Mysteries Of The Chained CPI Explained

Posted on


Ezra Klein says that one deficit reduction measure under consideration is to start using the Bureau of Labor Statistics’ Chained Consumer Price Index (C-CPI-U) to index Social Security benefits and tax brackets, rather than the current indexes. This is appealing, I guess, because it’s simultaneously a benefit cut and a tax hike but you can claim that it’s just a technical change. Sadly, though, Ezra seems to think that explaining what it actually is would be too terrifying, saying that “[b]ecause I’d like at least a few readers to get to the end of this post, I’ll omit the gory details.”

I don’t actually think the details are all that gory.

Or, rather, calculating any cost of living index is a fairly gory business that involves lots of people running all around the country writing down the price of various things. And I also find the difference between the CPI-U (used for tax brackets) and the CPI-W (used for Social Security benefits) to be pretty obscure. But the difference between C-CPI-U and old school CPI-U is pretty easy to explain. Simply put, the current index assumes that if the price of beef skyrockets, then consumers just face a higher cost of living in the form of expensive beef. The chained index, by contrast, assumes that beef is a sub-element of a larger “meat” product category such that if beef gets more expensive consumers will respond in practice by buying more pork. Consequently, while the higher price of beef does push up the cost of living, on the chained view it doesn’t push it up as much as it would under the unchained view.

Whether this is technically “correct” or not strikes me as a somewhat metaphysical dispute not amenable to any kind of real resolution. It’s of course true that if beef rockets up in price, people will respond by buying more pork. But beef isn’t pork. We could eat nothing but rice and beans all day, but the fact that we don’t have to is part of what makes it nice to live in a non-impoverished country. The important thing about inflation indexing as applied to Social Security is that in practice old people consume much more health care than average people, and old people also tend to be relatively uninterested in high-tech gadgets. This mismatch between the consumption patterns of the elderly and economy-wide trends is much more significant than the chaining issue. On the tax side, however, one of my fondest hopes would be to eliminate the inflation indexing of the brackets altogether so anything that cuts the index rate to a lower level would be welcome.