Is the chained CPI regressive?

12.19.12 Biggs Graph SS Chained CPI

Recently, I responded to claims from Paul Krugman that raising the Social Security retirement age was “hugely regressive” because higher income people live longer. Krugman’s claim was incorrect, in part because low earners are more likely to claim Disability Insurance benefits, which wouldn’t be affected by raising the retirement age. But based on Krugman’s logic, the latest Social Security reform proposal, to base Cost of Living Adjustments (COLAs) on the lower “chain weighted” version of the Consumer Price Index, should be progressive: Rich people live longer, and the longer you live the more a lower COLA affects you. But a “diet COLA” turns out not to be progressive, and for the same reason: Disability.

Using the Policy Simulation Group’s Social Security models, I projected lifetime benefits under current law for members of the 1990 birth cohort. I then projected benefits under the assumption that COLAs would be 0.3 percentage points lower, approximately the difference between the chained CPI and the CPI-W that’s currently used to calculate COLAs.

The blue bars in the graph above show the reduction in all Social Security benefits by lifetime earnings quintile (the 1st quintile represents the bottom 20% of the earnings distribution, and so on). From this point of view, a COLA cut is regressive: The bottom earnings quintile sees its lifetime benefits cut by 4.4% while the top quintile loses only 3.7%. But why? I suspected the main reason was disability: While poor retirees don’t survive as long as rich ones, many truly low-income individuals will spend a very long time collecting DI benefits, since they may claim while young but—due to the low mortality of disabling ailments such as depression and back pain—collect into old age. The longer you claim benefits, the bigger the impact of lower COLAs.

So I tried simulating a diet COLA that would apply only to retirement benefits. This wouldn’t exempt the disabled entirely, since under current law DI beneficiaries switch to the retirement program at age 66. But they’d receive the standard COLA based on the CPI-W at least through the full retirement age. This produces the results you’d expect: The lowest quintile of earners see only a 2.7% lifetime benefit cut while the top fifth of earners see their benefits cut by 3.4%.

So a COLA fix can be progressive and it can protect the disabled. But that comes at a cost: The all-inclusive COLA fix would cut Social Security’s 75-year deficit by around 19%, while exempting disability benefits would reduce the shortfall by only around 16%.

6 thoughts on “Is the chained CPI regressive?

  1. re: ” he all-inclusive COLA fix would cut Social Security’s 75-year deficit by around 19%, while exempting disability benefits would reduce the shortfall by only around 16%”

    How many other budget items have a budget horizon of 75 years? How about DOD health care and pensions? Are they look at 75 years ahead?

  2. I believe changing COLA to chained CPI might be regressive, and for a reason I have not yet seen raised.

    Consider purchasing power. The very concept of chaining assumes that consumers will substitute cheaper goods for the goods they purchase currently.

    But a person receiving a low Social Security benefit is likely to be lower on the substitution chain than is a person receiving a higher benefit. Persons near the bottom of the benefit structure may well already be at the bottom of the substitution chain – with NO further substitution option.

    The lower a recipient is on the substitution chain, the lower their real purchasing power – the greater the real cut in benefits – will be under chained CPI.

    Conversion to chained CPI will effect a pure cut on the benefits and purchasing power of those recipients at the bottom of the substitution chain

    • Terry, Chaining also means that people substitute more expensive goods for cheaper ones, since it’s about relative prices. Yes, if beef prices rise then people will buy more chicken; but if chicken prices rise, then people will buy more beef. On important question, I’d think, is how prices tend to change for staples (like chicken) and their more expensive substitutes (like beef). My guess — and it’s only a guess — is that the more ‘luxury’ goods tend to fall in price, at least in relative terms, but I don’t know that for sure.

      • According to Mortgage News Daily, just about half of all low-income renters spend at least half their income for shelter.

        This is (obviously) the largest item in these people’s spending, and they are also likely to lack cheaper substitution options for housing.

        For example, I am over 50 (thus with a short horizon for retirement age), earn minimum wage, and rent a room. Exactly what substitution am I expected to make for housing? I don’t have any cheaper available substitutes and will lose purchasing power under chained CPI.

        • Terry, the chained CPI should show lower inflation in cases where people can (and do) substitute between goods. If you’re right that it’s hard to substitute within housing, then the chained CPI shouldn’t show a large change there.

          • People with middle class incomes CAN substitute within housing, those financially constrained to the point of renting rooms cannot substitute within housing.

            Also, are homeowners relatively advantaged and renters relatively shortchanged under cost of living adjustments tied to CPI (chained or not)?

            Homeowners with fixed-rate mortgages have locked in their P&I payment and thus are immune from cost increases in this component when housing prices rise.

            Renters, on the other hand, must always pay current rents, which at most times are rising, and which often rise much faster than CPI.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>