Thanks to Dave Richardson for his discussion of the bias problem with the CPI. This issue comes up in Macro and now I have a detailed response. One quick addition, and Dave can tell us if I am wrong on this. The GNP or GDP deflator calculation process overcomes most but not all of the problems Dave discusses. The one that is still remainng is the new product/quality change problem. Thus, I always tell my students that if they are going to do any research on "real" values, they have to deflate using the GNP/GDP deflators. Using these deflators, real wages have still been going down. On the issue of whether the CPI overestimates or underestimates inflation, I have my students do a very simple (simplistic?) exercise. I have them calculate annual inflation rates for every year since 1940 using the GNP deflator and the CPI. I then have them count how many years during the period of 1940 to 1993 the CPI estimate of inflation is higher than the GNP estimate. These are the results, by decade: 1940's CPI est > GNP est. 4 years 1950's CPI est > GNP est. 2 years 1960's CPI est > GNP est. 2 years 1970's CPI est > GNP est. 5 years 1980's CPI est > GNP est. 5 years 1990's CPI est > GNP est. 4 years If we assume the GNP deflator is an "accurate" measure of inflation, the CPI overestimates less than half the time, but there is a trend toward increased frequency of overestimation. Doug Orr [EMAIL PROTECTED]
