Let's Fix The Cpi -- But Not Overfix It

Volume II, Number 15 December 24, 1996

At first, the Boskin Commission's conclusion that the CPI overstates inflation seems to point toward a policymaker's dream: a painless procedural fix for the deficit. But the magnitude of the Commission's estimate -- 1.1 percentage points -- has everybody in a quandary. Liberals fret that such a large reduction would starve the welfare state. Conservatives hesitate because it means a sizable tax hike. And both sides wonder whether the finding undermines the rationale for a balanced budget even as it makes it easier to achieve. After all, if real incomes are rising much faster than we thought, maybe it's OK to borrow from our kids.

The opinion of the Concord Coalition is as follows: While we believe that the Boskin Commission makes a good case that CPI overstates inflation, we are not yet persuaded that the overstatement is as great as it concludes. And while there is no justification for overindexing entitlements, we believe that setting COLA's beneath a fair measure of inflation is not the right way to balance the budget since it would unfairly penalize the oldest and poorest beneficiaries.

Accordingly, we recommend that Congress and the President take action to immediately adjust all federal indexing by 0.5 percentage points, a cut that would remain in force pending more study and a revision of the CPI itself. Although it's possible that the Boskin Commission is close to the mark (or conceivably, too low), we think that the whole business of defining and measuring changes in living standards is so murky as to justify, for now, a more cautious initiative.

The Commission's Argument

The Boskin Commission makes several excellent arguments for why the rise in today's CPI overstates inflation and therefore understates the growth in the standard of living of the average American household.

Its first argument is that the CPI fails to take into account how people adjust to relatively higher prices, often by substituting a less expensive product for a more expensive one. In averaging raw price data to calculate an index number, the CPI weights the price of each item in its market basket according to the share of total spending represented by that item in a base year. Because the weights used in translating price data into index numbers are "rebased" only every decade or so, the CPI overweights items whose prices are rising rapidly.

The energy crisis provides a classic example of this "substitution bias." When the price of energy soared after 1973, the CPI continued to weight it as if people's consumption habits had remained unchanged. But of course they hadn't. Americans adjusted by getting rid of gas-guzzling cars and closing their windows. While some of this substitution may have pulled down our living standard (some people like gas-guzzling cars and closing windows is an inconvenience), we were clearly better off than if we had not cut back on energy.

The Boskin Commission's second argument, which is of similar quantitative importance, is that the CPI low-balls quality improvements in a host of consumer products. Cars are safer and last longer than they did in the past. TVs are better. Medicine is more effective. To the extent that higher prices reflect these improvements, they do not represent inflation. In addition, the Commission faults the CPI for not adding new products to its market basket in a timely fashion and for ignoring price changes when consumers switch to discount outlets.

By adding up all these factors, the Boskin Commission arrives at a total bias in the CPI of 0.8 to 1.6 percentage points per year. It's "best estimate" is 1.1 percentage points. While this number might seem small, its budget impact would be large: a deficit reduction of about one- third by 2002. Excluding interest, three-fifths of this savings would come from lower entitlement outlays and two-fifths from higher income tax revenues.

And beyond 2002? True, this change does nothing to stem the growth in Medicare and Medicaid and would by no means render the coming age wave affordable. But it would make a significant difference in cash benefits, erasing some one-third of Social Security's long-term operating deficit and, according to SSA, extending the trust funds' "solvency" from 2029 to 2052.

A Vast Uncertainty

These savings sound impressive. So why don't we recommend that the Boskin Commission's number be implemented in its entirety? The reason is that it masks a vast uncertainty about how best to measure living standards. This question would be hard to answer even if we knew what is meant -- and quite honestly we don't.

To begin with, even if we accept the theory behind the Boskin Commission's arguments, there are problems with its empirical methodology. Take the issue of quality. Since it's seldom possible to compare the prices of improved and unimproved products directly, economists are reduced to imputing the cost of product inputs or outputs. Thus do they hope to learn how much of the price of today's 75-watt light bulb is attributable to its longer life, or how much of the price of today's VCR is attributable to its on-screen programming capability.

Such statistical artifice produces results that are all over the map, which is why the Commission's overall best estimate falls within a broad range. It might seem reasonable to take the mid-range figures in every case and let it go at that. But often, above and beyond the empirical problems, there are lingering questions about the theory itself.

First of all, let's return to the issue of quality. Maybe economists know what a better light bulb is. But there are many areas where no one knows, even in concept, which measures of quality actually translate into consumer satisfaction. How are we to gauge the value of what we buy when we visit a doctor or mail a college tuition check -- much less go to a barber or the movies? Do we look at exam results or job results? Years of life gained or quality of life gained? Such conceptual difficulties are growing more important as our economy shifts from manufacturing to services.

A second problem is how to handle new products. Many inventions seem almost arbitrarily expensive at the beginning, when they are valued highly by a few people, but drop rapidly in price once they are manufactured in volume and popularized. The current CPI, which uses only base-year weights, isn't influenced by this dynamic. But any index including end-year weights would be. Take computers. In 1980, a handful of scientists would have paid a fortune for a Pentium. Does this imply that a vast deflation has gripped the U.S. economy -- or, to put it the other way around, that millions of Americans now have a machine sitting on their desk that's worth far more than their house? And does this imply that today's poorest Americans are better off than their parents simply because they own some high-tech gewgaw that would have bankrupted Mom and Dad?

A third problem is the economist's assumption that our social context remains forever fixed and that only the marketplace changes. In this framework, social expectations about everything from the clothes we wear to the cars we drive have no effect on how much satisfaction we derive from identical consumer products. But this is absurd: Satisfaction depends on social norms. The same polyester suit your grandfather viewed with wonder you may view as a disposal challenge. Indeed, so powerful are these norms that they often leave us with no choice but to buy goods and services that conform -- even if it means doing without other things some of us might value more. Nineteen-fifties style health care is not only unavailable in the 1990s, it's illegal.

Ultimately, the Boskin Commission's critique of the CPI boils down to this: It measures changes in the price of a fixed market basket, not our "satisfaction." But if this is the problem, then there is no satisfactory solution.

The Boskin Commission admits as much when it explains that a true cost of living index would have to take into account such imponderables as the environment and crime. We commend the Commission for not shying away from where its logic leads. But once we embark on this path, we leave the realm of the quantifiable behind. The Commission speculates that recent improvements in the environment have offset the negative impact of higher crime rates. Maybe this is so and maybe not. But how do we measure changes in the moral climate of American society? Do increases in charitable giving offset higher rates of child abuse?

Let's Get It Right

Let us repeat: There are excellent reasons to believe that the CPI overstates inflation. But this does not mean that one cannot make other plausible estimates than the Boskin Commission has -- or indeed, that the very concept of a standard of living is well understood.

The political temptation of a technical revision in the CPI is that it seems purely procedural. But it's not. It's a choice -- and one that could cause a lot of pain. A 1.1 percentage point reduction in COLAs would eventually reduce the Social Security benefit of a 95-year-old by roughly 30 percent beneath current law. Let's not do this unless we're very sure we've got it right.

We therefore favor making an ad hoc adjustment to federal indexing of 0.5 percentage points -- an amount most economists would regard as a minimum estimate of CPI's bias. A technical revision of the CPI should await further study. We're not sure what that study will show. But we do suspect that the widespread perception U.S. living standards have stagnated is not a mass delusion, and that Americans really do have reason to be worried about their own and their children's future.


The Concord Coalition web pages were designed by Marla Parker and Krista Reymann. These pages are now maintained by Craig Cheslog. . Last updated: 24 Apr 1997