By Alicia H. Munnell
Alicia Munnell, the director of the Center for Retirement Research at Boston College, is a weekly contributor to “Encore.”
Two prominent commissions recently proposed introducing a “more accurate” consumer price index (CPI) to adjust Social Security benefits each year. This “more accurate” measure is projected by Social Security’s Chief Actuary to rise about 0.3 percentage points more slowly than the current index and thus would likely result in lower cost-of-living adjustments (COLAs) for seniors. But the current CPI already does not reflect the spending patterns of the elderly, and an experimental price index that does reflect these patterns is projected to rise 0.2 percentage points more rapidly than the current index. So, a fair discussion of the COLA must recognize these offsetting biases. It also should question whether low-income elderly really have the flexibility to change their purchases significantly – which the proposed new CPI assumes — in response to price changes.
Some background might help. When the Social Security COLA was first introduced in 1972, the Bureau of Labor Statistics (BLS) had only one price index; it was the CPI-W for urban wage earners and clerical workers, which covers about 32% of the population. In 1978, the BLS expanded the sample to all urban residents and created the CPI-U, which covers about 87% of the population including most retirees. In 1988, the BLS added the experimental index – the CPI-E, which reflects the spending patterns of persons 62 and over. As shown in Table 1, the CPI-E has increased about 0.27 percentage points faster each year than the CPI-W.
Table 1: Average Annual Percentage Change in The Experimental Consumer Price Index for Elderly Americans (CPI-E) and in the CPI for Urban Wage Earners and Clerical Workers (CPI-W), Dec 1982- Dec 2010
|CPI – E||CPI – W|
|Average annual change||3.12%||2.85%|
Economists argued that these indices overstated inflation, because they did not account for how people change their buying habits in response to a price increase. The theory is that by shifting to a close substitute product or service, people can lessen the increase in their cost of living and be just as happy. In 1999, the BLS incorporated consumer substitution among similar items in all three indices, such as buying fewer Granny Smith apples when the price rises relative to Golden Delicious. This change reduced the rate of growth in all the price indices.
What remains is substitution among dissimilar categories, such as food and home heating. The “chained” version of the CPI-U, which was created in 1999, allows for this type of substitution. While the standard CPI-U and CPI-W adjust the weights used for the 211 broad groups of goods and services in the market basket every two years, the chained CPI is designed to reflect monthly changes. As shown in Table 2, the chained CPI-U has increased about 0.33 percentage points slower each year than the CPI-W.
Table 2: Average Annual Percentage Change in Chained CPI-All Urban Consumers (chained CPI-U) and CPI for All Urban Consumers (CPI-U), Dec 1999- Dec 2010.
|Chained CPI – U||CPI – U|
|Average annual change||2.10%||2.43%|
If we were starting with a price index for Social Security that properly reflected the spending patterns of the elderly, then moving to a chain-weighted index might improve accuracy. The problem is that the current index understates the price increases experienced by the elderly, since, for example, it does not reflect the fact that older people spend much more on health care where prices are rising rapidly. So moving to a chain-weighted index without correcting for spending patterns is a reduction in benefits.
Moreover, some experts question whether low-income elderly really have the flexibility to change what they buy in response to price changes and still maintain their well-being – the key assumption behind the chained CPI-U. Their consumption may be too near subsistence to spend less on food and more on home heating when the price of food rises.
The COLA is probably fair game in restoring balance to Social Security. It is the only way to have current retirees contribute to the effort. The Greenspan Commission in 1983 recommended delaying COLA payments by six months. Perhaps some such change could be included in a package of proposals.
But using a diet COLA to adjust benefits for all time is not a good idea. It is a relatively small change for young retirees, but results in a substantial benefit cut for the old. A COLA that is 0.3 percentage points lower would produce a monthly benefit that is about 6.5% lower by the time a retiree reaches 85. To compensate, the commission proposals include a one-time 5% benefit increase around age 85. This adjustment helps at that age, but then the cut continues. Even with the adjustment, the COLA change eliminates one fifth of the Social Security 75-year shortfall; it’s a benefit cut.
So let’s be upfront. If the COLA is meant to reflect the best estimate of the increase in the cost of living for the elderly, switching to the chained CPI-U is not the answer. (The chained CPI is also difficult to use because the final value would not be available until two years later.) If any adjustment is made to the nature of the COLA – as a opposed to, say, a one-time delay – it should take into account both the projected 0.3% overstatement due to not accounting for the substitution effect and the projected 0.2% understatement due to not reflecting the spending patterns of the elderly. And then we should worry whether people who buy only necessities can really change their spending on food and fuel when the price of fuel rises without loss in well being.
NOTE: Source of tables is the U.S. Department of Labor, Bureau of Labor Statistics
(To see how an inflation/COLA change would affect your payouts, change the “Inflation” assumption in “Advanced Settings” in our new retirement calculator.)