The Chained CPI

President Obama has made a budget offer to the Republicans that includes adoption of the so-called "Chained" Consumer Price Index (CPI) for computing the cost of living adjustment (COLA) for Social Security benefits. The universally accepted judgment is that adopting the Chained CPI would reduce Social Security benefits, so there's been an uproar from some progressives and from the American Association of Retired Persons.

 I'd like to explain why I don't think the uproar is warranted on this issue.  But first, a little background.

Social Security beneficiaries have not always received an annual COLA. In fact, the first COLA was a one-time thing in 1950 – ten years after the first checks were paid – but it was a whopper, relatively speaking. Ida Mae Fuller, the representative "first person to receive a Social Security check," saw her monthly check increase from the $22.54 it had been since 1940 to $41.30.

For the next 25 years, every Social Security COLA required a separate legislative action. There were COLA's in 1952, 1954, 1959, 1965, 1968, 1970, and 1971. In 1972 Congress passed legislation that provided for annual COLA's beginning in 1975, providing the CPI-W (explained below) had risen by a specified percentage. In 2009 and 2010 there were no COLA's because there was essentially no inflation, and the triggering percentage had not been reached.

How does the Chained CPI differ from the current way of computing the annual COLA for Social Security benefits? Why would it reduce benefits, and by how much?

Peter Orzag, former director of the Congressional Budget Office and Obama's first director of the Office of Management and Budget has some answers for us.

Once an American begins to claim Social Security benefits, his monthly checks increase each year in line with a consumer price index called the CPI-W (the “W” is there because the index was created to measure inflation for workers). The federal tax code, for its part, is indexed to a related measure, the CPI-U, which is the inflation measure that receives the most attention each month.

The Bureau of Labor Statistics, which calculates both indexes, also publishes the chained CPI, which is a more accurate measure of inflation [in Orzag's opinion - ed.] because it better reflects how people change what they buy in response to price increases. When the price of apples rises relative to oranges, for instance, people eat more oranges, and the chained CPI accounts for this substitution, reducing the measured inflation rate. [It can easily be argued that this makes the chained CPI a less accurate measure of inflation, as people substitute sawdust for wheat flour, but that's not the point of this post. -ed.]

The result is that the chained CPI rises more slowly than either the CPI-W or the CPI-U. Switching to the chained index would therefore cause Social Security checks to grow more slowly. And if the Internal Revenue Service switched to the chained CPI as well, the cutoff lines for tax brackets would rise more slowly, pushing more Americans into higher marginal tax brackets and thereby raising revenue.

How big of a difference is there between the CPI-W and the chained CPI?

Official budget estimates suggest that switching to the chained CPI would save the federal government about $125 billion on Social Security benefits and about $40 billion in other indexed benefits (such as federal civilian and military pension payments) over the next decade, and raise about $125 billion more in tax revenue. It would also save about $30 billion in health programs and nearly $20 billion in refundable tax credits. That adds up to total deficit reduction of about $340 billion. The Social Security actuaries suggest that it would also reduce the 75-year actuarial gap in the program by about 20 percent.

But "Social Security benefits even 20 years after retirement would be reduced by less than 2 percent." [My emphasis.]

What neither side seems to have noticed, however, is that the difference between the chained CPI and the standard CPI has been diminishing. That means the impact of switching indexes may not be as great as many assume. The change may still be a good idea, but it probably won’t matter as much as expected.

The current method of computing the annual COLA was not handed down to us on stone tablets. In fact, it's a wonderful thing that we have COLA's at all; as we have seen, it was not always so. It's best that we not become unstuck from the reality-based world, as did some who bemoaned the lack of a "raise" in their Social Security in 2009 and 2010.

In a post last month I liked a NY Times proposal that Congress instruct the Bureau of Labor Statistics to construct a "statistically rigorous index of inflation among retirees." That seems fine to me. Make it statistically rigorous, then let the chips fall where they may.

Until then, I will not get exercised about a change that reduces benefits by 2% over 20 years. There are bigger battles to fight.

A good commentary from the Center on Budget and Policy Priorities here.