Some Comments on the Chained CPI Controversy

I have perhaps a unique perspective on the Administration’s proposal in its FY 2014 Budget to use the chained CPI-U (“C-CPI-U”) for “most” government programs indexed to an inflation measure, including Social Security, and for the indexation in the Internal Revenue Code (see last page of this link). As a Treasury official, I once had to decide what inflation measure would be used for a new security we were developing – inflation-indexed bonds and notes, which are now called Treasury Inflation-Protected Securities (“TIPS”). While various indices had been proposed, the choice was not all that difficult – the CPI-U (all urban consumers) was the most recognized measure of inflation, though Social Security is indexed to a somewhat narrower index, the CPI-W (urban wage earners and clerical workers.)  However, while the headline monthly inflation number uses the seasonally adjusted version of the CPI-U, TIPS use the non-seasonally adjusted number because it is not subject to revision.  In order for there to be a well-functioning market for these securities, there needs to be finality in the index numbers. On trade and settlement dates, the nominal values of the principal and accrued interest need to be known with certainty.

One concern those of us who worked on TIPS had was the possible perception that the U.S. government had a conflict of interest in indexing securities to an index the U.S. government produces. We hoped people would understand that the Bureau of Labor Statistics (“BLS”) was not under the influence of the Treasury and staffed by objective economists and statisticians not subject to political influence. It did not help matters, though, that the Federal Reserve was at the time trying to persuade the BLS that the CPI needed technical fixes. All the technical fixes the Federal Reserve was proposing would lower inflation as measured by the CPI. While at least some of the Fed's suggestions were likely justified and the BLS made some changes, I did find it peculiar that they all went in the same direction -- lowering reported inflation. Were there no arguably necessary technical fixes that went in the opposite direction? And didn't the Fed's lobbying of the BLS seem a little like a student complaining about his grade to his teacher?

This brings out two concerns about the Administration’s proposal. First, it is not clear how the Administration plans to handle revisions in the chained CPI. The revisions are necessary because the calculations are based on expenditure data which are only available with a lag. According to the BLS, the initial data for the current calendar year is subject to two revisions. The next calendar year, interim data will be published for the previous calendar year, and, the year after that, final values will be published. If Congress and the Administration decide to use the C-CPI-U, they will have to decide whether or not they will ignore revisions to the initial data.

While this first concern is a technical problem, the second concern is more political. The Administration and supporters can protest all they want that using the C-CPI-U is justified on the grounds that it is a more accurate measure of inflation. No one will believe it. Even the Administration seems to contradict this premise, because it has proposed “protections” from the effects of indexing Social Security to the C-CPI-U for the elderly (starting at age 76). Also, means-tested benefit programs would not be indexed to the C-CPI-U.

Patrick Brennan, who writes for the National Review, also argues that the real reason to use the C-CPI-U is not because it may be a more accurate measure of inflation. In a recent article, he states that those who argue over which index to use for Social Security, perhaps a chained CPI for older consumers, are “missing the point: Chained CPI has been proposed because it is expected to slow one of the ways in which Social Security benefits are scheduled to increase, though it also happens to reflect that those increases were probably more generous than intended. Debating the most accurate measure of inflation is merely the most politically palatable way of limiting how much we are willing to promise in retirement benefits to every American. That kind of limitation has to happen somehow, unless Americans would prefer significantly higher taxes, much less spending on other federal priorities, or permanently higher levels of debt.” (Of course, as many of pointed out, there are other ways to improve Social Security finances, such as raising or eliminating the ceiling on the amount of salaries or wages to which it is applied.)

A perhaps somewhat more objective observer, Peter Coy, the economics editor of Bloomberg Businessweek, writes in another recent article that the chained CPI proposal is “presented as a technical, politically neutral fix, but make no mistake: The Obama administration’s proposal to change the basis for Social Security raises to ‘chained CPI’ is all about saving money by slowing the growth rate of benefits. Whether you think that’s a good thing or a bad thing depends on whether you believe workers have been paying too much to support their elders.”

I would also point out that there are people who feel that the current CPI under reports inflation. Many of the people who think this are probably not liberals and generally distrust the government. It is true, though, as anyone who has been involved in creating or calculating economic indices knows, there is no perfect index, whether it be an inflation, foreign exchange, stock market, or other index. In producing a general inflation measure, there will always be an element of judgment about how to do it.

Also, choosing the right index is a judgment call. If the purpose of indexing Social Security benefits is  to protect those receiving them from increases in prices, an argument can be made that a chained CPI based on the basket of goods and services older people consume would be justified. This might, though, mean indexing different parts of Social Security differently, as the Administration already proposes to some extent, since not all who receive Social Security benefits are elderly.

While liberals’ problems with the chained CPI proposal are obvious, it also poses a dilemma for Republicans. Since tax brackets would be indexed to an inflation measure, this would serve to increase taxes over time. Contrary to some commentary I have seen, this would even apply to those in the top marginal bracket. While these high income taxpayers would not see an increase in their marginal tax rate over time because of the chained CPI proposal, their average tax rates would increase as the lower brackets moved up more slowly than they would under current law. It is not politically possible for Social Security to be indexed to the C-CPI-U without doing the same to tax brackets.

While a grand budget bargain seems unlikely, those who are concerned about the chained CPI proposal because of its effect on Social Security benefits, taxes, or both, should not be entirely complacent.  There could be a smaller budget bargain that includes this.

Finally, with respect to TIPS, the index for the existing securities cannot be changed, since this is part of the terms and conditions for these securities. Also, as I indicated above, there would be problems with indexing new securities to an index subject to revision. The Treasury Department, though, might have a bit of a public relations problem if the chained CPI proposal were to become law. It might have to justify indexing securities designed to “protect” investors from inflation to an index that indicates higher inflation than that used by the government to calculate Social Security benefits or tax brackets.