August 19, 2010 7:35am |
The battle to avoid deflation in the developed world could prove to be a long one, with twists and turns which could last for many years. In July, the core CPI data in the eurozone were somewhat firmer than expected, as were the core PPI data in the US. This has led some economists to suggest that underlying price pressures are beginning to rise again, and that the deflation scare is over. Would that that were true. Some interesting new evidence from the IMF suggests that while outright deflation might be avoided, at least for a time, the developed economies could soon get stuck in a kind of limbo land, with inflation remaining unhealthily close to zero for a very long period.
The standard theory of the inflation mechanism, used by all the major central banks, assumes that the rate of inflation is driven by the size of the output gap in the economy (i.e. the extent of spare capacity) and by expected inflation. (In some formulations, the change in inflation appears on the left hand side of the equation - see this useful and accessible summary by John Duffy at the University of Pittsburgh.) When the output gap rises, we would therefore expect the inflation rate to decline. At present, most estimates of the output gap in the developed economies are historically very high at around 4 per cent, so it would appear to be fairly certain that core inflation will continue falling from the 1 per cent now reached in the US and Europe. Deflation, here we come?
It is important to ask how this apparently strong conclusion could be proved wrong. Maybe we have over-estimated the output gap, but surely not by as much as 4 per cent. Maybe expectations of inflation have risen in response to the printing of money by central banks, but none of the direct sources of expectations data confirms this. Maybe high commodity prices have been over-riding the effects of the large output gap, but oil prices have been stable or falling recently.
That leaves the heart of the matter, the relationship between inflation and the output gap itself. The key question is whether this is sufficiently reliable to support the strong conclusion that deflation is now a severe threat. The econometric evidence on this is voluminous, but my 10 cent summary would be as follows. Yes, the predicted relationship does emerge from the data, but it has been much weaker in the past decade than it used to be. And at very low rates of inflation, it may break down altogether.
Andre Meier at the IMF has just published a paper examining all of the episodes in the past 40 years in which developed economies have experienced persistently large output gaps (which he calls PLOGs). This is an extremely useful exercise, in the spirit of quantitative economic history, rather than full blown econometrics, and probably all the better for that. Meier identifies 25 separate episodes (prior to the current examples, of which there are another 15) in which a national output gap has exceeded 1.5 per cent for at least 8 consecutive quarters. In shorthand, we can label these episodes as deep recessions.
As the standard theory would predict, inflation fell in virtually all of these episodes. Interestingly, Meier finds that the extent of the decline in inflation is proportional to the rate of inflation in the year before the deep recession starts (see graph). 
This implies that countries which start out with a very high rate of inflation (e.g. the UK from 1980-83) see the biggest drop in inflation when a deep recession occurs. As a rough rule of thumb, the inflation rate seems to drop by about one fifth of the original rate for every year that the episode lasts, with much of the improvement coming in the first couple of years.
Since many developed economies, including the US, embarked on their present deep recessions with inflation at around 2-2.5 per cent, this rule of thumb would indicate that inflation should be dropping at about 0.5 per cent per annum, which is almost exactly what has been happening. So far, then, so bad.
But the really interesting thing is that this rate of decline in inflation appears to peter out altogether as inflation itself approaches zero. Admittedly, there are only two such episodes in the database - Japan in 2001-03 and Sweden in 1992-94. But they are nonetheless important, because we do not have much else to cling on to. These episodes of downward rigidity in inflation are probably explained by the fact that price and wage setters are extremely reluctant actually to cut prices or nominal wages in absolute terms, so a really massive shock (perhaps even bigger than the recent credit crunch) is needed to force the overall inflation rate below zero.
Here is my (highly) speculative conclusion. A recession as deep as the one we are experiencing will continue to put downward pressure on inflation, but this will get less and less obvious as we approach zero. There will be times when higher commodity prices, or declining exchange rates, will lead to temporary increases in inflation, counter to the dominant deflationary trend.
An even bigger recessionary shock might be needed to turn inflation meaningfully negative. This probably can be avoided, enabling some people to continue denying that deflation is a serious threat. Nevertheless, in this netherland of zero inflation, many of the adverse dynamics of debt deflation (identified by Irving Fisher in the 1930s) will be working away, insidiously, under the surface.
If deflationary forces go undercover, as they did in Japan, then they may become particularly difficult to defeat.
