Cruel Statistics:
The Impact of Changes in the
Consumer Price Index on Older Women
by Dean Baker and Mark Weisbrot
May 12, 1999

In the last four years the Bureau of Labor Statistics (BLS) has made a series of changes in the methodology used to construct the consumer price index (CPI). The net effect of these changes has been to substantially reduce the CPI’s measure of inflation relative to the actual rate of inflation. According to the estimates of the Council of Economic Advisors (CEA), the cumulative impact of these changes has been to lower the measured rate of inflation by 0.7 percentage points annually. These changes will have a direct effect on many people, most importantly Social Security beneficiaries, because the annual cost of living adjustments are tied directly to the CPI. In particular, older women are likely to be affected by the changes in the CPI, because they disproportionately depend on Social Security for their income. Also, since women have longer life expectancies, and the effect is cumulative over time, they are likely to see the largest reduction in benefits. Since many older women are already in or near poverty, the lower benefits that will result from the changes in the CPI will hit this group quite hard. It is important that Congress adopt measures that will counteract this effect.

Social Security and the CPI

In 1973 Congress voted to tie the Social Security benefits received by retired workers to the CPI. Each year, the benefits received by workers who are already retired rise by exactly the rate of inflation shown by the CPI-W.[1] This link only applies to workers after they are retired. The initial level of benefits at the time of retirement is indexed to the average wage in the economy. In years of high inflation, such as the late seventies, this annual cost of living adjustment has played a very important roll in maintaining the standard of living of retirees.

The Bureau of Labor Statistics (BLS), which produces the CPI, regularly modifies the index when research demonstrates ways in which the index can be improved. In the last four years, BLS has put in place a series of changes to the CPI, most of which have had the effect of lowering the CPI relative to the true rate of inflation. According to the estimates of the Council of Economic Advisors, the cumulative impact of these changes has been to lower the annual rate of inflation reported by the CPI by 0.7 percentage points (Economic Report of the President, 1999, p. 94). This means that if the old CPI would have reported a 3.0 percent rate of inflation, the CPI presently in place would report a 2.3 percent rate of inflation.[2]

This change in the CPI has the effect of lowering the size of the annual cost of living adjustment by exactly the same amount, 0.7 percentage points. In the case of a retiree receiving a $500 per month Social Security benefit, the change in the CPI in this case will mean that instead of getting a $15 per month cost of living benefit, raising their monthly check to $515, they will get an $11.50 increase, which will place their benefit at $511.50. This one year change may not seem of much consequence, but the impact will increase through time. After ten years, if the CPI had shown a 3.0 percent annual inflation rate, the monthly benefit would have risen to $672. However, with the new CPI, the benefit would only be $628, a reduction of 6.5 percent. After twenty years, the benefit reduction resulting from the new CPI would be $115, a cut of 12.7 percent.

For better off retirees, these cuts may not be of much consequence. They have other sources of income, and getting a somewhat smaller Social Security check will not affect them much. But poorer retirees are almost completely dependent on their Social Security checks. Elderly poor families rely on Social Security for 64 percent of their income. Poor elderly people living alone rely on Social Security for 82 percent of their income. For these people, a 12.7 percent reduction in the size of their Social Security benefit is almost the exact same thing as a 12.7 percent reduction in their income.

Since people already get poorer as they get older, this reduction in the size of annual cost of living adjustment will amplify this process. In particular, women retirees, who get lower benefits than men because they were paid less in their working lives, and on average live longer than men, will be hit particularly hard by the changes in the CPI. While only 11 percent of the people over 65 live below the poverty line, 20 percent of older woman living alone are below the poverty line.[3] Another 13 percent of these women are classified as near poor, with incomes that place them at less than 125 percent of the official poverty line. The oldest women are also the poorest. The median income for women age 65-69 was $11,630 in 1996. This drops to $9,417 for women over age 85.

The lower benefits that will result from the changes in the CPI will hit this vulnerable segment of the population quite hard. If the new CPI had been in place five years ago, the income for a typical woman age 65-69 would be approximately $370 lower at present. If the new CPI had been in place twenty years ago, the income for a typical woman over age 85 would be approximately $1100 lower today. This a considerable cut for a group with a median income of $9,417. It would leave most of this group either below the poverty line of $7,525, or just slightly above it.

It is important to recognize that the impact of the CPI changes is independent of whether or not the new measure is more accurate. If we assume that the new CPI is correct, then the old measure overstated inflation by 0.7 percentage points each year. This means that the annual cost of living adjustment had been equal to the true rate of inflation, plus 0.7 percentage points. Even after receiving a cost of living adjustment that exceeds inflation, 20 percent of older women were still in poverty in 1996. From the standpoint of viewing the new CPI as correct, the nation has implemented a policy change. Instead of giving cost of living adjustments that exceed the true rate of inflation by 0.7 percentage points (a CPI plus 0.7 percentage point formula), it will only increase benefits each year by the actual rate of inflation. If 20 percent of older women living alone were poor even after receiving their CPI plus 0.7 cost of living adjustment, then the number of poor elderly women can only increase further with a COLA that is just equal to the CPI.

It is also worth noting that there is evidence that elderly actually experience a higher rate of inflation than the rest of the population. BLS has calculated an experimental index based on the consumption patterns of the elderly since 1984. This index has shown an average a rate of inflation that is 0.4 percent higher than the CPI-W used to adjust Social Security. There are also some types of costs borne by the elderly, most importantly higher co-payments in Medicare plans, that are not even counted in the CPI.[4] As health care expenditures rise in coming years, and the portion borne by the elderly increase, as is widely expected, many retirees will be seriously squeezed.

Ways to Fix the Problem

There is no single solution to the problem of poverty among older women, but there are many steps that go in the right direction. This paper notes two:

1) raising the benefits received by a surviving spouse, and

2) making the basic benefit formula more progressive.

Both of these proposals could substantially increase the Social Security benefits received by older women without adding significant costs to the program.

The first proposal was proposed by several members of the 1994-6 President’s Advisory Council on Social Security. At present, a surviving spouse (usually a woman) is allowed to keep either his or her own or his or her spouse’s Social Security benefit, whichever is larger. When both of them were living, the couple would have received both checks. If the couple had roughly equal benefits, then this means that the surviving spouse will have to get by on half of their previous level of benefits. The proposal suggested by members of the Advisory Council would have raised this to two-thirds of the combined benefit. While this could help many elderly women, it carries a relatively small price tag. According to the Advisory Council’s estimate, the additional expense would be 0.39 percentage points of payroll. Assuming that the changes in the CPI have had the effect estimated by the CEA, this change would take back slightly more than one-third the resulting savings to the program.

The second proposal would make Social Security’s benefit structure even more progressive. At present, workers receive a benefit equal to 90 percent of their average lifetime wages up to $6060. They receive 32 percent of their average wage for amounts in excess of $6060, but less than $36,516. For amounts over $36,516 they receive 15 percent, up to the current limit. It would be a relatively simple matter to boost the bottom payback rate to 100 percent. This could be offset with a corresponding reduction in the second payback rate to 15 percent, up until $9,625. At this point, a worker would be exactly as well off under the new formula as the old formula. For incomes above this point, the old formula could be kept in place, leaving these workers unaffected.[5] The low wage earners who would be helped by this measure are mostly women. This is due to the fact that women earn less on average in the years in which they do work, and also that women are far more likely to have years of zero earnings in which they leave the paid labor force to take care of children or parents.

The cost of this measure would be minimal, approximately 0.10 percent of payroll. It should also be relatively easy to implement administratively, since it involves just a one time change in the benefit formula.[6] At the same time, it would make a real difference in the lives of the poorest among the elderly, largely offsetting the impact of the recent changes in the CPI.

Conclusion

The Bureau of Labor Statistics has made a series of changes to the CPI over the last four years which have the effect of lowering its measurement of the inflation rate. Since Social Security benefits for retirees are indexed to the CPI, this means that Social Security benefits will be rising less rapidly in the future than in the past. Older women, who often already live close to or below the poverty line, will be particularly hard hit by this reduction in Social Security benefits. Unless policies are implemented, like the two suggested here, many more elderly women will live out their lives in poverty in the future.

Dean Baker is a Senior Research Fellow at the Century Foundation in New York and the Preamble Center in Washington, D.C. Mark Weisbrot is Research Director of the Preamble Center.

References

Baker, D., 1997. Getting Prices Right: The Battle Over the Consumer Price Index. Armonk, NY: M.E. Sharpe.

Economic Report of The President, 1999. Washington, D.C.: U.S. Government Printing Office.

Social Security Administration, 1996. Income of the Aged Chartbook, 1996. Washington, D.C.: U.S. Government Printing Office.

[1] The CPI-W is an index that was explicitly designed to track the rate of inflation experienced by wage and clerical workers. It reflects the purchasing patterns of these groups. The purchasing patterns of retirees and salary earners are not reflected in the market basket tracked by the CPI-W. On average, the CPI-W has shown a rate of inflation that is 0.1 percentage point below the level in the more commonly reported CPI-U.

[2] There was considerable political controversy surrounding the CPI a few years ago, when a commission established by the Senate Finance Committee issued a report concluding that the CPI substantially overstates the true rate of inflation, and recommending that Congress establish an ad hoc committee to determine the rate of inflation that would be used to adjust Social Security and other government benefits, as well as income tax brackets. At the time, BLS resisted considerable pressure to make ad hoc changes in the CPI that would have lowered its measured rate of inflation. While there are issues that could be raised by some of the changes that have since been instituted, all are supported by solid research. If there has been a downward bias introduced by these changes it is the result of the fact that there has been more pressure to eliminate sources of upward bias in the index than downward bias. If one assumes that the CPI will always be an imperfect index, with sources of bias in both directions, then an effort to eliminate exclusively sources of upward bias will lead to an index that understates inflation. For possible sources of downward bias see Baker 1997.

[3] This data can be found in Social Security Administration 1996.

[4] Higher Medicare co-payments would be treated as a reduced subsidy, not an increase in prices.

[5] The change in the formula should not create any complications for beneficiaries. The benefit would be calculated by the Social Security Administration, as it is presently. The beneficiary would never have to know the specifics of the formula unless she wanted to independently calculate her benefits.

[6] It could be argued that it would be appropriate to phase in this increase among current retirees, since this group will be getting a lower COLA each year than that which they had become accustomed to.