This week the Social Security Administration announced a 2.8% Cost of Living Adjustment (COLA) will be added to benefit checks beginning in January 2019, the biggest increase since 2012. Even then, progressive groups complain that COLAs aren’t high enough to keep up with seniors’ cost of living. In fact, the COLA overstates increases in retirees’ costs of living, based on conceptual errors in how the Consumer Price Index treats the homes retirees own. Rising housing prices are interpreted as making retirees poorer, such that retirees need a larger COLA. In fact, because the vast majority of retirees own their homes, rising home prices have no direct impact on their cost of living and in fact make retirees richer. If the CPI properly treated retiree-owned housing, this year’s 2.8% COLA would likely have been about 0.3% smaller. A COLA that accurately accounted for retiree housing costs could reduce the long-term Social Security deficit by 20%.
Social Security began granting regular COLAs in the early 1970s, to maintain the purchasing power of Social Security benefits in the face of rising prices. Prior to 1972, Congress passed ad hoc benefit increases when necessary. COLAs are calculated using the Consumer Price Index for Urban Wage Earners and Clerical Workers, called the CPI-W. The Bureau of Labor Statistics tracks changes in the CPI-W from the third quarter of one year to the third quarter of the next, and benefits are adjusted beginning in the following January.
The BLS constructs the CPI-W by first tracking changes in the prices of everything from furniture and bedding to sugar and artificial sweeteners. The BLS then weights these price changes based upon how much households spend on particular items, using data from the Consumer Expenditure Survey. The change in the weighted average of these prices is the increase in the CPI, on which the COLA is based.
Except for homes, where the process is different. If you own a home you’re not spending directly on the home itself. Instead, the BLS calculates how much that home would rent for, and then adds that “implicit rent” to the other spending captured by the CPI. If home prices rise, then implicit rent rises and retirees are assumed to be “spending” more on housing.
Except they’re not. Only 17% of Americans age 65 and over rent their homes, with those costs making up less than 5% of total household outlays. The remaining 82% own their homes, and so they’re not paying that implicit rent. Rising home prices make retiree homeowners richer, but the CPI acts as if retirees are made poorer.
These effects aren’t trivial, because housing is such a big part of the CPI. In the CPI-W, which is used to calculate Social Security COLAs, “implicit rent” on owner-occupied housing makes up 20% of household “expenditures,” even though these are not truly expenditures at all. For those who favor the CPI-E, which focuses on households aged 62 and over, the problem is even bigger: owned housing makes up 31% of “expenditures” in the CPI-E.
None of this would matter if home prices changed along with the prices of other items in the CPI. But they haven’t. Over the past two decades, the overall CPI-W grew at a rate of about 2.2% per year. Over that same 1998-2018 period, the implicit rental price of owner-occupied housing grew by 2.6% annually. Which means that prices for the things retirees actually buy grew by less than 2.2%. While it’s not easy to do this precisely, if you look at the CPI net of shelter costs it grew by an annual rate of 1.9%, making for an annual difference of 0.3%.
The median retiree household has a home valued at about $140,000, according to the Federal Reserve’s Survey of Consumer Finances. If home prices increased by 2.6% in a given year, that implies an increase in household net worth of $3,640, an amount that far exceeds the average annual COLA of about $450 that will be paid beginning in January of 2019.
In theory, a retiree household could use a home equity loan to borrow against its more-valuable home to offset price increases for other things they buy. If they did this, their standard of living could increase even before the COLA is paid because their net worth is rising more in dollar terms than the cost of living. In practice, most retirees don’t extract home equity to pay day-to-day costs. But even if they don’t, the fact that non-housing prices have risen more slowly than housing prices means that the rate of inflation on the things retirees actually buy is lower than a CPI which is heavily weighted toward housing.
Retiree groups complain that the CPI-W used to calculate Social Security COLAs doesn’t account for the larger amounts that retirees spend on health care, because the CPI-W is calibrated to the spending habits of the working-age population. The CPI-E, which is calibrated to spending by households 62 and over, shows inflation about 0.2 percentage points higher than the CPI-W. And yet, the roughly 0.3 percentage point upward bias introduced by the CPI-W’s treatment of owner-occupied housing is more than enough to offset this. In other words, it’s not at all clear that an “accurate” CPI for retirees’ cost of living would produce higher COLAs rather than lower ones.
The CPI’s treatment of owner-occupied housing also is an entirely different issue from that of the so-called “chained” CPI. The chain-weighted CPI seeks to capture how household purchases change in response to changing prices: if, say, chicken rises in price, households may purchase more beef. The BLS appears to believe that the chain-weighted CPI is a more accurate measure of the cost of living than the non-chain weighted CPI-W used to calculate Social Security COLAs. But the housing-related issues raised here are in addition to the questions the chained CPI seeks to resolve.
A three-tenths of a percentage point average increase in the CPI-W due to owned housing wouldn’t make a huge difference in a single year. But over a 20-year retirement it can lead to benefits about 8% higher than what retirees would receive if we removed the distortion caused by owner-occupied housing. Put another way, reducing COLAs by three-tenths of a percent would cut Social Security’s long-term funding shortfall by about 20%. That’s real money.
Some of this may be counterintuitive and hard to grasp. But just imagine if the Social Security COLA were set up to compensate for increases in the prices of stocks and bonds. The obvious response would be, “But retirees don’t buy stocks and bonds; if anything, they sell them.” The COLA’s treatment of housing isn’t quite as extreme, because it’s harder to pull cash from a house than it is to sell a stock. But, at the least, an increase in the price of retiree-owned housing shouldn’t be treated as a cost for retirees.
The International Monetary Fund warns that “The treatment of owner-occupied housing in consumer price indices (CPIs) is arguably one of the most difficult issues faced by CPI compilers. Depending on the proportion of the reference population that are owner-occupiers, the alternative conceptual treatments can have a significant impact on the CPI…” Given the needs of the retiree population, and the effects of COLA payments on Social Security’s finances, policymakers should pay close and objective attention to how inflation for retired Americans is measured.
Ideally, Social Security COLAs should be based on an index measuring changes in the prices of things that retirees actually spend their money on. Measuring changes in the prices of things they bought prior to retirement doesn’t make much sense, and can indeed produce results showing retirees growing poorer when they’re in fact growing richer.