Wolfgang Munchau, who writes for the Financial Times as well as the blog Eurointelligence, ruminates about inflation statistics and argues that economists and statisticians may be going down the wrong path in dismissing consumers’ subjective perceptions. He also has considerable doubts about hedonic adjustments (basically, the methodology for adjusting for the fact that computers and other devices have become cheaper for the same performance, and that even seemingly mature products, such as cars, have features they lacked a decade ago (think GPS and more self-diagnostics).
While his post is helpful, he misses some of the other adjustments that, at least in the US, lead to the official understatement of inflation. The Boskin Commission in late 1996, which was chartered to adjust the CPI calculation (remember, at this point Social Security payments were indexed to CPI) rather conveniently concluded that CPI overstated inflation by 1.1% in 1996 and roughly 1.3% per year in prior years. And of course, the CPI methodology was then adjusted to produce lower numbers, therefore reducing Social Security payment increases.
What did the Boskin Commission think was out of line? According to Wikipedia:
The report highlighted four sources of possible bias:
Substitution bias occurs because a fixed market basket fails to reflect the fact that consumers substitute relatively less for more expensive goods when relative prices change.
Outlet substitution bias occurs when shifts to lower price outlets are not properly handled.
Quality change bias occurs when improvements in the quality of products, such as greater energy efficiency or less need for repair, are measured inaccurately or not at all.
New product bias occurs when new products are not introduced in the market basket, or included only with a long lag.
So the Boskin report would have us believe that if I switch from steak to hamburger because beef prices are up, we should only capture the change in how I consume (ie, inflation is new hamburger/old steak price, not new steak/old steak). That is patently bogus. Similarly, the outlet substitution seems rife for abuse (“Ooh, the number is going to be really bad this month! Can we find anywhere selling X cheaper so we can put that in the model instead?”).
There is a debate, usually in blogs, and usually with a whiff of conspiracy, about whether our inflation numbers are real, forged, or statistically so skewed as to underrepresent the true rate of inflation by quite a wide margin. I want to pick on this debate in this entry, not so much in support of one of those conspiracy theories, but in support of a more wideranging debate about how we measure inflation….
In the last five years, we have observed a phenomenon that we were not familiar with before, the phenomenon that people “feel” inflation to be higher than officially measured indices tell us. We hear a frequently used explanation: Cognitive science tells us that we give a higher weight to prices we actually see in supermarkets, or at petrol stations, than to prices with no explicit tags on them, such as rents, or telephony. The explanation is that the felt inflation is a purely psychological phenomenon.
You probably all remember that we heard exactly the same argument when we switched from national currencies to the euro. We felt there was substantial inflation, and as it turned out some shopkeepers used the confusion to raise prices, so there was a modest amount of real inflation. But if this had been a change-over phenomenon – we are talking 2002 – it would gone away. It did not.
I myself thought at the time that I was facing a significant rise in costs…. Now you can say: Well this is just you. You are not average. This does not apply to Mr and Mrs Average, and, in fact, I did believe this too.
But Mr and Mrs Average kept on complaining. The raise in euro prices was a factor during the 2005 No Vote in the Dutch referendum on the European constitution. It was almost certainly not the decisive factor, but people mentioned it when asked. In France, in particular, the biggest economic debate today is not the subprime crisis, but the apparent loss of purchasing power, which is economic illiteracy for a “rise in inflation”…..
Experts, and this applies to economists just as much as any engineer or scientist, often dismiss public comments about their subject area with varying degree of snobbish arrogance. This is particularly true about the debate about inflation. It is all in our heads, they say. The numbers don’t lie. The statistics are correct. We are unstable, not the index.
Well, I have my doubts – and this is not a psychological argument, but a statistical one. The first thing to notice is that inflation is not an observable real world variable, such as the number of widgets produced by a factory. Inflation is a statistic – technically a mapping from a probability space of random events into the positive real numbers. To arrive at a statistic, i.e. a number, we have to take multiple decisions, such as which sample of goods to include in our basket, since we cannot measure the universe of prices. We also have to choose a method how to weigh the results mathematically. You might remember the Paasche or Laspeyres price indices taught in Economics 101. In particular, we have to choose what to put into the basket, and what not.
In the 1950s, this exercise was easy. In the UK, I was told by someone who was actually involved in this exercise that they had chosen a typical working class family, and looked at their consumption basket, which was relatively uniform by today’s standards. They would pay rent, consume a certain amount of energy, obviously much of the spending went into foods, household goods, and some durables. The RPI, the retail price index, is still used today by ordinary people as their favourite measure of inflation (and also by wage negotiators). It has been significantly higher than the CPI, the index targeted by the Bank of England.
The reason for this discrepancy is, of course, related to what we put into the basket and to the adjustments we choose to make. We make lots of adjustments. If the price of a family computer at your local hardware costs €1000 today, and €1000 in one year’s time, we calculate this as a fall in prices, because the quality of the computer has presumably increased. I have problems with this now ubiquitous concept of hedonistic pricing because we are double-counting. The improvement in quality is the result of a rise productivity – which is a real variable. So the improvement in quality raises nominal growth in the numerator, and it lowers the price in the denominator, in other words, we double-count the effect. It may well be that we have been consistently underestimating the rate of inflation, and overestimating the rate of real productivity growth. Since the US uses the hedonic pricing more consistently than the Europeans (I think, please correct me if I am wrong on this one), the problem would be worse in the US than in Europe.
There is a website called Shadow Government Statistics, for whose accuracy I cannot vouch, which claims that the pre-Clinton era inflation index shows current inflation at close to 8%, while opposed official CPI inflation is only half that level. Here is the chart. What makes me a bit doubtful is that the higher series is an almost perfect image of the lower series (just follow it turn for turn), so that it may be calculated as actual inflation x%. That would not be a very acurate way to do this.
But let us suppose for a moment that series is correct. If US inflation were really 8%, this would mean that interest rates in the US have been negative at all times in the last 10 years. It would mean that 10-year treasuries, which yield only a little over 4%, are massively mispriced, that a bond price crash of historic proportion would beckon, essentially wiping out a large amount of China’s and Russia’s wealth – countries that have heavy investors in the US. It would be a global economic catastrophe. So we are not going to switch back with ease and pleasure. There are many vested interests in not doing so.
I do not want to discuss the merit of this particular statistic – which I cannot – but I believe strongly that the Fed is absolutely wrong to target a core-inflation index (and it is not even doing that with any great conviction and success). Core inflation is supposed to be more stable, as it excludes volatile categories of food and energy, but both categories have not been volatile, but persistently rising. To exaggerate a little (well, ok, a lot): All the troublemakers are taken out of the basket, the rest is adjusted.
But if some of the criticisms of the modern inflation indicators are even remotely correct, it would not only mean that we are about to return to a 1970s period of stagflation, with its double-digits inflation rates in the US and in some European countries. With the Fed now swamping the market with cheap money as though there is no tomorrow, it could be a lot worse than that.
One reader wrote to me that the 8% estimate for US inflation is probably still too optimistic, as it does not fully take into account the rise in wheat and other commodity prices, for example. Another important side effect of a potentially misjudged inflation series is that US growth is actually not higher than European growth – a claim that has lead to much soul-searching over here – as we are deflating nominal GDP growth by an excessively modest indicator. As for the apparently superior performance of the British economy, just try to deflate all those nominal prices by RPI, not the actual GDP-deflator used, and the economic miracle disappears.
There is surely some of this going on in the euro area as well, but the effect is probably less extreme, I think. At the very least, the ECB is not taking oil and food out of the price index, but I think we do use hedonistic pricing too. I have not seen any estimate of German or French inflation in 1980s, or early 1990s terms, and would be very interested if readers could alert me if such estimates exist. My gut instinct tells me that our inflation rate also understates the true rate of inflation, but perhaps to a lesser degree than in the US. But that assertion only cries out to be verified, or to be dismissed.
Even if we are sceptical about some of those numbers, let us at the very least have an honest debate about inflation. While an artificially depressed inflation indicator may make life a lot easier for a central bank, we know we cannot fool all of the people of the time. This was just tried in the credit market. Another catastrophic Ponzi game would eventually come unstuck. The last think we want after this credit crisis is over, is for central banks to put up nominal short term rates to 20% to contain runaway inflation. Contrary to popular wisdom in the US, it may be better not to cut them now, as opposed to cutting now, and hiking later.