The Cpi: Washington's Hot Potato

Why fixing errors in the CPI is so politically risky

It looks like a pot of gold--but it smells like a barrel of trouble. On Dec. 4, a Senate advisory panel concluded that the consumer price index, America's main gauge of inflation, overstates price hikes by 1.1 percentage points a year. Correcting that error--in effect, knocking down today's official 3% inflation rate by a third--could slash almost $200 billion from the federal deficit over the next five years. It also would slow the growth of wages and benefits for millions of workers and retirees, and even change America's view of how the economy has done over the past two decades.

But estimating errors in the CPI is one thing. Fixing them will be a far tougher task. The problem: Nearly half the budget savings from a CPI correction would come in the form of smaller annual cost-of-living adjustments in Social Security and other federal retirement benefits. And while the panel's findings imply that retirees have been reaping a COLA windfall, lobbies for the elderly don't see it that way. "Seniors watch very closely, and they know what to do if Congress tries to cut COLAs," warns Evelyn M. Morton of the American Association of Retired Persons. Both the AARP and organized labor already are mobilizing to block any change.

Those groups may find some unlikely allies: hard-line antitax conservatives. Since 1981, the CPI has been used to adjust federal income tax brackets and personal exemption amounts to protect taxpayers from inflation-induced "bracket creep." For a family of four in the 28% bracket, a 1.1-point drop in the COLA would mean an extra $86 in taxes. Budget hawks argue that over-indexing has given taxpayers, like retirees, a windfall. But GOP conservatives are likely to paint a CPI revision as a backdoor hike in taxes on families.

UNEASY EMBRACE. Little wonder, then, that GOP leaders and the White House greeted the CPI report coolly. Fresh from their battles over slowing Medicare spending, Republicans insisted that President Clinton would have to sign on to COLA cuts first: "Silence is not leadership," says Senate Majority Leader Trent Lott (R-Miss.). But silence may be all the GOP gets from Clinton for now. Administration economists say they'll need months--maybe years--to study whether COLAs have been too rich and whether the index needs to be replaced. CPI revisions "won't play a role" in the 1998 budget talks, says Treasury Secretary Robert E. Rubin.

So President Clinton and GOP leaders will find themselves in an uncomfortable embrace. If they want the savings from COLA cuts, they must take a public plunge together. But if they pass them up, they'll lose one of their best hopes for a budget-balancing deal.

This high-stakes game was launched by Federal Reserve Chairman Alan Greenspan. Shortly after the GOP captured Congress in the 1994 elections, the Fed chief testified that the index could overstate inflation by as much as 1.5 percentage points, adding billions to the budget. Senator Daniel P. Moynihan (D-N.Y.) took the hint and lobbied the Senate Finance Committee to appoint a commission to study the CPI and measure its biases. Stanford University professor Michael J. Boskin--President Bush's chief economist--was picked to head the group.

In its report, the commission concludes that the CPI is too simple an index to measure prices accurately in a rapidly changing economy. Each month, the Bureau of Labor Statistics surveys prices in stores, auto dealerships, doctors' offices, and other outlets. It plugs the prices into formulas that give each price a weight based on how much of the average consumer's budget is spent on that item. New cars, for example, make up 4% of the CPI market basket. The budget breakdown is based on a snapshot of consumer spending, last taken in 1982.

This "fixed-weight" approach "guarantees you'll have errors in a flexible, dynamic economy," says Boskin. As some goods become more expensive, consumers substitute items whose prices are rising less quickly: Golden Delicious apples for Granny Smith, or chicken for beef. And as discounters edge out downtown department stores, the BLS's calculations don't credit shoppers with the full savings. The Boskin panel figures these two biases add half a percentage point to the CPI's rate of growth. To correct that, the group calls for scrapping fixed weights and updating the consumer market basket more often.

The panel also identifies an even bigger set of errors in the BLS's adjustments for quality improvements. Today's cars are safer, more fuel-efficient, and cleaner than those of 1983. So while auto prices rose 73% from 1983 to 1995, the CPI for new cars rose only 39%. The difference, BLS says, is due to higher quality and added features on today's cars. But the Boskin group concludes that not enough credit is being given to quality improvements--not only in cars but in medical care and consumer electronics. Using studies by panel member Robert J. Gordon of Northwestern University, the group says that fully recognizing quality gains would shave an additional 0.6 points off the CPI.

A NEW POVERTY LINE. Are they right? BLS statisticians acknowledge that their methods overstate some price changes--though they insist the errors are much smaller than Boskin says. Quality improvements are especially tricky to measure. But the Boskin panel's estimate is in line with other studies, such as a DRI/McGraw-Hill report to the BLS that pegs CPI bias at 1.2 points. And it's far below other estimates, such as a Federal Reserve Bank of Philadelphia study that says that errors total 2.75 points--virtually all of today's 3% inflation.

Whatever numbers are used, rejiggering the CPI will affect a vast array of other government statistics. The impact is greatest on real, inflation-adjusted wages and incomes. Using the official CPI, real median family income rose only 4% from 1973 to 1995; using Boskin's figures, family income rose 36%. Average hourly earnings rose 13%, Boskin says, rather than falling 13% as originally reported. The new figures "will put the problem of poverty in a rather different light," says panel member Dale Jorgenson of Harvard University, because the poverty line, adjusted each year for price hikes, would be lower.

To be sure, the spreading income gap between low-wage workers and executives won't grow or shrink with these new figures. The CPI measurement gap appears to have been fairly consistent during the past 20 years, and there's no evidence that any errors have affected income measures differently for different groups of consumers--low-income vs. middle-income households, for instance. "Consumers won't say, `Now I'm feeling better, because the statistics say I'm better,"' says F. Thomas Juster, a University of Michigan economist.

The debate over how fast the economy can grow, too, won't go away. Boskin's and DRI's figures imply that today's economy is growing faster than official measures indicate, by 0.5 to 0.8 points: Growth in real gross domestic product, measured at a 2.9% rate in the first nine months of 1996, would be adjusted up to as much as 3.7%. But the measurement error doesn't explain the economy's slowdown since the 1960s, when GDP gained 4.7% a year, because a similar adjustment would be added to historical growth rates. "It's clear that real growth is underestimated now--but it also was underestimated before," says panel member Zvi Grileches of Harvard. Business leaders who champion tax cuts and deregulation to boost growth won't be deterred by any CPI revisions.

A restated CPI would mean that productivity is growing faster than we think--but still not as fast as it used to. DRI estimates that the 1990s' 1.2% growth rate in output per worker-hour should be revised up to 2%. But that's still far below the 1960s' rate, which DRI would revise up to 3.2%. The Fed, for one, has long adjusted to that problem anyway. Following Greenspan's lead, the central bank has largely ignored the CPI, steering instead by other signs of wage and price pressures. Says Fed Governor Lawrence B. Lindsey: "We adjusted our thinking for CPI bias long ago."

FIVE-YEAR PHASE-IN. Business, however, has not. A new, slower-growing cost-of-living index could eventually find its way into lower raises for workers. Some 60% of the big-company managers surveyed by Buck Consultants Inc. say that CPI inflation is still important in setting raise amounts. "I'm not sure the CPI alone can change things," says Gordon M. Bethune, CEO of Continental Airlines Inc., "but it can sway things over time."

Over a long time. The Boskin report's 16 recommendations for changing inflation measures would require as much as five years to phase in, says Grileches. Even then, the BLS would be far from realizing the full adjustments for quality changes. So Congress and the White House won't have an official index to justify any benefit slowdown before their 2002 goal for balancing the budget.

But they need budget savings now. Budget negotiators had hoped that the Boskin panel would give them a clean, scientific way to slow down COLAs. Instead, the slow pace of CPI reform will mean that politicians will have to trim COLAs on their own, by pegging them to something less than the CPI. That means looking seniors in the eye and explaining why. The Boskin report may justify COLA reductions to economists. Convincing the elderly that they're not getting a raw deal will be one tough sale.

Before it's here, it's on the Bloomberg Terminal.