Home Mail Articles Stats/current Supplements Subscriptions Links

The following article appeared in Left Business Observer #70, November 1995. It retains its copyright and may not be reprinted or redistributed in any form - print, electronic, facsimile, anything - without the permission of LBO.

January 1997 update

In December, Boskin's commission issued its final report, declaring that the CPI was overstated by 1.1 percentage points. They did no original research.

Here's how Boskin's commission was formed. In the summer of 1995, the Senate Finance Committe held hearings on the CPI. Of the 15 or 20 economists who testified, 5 of the 6 with the highest estimates were put on the panel; the 6th was a Canadian.

For Economic Policy Institute's analysis of Boskin's revisioning of the CPI, click here. The BLS's own response is here.

The politics of counting

In normal times, statisticians do their work in peace, free of the scrutiny of politicians and opinioneers. But these aren't normal times. Under its new Republican masters, the once-honest Joint Tax Committe of Congress is now playing games to disguise the class effects of proposed tax and budget cuts. This is fairly straightforward deviousness. More complex manipulations are also underway. It's now almost universally believed that the consumer price index (CPI) overstates inflation, and that our entitlement programs guarantee that newborns will end up paying 84% of their lifetime income in taxes. The first assertion, on which Fed chair Alan Greenspan and the AARP concur, is based on only shreds of evidence; the second is spun from techniques so flawed you'd almost think their concoctors set out to prove what they wanted to. Whatever the spirit in which the numerical exercises were first undertaken, both are being used to screw the weak.

What follows owes a lot to Dean Baker of the Economic Policy Institute, author of two recent dissections of the numbers.



The charm of redefining the consumer price index (CPI) down is that it would save lots of money. Social Security checks would get a little less bigger every year, and tax brackets, indexed for inflation, would creep up more slowly, boosting revenues. Were the index nudged downward by one percentage point a year, it would reduce the cumulative federal deficit by $634 billion through 2005. Dazzled by the political arithmetic - benefit cuts and tax increases without a vote! - the Senate Finance Committee assembled a panel of worthies, led by Bush's chief economist Michael Boskin, to study the index; all were known to be keen on the idea of ratcheting it down. The Committee and its experts concluded, on the basis of not much evidence, that the overstatement is 1%.

Everyone's list of reasons why the CPI exaggerates inflation includes these: (1) The Bureau of Labor Statistics (BLS), which makes the index, doesn't incorporate new products in its sample quickly enough. The BLS addressed this complaint with a technical change last January. [For a sample of the BLS's internal critiques of the CPI, click here, and enter "CPI" as your search term.] (2) The index doesn't adjust enough for substitutions; if beef gets more expensive, people eat chicken or beans, but you'd never know this from the CPI, since it retains the spending patterns of its base year, now 1982. But substitutions may be made grudgingly, so there may be an intuitive point to retaining a fixed market basket. (3) The BLS hasn't kept up with discount retailing, from Sam's Club to home shopping. (4) The BLS under-adjusts for quality improvements, which are conceptually the same as a price decrease. (5) Gadgets that plummet in price, like VCRs and computers, aren't weighted heavily enough. Taken all together, a 1994 Congressional Budget Office study put the overstatement at 0.3-0.4%.

Countervailing factors

Almost no one, though, looks at reasons the CPI might understate inflation, but there are a few. (1) The BLS's own research suggests it may be undercounting inflation in health insurance premiums. (2) Use of 1980 population figures to blend local figures into a national mix underweights rapidly growing areas, which are the most likely to have rising prices, and underweights stagnant areas, where prices are likely to be tamer. (3) Ironically, since a major aim of the CPI rethink is to cut Social Security payments, the existing CPI may understate general inflation because it understates inflation experienced by the elderly. An experimental BLS price index for over-62s increased 0.3-0.4 percentage points more than the official price indexes annually between 1987 and 1993. Pensioners haven't had it so good after all.

And the final reason the CPI may be understated is that (4) the BLS overadjusts for quality improvements. It's easy enough to say that if something stays the same in price but suddenly gets snazzier or shoddier, then it's as if its price has changed. A price index should account for this. But in practice, quality changes are very hard to measure, both with goods and services. Sure the MegaWarehouse is cheaper, but is shopping in a barn a decline in quality? Cheesy jewelry only a phonecall away - progress? Certainly people feel that the quality of lots of services has declined.

It might seem easier to adjust goods for quality changes, but there too the details are devilish. Take two important items, clothes and cars. Fashion changes clothing all the time - what are real quality changes and what are ephemera? And how do you adjust cars for quality? By some measures, cars have gotten a lot more expensive over the last 20 years - but the CPI says no. The number of workhours required for the average worker to earn enough to buy the average car has risen 69% since 1973 - but according to the "new car" portion of the CPI, the number should have fallen. A good bit of that difference is accounted for by quality adjustments. Bigger engines were counted as quality improvements in the 1950s and 1960s, but smaller engines have been rated as improvements since the 1970s. Anti-theft devices are counted as improvements, but they're no substitute for lower crime rates, and they don't seem to deter that much theft. Pollution control and safety devices may be desirable, but there's no doubt they impose real new costs that a price index shouldn't discount.


Bellowing yahoos

None of this is to argue that the statisticians are incompetent or devious. Quite the contrary, the U.S. statistical agencies are admirable, among the best and most open in the world. Their work is very hard, and they don't get much financial or political support for it. Computing the CPI is among the toughest of all their pursuits; it blends thousands of goods and services bought by 264 million people into a single representation of household buying power.

In his testimony before the Committee, one of the panel members, Harvard economist Zvi Griliches, noted that while the Senate Republicans were out to rework the CPI, their colleagues in the House were cutting funds for economic research and government statistical agencies. A serious rethink of the CPI - not that any is really needed - would be a massive undertaking for an agency that can't keep up with the work it already has. No one in Congress would ever fund it. Instead, Congressional yahoos wonder aloud why we need "all those questions" on the Census; it's not a climate friendly to serious inquiry. Griliches also said this: "The perception that we are trying to evade our national responsibility of being our 'brother's keeper' leaves me feeling sad about my adopted country." He overcame his regrets to lend his blue-chip name to the Committee report.

Despite the pressure, the BLS isn't going to change the CPI until its normally scheduled rebasing late this decade, an event that occurs once every 15 years or so. But it's easy to suspect that the Bureau will do its best to keep the reported number down to please its paymasters.

Curiously, though the AARP says publicly that it opposes a radical overhaul of the CPI, a Washington source says that it's invisible in the lobbying wars, where it counts. There is speculation that this silence was won by Sen. Alan Simpson's threats to take away AARP's tax exemption if it engaged in too much politicking.


Tax shocker!

War against the old is also being fought on another numerical front, accounting for time. Economist Laurence Kotlikoff has made a brand name of himself - appearing even in Rock the Vote documents - by promoting something called generational accounting (GA). GA tries to put hard numbers on one of the austerity cops' favorite images, the deficit as theft from the future. Kotlikoff and his frequent collaborators Jagdeesh Gokhale and Alan Auerbach (call them AGK) are the source of the claim that future generations - represented in TV ads by cute, scared X-ers - will face a lifetime tax rate of 84%, compared to 24% for those born in 1900 and 34% for those born in 1993.

Robert Haveman took GA apart in the Journal of Economic Perspectives (Winter 1994). As Haveman wrote, only governmental takings enter into the generational accountants' spreadsheets; the positive effects of today's spending on education, research, health, and infrastructure on future lives is ignored. AGK assume that today's citizens need not pay off their generational debts in their lifetimes, but tomorrow's citizens must; generations of the future are denied the privilege of passing the buck. That assumption alone produces a huge jump in the tax burden between those born in 1993 (34%) and those born in 1994 and beyond (84%). Finally, despite the appearance of precision, GA numbers are actually the products of complex estimations that "are calculated using a myriad of judgments that are open to question," in Haveman's soft words. Proof of that lies in the wild variations among estimates made by different number-crunchers, and even by the same team of jugglers working at different times.


Deep geekiness

Three of the questionable judgments are the interest rate used to value the future, the distribution of costs and benefits, and the evolution of government spending.

Discounting the future is one of economics' fundamental techniques. In this case, one of the crucial issues is how you compare future Social Security and Medicare benefits with the amounts workers pay into the system during their working lives. What is $100,000 in 2005 worth in 1995? What you'd have to set aside in 1995, at prevailing interest rates, to have $100,000 a decade later. The higher the rate, the less you have to set aside. Over many years, the choice of interest rate makes all the difference in the world. To have $100,000 25 years hence at a 5% interest rate you'd have to deposit $29,530 today; at 10%, only $9,230. In jargon, the present value of $100,000 in 25 years at a 10% discount rate is $9,320. Thanks to interest compounding, doubling the interest rate cuts the present value by two-thirds.

In GA, choosing the right discount rate is critical, but which one? The higher the rate, the worse a deal it looks like young people are getting from the government (since the higher the rate, the lower the present value of their future benefits compared with the taxes they'll pay for them). Since we're dealing with the federal budget, the intuitive discount rate would be the rate the U.S. government pays on long-term bonds; over the last 76 years, that has averaged 2.2% (after inflation). Despite lip service paid to the virtues of a low rate, AGK choose a high one of 6%, for reasons they never make fully clear. This stacks the deck from the outset.

Other problems are easier to explain. AGK allocate spending on the elderly to the elderly alone - as if their children who are freed from caring for them are not better off because of public health and pension programs. But they allocate spending on education to society as a whole, not the young. They count Medicare payments to doctors and hospitals as income to the elderly; no doubt the sick would prefer to be healthy. And they assume that present Medicare and Medicaid spending growth will continue - though of course it won't.

Canonical GA accepts preposterous assumptions on the growth in medical spending - ones that forecast that the average family in 2030 will spend 97% of its income on health care. Obviously, life as we know it would end before that could happen. One way or the other, health spending will not rise until it consumes 97% of our income. If health inflation were to settle down to the general rate of inflation, the generational problem virtually disappears, even at high discount rates.

Dean Baker re-ran the generational accountants' numbers using interest rates a lot lower than their 6%. At 6%, someone born in 1989 takes a net loss of $20,900 (paying that much more in taxes than getting in Medicare and Social Security benefits); at 3%, the accounts about balance. Allocating education spending to the young yields a lifetime tax burden of 57% at the 6% discount rate, and 27% at 2%. If you count education as a benefit to the young they carry into the future, and allow modest levels of borrowing, then lifetime net tax burdens fall profoundly, to 23% and below, depending on the discount rate. Sensible assumptions make the generational problem disappear.

Deconstructing AGK's devious arithmetic doesn't mean that the deficit isn't a problem. Debt increases capital's power over the state; if you have to sell lots of bonds, you have to keep your bankers happy, and the servicing of debts takes priority over more fetching possibilities. Because the government borrows from rich people rather than taxing them, interest on the federal debt means a transfer from the bottom 90% of the population, who own few if any Treasury bonds, to the richest 10% who own them in quantity. By playing games with time, GA completely obscures the fact that debts and interest payments occur between classes of people in the present. GA replaces class war with a spurious generational war.

Home Mail Articles Stats/current Supplements Subscriptions Links