You may have noticed in the press that the Congressionally-appointed taskforce appointed to find biases in the Consumer Price Index (CPI) has - lo and behold - found some. The press is filled with statements such as "most economists believe that the CPI overstates inflation." Of course, most economists have never looked at the CPI's methodology and at most vaguely remember something from grad school about substitution effects (which empirically are not all that important). The politics behind all of this are simple enough. Social Security benefits are indexed to the CPI, as are income tax brackets. Anything that makes the CPI rise more slowly cuts "entitlements" - mainly Social Security - and raises taxes, thus reducing the federal deficit. If all this can be done by pressuring govt. statisticians, Congress does not have to undertake unpleasant policies. Hence, you will find bipartisan statements of support for the taskforce's findings. (An added attraction to some is that because CPI components are used to estimate real output, productivity, and real wages, sluggish growth in these measures can be made to vanish from official data. The electorate, officially at least, will always be better off than it was 4 years ago.) Sadly, there are many things wrong with the idea of tweaking the CPI to achieve political goals covertly. The first is, of course, that we should not play with official data to achieve political goals. Any time it wants to, Congress can adjust tax rates and social security benefits. We are starting down a very slippery slope here if in fact diddling with the CPI takes place. You can already see the results. The taskforce's views were in fact made public months ago. It announced that it thought the CPI overstated inflation by around 1% per annum. The Bureau of Labor Statistics (BLS) came back with a statement that in its view the bias was only a couple of tenths. In effect, we have a kind of public bargaining already set in motion. Congress/taskforce: "We demand 1.1%." BLS: "Sorry, we can only afford 0.2%." Since Congress controls the BLS budget, however, the bargaining parties are not evenly matched. In any event, the entire process fails the famous "smell test" that should be applied to public policy. Second, the notion that the indexing must be done precisely implicitly assumes that the base to which the index is applied is somehow perfect, precise, and optimum and will therefore be distorted if perfect indexing is not applied. But this is ridiculous. Social Security benefit formulas and tax rates are essentially arbitrary political decisions. They could have been higher or lower than they are. The idea that these levels are sacred and must be precisely preserved in real terms is absurd. If there is anything imprecise in the process, it is the determination of base tax rates and benefits, not the comparatively minor impact of indexing with an imperfect CPI. Third, the CPI is filled with anomalies. And not all of these bias the measured inflation rate up. Example: If auto manufacturers are mandated to put pollution controls on their factories, and they raise car prices as a result, those price hikes are included in the CPI. But when the govt. mandates the placement of a catalytic converter on the car itself, the resulting price hikes are not included. If you went line by line through the CPI, you would find lots of such oddities. There is no perfect inflation index in the real world. (By the way, a BLS study a few years back found that if you were to construct a CPI for retirees under Social Security, you would have to put more weight on health services. Since health service prices have historically risen faster than other prices, the official CPI can be said to be biased down for purposes of Social Security indexing.) Fourth, the purpose of indexing is to guard against the impact of sudden inflation shocks, e.g., middle east oil crises. If inflation were perfectly predictable, why would we need indexing? So the question to be asked is whether, in a reasonable way (not a perfect way!), the CPI inflation rate will jump up (or down) when there is sudden inflation or dis-inflation. The taskforce worried about such things as quality adjustments and locations of purchase. But since we don't have sudden shocks in the quality of goods and services or the kinds of stores from which consumers buy, these considerations are irrelevant. There is lots more to be said about U.S. official statistics in terms of quality, responsiveness, and efficiency of administration. But on this matter, the message should be "HANDS OFF THE CPI!" ---------------------- Daniel J.B. Mitchell [EMAIL PROTECTED] Professor at UCLA Anderson Graduate School of Management School of Public Policy & Social Research Mailing address/phone/fax: Anderson Graduate School of Management U.C.L.A. Los Angeles, California 90095-1481 USA Office phone: 310-825-1504 Home phone: 310-829-1246 Fax: 310-829-1042