For an organisation blessed with as many economists of good repute, both on its policy-making committee, the FOMC, and on its staff, the Fed is remarkably obtuse when it comes to the role of core inflation in designing and publicly explaining its policies. Core inflation is the inflation rate of a subset of the standard bundle of goods and services used to track the behaviour of the general price level. It excludes food and energy products (oil, gas, coal, power etc.). Headline inflation is the rate of inflation of the whole (supposedly representative) bundle of goods and services – core and non-core. The Fed is confused and/or confusing about matters other than the role of core inflation in designing, implementing and explaining monetary policy. The Fed’s effectiveness is hampered by at least two other flaws. The first is a design flaw not of the Fed’s making. The Federal Reserve Act assigns the Fed three co-equal fundamental objectives: maximum employment, price stability and moderate long-term interest rates. Of this triple mandate for the Fed, the first objective is meaningless (maximum subject to what? whatever happened to full employment?), the second is sensible but only qualitative, and the third is weird. Fed officials, from the Chairman down, tend to ignore the third objective and speak of the dual mandate of the Fed rather than its triple mandate. When pressed, they mutter that moderate long-term interest rates (nominal/real?) really are not a macroeconomic monetary policy objective but refer to the financial stability role of the Fed. But that’s not what the Federal Reserve Act says. Of the three fundamental objectives, the only one to make economic sense, price stability, does not have a quantitative operational expression in the form of a price level target or an inflation target. Again the Fed cannot be blamed for this. Congress would not wear a numerical inflation target without a numerical (un)employment target. The second pointless idiosyncrasy of the Fed is that it pays more attention to a deflator-based index of consumer prices, the Personal Consumer Expenditure (PCE) deflator than the rest of the world’s central banks, who tend to focus on Consumer Price Indices (CPI). The PCE deflator is also unlikely to resonate well with the general public, who in all likelihood have never heard of it. The US Federal government’s index-linked debt is linked to the CPI, not the PCE deflator. It’s not clear that there is unanimity among the FOMC-12 (or even among just the 7 Board members) about whether the headline inflation measure they care about is the CPI or the PCE deflator. Greenspan was a PCE deflator man, but that was then. Bernanke has never expressed a strong preference for a deflator measure over a CPI measure. This matters, because a deflator measure of inflation (core or headline) is a Paasche or current-weighted price index while the CPI is a Laspeyre or base-weighted price index. Current-weighted price indices show lower inflation rates if there are any relative price changes than base-weighted price indices; historically, the PCE deflator rate of inflation has been about 0.5 percent per year below the corresponding CPI measure. Uses and abuses of core inflation It is essential to differentiate between what price index the Fed ‘cares about’ (if not ‘targets’) and whatever is the best predictor over the relevant policy horizon of the price index the Fed cares about. There is no doubt that the Fed cares about headline inflation and not about core inflation. It does not believe that Americans don’t eat, drink, drive, or use heating and air conditioning. Even if we knew what the FOMC’s preferred headline inflation measure was (PCE deflator or CPI), the location of the comfort zone (the Fed’s Congress-proof euphemism for inflation target) is ambiguous. Bernanke in his pre-Chairman days appeared to have a PCE deflator comfort zone centered symmetrically around a 1.5 percent per year inflation rate. Mishkin appears to want to shift this up by about 0.5 percent. Given the historical difference between the CPI and PCE deflator rates of inflation, a PCE deflator comfort zone centered on 1.5 percent would correspond to a CPI comfort zone centered on 2.0 percent. Two percent is also the Bank of England’s inflation target, and ‘below but close to two percent’ is the medium-term inflation target of the ECB. The Bank of England’s and ECB’s CPIs are quite different from the US CPI and it’s likely that two percent inflation for a Eurozone-style CPI would mean a somewhat higher rate of inflation for a US-style CPI. Mishkin’s apparent preference for the center of the comfort zone could therefore well be closer to the UK and Eurozone inflation targets than Bernanke’s pre-Chairmanial comfort zone center. It is a strange spectacle, though, to watch the members of a monetary policy making committee argue in public about the numerical value of an inflation target that the institution they serve cannot admit to having. So if the Fed does not care intrinsically about core inflation, it must care about it instrumentally, that is, as an indicator or predictor of future headline inflation in the medium term – the one to three year horizon over which monetary policy decisions today have their effect. It may well be that historically, that is, using data for the Greenspan-Bernanke period 1987-2007, core inflation has been a better predictor of headline inflation than headline inflation itself. So-called ‘Granger causality’ tests of incremental predictive content support the view that, using monthly data for the period 1987M01-2007M05, core inflation Granger-causes headline inflation, that is, it ‘helps predict’ future headline inflation (for both the CPI and the PCE deflator) even when the predictive content of past headline inflation for future headline inflation is allowed for. Conversely, headline inflation does not Granger-cause core inflation. However, over a longer period (1957M02-2007M05), we find that headline inflation Granger-causes core inflation and that core inflation Granger-causes headline inflation; in fact, headline inflation Granger-causes core inflation at a higher level of statistical significance than core inflation Granger-causes headline inflation. Of course, even for the period 1987-2007, the fact that core inflation Granger-causes headline inflation does not mean that past headline inflation does not help predict future headline inflation. In fact, past headline inflation does help predict future headline inflation, but so does core inflation, even after allowing for the predictive content of past headline inflation. Core goods and services tend to be subject to nominal price rigidities – they are Keynesian goods and services if you want. I’ll call them sticky price goods. These prices often are governed by long-term contracts, that do not have inflation indexation clauses and other nominal contingencies built into them. Non-core goods, like agricultural commodities, oil, natural gas, metals etc. tend to have their prices set almost continuously in markets that look and behave like auction markets. Let’s call them flex-price goods. Not surprisingly, for much of the 20th century, core inflation therefore has been both less volatile and more persistent than the inflation rate of non-core goods. The application of statistical methods to data generated by economic processes is a most hazardous exercise when you don’t understand the economic, social, political and technological phenomena in question. Too many technically highly qualified statisticians and econometricians are babes in the woods when it comes to understanding the economic, social, political and technological fundamentals driving these processes. The ruling conviction at the Fed that even today, current core inflation is the best predictor of headline inflation in the medium term (say the next 1 to 3 years) that the Fed operationally cares about, is a leading example of bad and dangerous use of statistics – the statistical analogue of driving a car while looking only in the rear-view mirror. I am happy to accept as a fact the historical empirical regularity that, for the period 1987- 2001, say, core inflation may have been a useful predictor of future medium-term headline inflation. Then the world changed: a rapidly growing fraction of 2.3 billion Chinese and Indian consumers and producers became fully engaged in the global economy. They entered as suppliers of core goods and services and as demanders of (non-core) commodities. The result has been a major, persistent and continuing increase in the relative price of non-core goods to core goods. That means a long-lasting excess of the inflation rate of non-core goods over the inflation rate of core goods. The evidence for this is in Charts 1 and 2 below, which show the behaviour since 1970 of the ratio of the headline price index to the core price index for both the CPI and the PCE deflator. Apart from the wild decade of the 1970s, with its two oil price shocks, the increase in the relative price of non-core goods to core goods since 1998/99 is unprecedented, as pointed out to me by John Makin of Caxton Associates. Acting as a bad statistician who does not realise that the data generating process of the time series he relies on for his forecasts has been subject to a structural break, the Fed continues to treat core inflation as a good predictor of medium-term headline inflation. As a result, inflation got away from the Fed and the Fed did not even realise it. The point has been made well by Charlie Bean, Chief Economist of the Bank of England and an internal member of its Monetary Policy Committee (see http://www.bankofengland.co.uk/publications/speeches/speaker.htm#bean). It has also been made repeatedly by Mervyn King, Governor of the Bank of England, and by Governing Council members and staff of the ECB. But the Fed keeps blundering on – a living example of the dangers of Groupthink and confirmation bias. It is clear that the current phase of the globalization process, with Chindia entering the global markets massively as suppliers of core goods and as demanders of non-core goods, has by no means run its course. Hundreds of millions of rural labourers in both China and India will move in the years to come from the remnants of the old autarkic national economies into the globally integrated economy. This is therefore a time par excellence when looking out of the window and recognising or anticipating a structural break is especially important. Forecasters should predict a continuing, structural increase in the relative price of non-core goods to core goods, regardless of what the past time-series properties of core and headline inflation may have been. Of course, a cyclical downturn in the global economy would bring about a lower relative price of non-core goods, but that would not refute the reality of the underlying trend towards stronger commodity prices. The argument so far, that an increase (decrease) in the relative price of non-core goods to core goods will raise (lower) headline inflation relative to core inflation would be true even if the change in the relative price of non-core to core goods did not have any effect, even temporary, on the rate of headline inflation. However, because core goods are sticky-price goods while non-core goods are flex-price goods, a change in the relative price of non-core to core goods will, cet. par., have an effect on the rate of headline inflation. The cet. that should be held par. for this statement to be correct, are current and future nominal policy interest rates. Specifically, the following proposition holds:
When core goods are subject to nominal price rigidities but non-core goods prices are flexible, a relative demand or supply shock that causes a permanent increase (decrease) in the relative price of non-core to core goods will, for a given path of nominal policy rates (short-term nominal interest rates), cause a temporary increase in the rate of headline inflation, as well as a temporary reduction in the rate of core inflation.
This pattern is clear from Charts 3, 4, 5 and 6, which plot the difference between the headline inflation rate and the core inflation rate on the horizontal axis against the rate of headline inflation. This is done, in Charts 3 and 4 for the CPI over, respectively, the 1957-2007 period and the 1987-2007 period. It is repeated in Charts 5 and 6 for the PCE deflator over, respectively, the 1959-2007 and the 1987-2007 periods. Therefore, when there is a continuing upward movement in the relative price of non-core goods to core goods, core inflation will be poor predictor of future headline inflation for two reasons. First, even if headline inflation were unchanged, core inflation would, for as long as the upward movement in the relative price of non-core goods continued, be systematically below both non-core inflation and headline inflation. Second, for a given path of nominal interest rates, the increase in the relative price of non-core goods will temporarily raise headline inflation above the level it would have been if there had been no increase in the relative price of non-core goods to core goods. When the increase in the relative price of non-core goods comes to a halt, headline inflation will not decline below the level it would have been at without the increase in the relative price of non-core goods. It would take a reversal of the increase in the relative price of non-core goods for headline inflation to fall below the path it would have been on in the absence of the increase in the relative price of non-core goods. The implication is that for many years now (starting around the turn of the century), the Fed has missed the ball completely on the implications of the global decline in the relative price of core goods for the usefulness of core goods as a predictor of future headline inflation. Medium-term inflationary pressures have been and continue to be higher than the Fed thinks they are. If the Fed ever recognises the truth, the Federal Funds rate will have to be higher than the Fed and the markets believe today.