Economists on both left and right mostly agree that the current standard measure of inflation, CPI-W, slightly overstates the actual growth in the cost of living. The reason is something called “upper level substitution bias,” which means that instead of always buying a standard basket of goods and services, people change their buying habits over time as prices change. When the price of hamburger goes up, they eat more chicken. When the price of chicken goes up, they switch back to hamburger.
A version of CPI that takes this into account is called chained CPI, and overall it’s considered a more accurate reflection of actual inflation. But technical merits aside, there are always winners and losers when you make changes to statistics like this. One big loser would be Social Security beneficiaries. Initial Social Security benefits upon retirement are 
calculated based on wage levels, so they’d be unaffected by a switch to chained CPI. But annual COLA increases would be affected, and they’d be lower than they are now. Michael Hiltzik suggests two reasons this is unfair. First:
It’s not at all certain that elderly persons on fixed incomes can make the sort of lifestyle changes contemplated by the chained CPI….That’s because a larger portion of seniors’ spending is concentrated in medical goods and services, which aren’t as amenable to substitution as, say, oranges for apples.
….Indeed, the BLS has recognized that elderly consumers are a special case by developing an experimental CPI, known as the CPI-E, just for those 62 and older. Among other differences, the index overweights medical care as a factor in seniors’ spending….The CPI-E rose nearly 7% faster than the standard CPI from 1998 through 2009, according to government estimates. It also tells you why, from the standpoint of seniors’ real cost of living, the chained CPI is a rip-off.
No measure of CPI is perfect for everyone: if the price of gasoline is skyrocketing and you have a long commute, then your personal cost of living will rise faster than official inflation figures. Likewise, because healthcare costs are rising faster than most other goods, people with a lot of medical problems face higher inflation than those who are healthier. From a statistical point of view, then, the best you can do is choose a measure of CPI that’s most accurate in general.
Still, the CPI-E issue is a serious objection: it applies to a very large group, and it applies to a large group that typically has modest incomes. Ideally, it would be handled by broadening the scope of Medicare, not by deliberately using an innacurate measure of general inflation, but broadening the scope of Medicare is hardly on the table right now. Given that reality, the net result of this change would be to cut Social Security benefits by calculating inflation less accurately for seniors.
The second objection to chained CPI is more frivolous:
If you use the chained CPI instead of the standard CPI for the annual adjustment in income tax brackets, over time that will create an effective tax increase, especially for wealthier taxpayers….What do the agents of the wealthy say about that? Let’s ask the right-wing Cato Institute, which cherishes both a sedulous admiration for free enterprise and a long-standing hostility to Social Security. Cato last year called switching to the chained CPI for Social Security a “sound and overdue reform.” But when it came to using the chained CPI to adjust tax brackets, Cato called that “a very bad idea.”
….It’s a measure of the cynicism that guides debate in the nation’s capital that an “overdue reform” that would take $112 billion from the needy can be regarded as “a very bad idea” if it costs the rich $72 billion — and that no one pauses to ponder the rank injustice involved. Must be that they can’t make out their own words over the purring of those Mercedes engines.
Obviously Michael is right about this. If BLS adopts chained CPI as its new official measure of general inflation, then the change should be global throughout the government. Anything else is just obvious special pleading.
On a broader note, regular readers know that I’m generally in favor of Social Security reform. But I’m in favor of it only in the context of a broad-based reform that includes a mix of small, phased-in benefit cuts and small, phased-in revenue increases. A move to chained CPI could be part of that — one that has the benefit of also affecting the current elderly, instead of dumping the entire burden only on future generations — but no one should favor it in isolation. If this is something we end up doing, it has to be done as one piece of a complete package. Otherwise we’ll get the benefit cuts, Republicans will refuse to pass the corresponding revenue increases, and Social Security will remain fiscally unbalanced and endlessly under attack. A complete deal to fix Social Security all at once is the only kind of deal anyone should countenance. Piecemeal “reform” is a chimera.
