Fun With Inflation

I was considering trying to do a post on how the Bureau of Labor Statistics puts its inflation indexes together, but the BLS FAQs here and here are really quite good and people should just read them.

I would just make a few quick points. Even the broadest measure of inflation, the CPI-U (“for all urban consumers”) doesn’t really try to measure the prices paid by the 20 percent or so of consumers who don’t live in Census Bureau-designated Metropolitan Statistical Areas or Urban Places. There’s a sound logistical reason for that, but 20 percent is a pretty large number. The other thing is that to construct the index you need to decide how much weight to give to different categories of expenditure. They do this by using the data from the Consumer Expenditure Survey. And one thing you’ll see if you look up the CES data tables is that, just as intuition would lead you to believe, different kinds of households buy pretty different stuff. For example, if you compare households by education level, people with bachelor’s degrees spend a way lower proportion of their income on housing and tobacco products than do non-college households. People in the 25-34 age bracket (that’s me) spend a much higher proportion of income on food away from home than do people in the 55-64 age bracket.

None of that is a “problem” with the measurement, but it serves as a reminder that any particular family’s experience of price fluctuations can diverge substantially from the national average. It also seems to lead to the conclusion that Social Security benefits should be tied to a special index linked to seniors’ consumption basket since the price of things seniors don’t buy isn’t really relevant.

Last, the BLS tries (hard!) to make calculations about changes in the quality of goods. But when you think about it, this isn’t really possible to do in a definitive way.