Penny-Wise and Pound-Foolish on the CPI

On Dec. 4, a congressional advisory commission that I chair released its final report, titled “Toward a More Accurate Measure of the Cost of Living.” My four eminent economist colleagues and I were charged with assessing the extent to which the government’s main inflation gauge, the consumer price index, accurately measured inflation and, if not, what to do about it.

The commission’s unanimous conclusion was that changes in the CPI overstate changes in the true cost of living by about 1 percentage point per year. This may seem like a small amount, but when compounded over time, it matters a lot. Because one-third of the federal budget and parts of the tax code are adjusted automatically by changes in the CPI, the impact on the budget over a dozen years of this overindexing amounts to $1 trillion.

The overstatement of inflation obviously also sends confusing, inaccurate signals to private-sector workers, savers, investors and renters. Some private contracts are explicitly tied to the CPI, and many more are implicitly.

But perhaps the most important ramification is that our basic measures of economic progress have been understated for many years. This is because everything we measure in dollars of the current period must be converted to so-called constant dollars by adjusting for inflation in the intervening period to compare real purchasing power between, say, 1973 and 1996. Thus, the inflation factor is crucial in measuring real economic progress over time. The more accurate inflation gauge suggests that we have been doing somewhat better than previously thought. Rather than declining, real wages have increased since the 1970s, albeit at a slower rate than during the 1950s and 1960s.

The culprit here is much more the economy as hero than the government’s statisticians as laggards. Measuring inflation in a dynamic market economy is complex, as prices of tens of thousands of goods must be measured accurately, purchasing patterns ascertained, trends in retailing evaluated and, most important, the continuous quality change and introduction of new products providing the bulk of Americans more and better options must be factored into the index.

Unfortunately, the current mathematical formula and procedures used in calculating the CPI either ignore these factors or account for them only partially or with a long lag. Many new products are introduced a decade or more after they enter the market, after prices have fallen and quality has improved substantially--VCRs and microwave ovens, for example. Cellular telephones won’t be in the index until 1998, after competition from personal communications services and the advent of digital cellular technology.

Consumers also change what they buy and where they buy it in response to changes in prices. Economists call this substitution, as in switching to Golden Delicious apples from Granny Smith if prices change.

The commission made many recommendations that would allow the Bureau of Labor Statistics to partially solve some of these problems. Although some can be done quickly, others will take much longer and require more resources, which the president and Congress should provide.

Because the budgetary ramifications are so important, far too little attention is focused on the general problems of our economic statistics that are a major input into citizens’ understanding of the state of the economy, economic progress over time and comparisons with other countries. The components of the CPI are fed into the national income accounts and thus part of the bias carries over to the measures of real personal consumption expenditures, by far the largest component of real gross domestic product.

Other statistical gauges calculated in other statistical agencies also suffer from serious problems. Partly this reflects tremendous changes in the economy. As the economy evolved from an agrarian society, in which agricultural output and employment dominated, to a manufacturing one, problems with measurement arose, but not nearly so severe as those in recent times. It was not that much more difficult to measure tons of steel than bushels of wheat.

But as an ever larger share of what we produce and consume in most advanced economies is far more complex to measure, problems naturally arise. We spend a much larger fraction of our resources on medical care than we did decades ago. But should we be pricing the inputs, such as doctor salaries and hospital stays, or the output, such as the price of an episode of treating heart problems or cataract surgery?

Medical care is a good example of quality change and new products not being reflected adequately in the price indices. Indeed, a sizable chunk of the rapid medical inflation of the last 15 years was quality improvement, not price increases, for the same products.

I strongly favor proposals to make the government a more efficient and effective provider of necessary public goods and services. It should be adopting private-sector business methods. While our report indicated areas of cost savings, it is important to place this in perspective: Virtually every private enterprise in America is spending heavily in information technology, from buying hardware from Sun Microsystems and Intel and software from Oracle and Microsoft to hiring and training human capital from our nation’s colleges and universities.

These statistical agencies need to be doing the same. We’re being penny-wise and pound-foolish not to provide the guidance and resources necessary for the basic economic information on everything from making monetary, tax, regulatory and budget policy to calibrating private contractual arrangements to providing our citizens with as accurate a measure as feasible of where the economy is, where it has been and where it is going.

Despite the best efforts of many hard-working, talented people in the statistical agencies, the current system is close to a crisis. We can make the necessary investment to update and improve it now, or continue to suffer ever worsening consequences. That should be a no-brainer, even in tight budgetary times.