Monday, May 2, 2011

New NASI paper: “Should Social Security’s Cost-of-Living Adjustment Be Changed?”

The National Academy of Social Insurance has issued a new Fact Sheet titled "Should Social Security's Cost-of-Living Adjustment Be Changed?" by Benjamin W. Veghte, Virginia P. Reno, Thomas N. Bethell and Elisa A. Walker which looks at potential policies to reduce or increase annual Cost of Living Adjustments (COLAs) for Social Security benefits.

Overall it's a pretty good review of the arguments given by both sides. Most economists agree that conventional CPI measures – like the CPI-W used to calculate Social Security COLAs – tend to overestimate true increases in the cost of living because they don't account for how individuals can shift between different items to account for changing prices (e.g., if apples increase in price we may buy more oranges; if chicken rises in price we may buy more beef, etc.). A version of the CPI that's designed to account for these changes – called the "Chain Weighted" CPI, or C-CPI – generally shows inflation around 0.3 percentage points lower than the CPI-W.

On the other hand, neither the CPI-W nor the chained CPI account for the particular purchasing patterns of Social Security beneficiaries, which can differ from those of the working age individuals who are the base for the CPI. An experimental CPI for the elderly, called the CPI-E (E stands for "experimental," not elderly) generally shows inflation around 0.2 percentage points higher than the CPI-W. The main reason is that seniors spend more on health care and health costs have risen faster than other categories. However, the CPI-E shares the same overstatement of inflation that experts see in the CPI-W.

So what to do?? The obvious answer – to me, at least – is to create a chain-weighted version of the CPI-E – that is, a C-CPI-E – that accurately accounts for the purchasing habits of seniors. As it happens, in a forthcoming Social Security reform proposal I'll be proposing exactly that. This new C-CPI-E would account for both the upward and downward biases of the CPI-W and would produce annual COLAs around 0.1 percentage point lower than those generated under current practices. This wouldn't make a huge dent in Social Security's deficit, but that's not the point. The point is to get COLAs right and this dual approach seems to get closer than either the current method or the two proposed alternatives.

It's unfortunate that NASI's Fact Sheet seems to come down on the side of the CPI-E while more or less ignoring the overstatement of inflation that most experts see in that measure.


WilliamLarsen said...

Changing the CPI calculation first requires fixing or at least acknowledging Social Security's problem.

When Social Security began, there was no COLA, but congress increased benefits many times before COLA legislation passed. In fact the increases out stripped inflation by nearly 50% over that time period.

When determining what the tax rate should be to pay for such a change, adding COLA, no one did this. They simply legislated the change with no cost analysis.

It is true that seniors spending habits are different. They buy less new furniture, cars, homes, fewer groceries, have no kids to support. However, they spend more on medicines. To me if you plan on retiring it is not up to Social Security to determine your increase or support the life style you want. It is up to you to look at your financial situation long before you retire and plan accordingly.

The problem is neither congress or most seniors have ever done this. To me we simply repeals the social security act and we no longer have to worry about how to save something that workers would rather opt out of or worry about COLA calculations.

bveghte said...

Thanks for your careful reading of and thoughtful comments on our Fact Sheet. We agree that the goal of any COLA reform should be to more accurately keep pace with the cost of living for Social Security beneficiaries.

A good case can be made for a chained CPI-E, as you propose. Indeed, the choice of CPI on which to base the Social Security COLA is independent of the choice to chain or not chain the chosen index. As you point out, the CPI-E clearly best accounts for the purchasing patterns of seniors, but it needs to be improved to be statistically robust. Basing the COLA on a chained version of the CPI-E would require funding for BLS to increase the Consumer Expenditure Survey’s sample size.

Our Fact Sheet evaluates current policy proposals, but does not make any policy recommendations. At the moment a chained CPI-E is not on the table. But it is worth considering and we look forward to reading your proposal.

With regard to the purported overstatement of inflation that you note that most experts see in the CPI-E, while this is indeed the consensus among economic theoreticians, empirically this is an open question. The claim that non-chained indexes overstate inflation for Social Security beneficiaries is based on the assumption that when seniors and the disabled change their purchasing patterns in response to inflation, they indeed maintain their standard of living. There is no empirical research to support this claim, and hence no expertise on this point. The Consumer Expenditure Survey does not currently ask anything about the change in the subjective standard of living that results from such “higher-level” substitution behavior, i.e. when the elderly or disabled buy less home heating oil and more sweaters or fewer apples and more oranges. The legislative intent of the COLA is vague on this point: it is simply to ensure that the purchasing power of Social Security benefits is not eroded by inflation. It is thus an open question whether the purpose of a COLA is to enable seniors to maintain their standard of living or whether they are expected to switch annually to less preferred, less expensive market baskets in response to inflation, and receive a COLA sufficient only to purchase this less preferred market basket.

WilliamLarsen said...

COLA is to maintain the buying power of the basket of goods in the year a person turns 60 (the year wages are used to determine the SS-OASI benefit). That means you may maintain the standard of living in the year you retired and buy those goods and services available in that year.

As new services and products come on the market, COLA does not account for this and is not intended to MAINTAIN A STANDARD OF LIVING as the STANDARD OF LIVING for the economy improves.

COLA maintains buying power. What it appears to happening is that now people equate COLA with Standard of living.

To maintain the standard of living of the economy requires a much higher payroll tax than any person paid prior to 1970.