When the Advisory Commission to Study the Consumer Price Index released its report in early December claiming that consumer inflation is being overstated by 1.1 percent a year, its findings were generally treated by the press as if they were technically irreproachable. The Boskin Commission, appointed by the Senate Finance Committee and named after its chairman, Michael Boskin, President Bush’s chairman of the Council of Economic Advisers, was repeatedly described as “blue-ribbon,” “distinguished,” and “bipartisan.” The report was said to have met the highest academic standards. Many people I spoke to, including economists, took it for granted that the Commission was at least approximately right, and recently Alan Greenspan, the Federal Reserve chairman, spoke approvingly of its conclusions.1 All agreed that the consequences were enormous. To take one example: because the payments for many federal programs, the largest of which is Social Security, are linked to the Consumer Price Index (CPI), it would cost the government an unnecessary $1 trillion in additional spending between now and 2008 if the CPI were indeed overstated by 1.1 percent.
But without working directly on the problem of consumer prices, even the most thoughtful economists, not to mention members of the financial press, may not be able to appreciate fully the labyrinthine complexity of the CPI. In fact, the Boskin Commission’s conclusions are based on a high degree of conjecture and speculation, as I believe most interested observers would conclude if they actually read the ninety-page report. The Commission conducted little original research and much of the evidence it cites is old or covers only limited periods of time. Even a strong supporter of the Commission’s work, William Nordhaus, professor of economics at Yale, warned at the recent convention of the American Economics Association in New Orleans that some of the most important of the Commission’s conclusions are “highly subjective.” The adjustments in the index for the price for food, according to Nordhaus, “have no quantitative support; housing is a guess; and medical care is based on a few preliminary studies comprising a tiny fraction of the expenditures.” A member of the commission, Harvard professor Zvi Griliches, was quoted in a New York Times article as conceding that the numbers are “squishy.”2
The CPI is the nation’s main measure of consumer inflation. The Bureau of Labor Statistics (BLS), which compiles the CPI, tracks the prices of about 80,000 goods and services each month in seven major groups, such as food, clothing, and medical care, and the rents of about 5,000 apartment units. It calculates the relative amounts consumers spend in each of these categories, and the overall average is turned into an index in which 100 is equivalent to the average of prices in the period 1982-1984. The change from year to year in this index is consumer inflation. In 1996, the CPI rose to 158.6, a rise of 3.3 percent from the previous year. If the Boskin Commission is right and the CPI is overstated by 1.1 percent, then consumer inflation in 1996 was only 2.2 percent.
The Boskin Commission believes the CPI overstates true inflation in several ways. First, it claims, the index is updated too slowly to reflect the consequences when consumers substitute one product or one brand for another as prices change. If the price of beef goes up, more consumers will buy chicken instead, but the CPI will not take into account that they are paying lower prices for chicken. The Boskin Commission concludes that the CPI is really .4 percent a year lower than reported as a result of the effect of substitution and related technical concerns. This is the best-documented of the Commission’s claims.
As for the rest of the Boskin Commission’s estimated overstatement of the CPI (within a range of .8 percent to 1.6 percent), it attributes another .1 percent to the failure of the CPI to allow for the proliferation of new discount retailers who charge significantly lower prices. The largest share of the adjustment—.6 percent—the Commission attributes both to improvements in the quality of existing and the introduction of new products that are not recorded by the CPI. An unspecified part of this it also attributes to the CPI’s failure to take account of the decline in the prices of new products—such as cellular phones—fairly soon after they are introduced. These proposed adjustments for quality improvements are the most subjective of the Commission’s claims.
But in addition to the speculative nature of the Commission’s conclusions about quality, what has also been missed in most press accounts is the degree to which the BLS already makes a large adjustment in the CPI for such quality improvements—in fact, a considerably larger one than the .6 percent recommended by the Commission. Prices are reduced for quality because as products improve, the consumer is able to buy more well-being for every dollar spent. For example, if automobiles are much better than they once were—more efficient in their use of fuel or more durable—an increase in the price of those cars should justifiably be reduced by some amount when computing the cost of living. Consumers, it can be said, are buying “much more car” than they once did. Similar adjustments should ideally be made for improvements in the quality of food, clothing, consumer electronics, health care, and so on.
The BLS, however, has its own procedures to take account of such changes in quality, and these have been largely ignored both by the Boskin Commission’s report and the comments on it. Several of the most important of these procedures were significantly revised and improved in the 1980s, and last year, the BLS conducted a study to determine the size of adjustments made to the CPI as a result of improved quality. The study, completed only this November, included about two thirds of all the goods and services in the CPI during 1995. When we extrapolate for the entire basket of goods and services, the adjustments for improved quality total about 2.2 percent of the index. 3 The size of this adjustment surprised every economist I spoke to before writing this article. In other words, had the BLS made no adjustments for quality and merely recorded price increases, the CPI would have risen by about 4.7 percent in 1995 rather than the reported 2.5 percent.
Though the Boskin Commission had access to this study, if only a few weeks before publication of its report, it made no mention of it. The Commission says that the improvements in quality it is discussing are for the most part above and beyond what the BLS takes into account. But this is not necessarily the case, and it is one of many questions that require further investigation. The BLS is reviewing its study and may, I am told, reduce somewhat the estimate of its adjustments for quality.4 But even so, the adjustment is so large that some economists believe the BLS may actually overestimate improvements in quality by as much as, or more than, the amount that the Boskin Commission thinks quality has been underestimated.
None of this is to say that the Boskin Commission hasn’t summarized much useful information about how information on prices is gathered today. Its stated objective, which is to turn the CPI into a measure of the true cost of living rather than merely of price changes, is commendable. But after reading the Boskin Commission report and several other analyses of the CPI—one of the most important of which, “Bias in the Consumer Price Index: What is the Evidence?” is by a BLS staff member, Brent Moulton—it seems to me that any immediate acceptance of the Commission’s main conclusions is decidedly premature. After more research is completed, the Commission could well turn out to be wrong.5
Careful appraisal of its conclusions is all the more important when we consider how serious the consequences of either overstating or understating the CPI can be. As the Commission points out, fully one third of the federal budget is indexed to the CPI. The annual cost-of-living adjustments to Social Security benefits are determined by the annual change in the CPI, as are those in other federal pension programs. Income tax brackets are also adjusted according to the CPI. If the CPI is overstated by 1.1 percent, according to the Boskin Commission, the federal government will have a net additional expenditure of $148 billion more than it should in the year 2006 alone. The cumulative outlay would come close to $700 billion over the next ten years and, as noted, reach more than $1 trillion in 2008. It is not hard to understand why the Senate Finance Committee is interested in the subject, or how politically convenient it might be for Washington’s budget balancers if there were a broad technical consensus among economists that the CPI overstates inflation.
But if the CPI is incorrectly reduced by as much as 1.1 percent a year, it would harshly penalize many Americans. The AFL-CIO figures that an average retired couple would lose $2,360 over the next five years if the cost-of-living adjustment for Social Security were reduced by 1.1 percent a year. The average widow or widower would lose $1,470. Many private corporate pensions are tied to the CPI as well, as are private wage contracts and rents. Nor is it apparent that the CPI, which measures average inflation for all consumers, is applicable to the elderly. In a recent study, the BLS found that actual price increases for the elderly may exceed the current CPI by .2 percent or .3 percent a year.6 A reduced CPI would also lower the poverty threshold, removing many poor Americans from benefit programs. In fact the cost-of-living increases for the poor may be significantly underestimated by the CPI. A study by the economist Trudi J. Renwick finds that the increase in the cost of living of single-family households at or below the poverty line has been understated by the CPI by 50 to 100 percent since 1983.7
The CPI is also essential for measuring the performance of the economy. It is used to determine how much total expenditures on consumption should be discounted for inflation in the Gross Domestic Product, and it therefore affects the federal government’s computations of economic growth and productivity. It is also used by the government to discount workers’ wages to determine “real wages.” A reduction in the CPI would imply that the economy and real wages have grown faster than we thought. On the other hand, because the overstatement of CPI would apply approximately as much to the years before the early 1970s as after, the widely reported slowdown since the early 1970s in the growth of Gross Domestic Product, productivity, and real wages would still have taken place even if the Boskin Commission proves right.8 A lower CPI between 1945 and 1973 would simply imply that real economic growth was significantly higher in those years than reported.
What do we actually know, then, about the accuracy of the CPI in light of the Boskin Commission report? The BLS updates the proportional weights assigned the CPI’s broad categories of goods and services only every tenyears; it does so on the basis of surveys of consumer expenditures. While 20 percent of the items tracked are now changed each year, this is too slow to reflect the lower prices that result from the kinds of product substitution I earlier described. Despite revisions in the BLS’s procedures, major new products still sometimes can be ignored, or they may enter the index only after they have been on the market for years. The BLS may therefore overlook improvements in their quality and it may also occasionally miss the rapid price declines that usually occur in the initial years after the introduction of such new products. In the most glaring example, cellular phones are still not included in the index, though their prices have fallen by as much as 90 percent since they first came on the market. Still, as we shall see, the consequences of these procedural oversights are often not as large as they intuitively seem. Moreover, current BLS procedures capture more of the price effects than some critics have claimed.
The Boskin Commission makes its strongest points in its comments concerning the effects of substitution. The CPI is not designed to take account of the bargain-hunting habits of some consumers. The BLS simply assumes that consumers will continue to buy the same amount of beef, even if its price has risen compared to chicken. Some economists used to take the view that such substitution would usually result in a drop in well-being, and that a deterioration in quality would precisely offset the price decline. No doubt, this is still partly true. After all, the consumers in question preferred to eat beef, not chicken, and would have done so if the price had not risen. But consumers also regularly substitute cheaper brands or products that don’t seem to them qualitatively different. They may switch from Budweiser to Miller beer, from corn flakes to Cheerios, or from oranges to tangerines. In these cases, price declines that result from substitution should be included in the CPI.
The Boskin Commission has proposed several technical changes in order for the CPI to take such effects of substitution into account. Some of them are already used by the BLS’s fellow agency, the Bureau of Economic Analysis. Still, some losses in quality may be ignored. The technical changes also involve oversimplifications that make adopting them unwise in some cases. For example, to use one of the Commission’s recommended averaging techniques would require the assumption that any increase in prices would always be matched by a proportionate decrease in consumption—i.e., a 10 percent increase in the price of beef will result in a 10 percent decline in the consumption of beef. But this is not always the case. The BLS will probably adopt Commission proposals that amount to about .2 percent of the .4 percent proposed adjustment for substitution effects.
The Commission has also contended that the CPI is overstated by .1 percent because the BLS doesn’t take sufficient account of sales in the many new discount outlets throughout the US. The BLS now assumes that the reduced services offered by these stores offset any decline in price. The Commission replies that a growing number of customers use them anyway, suggesting that price discounts are of greater value than the quality of services lost. But a persuasive analysis by Brent Moulton of the BLS shows that total price savings from using discount outlets may amount to considerably less than the Boskin Commission assumes. It seems likely that any overstatement of the CPI owing to neglect of discount purchases is only a fraction of the .1 percent proposed by the Commission.
As has been said, the main and most contentious claims of the Boskin Commission are that the BLS underestimates improvements in quality and fails to record the price declines in new products, to which the Commission attributes .6 percent of its 1.1 percent estimate of the overstatement of inflation. In several of the most prominent cases cited by the Commission, however, there is good reason to believethat improvements in quality were not underestimated by the BLS and that it may in fact have even overestimated them. A closer look at a few of these cases also makes clear just how speculative such judgments about product quality necessarily are, especially in view of the state of research on these subjects.
The BLS, for example, already makes a substantial adjustment for the improved quality of cars. Prices of cars are 4.13 times higher than they were in 1967, according to the BLS; but the car component of the index has been raised by a factor of only 2.72 because the BLS believes that improvements in quality offset the rise in prices. In other words, the BLS figures the consumer gets about 50 percent more satisfaction from the car today than in 1967.
But despite this substantial adjustment, the Boskin Commission claims the allowance for quality should be greater still because the BLS does not adjust the CPI for the increasing durability of cars. This turns out not to be true. In fact, the BLS does take account of annual improvements in autos, many of them adding to the durability of cars, and the Boskin Commission simply failed to note that it does so. Some economists even believe the BLS overadjusts for the higher quality of automobiles by accepting too readily the manufacturers’ claims of improvements.
What’s more, the Commission’s assessment of the improved quality of new cars rests not on direct evidence of improvements in durability or a survey of consumers’ experience but entirely on the fact that consumers are holding onto their cars longer. The Commission points out that the average age of cars on the road was increasing in the 1970s and in the 1980s. But the reason cars are being held longer probably has less to do with increasing durability than with the fact that prices of new cars have been rising significantly faster than have the slow-growing incomes of buyers over these years. Also, many cars are now used as the second and third vehicles in the family. In fact, according to independent surveys of the industry, the quality of cars did not begin to rise significantly until the mid-1980s, yet the age of cars on the road increased most rapidly in the 1970s.9 The quality adjustment for new cars may, if anything, be too high, meaning that inflation is understated for cars rather than overstated.
Similar examples of the Commission’s flawed conclusions are not difficult to find. In housing, the Boskin Commission notes that the CPI actually has understated rent increases by about 20 percent since 1976, but claims that this is offset by a 20 percent increase in the size of apartments. Apartment dwellers, however, don’t pay 20 percent more in rent for a 20 percent increase in space, as almost anyone who has ever rented an apartment surely knows. Moreover, the Commission then assumes the BLS has not taken into account other improvements to apartments such as central air conditioning, to which it arbitrarily assigns a value of .25 percent a year. In fact, the BLS has included substantial adjustments for quality improvement, and it may have understated inflation in housing costs rather than overstated it, contrary to what the Commission claims.
Similar questions can be raised about some of the Commission’s conclusions concerning other major products. BLS economists claim that the Commission’s assessment of clothing costs mistakenly ascribes many price increases to quality improvements when in fact the prices are driven up by changes in fashion. The judgments of economists about quality often seem arbitrary and arguable. For example, the Commission believes the CPI does not adequately account for the improved quality of drivers’ lives now that they can use credit cards in self-service gas stations. Roseanne Cahn, an economist at the investment banking house CS First Boston, considers this a loss of quality because drivers now have to fill their own tanks in cold weather, often getting dirty as they do so. Full-service gas stations are harder to find. The Commission also says the CPI is overstated because it doesn’t account for the increased variety of fruits and vegetables available at stores even in the off-season. Cahn says this produce is often old and of poor quality.
On the other hand, the Commission’s findings about improvements in quality and new product effects in consumer electronics and health care are telling, if perhaps overenthusiastic, and require more study. It should come as no surprise that the quality of TVs, stereos, VCRs, camcorders, and PCs has improved while prices have often fallen precipitously. Recent work by the BLS suggests that such improvements in quality may be underestimated. In addition, such products are underweighted in the CPI at about .8 percent of expenditures because the updating procedures are slow. This should be corrected. Yet we should keep in mind that these items still only account for 2 percent of all consumer purchases today. It is also likely that improvements in health care may well be outpacing even rapid price increases for some medical services that the BLS does not adequately account for. But there is little research to substantiate these claims or to enable analysts to make specific estimates of improvements in quality. And the BLS has changed some of its procedures.
Is the Commission’s .6 percent estimate of improvements in quality too high? In anticipation of criticism that it is, Boskin and other Commission members claim that there are many kinds of unrecorded quality improvements that they did not bother to cite. In a letter to Boskin last December, Robert Gordon, a commission member and professor of economics at Northwestern University, wrote that “…the report’s final estimate of a 1.1 percent annual rate of upward bias in the CPI is doubtless an understatement.” It strikes me as odd that the authors of a presumably objective, technical report would choose to leave out so much that seems relevant. But if the sources of “upward bias” are as subject to speculation as some of the claims made in the published report, then one can hardly take this assertion at face value.
Another reason that some economists, including William Nordhaus, believe the Boskin Commission’s estimates have been conservative is that the Commission does not take adequate account of the decline in the prices of new products. The Commission is vague, at best, about how this factor is included in its estimates, and in fact there is no firm evidence about the effects on the CPI of such changes in price. No one would deny, for example, that the decline in the price of VCRs created benefits for buyers who were originally willing to pay $500 or $1,000 for a VCR and now can buy one for $200 or less. But when prices are high and the rate of decline is most rapid, relatively few people are buying the product.
Moreover, there may not be as many of these products as we seem to believe. All but a few of the new products that come onto the market, in fact, last at most a year or two, but those that last longer are the ones that capture economists’ attention. Most of the anecdotal examples cited to substantiate the effect of new products, for example, are for electronics products which, as noted, account for only 2 percent of all consumer purchases.10 Finally, the BLS does indeed take account of many of these effects, and the Boskin Commission made no attempt to determine specifically what proportion were included or excluded by BLS procedures.
What is particularly disturbing about the Commission’s work is that it also makes no effort to analyze and estimate the deterioration of quality in products and services. It does not mention the overcrowding and overbooking of airplane flights, for example. The Commission does not try to assess how traffic congestion, reduced rail service, poor highway and bridge maintenance, and the aging of mass transit have reduced consumer well-being. In most of these cases the consumer is paying for something (including bridge tolls) and getting less than he used to. In its rather euphoric treatment of health care, the Commission takes no account of the overcrowding of public hospitals that results in often dangerously long waits for service, especially as the proportion of Americans who have no health insurance rises. The Commission mentions that services at HMOs may be deteriorating but does not pursue the subject.
The Commission members, moreover, hardly seem aware of the general tendency of many products and services to require more effort from consumers these days, whether in dealing with computerized telephone exchanges, choosing a long-distance phone service, managing one’s own direct-contribution pension funds and personal investments, or buying from do-it-yourself stores such as Ikea. The Commission implicitly argues that if consumers are willing to invest the time and effort required to enjoy price discounts at stores such as Ikea, then the price reductions offset the loss in the quality of their lives. But this disregards how much consumers living in a slow-growth era need to save money.11 The decline in the quality of free public education is simply not mentioned; nor is the costly shift to private education that has resulted. Some of these changes would be difficult to incorporate in the CPI, but the Commission should at least have taken account of them. Instead, in a short but sweeping statement, it dismisses deterioration in quality as essentially inconsequential compared to the improvements that have taken place.
The Commission’s assessment of improvements in quality seems both overoptimistic and lacking in historical perspective. The increasing variety of products and services today may influence our lives much less than the Boskin report assumes, especially when they are compared to the new products of earlier generations. Do the benefits to the standard of living of “call waiting,” improvements in voice quality, and even the cellular phone compare in any way to the benefits from the original installation of telephones earlier in the century? The MRI, new surgical techniques, and similar innovations may well have improved our well-being more than rising prices for these services have reduced it. But how do these recent improvements compare to the introduction of the polio and smallpox vaccines and antibiotics, or what we learned from Pasteur’s germ theory in the nineteenth century? Compare the time it takes for a doctor to administer a polio vaccine to the time it took to treat a lifelong victim before the 1950s.
The Commission makes much of rising life expectancies during the past twenty years. But the average life expectancy had been rising throughout the century even when medical expenditures were a far lower proportion of family budgets or the overall GDP. The VCR makes it cheaper for a family to see a movie; but how does this compare to the introduction of television in the 1940s or the nickel movies that entertained millions of American workers during the 1930s? Would a similar analysis done by economists then tell us that the Great Depression wasn’t really all that bad once we took the quality of such new products into account?
In some cases, the benefits of new products are almost surely less than we think, despite the glamour of high technology. As Moulton writes, “If the price of a new good falls dramatically, consumers will tend to apply it to low-valued uses (for example, computers used for playing games, lights left on in unoccupied rooms).” Consumers may take liberal advantage of cash machines simply because they are free, for example, much the way many have been jamming the Internet since unlimited-time options have become available. But what would we actually pay for such conveniences?
If the Commission took account of the kinds of omitted factors I have mentioned, it might argue that additional improvements in quality and the effects of new products would still result in a downward adjustment of .6 percent or even more in the CPI. The Commission believes, as noted, that there is little overlap with the quality adjustments already amounting to 2.2 percent or so of the CPI that have been made by the BLS. But Charles Hulten, an economics professor at the University of Maryland, and a member of the respected National Bureau of Economic Research, as are Gordon, Griliches, and Boskin, thinks that there are systematic reasons to suggest that the BLS’s own adjustment may be substantially too high, by as much as .5 percent or so. As a result, Hulten and some other economists think that it is possible that, once all the conflicting figures are taken into account, there may be no upward bias in the CPI caused by neglect of quality improvements and new products.12 In other words, the CPI may hardly be overstated after all.
The attempt of the Boskin Commission to measure the true cost of living raises deeper questions than the Commission chooses to address about how to define the standard of living. Telephones and cars may once have been luxuries that made substantial improvements in the quality of life; when they were first introduced, such improvements undoubtedly raised the standard of living. But the phone and the car eventually became necessities for almost all Americans in order merely to live and work in the modern world. Similarly, new products such as laptop computers and cellular phones confer status that people may eventually build into their expectations. The Boskin Commission assumes that whenever consumers buy such a product, their standard of living rises. But to many consumers, doing without them means that their standard of living has fallen.
In a particularly useful book, American Standards of Living, Clair Brown, an economist at the University of California at Berkeley, reveals in detail how Americans tend over time to make purchases that raise their status. Once status has been attained by new purchases, it seems that new sources of status must be found. For example, Brown found that a group she defines as wage earners, the lowest-paid group she considers, spent 40 percent of their budget on food eaten at home in 1918 compared to only 10 percent in 1988, when eating out had become a standard practice. In 1935, very few of her wage earners purchased a new car. But by 1960, almost all such wage earners were able to buy cars. If today’s wage earners couldn’t eat out or buy a car, they would feel that their standard of living had fallen.
In my own case, I recently bought Windows and Microsoft Word for my PC, not because I needed the additional features, but to make it easier to work with others who use these formats. I must admit as well that I was a little tired of being laughed at when I told colleagues I was still using Wordstar 2000, though this old-fashioned word-processing program seemed to meet all my needs. The Boskin Commission would treat owning Windows and Word 6.0 as an increase in my well-being, especially since prices dropped. But I consider it an additional cost of simply standing still.
Economists find such issues nebulous and tend to avoid them, but sociologists, psychologists, and historians find them all too real. It would be interesting, if expensive, to build on Clair Brown’s work and devise, quite apart from the current CPI, an experimental basket of goods and services that represents a middle-class standard of living—or, ideally, separate packages for different groups of workers, from the outright poor to the better-off. The objective would be to find out how much more costly a middle-class life has become. (The BLS once published an index for a typical family of four, but it was discontinued for lack of funds.)
One of the more interesting consequences would be that the adjustments for improvements in the quality of products would not be as great for a middle-class index as they are for the current CPI. Many of these improvements or new products have simply become embedded in the ever-changing middle-class way of life, and no longer represent a rise in well-being so far as status or the practicalities of dealing in the modern world are concerned. Also, today’s quality improvements are probably skewed towards higher-income workers who can afford camcorders, CD players, PCs, cable TV, and diagnosis by MRI. Certainly, the savings from the rapid drop in prices are enjoyed mostly by well-off consumers who could afford to pay $1,000 for the very first VCR. Adjusting the CPI for all workers, rich and poor, young and old, based on such quality adjustments would be seriously biased against those segments of the population who can’t afford such products.
A new approach such as the one I have suggested would also provide a useful way to compare the cost of living over long periods of time. As the CPI now stands, historical comparisons are difficult to make. In a perceptive paper called “Getting Prices Right: A Methodologically Consistent Consumer Price Index, 1953-94,” Dean Baker, an economist at the Economic Policy Institute, notes that reducing the CPI implies that real wages have grown much faster than economists realized over the decades. This must mean, however, that wages were much lower in earlier years than has been realized. If we were to accept the Boskin Commission’s claims, Baker points out, most Americans were living below the poverty line during the 1950s. This is highly implausible. 13
The Boskin Commission apparently wants its highly debatable findings to be taken as convincing conclusions. In a final, somewhat opaque recommendation, the Commission seems, at least, to be urging Congress and the President to reduce the indexing of Social Security’s cost-of-living allowance, whether or not the BLS adjusts the CPI. Boskin singled out this recommendation at the annual meeting of the American Economics Association in New Orleans in January by flashing it up on an overhead projector for the large audience.14
Such a demand for major change in public policy seems presumptuous. It should be clear from the recent detailed criticism of the Commission’s report that further study of the entire question of living costs is required; and the Boskin Commission itself recommends that more research be done on the factors that determine consumer satisfaction. Until such research is adequately funded and carried out, it would be irresponsible to make changes in the indexing of Social Security that are not supported by the BLS. Nor should the BLS, which is staffed by highly competent and diligent economists, be expected to adopt the same subjective approach that the Boskin Commission often has. National statistics-gathering agencies should try to be as objective as possible and to resist political pressures. All five Boskin Commission members testified in Senate Finance Committee hearings that they already believed inflation was significantly overstated before they were appointed. Other economists who testified that the CPI may not be overstated, such as Joel Popkin, a former BLS price chief and long a private economic consultant, were not called upon by the Commission or its supporters. No dissenters were invited to participate in this “bipartisan” report. Any further inquiry should include a broader representation of economists. A more objective examination of the issues now seems imperative.
March 6, 1997
The Commission members include Michael J. Boskin, professor of economics at Stanford University; Ellen R. Dulberger, director of marketing strategy for IBM Personal Computer Company; Robert J. Gordon, professor of economics at Northwestern University; Zvi Griliches, professor of economics at Harvard University; and Dale Jorgensen, professor of economics at Harvard University. ↩
Louis Uchitelle, “Balancing Quantity, Quality and Inflation,” The New York Times, December 18, 1996. This article and another by John M. Berry, “CPI Report Coming Under Fire,” The Washington Post, December 19, 1996, are noteworthy exceptions to the press’s uncritical reception. ↩
Brent R. Moulton and Karen E. Smedley, “Addressing the Quality Change Issue in the Consumer Price Index,” Bureau of Labor Statistics, Washington, D.C., draft of research in progress, January 1997. The BLS studied nearly 70 percent of the goods and services in the CPI and found that it made quality adjustments amounting to 2.56 percent in 1995. Assuming that some but not as much quality adjustment was made for the remaining 30 percent, a private economist, Charles Hulten, estimates that the entire CPI was adjusted by the equivalent of about 2.2 percent. ↩
The BLS is currently combing the data for possible distortions in its sample. But it doubts any adjustments will total more than .5 percent and thinks it is probably less. ↩
Brent Moulton lists a number of other estimates of the overstatement of the CPI, but all are less detailed or comprehensive than the Boskin Commission report, and some are admittedly merely educated guesses. All were published before the degree of quality adjustment made by the BLS was known. The Congressional Budget Office, for example, estimated the overstatement between .2 percent and .8 percent a year. Jim Klumpner of the Senate Budget Committee estimated the overstatement between .3 percent and .5 percent a year. Though not mentioned by Moulton, the highest estimate I have seen was nearly 3 percent by Leonard I. Nakamura of the Federal Reserve Bank of Philadelphia, whose methods were the least orthodox of those cited. Moulton notes that the literature in general is biased by examination of goods and services where there is a high likelihood that quality improvements will be significant. ↩
Kenneth J. Stewart and Joseph Pavalone, “Experimental Consumer Price Index, Americans 62 years of Age and Older,” Bureau of Labor Statistics, 1996. The main reason for a higher cost of living for the elderly is their out-of-pocket health costs, which have risen rapidly in recent years. The Boskin Commission makes note of this but claims there may be offsets such as elderly price discounts. It did not actually do research on the subject, however. ↩
Trudi J. Renwick, “Basic Needs Budget: 1996,” Public Utility Law Project of New York, Inc., 1996. ↩
Boskin himself made this point at the annual meeting of the American Economics Association in New Orleans in early January. ↩
The rate at which cars were scrapped only began to fall in the 1980s, according to R.L. Polk & Co., Detroit, Michigan. Dependability began to rise significantly only in the mid-1980s. See “Vehicle Dependability Surveys,” J.D. Powers and Associates, Agoura Hills, California. ↩
Two exceptions are: Jerry A. Hausman, “Valuation of New Goods Under Perfect and Imperfect Competition,” NBA Working Paper No. 4970, 1994; and William D. Nordhaus, “Do Real Output and Real Wage Measures Capture Reality? The History of Lighting Suggests Not,” Cowles Foundation Discussion Paper No. 1078, 1994. ↩
The Commission argues that if market shares for these discount services grow, then price discounts must exceed quality lost. But it is more likely that consumers are simply willing to give up more of their free time than they once did in order to save money. The best way to ascertain what is going on may be through studies of how consumers use their time. The Commission has recommended that such time-use studies be undertaken, a highly worthwhile suggestion. Nevertheless, in the absence of such research, the Commission is all too willing to emphasize the advantages of discount outlets. ↩
Charles R. Hulten, “Quality Changes in the CPI: Alternative Measures of the Bias,” revision of a paper prepared for the Federal Reserve Bank of St. Louis, Fall Policy Conference. ↩
To counter the implausibility of this conclusion, the Commission suggests that the overstatement of the CPI might have been somewhat less in the years before 1970. But in view of the changes made by the BLS to correct for quality bias and other problems since the early 1980s, it seems more likely that, if anything, the CPI was more overstated in the 1950s and 1960s than in recent years. Also, the BLS conducted studies in 1983 and 1984 similar to its 1995 study, and found that the adjustment for quality was only 1.1 percent and 1.2 percent; this suggests the adjustments it makes in quality are significantly greater today than they were in the past. ↩
The recommendation reads as follows: “Congress and the President must decide whether they wish to continue the widespread overindexing of various federal spending programs and features of the tax code. If the purpose of indexing is fully and accurately to insulate the groups receiving transfer payments and paying taxes, no more and no less, they should pass legislation adjusting indexing provisions accordingly.” In other words, cost-of-living allowances should apparently be indexed to the CPI minus X, where X is to be determined by Congress. ↩