Why the Consumer Price Index Understates Inflation
A panel of economists appointed by the
Senate Finance Committee and led by Michael Boskin,
the former chief economic adviser to President George Bush, reported that the
consumer price index (CPI) is overstated by 1.1 percent. If this is the case,
the entitlements provided by the government are too high and tax collections,
when adjusted for inflation, too low.
The CPI is based on surveys of consumer purchases and their prices. It is weighted by groups representing categories of purchases (e.g., food or energy). The commodities included in the market basket are chosen somewhat arbitrarily because choices are made concerning what to include after consumer surveys are made. Moreover, it is arbitrary to some extent how often the basket is changed.
The price of the market basket is a national average. Most consumers are not average consumers, although there may well be a group of modal consumers, for whom the figures are reasonably representative. It would be too complex to attempt to fine-tune these figures for different segments of the population.
However, when at least one of the purposes of a change in the method of determination is to change the income of a particular segment of the population (e.g., the entitlements of Social Security recipients), then it may not be amiss to argue that even the current method of determining the CPI likely understates the impact of price inflation on the general population and even more so on the elderly. In part, this stems from minimizing the extent to which prices depend upon preferences--although utility (or expected satisfaction), along with scarcity, is a foundational aspect of price theory--and in part on how changes in the mix of commodities affect the standard of living negatively.
Utility entails a preference ordering--in other terms, expected satisfaction- -which has personal, social, cultural, and temporal characteristics. If this were not so, we could explain neither the wide social and cultural differences in preferences among nations, ethnic groups, and social classes nor their variation over time. Preferences, and hence expected satisfaction, in addition to being one element in the foundation of price are also part of the essential foundation of the concept of a standard of living. This latter concept is as variable as the former.
WHAT ABOUT PRODUCT SUBSTITUTION?
Let us turn to the more specific aspects of Boskin's claim. Some of his team's conclusions rest on what they claim is a bias in the survey resulting from improper accounting of product substitution, qualitative improvement in products, and the introduction of new products.
First, consider product substitution. If a purchase is made at a cheaper outlet, the price to the consumer for the same product--assuming service and dependability are without value--is lower. If a cheaper plastic screw will adequately substitute for a more expensive metal screw, there is no doubt that this part of the standard basket would be cheaper. If a new drug will avoid an operation, the cost of living has gone down.
But that is not true of even a partial substitution of chicken for beef, an example Boskin uses, even if chicken is equally nutritious. Price is a function of utility and scarcity. Unless aesthetics is considered, one screw can usually be considered equivalent to another screw. The measure of equivalence in this case is externally objective. This is not true for food.
The current system of calculating the CPI uses the price increases of the mix of commodities in the previous basket (e.g., four pounds of beef and two of chicken) as the indicator of inflation. If one considered the price increases of the mix of commodities in the present basket (e.g., two pounds of beef and four of chicken) to measure inflation, inflation would seem lower because chicken is cheaper than beef. This, however, is not a convincing choice to make.
The person who prefers beef to chicken suffers a decline in purchasing power and, thus, in his standard of living when the price of beef rises and he must buy more chicken and less beef. This decline is measured by his standard of expectation concerning the appropriate ratio of beef and chicken, which in the example is four to two. His experience of inflation is based on the incontrovertible fact that it is the increase in the price of four pounds of beef and two of chicken that forced the change in his buying pattern. This is the measure of the purchasing power that he has lost and, hence, is his measure of the cost of inflation.
Although the items in the basket are changed only every 10 years, the actual proportions of purchases from among the items shift from year to year. It is unlikely that individual utility schedules shift this fast. If they do not, the rate at which the cost of living increases is underestimated, even if one determines the increase in the cost of living on the basis of each previous year's mix of purchases, for the mix will change faster than memory and expectation. The rate of inflation cannot be divorced from experienced preferences, which are an essential part of the foundation of price theory. Thus, even the present method of determining price inflation likely underestimates it.
QUALITATIVE IMPROVEMENTS
Next, consider qualitative improvements in products, which Boskin says reduce the CPI. This is at least partly inconsistent with his earlier argument with respect to certain types of substitution, for the change from beef to chicken is, for the beef lover, a qualitative decline. Hence, it raises the CPI more than Boskin's methodology allows.
Boskin is correct in holding that qualitative improvements such as increased durability lower the CPI. However, it does not follow that all qualitative improvements will lower the CPI. After some major qualitative improvements in a type of product become common, the cultural standard of expectation changes, thus changing individuals' utility schedules for the former products.
Consider the change from black-and-white TV to color TV. After, but only after, the more expensive color sets became common, not having one became a deprivation. If the color set was more expensive than the black-and-white, the CPI in fact went up by the extent of the change.
A third argument made by the economists for a reduction in the CPI is a decline in the price of a new product after its initial impact on the marketplace. The example sometimes used is that of a computer that costs $2,000 when initially introduced and that some years later, after production reaches economies of scale, can be sold for $800. This example, however, is not a case of reduction in the CPI.
A new product should not be considered part of a basket that is based on the purchases of the average consumer until it reaches an economy of scale that makes it widespread. Until then it is not yet part of the general standard of living. However, if--after economy of scale is achieved and the product's addition to the basket has already been taken into account--production efficiencies make it cheaper, then the CPI should be considered to have declined.
However, consider the sequence. When first introduced, it did not increase the CPI, because it was not part of a common basket. When it became part of the basket, the CPI soon went up because it added to the expected demands on income. After becoming part of the basket, if further reductions in price occurred, the CPI went down.
Boskin's mistake with respect to his treatment of the standard of living rests on a failure to distinguish between two different uses of the concept. One standard is internal to a living system and reflects the experience of ongoing life. If the cohort with which I compare my standard of living lacks a new product such as a computer or a microwave oven and my income rises enough to maintain my other purchases and also to acquire the new products, I will perceive an increase in my standard of living.
If my cohort has them and I cannot afford both them and my previous mix of purchases, my perception will soon be that my standard of living has declined. If all can afford both, there will be a short initial experience of a rise in the standard of living, which will soon be eliminated. The other concept that Boskin uses is external and asks whether one is better or worse off than, for example, 20 years ago with respect to possessions and their longevity.
MAINTAINING STANDARDS OF LIVING
Simple illustrations will make the validity of the first concept for CPI calculations clear. A family on food stamps today is living luxuriously by the standards of much of my childhood. I would be laughed out of court if I made that argument, however. That illustrates the principle. Now consider a family that can afford the same mix of commodities it could buy in 1987, while its previous cohort can afford a better basket.
The clear perception will be one of deprivation and decline in standard of living, which is, and must be, at least partly a comparative concept. Otherwise we could move a Stone Age family to our society, maintain their previous mix of commodities, and tell them they should not feel poorer than in the past, because their standard of living has been maintained.
Boskin argues correctly that in most ways we live better than we did 20 years ago. He is correct that we possess more and live longer. And he would be correct to tell the Stone Age family that it has suffered no decline in that sense. If he wants to refer to this as a standard of living, he should at least distinguish between standard of living (1) and standard of living (2).
Boskin's concept involves an external and abstract comparison between different eras or systems, whether temporal or national. The concept relevant to the internal process of life is the inside-system comparison with other members of society, the same framework required for price determination and rate of inflation.
If some natural disaster produced a lower standard of living in Boskin' s sense, we would be very upset. That is exactly my point, however. We measure standard of living, and our satisfaction with it, from a perceptual/cultural base. Should the disaster persist long enough for us to acclimate, that would be our new base. We would feel less deprived. Boskin is mixing apples and oranges.
If a person maintained his standard of living only in Boskin's sense, then in an effective economy he would be falling further and further behind in the cultural sense that is relevant to within-system standard of living accounts. In the price sense--which includes preference, and its foundation in expected satisfaction--Boskin is wrong, and strangely so for an economist.
Consider the entire range of qualitatively improved and new products that become widespread as they affect those on entitlements. They are common and, thus, set the standard for expectations. Now the retired and the poor on welfare are told that their standards of living, modified by a lower CPI increase, according to Boskin's criteria, are as good as they previously were. However, they are now worse off than they were before, compared with most working adults. And Boskin-type adjustments will make the situation increasingly worse.
Because the retired and the poor start from a smaller base--and for the retired a reduced base--the multiplicative differences from the rest of the population resulting from successive Boskin-type CPI adjustments increase dramatically over time. Only these unfortunate people's early death can minimize this. Those in lower income brackets with long lives will become virtual outcasts from their own society.
Quality also has declined in a way Boskin fails to recognize. Both beef and chicken are tasteless to me, because I remember how they used to taste. To match the meat I used to eat I would have to get extremely expensive open-range meat. I do not know where in Chicago I can get a good loaf of bread or a real onion roll at any price. Because I have a much higher income than most elderly, the impact on me is modest.
Let us admit that we need CPI calculations but that there is no perfect way to make them. Any mode of doing so will have imperfections and will be misleading to some extent. There is no paradigmatic method of determining the base or when it should change or even whether a consistent use of a particular base is appropriate for all segments of the population or all economic problems. However, the recalculation Boskin and his group made at least partly misleads with respect to its effects on those who receive entitlements.
If we must cut back on entitlements--and I agree we must--we should at least do so for the right reasons. Although I am not in a position to calculate the increase in the CPI, the flaws in Boskin's arguments and my impressions strongly suggest that the current CPI calculations understate the rate of inflation, and do so particularly for the elderly. In any event, if the method of calculating the CPI is to be changed, this should be done to improve its use in predicting the consequences of different economic policies, not to reduce entitlement payouts. We already know what the consequences of failure to do so will be.
Morton A. Kaplan is Distinguished Service Professor of Political Science
Emeritus at the
Morton A. Kaplan, Why
the Consumer Price Index Understates Inflation. Vol. 12, The World & I,