 and 4. 






The announcement by the Fed Chairman of a new communications and transparency strategy for the Fed hides a decisive break with the way in which the Fed conducted monetary policy in the past. It amounts to the de-facto adoption of a flexible inflation targeting framework by the Fed, with the role of the numerical inflation target taken by the central tendency of the FOMC members’ inflation forecasts at a three-year horizon. It also represents another nail in the coffin of core inflation as the Fed’s preferred inflation indicator/predictor, and its replacement by a less mechanically statistical, more robust and thoughtful economic analysis of the fundamentals driving headline inflation in the medium term.


Figure 1 source: Bureau of Labor Statistics, Bureau of Economic Analysis

Figure 2 source: Bureau of Labor Statistics, Bureau of Economic Analysis

Figure 3 source: Bureau of Labor Statistics, Bureau of Economic Analysis

Figure 4 source: Bureau of Labor Statistics, Bureau of Economic Analysis


Bean, Charlie (2006), “Comments in response to a paper by Ken Rogoff – "Impact of Globalization on Monetary Policy" given at the Federal Reserve Bank of Kansas Symposium in Wyoming, USA on 26 August,

Bernanke, Ben S. (2005), “Remarks by Ben S. Bernanke at a Finance Committee luncheon of the Executives’ Club of Chicago”, Chicago, Illinois, March 8.

Bernanke, Ben S. (2007), “Federal Reserve Communications”, speech by Chairman Ben S. Bernanke at the Cato Institute 25th Annual Monetary Conference, Washington, D.C., November 14.

Buiter, Willem H. (2007b), “The Fed still hasn’t recovered the plot: core inflation versus headline inflation again”, .

Buiter, Willem H. (2007a), “The Folly of the Fed or: Why is the Fed so Hardcore?”

Mishkin, Frederic S. (2007a). "Inflation Dynamics," speech delivered at the Annual Macro Conference, Federal Reserve Bank of San Francisco, San   Francisco, March 23,

Mishkin, Frederic S. (2007b), “Headline versus Core Inflation in the Conduct of Monetary Policy”, Speech given at the Business Cycles, International Transmission and Macroeconomic Policies Conference, HEC Montreal, Montreal, Canada, October 20.

[1] Mandy Rice-Davies (born 21 October 1944) is famous mainly for her minor role in the Profumo affair which discredited the Conservative government of British Prime Minister Harold Macmillan in 1963. While giving evidence at the trial of Stephen Ward, Rice-Davies made the quip for which she is most remembered and which is frequently used by politicians in Britain. When the prosecuting counsel pointed out that Lord Astor denied having an affair or having even met her, she replied, "Well, he would, wouldn’t he?" (Source: Wikipedia). 

[2] Federal Reserve Act [12 USC 225a. As added by act of November 16, 1977 (91 Stat. 1387) and amended by acts of October 27, 1978 (92 Stat. 1897); Aug. 23, 1988 (102 Stat. 1375); and Dec. 27, 2000

(114 Stat. 3028).]

[3] In addition, the frequency of the forecasts was increased to quarterly from semi-annually, and there will be summary reports, containing explanations for the individual forecasts.

[4] This assumes that all differences in forecasts at a 3-year horizon reflect differences in inflation targets only, and not differences in models (views on the monetary transmission mechanism), or the (perceived) inability to achieve the inflation target (in expectation) at a three-year horizon, which could lead to different combinations of inflation and unemployment being targeted not because of differences in inflation targets, but because of different weights being placed on deviations of inflation and unemployment from their target values.

[5] “My own guess is that core PCE inflation in 2005 will be slightly higher than its 2004 rate of 1.6 percent, though likely remaining within what I think of as the "comfort zone" of 1 to 2 percent”, Bernanke (2005).

Maverecon: Willem Buiter

Willem Buiter's blog ran until December 2009. This blog is no longer active but it remains open as an archive.

Professor of European Political Economy, London School of Economics and Political Science; former chief economist of the EBRD, former external member of the MPC; adviser to international organisations, governments, central banks and private financial institutions.

Willem Buiter's website