A Project of The Annenberg Public Policy Center

Chained Explained

A more accurate inflation adjustment could cut $300 billion from the deficit.


Using a more accurate cost-of-living adjustment for federal benefit payments and tax brackets would cut the federal deficit by perhaps $300 billion over the next 10 years. But it faces opposition from both right and left.

Economists generally agree that the “chained” Consumer Price Index is a more accurate way to measure the cost of living than the traditional CPI. Republican leaders and some Democrats favor it as part of the solution to the “fiscal cliff” dilemma, as does at least one leading liberal group.

Nevertheless, the AARP objects to using it to make cost-of-living adjustments for Social Security, because it would result in slightly lower yearly increases, and Grover Norquist’s Americans for Tax Reform objects to the slightly higher federal income tax liabilities that would result from using it to adjust tax brackets for inflation each year.

Leading economists have pointed out for at least half a century that the traditional CPI goes up faster than the average person’s actual cost of living. The flaw in the traditional CPI is that it simply measures prices of goods and services in a “market basket” that doesn’t necessarily reflect what people are buying in any given month. When the price of beef spikes, consumers often switch to pork or chicken, but the traditional CPI assumes they keep buying and paying for the higher-priced meat.

The “chained” CPI adjusts for such switches. Thus, it more accurately measures what the average consumer actually spends for goods and services. It also rises somewhat more slowly than the traditional CPI.

Using the “chained” CPI to adjust federal benefit payments and tax brackets would automatically trim the growth of future spending — and increase future revenues — without making any other changes in entitlements or tax laws. According to the Congressional Budget Office:

  • The “chained” CPI would rise by about 0.25 percent less each year, producing slightly smaller annual cost-of-living increases for Social Security. The $21 average increase that Social Security retirees will start getting in January, for example, would be $2.40 less if the chained CPI had been used to calculate it.
  • Over time, such small changes would compound. After 10 years of reduced cost-of-living increases, a Social Security beneficiary would be getting about 3 percent less per month than under the current system, for example. And by age 95, a Social Security pensioner retiring today would be getting about 8 percent less.
  • The same would also be true for annual cost-of-living adjustments for retired federal workers, retired military and veterans’ benefits.
  • On the tax side, using “chained” CPI to adjust federal income-tax brackets for inflation would bring in an additional $72 billion in revenue without any increase in rates, CBO estimates. That’s because bracket thresholds would rise more slowly, leaving more income in brackets taxed at higher rates.

CBO estimated that using the “chained” CPI for all federal programs (including some it didn’t specify) would save $145 billion in spending over 10 years, and bring in $72 billion in added revenue, for a total of $217 billion in deficit reduction. Another, independent estimate, adding in savings from interest payments for money that would not have to be borrowed, brings the total 10-year deficit reduction to nearly $300 billion.

Many who oppose using the chained CPI assert that seniors experience a more rapid rise in their cost of living than others. But over the years, economists have found no clear evidence of that.

Seniors do spend more on medical costs than younger persons, and medical costs rise more rapidly than other prices. But seniors also spend less on such things as insurance, college tuition and gasoline.

It’s true that the Bureau of Labor Statistics calculates an unpublished index (the CPI-E, for “elderly”) that attempts to measure consumer prices paid by those age 62 and over. But the results are inconclusive.

The experimental index rose slightly faster than the traditional CPI for a number of years. But since 2006, the “elderly” index has gone up a little more slowly than the index used to adjust Social Security benefits.

The BLS concedes that the experimental index suffers from multiple shortcomings. It doesn’t measure prices where seniors actually shop, or the goods and services they actually buy, for example. A 2002 report for the National Research Council also noted those deficiencies, and found the CPI-E unsuitable for indexing benefits. The panel said a valid measure of spending by seniors should not be based on “speculation and conjecture.”


The “Chained Consumer Price Index for All Urban Consumers,” or C-CPI-U, attempts to correct a problem that leading economists have been pointing out for half a century: The traditional CPI, they say, goes up faster than the prices ordinary people actually pay.

‘Substitution Bias’ in the CPI

At issue here is something called “substitution bias.” Put simply, the traditional CPI tallies up the current prices of items in a “market basket” of goods and services supposedly bought by the average consumer in any given month. But what goes into the market baskets of real-life consumers today isn’t necessarily what went in when the Bureau of Labor Statistics surveyed consumers to find out what they were buying. If the price of beef goes up and shoppers buy more chicken, their cost of living rises by less than the CPI reflects.

BLS used to conduct expenditure surveys only once per decade. Now it updates its market-basket weights once every two years, but still with a considerable time lag. The most recent update was announced in January 2012, when BLS started using weights reflecting what consumers were buying in the 2009-2010 period, updated from weights determined in 2007-2008, which it was using as late as December 2011.

The “substitution bias” was noted more than 50 years ago, in a 1961 report to the Bureau of the Budget, by a panel of economists headed by Nobel Prize-winner-to-be George J. Stigler. Among the problems causing what the Stigler report said was a “systematic upward bias” in the CPI, was this:

Stigler Report, 1961: Since consumers will substitute those goods whose prices rise less or fall more for those whose prices rise more or fall less—and within limits they can do this without reducing their levels of real consumption—the fixed-weight base CPI overestimates rises in the cost of equivalent market baskets.

Slow Change

The BLS has made many improvements in the accuracy of the CPI in response to the Stigler report, though not always quickly. It took more than 20 years for the BLS to change the way it calculated the cost of owner-occupied housing along lines suggested by Stigler, for example.

The CPI then based home ownership costs mainly on prices of homes sold, and interest costs for new mortgages taken out in any given month, even though the vast majority of homeowners don’t buy a home or take out a new mortgage in that month. Stigler recommended using instead what a homeowner would pay to rent his or her house.

In 1979, when mortgage rates first spiked to over 11 percent, the CPI put overall inflation at 13 percent. But relatively few people actually took out mortgages at the sky-high rates. The CPI thus overstated inflation by 2.5 percentage points compared with what it would have been had BLS used instead the cost of renting equivalent quarters, according to a 1981 report to Congress by the U.S. General Accounting Office.

Finally, in January 1983, the CPI began using the rental equivalent approach Stigler had suggested more than two decades earlier. A BLS paper later concluded that the CPI was higher by 0.6 percentage points each year between 1967 and 1982 than it would have been had the bureau used the rental-equivalence method that it now uses.

Boskin Report

But flaws remained. In 1996 another panel of distinguished economists (including one who had been a member of the 1961 Stigler commission) estimated that the CPI was still overstating the true rise in the cost of living by between 0.8 percentage points and 1.6 percentage points per year, with a “best estimate” of 1.1 percentage points. The panel was appointed by the Senate Finance Committee, and headed by Michael J. Boskin, former chief economic adviser to President George H.W. Bush.

Among other things, the Boskin panel recommended that BLS take steps to eliminate the substitution bias in the CPI, and to better account for the goods consumers actually buy. And the BLS made some effort in that direction.

Starting in 1999, the BLS started using a “geometric mean” formula in the CPI to remove the effects of substitution bias within product categories (for example, when consumers substitute one brand of ice cream for another because of price changes). That adjustment alone was predicted to reduce the growth of the CPI by about 0.2 percentage points each year.

But the larger problem of substitution bias across different product categories (for example, when consumers switch from apples to oranges when apple prices rise, or orange prices decline) remained.

Further efforts to eliminate that bias were recommended in a 2002 report by a panel of experts assembled by the National Research Council, an arm of the congressionally chartered National Academy of Sciences. An editor of that report was Charles Schultze, who was chief economic adviser to President Jimmy Carter.

Also, starting in July 2002 the BLS began publishing the new “chained” CPI with data going back to 1999. This index, the C-CPI-U, uses a formula designed to account for changes in purchasing patterns that consumers make in response to changes in relative prices across categories. So when pork prices increase or beef prices decline, the C-CPI-U assumes that consumers buy less pork and more beef. These spending weights are changed monthly, “chained” to the spending weights used the previous month.

But by law, annual cost-of-living increases for Social Security and other federal benefits remain linked to a variant of the traditional CPI, the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W). And federal tax brackets are adjusted based on the somewhat broader Consumer Price Index for All Urban Consumers (CPI-U), which is the index most commonly cited in news accounts of government reports on inflation.

Simpson-Bowles Report

In December 2010, a majority of the National Commission on Fiscal Responsibility and Reform, headed by former Sen. Alan Simpson and former White House Chief of Staff Erskine Bowles, recommended switching to the C-CPI-U for all federal spending and tax provisions that use inflation adjustments as part of a massive deficit-reduction effort. The report was endorsed by 11 of the commission’s 18 members, including both Democrats and Republicans. But the recommendation was not sent to Congress because the vote fell short of the 14 needed to trigger such action under the executive order establishing the panel.

In March 2011, the nonpartisan Congressional Budget Office estimated that switching to the C-CPI-U for adjusting Social Security benefits would automatically cut federal spending by $112 billion below current 10-year projections, and that using the “chained” index for federal workers’ pensions, veterans benefits and military retirees would slow spending growth by another $24 billion. Using the CPI-U to index all federal spending programs, including a few CBO didn’t specify, would cut spending growth by “about $145 billion” in total.

Also, using the index to adjust income-tax brackets would bring in $72 billion in additional revenue, CBO said. Taken together, the spending reductions and revenue increases would total $217 billion over a decade.

Even more deficit reduction was predicted by the Moment of Truth project (also chaired by Simpson and Bowles) in a report issued in May 2011. That report estimated a total of $299 billion in deficit reduction over 10 years, counting reduced interest payments resulting from lower spending, a factor not addressed by the CBO report.

Update, Dec. 13: The Moment of Truth project announced Dec. 12 that it had updated its 2011 report. Using newer budget figures, it said that using the C-CPI-U would cut the deficit “by over $235 billion over the next decade alone if implemented for 2014, and over $290 billion if implemented immediately.”

Gradual Effects on Benefits

Had the C-CPI-U been used to calculate the cost-of-living increase that will show up in next month’s Social Security checks, the increase would have been 1.5 percent instead of 1.7 percent.

The average monthly benefit currently paid to a Social Security retiree this year is $1,240, and the 1.7 percent increase means that will go up by $21 starting next month.

The C-CPI-U rose only 1.5 percent in the period used to gauge the COLA amounts, however. That would have meant an increase of $18.60 — a difference of only $2.40 a month. (The government rounds the percentage increases to the nearest tenth of a point, and rounds the dollar amounts down to the nearest dime when applying the increases.)

Over time, those small differences accumulate, and compound. But the effects would be gradual. CBO said that after years of smaller cost-of-living increases, somebody retiring on Social Security today would be receiving about 3 percent less in benefits at age 75 than he or she would receive under current law. By age 95, the retiree would face a reduction of about 8 percent.

Support and Opposition

A Washington Post editorial has described the idea as “a relatively easy way to save about $300 billion.” In current negotiations with the White House, Republican leaders in Congress are pushing use of the “chained” CPI as a relatively painless way to cut some of the deficit. And the idea has even been endorsed conditionally by the liberal Center on Budget and Policy Priorities, which said earlier this year that using the C-CPI-U is “a reasonable component of a comprehensive package to put the budget on a sustainable course,” provided that it applies to the tax code as well as to benefit programs, and provided further that certain concessions are made to ease the impact on very old or very low-income beneficiaries.

But the proposed change also faces opposition from predictable quarters. A tax expert at the libertarian Cato Institute says using the C-CPI-U to index the tax code is a “stealth tax increase,” and Grover Norquist’s Americans for Tax Reform opposed it as a “hidden” tax hike.

And on the left, the AARP argues that using it to calculate Social Security cost-of-living increases “targets the oldest, poorest Americans” and is “inappropriate and unwarranted.” A coalition of 85 labor unions and mostly liberal groups expressed “strong opposition” to using the C-CPI-U in a letter to Congress dated Oct. 16.

That letter argues that the cost of living for seniors rises faster than the CPI, which “does not adequately take into account health care costs.” But the fact is, economists generally find no solid evidence that the cost of living for seniors really does rise faster than for others.

The Trouble with the CPI-E

Those who oppose using the “chained” CPI for adjusting Social Security pensions regularly point to yet another cost index, the CPI-E (for “elderly”), which attempts to measure rising costs for persons age 62 and over.

It is true that seniors spend more of their budgets on their health care than do younger persons. But it’s also true that they spend relatively less of their budgets on other things — including education, transportation, food and clothing. So they are not as affected by the rapidly rising cost of college tuition, for example.

The CPI-E rose an average of 0.2 percent per year more than the CPI-U or the CPI-W between December 1982 and December 2011. But BLS notes, “recent trends show different results.”

Since 2006, the CPI-E has risen at the same rate as the CPI-U, the index most commonly cited in news stories, and used to adjust income-tax brackets. And more important, it has risen more slowly than the CPI-W, the index used to calculate Social Security cost-of-living increases. Over that most recent five-year period, the CPI-E and CPI-U both rose at an annual rate of 2.3 percent, while the CPI-W increased 2.4 percent.

BLS cautions that the CPI-E is an unpublished, “experimental” index, and that “any conclusions drawn from it should be used with caution.” BLS also concedes that the CPI-E has a number of shortcomings because it simply re-weights the price data collected for its regular price surveys, without attempting to collect some important data specific to seniors.

For example, the CPI-E makes no attempt to track what seniors actually buy. “Because the specific items sampled within selected outlets are chosen on the basis of total sales in the outlet—and not sales to the elderly—the items selected for pricing in each outlet may not be representative of the CPI-E population,” BLS states.

Nor does it try to track prices at the places where seniors actually shop. “[T]he outlets may not be representative of the location and types of stores used by the elderly population,” BLS says.

Another shortcoming that BLS readily admits is that the “elderly” index takes no account of “senior discounts” available on such purchases as movie tickets, car rentals, train tickets, public transportation, chain restaurants and so forth.

At a 1995 Senate hearing, BLS Commissioner Janet Norwood was asked whether prices rise more rapidly for the elderly than for others. “The real point is that we do not know,” she said. “And we do not know because we do not have prices that are collected for items that are purchased by the elderly.”

The 2002 study by the National Research Council (previously mentioned for recommending creation of a “chained” index) also noted deficiencies in the experimental “elderly” index. “[W]e see no rationale for switching to an index along the lines of the CPI-E for purposes of indexation,” the report said.

The NRC report said further that a valid index should not be based on “speculation and conjecture”:

National Research Council, 2002: Would a price index for the elderly behave differently than the overall CPI if data were collected on items and qualities consumed by the elderly and on the prices paid in outlets where the elderly shop? To have a definitive answer to this question, or even relevant evidence instead of speculation and conjecture, an index for the elderly would have to be constructed to reflect “items that are purchased” and “prices actually paid.”

We take no position here as to whether benefits for seniors are too high or too low, whether future cost-of-living adjustments should be higher or lower, or how income-tax brackets should be adjusted in the future. We also note that the cost of living for any individual or family may go up faster or slower than the national average.

But it’s just a fact that leading economists have said for many years that the current CPI overstates the true rate of inflation. So using it to index federal programs produces more spending and less revenue than a more accurate measure would justify.

— by Brooks Jackson

Correction, Dec. 12: As originally posted, our Summary contained an incorrect example of how a 0.25 percent difference in the CPI would affect the dollar value of a hypothetical benefit payment. We have removed the erroneous example.

Update, Dec. 13: We have updated this article to include our calculations showing how the use of the C-CPI-U would have affected the 1.7 percent COLA that Social Security recipients will begin receiving in January. It replaces the incorrect example we removed Dec. 12.


U.S. Bureau of Labor Statistics. “Note on a New, Supplemental Index of Consumer Price Change.” 26 Apr 2011.

Stigler, George J et al. “The Price Statistics of the Federal Government,” A Report to the Office of Statistical Standards, Bureau of the Budget, Price Statistics Review Committee of the National Bureau of Economic Research. 1961: p. 35

U.S. General Accounting Office. “Measurement Of Homeownership Costs In The Consumer Price Index Should Be Changed.” 16 Apr 1981.

Poole, Robert and Frank Ptacek and Randal Verbrugge. “Treatment of Owner-Occupied Housing in the CPI.” U.S. Bureau of Labor Statistics. 9 Dec 2005.

Boskin, Michael J et al. “Toward A More Accurate Measure Of The Cost Of Living: Final Report to the Senate Finance Committee from the Advisory Commission To Study The Consumer Price Index.” 4 Dec 1996.

Dalton, Kenneth V. and John S. Greenlees and Kenneth J. Stewart. “Incorporating a geometric mean formula into the CPI.” U.S. Bureau of Labor Statistics, Monthly Labor Review. Oct 1998.

Panel on Conceptual, Measurement, and Other Statistical Issues in Developing Cost-of-Living Indexes, Charles Schultze and Christopher Mackie, Editors, Committee on National Statistics, National Research Council. “At What Price?: Conceptualizing and Measuring Cost-of-Living and Price Indexes.” National Academies Press. 2002.

U.S. Bureau of Labor Statistics. “Frequently Asked Questions about the Chained Consumer Price Index for All Urban Consumers (C-CPI-U).” 6 Apr 2005.

U.S. Social Security Administration. “Cost of Living Adjustment.” 16 Oct 2012.

U.S. Bureau of Labor Statistics. “The Use of the CPI in Adjusting Federal Income Tax Brackets and other Federal Tax Parameters.” Focus on Prices and Spending. Aug 2010.

The Moment of Truth: Report of the National Commission on Fiscal Responsibility and Reform.” Dec 2010.

Congressional Budget Office.”Reducing the Deficit: Spending and Revenue Options.”Mar 2011.

Rosenberg, Adam and Marc Goldwein. “Measuring Up: The Case for the Chained CPI.” The Moment of Truth Project. 11 May 2011.

The Chained CPI, an easy way to save money.” Editorial. The Washington Post. 26 May 2011.

McKinnon, John D. “Both Parties Divided Over GOP ‘Cliff’ Idea: Proposal Would Reduce Federal Spending and Raise Tax Collections By Using a Different Index for Inflation Adjustments.” Wall Street Journal. 9 Dec 2012.

Ruffing, Kathy and Paul N. Van de Water and Robert Greenstein. “Chained CPI Can Be Part of a Balanced Deficit-Reduction Package, Under Certain Conditions.” Center on Budget and Policy Priorities. 22 Feb 2012.

Edwards, Chris. “Chained CPI: A Stealth Tax Increase.” Cato@Liberty, Cato Institute 28 Jun 2011.

Americans for Tax Reform. “Cooper-LaTourette Budget Resolution Raises Taxes by Nearly $2 Trillion.” 28 Mar 2012.

AARP Public Policy Institute. “Adopting a Chained CPI Targets the Oldest, Poorest Americans; The Longer You Live, the More Social Security You Lose.” Dec 2012.

AARP. “AARP Opposes Social Security Benefit Cut as Part of Lame Duck Deficit Discussions.” press release. 8 Nov 2012.

U.S. Bureau of Labor Statistics. “Consumer Price Index for the elderly: Relative importance of expenditure categories in Consumer Price Indexes for three population groups, December 2011.” 2 Mar 2012.

U.S. Bureau of Labor Statistics. “Current Price Topics: The Experimental Consumer Price Index for Older Americans (CPI-E).” Focus on Prices and Spending. Feb 2012.

U.S. Social Security Administration. “2013 Social Security Changes.” 16 Oct 2012.

U.S. Social Security Administration. “Application of COLA to a Retirement Benefit.” 16 Oct 2012.