To chain or not to chain

In an effort to obtain a Grand Bargain on deficit reduction, the Obama administration has offered to accept a Republican proposal for a new inflation index—a chained CPI—in setting the annual cost of living adjustment (COLA) for Social Security benefits. This new inflation index would also apply to the indexation of income tax brackets. Since a chained CPI is expected to show lower inflation, the change in indexation will mean lower COLAs and greater revenues over time. This is the first of two posts articulating why accepting a chained CPI for calculating the Social Security COLA is a bad policy choice. The other post will address the chained CPI proposal in the context of Social Security policy. This post addresses whether a chained CPI is simply a “technical fix,” as some maintain, to obtain an accurate measure of inflation. I pay particular attention to my disagreements with my friends at the Center on Budget and Policy Priorities (CBPP).

A better measure of inflation?

Let’s be straight, a chained CPI is not a more accurate measure of inflation for setting the COLA for Social Security beneficiaries. There is a good argument to be made for any given reference population that a chained CPI index is more accurate than an unchained index. However, this “any given reference population” is an important caveat: applying a chained CPI for average consumers to calculate price increases faced by Social Security beneficiaries is not an improvement in accuracy since the expenditures of Social Security beneficiaries, especially the elderly, are very different than the average consumer. As experts have pointed out, indices based on the spending patterns of workers or the general population likely understate the impact of cost increases faced by Social Security beneficiaries because seniors and disabled people spend a greater share of their incomes on out-of-pocket medical expenses than do other consumers, and health costs have risen faster than overall inflation in recent decades. This has been documented in the Bureau of Labor Statistics’ (BLS) CPI-E inflation measure which uses consumption weights specific to the elderly and had 0.2 percent faster inflation from 1982 to 2007 than the measure currently used to index Social Security benefits.

So, what this means is that there are two known biases to current Social Security COLA indexation, the failure to chain expenditures (which overstates inflation) and the failure to adopt weighting particular to Social Security beneficiaries (which understates inflation). Yet too many inside the Beltway only seem interested in correcting the first. And why this narrow focus? The only rationale for imposing a new inflation measure on the elderly that only addresses the chaining bias is to reduce benefits.

CBPP’s Jared Bernstein, formerly our colleague at EPI, recognizes this, saying “So, a smart way to go here is to pay the BLS to derive a more accurate price index for the elderly and then chain weight it.” Dean Baker, also a former EPI colleague, says the same, “If we want a more accurate index for adjusting Social Security benefits then we would have BLS construct a full elderly index.” Hundreds of economists said the same thing in a recent statement. A new index could reflect the consumption weights of Social Security beneficiaries, which includes the elderly, the disabled and some children as well. This would allow us to have a more accurate inflation measure in a year or two, but this more accurate measure is not what Grand Bargainers actually want. Instead, they want a policy change that necessarily reduces future spending and can be scored by the CBO now, something the chained CPI approach does do.

It is also clear that the main political actors supporting chaining the Social Security COLA index don’t behave as if the chained inflation measure is the most accurate. A prime example is the recent fiscal cliff deal in January, which set the future course for the inheritance tax and inflation-adjusted the amount of inheritance excluded from taxation using, guess what index, a traditional “unchained” CPI measure. Had the legislation used the chained index, then more inheritance would be taxed each year. So, why did the conservative Democratic and Republican Senators (and the administration) not apply a chained CPI index to the inheritance tax exclusion if it is a more accurate measure of inflation? Hmmmm.

I hate to say it, but some of my liberal friends, including those in the Obama administration, have a contradictory/illogical position. Kathy Ruffing, Paul Van de Water and Robert Greenstein argue in a recent CBPP paper that a chained index is more accurate, even if it does not use weighting specific to the elderly, but that the switch should only be done “if it is accompanied by several necessary adjustments to prevent significant hardship.” However, as elaborated below, if the index is accurate then there is no hardship.

In fact, a recent paper from Moment of Truth quotes a 2004 paper written by Bob Greenstein that makes this same point, saying: “this change [chaining the CPI] should not be regarded as a benefit cut or a tax increase. It should be regarded more as a technical change to achieve Congress’ stated goal of keeping pace with inflation in as accurate a way as possible.”

The CBPP paper also argues that the change in indexing should not be assumed to apply to all programs that rely on the current CPI for indexation or eligibility thresholds. Instead, it should just be applied to a handful of specified programs that will see large spending reductions because of the shift. Again, this just does not look like a “technical adjustment.”

Further, if this was all just a question of technical economics, a failure to adopt chaining should be seen as allowing people to claim a windfall—receiving more in benefit growth than is required to keep up with inflation. It is not logical to argue that a chained CPI is more accurate and simultaneously argue that we need to protect the vulnerable from its consequences. The administration’s wish to protect poor beneficiaries, veterans and the oldest old suffers from the same contradictions. Jared Bernstein also gets caught up in this contradiction, saying, “It’s a more accurate price index, but it would constitute a benefit cut.”

A more accurate index would be a benefit cut only in the most literal sense that benefits would be less than currently written into law, but such an adjustment would adequately and totally insure benefits against erosion by inflation—the purpose of the COLA. However, a chained CPI would be a benefit cut in both the literal and commonly understood sense of the term if you believe, as I do, that the failure to use the detailed consumption weights of Social Security beneficiaries leads a chained CPI to understate inflation for this population.

The economics could not be clearer: in regards to the Social Security COLA, there is no merit at all to the claim that adopting a chained CPI measured for the average consumer is a technical improvement. In addition, there is no liberal safe harbor where one can argue that a chained CPI is technically accurate but vulnerable populations should be shielded from the consequences.