Our flawed measures of inflation understate wealth and improvements in consumer well-being.
By Phil Gramm and John F. Early
Aug. 21, 2019 6:59 pm ET
What’s the best way to measure economic well-being in a country as large and diverse as America? One crucial tool is price indexes, which allow us to take common measures of flourishing—hourly earnings, median household income, poverty, gross domestic product and productivity—and remove the impact of inflation to allow meaningful comparisons across time periods.
Yet the economists, statisticians and agencies that create price indexes overwhelmingly agree that official measures of inflation are significantly overstated, leading to an understatement of America’s well-being. The government uses these same price indexes to adjust transfer payments and tax brackets, driving up federal debt.
The consumer-price index was conceived as “an upper bound on a cost of living index,” according to the agency that compiles it, the Bureau of Labor Statistics (BLS). George Stigler, a Nobel laureate and one of the past century’s top economists, chaired the 1961 Price Statistics Review Committee, which found a “systematic upward bias in the price indices.” In 1996 Michael Boskin, a former chairman of the Council of Economic Advisers, headed another blue-ribbon commission, which concluded that, despite improvements, the CPI was still overstating inflation by some 1.1 percentage points a year. Alan Greenspan, then chairman of the Federal Reserve, agreed and called on Congress to correct the overstatement. And in 2010 Erskine Bowles, a former chief of staff to President Clinton, co-chaired the Commission on Fiscal Responsibility and Reform and recommended a more accurate measure of inflation to index entitlement spending.
If consumers always bought the same amounts of the same goods and services, the CPI would provide an accurate picture of inflation. But when relative prices change, tastes shift or new products are introduced, consumers substitute relatively cheap or more-preferred items for relatively expensive and less-preferred ones. As cellphone prices have fallen, for instance, consumers have shifted away from landlines. The CPI calculations didn’t capture the price drop from that substitution.
Price indexes attempt to address new or improved items by splitting their prices into quality improvements and pure price changes. But extensive research has demonstrated that current methods of separating quality and pure price changes consistently overstate inflation.
The CPI for All Urban Consumers, or CPI-U, and its derivative CPI for Wage Earners and Clerical Workers, CPI-W, are the most widely used adjustments for inflation. They have been revised over time, but improvements have been applied only prospectively. A second official index, the CPI-U-RS (research series), incorporates many of the CPI-U improvements retrospectively, improving the accuracy of historical comparisons. A third index, the Personal Consumption Expenditure Price Index, or PCEPI, improves on the CPI-U-RS by accounting for changes in actual consumer expenditures in real time.
Despite the consensus that CPI-U overstates inflation more than CPI-U-RS, which in turn overstates inflation more than PCEPI, the government continues to base its calculations on a variety of indexes rather than use only the most accurate. BLS adjusts average hourly earnings with the CPI-W, which led it to find a 6% increase in real average hourly earnings between 1975 and 2017. The more accurate CPI-U-RS shows real hourly earnings rose 10%, and the still-more-accurate PCEPI shows real hourly earnings rose 23%, nearly four times the number BLS reports.
The Census Bureau uses CPI-U to inflate poverty thresholds and finds the incidence of poverty was unchanged from 1975 to 2017. Using the CPI-U-RS shows a decline in poverty of 14%, and using the more accurate PCEPI produces a 26% decline. The bureau, however, uses CPI-U-RS instead of CPI-U to deflate median family income, publishing its findings of a 21% increase, which would be smaller by more than half using the CPI-U and larger by two-thirds using the more accurate PCEPI.
Even PCEPI significantly overstates consumer price increases. Two recent studies, one by Bruce Meyer and James Sullivan and another by Brent Moulton, combine more than 50 credible studies documenting specific overstatements in consumer price indexes. For example, studies of personal electronic devices showed overstatements of between 3.6 and 5.8 percentage points annually because the value of new features was either understated or missed completely. Annual price increases for medical care were 3 percentage points too high because they didn’t account for the greater efficiency and improved outcomes from new drugs and procedures. No price index takes account of the fact that each month more than twice as many people get medical advice from WebMD as visit the doctor’s office in America. Shelter inflation was shown to be overstated by 0.25 percentage point annually because indexes ignored increased living space and added modern conveniences in homes.
When corrected for documented price overstatements, real average hourly earnings from 1975 to 2017 are shown to have risen some 52%, not 6%—an additional $6.77 an hour. Real median household income increased 68%, not 21%—$17,060 more annually. Gross domestic product grew 253% rather than 216%—$6,312 of additional output per capita. Productivity expanded 142% rather than 117%—$10 of additional value for every hour worked. And published poverty incidence fell by almost half. Combined with the 67% drop in poverty that comes from accounting for all government transfers, poverty incidence sank from 12.3% to about 2%.
By overstating inflation, CPI indexes inflate real spending in Social Security, federal employee retirement, military retirement and poverty programs like Supplemental Security Income. Overstating inflation also overstates the minimum-income threshold to qualify for Medicaid, food stamps and other subsidies. Tax brackets are indexed to chained CPI-U (C-CPI-U), an index similar to the PCEPI. If all transfer payments from 2000 forward had been adjusted with the same C-CPI-U now used for taxes, the federal debt would be $1.1 trillion, or 7%, lower. If the tax code and transfer payments had been indexed to an even more accurate index embodying all known price-index overstatements, revenues since 2000 would have been more than $1 trillion higher, transfer payments $2.6 trillion lower, and the national debt 23% lower.
Since PCEPI and C-CPI-U overstate inflation less than other official indexes, it would seem reasonable to adopt one of them immediately for all historical comparisons, official statistical estimates, and inflation adjustments for taxes and spending. Much of this change could be made administratively. The Trump administration and Congress should support research and development necessary to create a more accurate measure of inflation and enact laws requiring the use of that measure across the government.
We are having major debates based on measures that are demonstrably wrong. We are artificially inflating government benefits and cutting taxes based on bad measures of inflation. As a nation, we need to get our facts straight.
Mr. Gramm is a former chairman of the Senate Banking Committee. Mr. Early served twice as assistant commissioner at the Bureau of Labor Statistics. This article is adapted from a forthcoming book with Bob Ekelund, “Freedom and Inequality.”