A while back, Dean Baker wrote a piece in response to a NY Times review of Naomi Klein’s book The Shock Doctrine: The Rise of Disaster Capital. It’s all worth reading (as usual), but along the way Baker touches on the claim, made in the context of various attempts to “reform” Social Security, that the CPI has been overstating SS’s cost of living increases by a percent or so, and that scaling the COLA back would help “save” the program.
Did it seem plausible to you at the time? Baker thinks it shouldn’t have.
The Social Security debate provides the most obvious example. Consider the effort to change the post-retirement indexation formula in the mid-nineties, which had support from the Clinton administration, many prominent members of Congress (Senator Daniel Moynihan led the crusade), and many of the country’s most respected economists.
The argument was that the consumer price index (CPI) overstated the true rate of inflation by approximately 1 percentage point, so therefore Social Security benefits to retirees should rise each year by approximately 1 percentage point less than the rate of inflation shown by the CPI, rather than the CPI, as is the case under current law.
While the evidence for their claim was weak as I argued in my book, Getting Prices Right: The Debate Over the Consumer Price Index, there was a more basic issue that there were huge and unavoidable implications of this claim that none of its advocates were willing to accept. Specifically, if their claim was true, then most of the people who would see their benefits cut by the change in the indexation formula had grown up in poverty. Furthermore, the future generations who they wanted to protect by reducing the deficit were actually going to be far richer than we could possibly have imagined.
The logic is simple. If the CPI overstated inflation by 1 percentage point annually, then real incomes had been rising much more rapidly than the official data show. Instead of rising by about 2.0 percent annually over the prior forty years, if inflation had been overstated by 1.0 percentage point, real per capita income had actually risen by 3.0 percent annually. (That’s arithmetic – if nominal income had risen by 5.0 percent, and the real inflation rate was 2.0 percent, rather than the 3.0 percent shown by the CPI, then real income rose by 3.0 percent.) If real income had been rising by 3.0 percent instead of 2.0 percent, then we were much poorer 40 years ago relative to the present than the official data show. In fact, if we go back 40 to 50 years with this adjustment, the median family was below the current poverty line.
On the other side, if the yardstick against which we are measuring future income growth is overstating inflation by 1 percentage point annually, then we should adjust upward our projections for future income growth accordingly. This means that our children and grandchildren will be hugely richer than our current projections show – we can’t even think of any economic policies that we would expect to lift income growth by a full percentage point.
Anyhow, virtually all the leading lights of the economic profession were prepared to completely ignore the logical implication of their own claim about the CPI in the effort to force a reduction in Social Security benefits. The drive was only halted by the refusal of Richard Gephardt, then the leader of the Democrats in the House, to go along with the scheme. At the time Gephardt was considering a challenge to Vice-President Al Gore for the 2000 Democratic presidential nomination. There could have been no better issue for Gephardt in the Democratic primaries than the defense of Social Security against the guy who cut it. Therefore, the Clinton administration nixed the benefit cut.
Note that all the economists who lined up behind the cut to Social Security are not currently expressing concern about the enormous distortions in our official data that result from the fact that the CPI has not been “fixed.” (The CPI overstatement would contaminate all price and quantity data over time.) They also don’t adjust for this error in their own work.