As the markets gyrate, trying to decide on just what the next move is going to look like, I thought I'd answer a question a friend posed to me recently. He asked "what does the consumption basket look like today, compared with 25 years ago?"

The question was something of a challenge to me (well, more precisely to the BLS, who puts together the CPI) because my friend - like so many others - had heard that the BLS assumes that when the price of steak goes up, you eat more chicken, and that the CPI doesn't capture what may be construed as a decline in consumer welfare due to this effect.

The BLS doesn't in fact make such determinations. In 1999, the BLS started using a geometric mean formula that captures the fact that consumers in fact tend to switch their consumption patterns towards items that have fallen in relative price. But that only affects items in the same very low-level "basic" category (the example the BLS gives is "different types of ground beef in Chicago"), and not at the higher levels like chicken-for-beef.

Moreover, an important point is that consumers will buy more of the item whose price relative to other goods has declined (say, chicken) even if the consumer is compensated for that change. Suppose a consumer pays $10 per week for soda, $5 each for Coke and Pepsi (which he considers interchangeable). Now suppose the price of Pepsi doubles, but the American Society for the Promotion of Cola Products sends the consumer $5. With the additional $5 he could continue to buy equal amounts of Coke and Pepsi, but will he? Of course not; the consumer will spend his $10 all on the cheaper brand and keep the extra $5, or buy more total cola by spending the entire $15--but all on the cheaper brand. The consumer's utility/dollar is now much higher with one (formerly interchangeable) brand.

It is important that the BLS attempt to get the consumption basket to look as much like the average consumer's consumption basket as possible, in order that the price index resembles as closely as possible a true cost of living index. So it doesn't spend a lot of time guessing. Every few years, the BLS conducts a "Consumer Expenditure Survey" to collect information about spending habits from around 60,000 quarterly interviews and 28,000 detailed weekly purchase diaries gathered over a two year period. From these, the base weights are determined for about 200 item categories in dozens of cities and regions. These are then aggregated up into successively larger groups up to the eight "major groups" (Food & Beverages, Housing, Apparel, Transportation, Medical Care, Other, Recreation, and Education & Communication) and the overall CPI-U.

In short, the BLS does everything possible to avoid making value judgments about what is being spent. That being said, studies have assessed the aggregate effect of substitution adjustments to be around 0.3% per year.¹

So with that all said, how much have our spending patterns shifted in the CPI? Are we spending money dramatically differently than we did 25 years ago?

You can find the CPI weights for 1988 here, and compare them to current weights found on p. 11 of the monthly CPI report, for example here. Be aware that the categories sometimes change over time, so that for example in 1988 there was not an "Education and Communication" major category. But here are some comparisons:

In summary? Surprisingly, while we get a lot less for our money than we used to, our patterns of expenditure haven't changed much, in aggregate. Sure, some of us are spending less on education while others spend more, simply because our patterns change as we go through life (and the BLS has some age-related indices as well). We spend, not surprisingly, less of our money today on clothes, food, alcohol, cigarettes, housing, and transportation - even with higher energy prices. Overall, though, we spend more on energy itself, on medical care, and on tuition. You can see the effect of the 'home ownership movement' in the greater basket weight of "Owner's Equivalent Rent" compared to "Primary Rents". But overall, the spending patterns are evolutionary, not revolutionary.

Oh, and we do spend less, as a portion of our total expenditure, on beef and veal! But we also spend less on chicken, pork, fish, and "other meats" (shudder).

¹ I won't talk here about quality adjustments, including so-called "hedonic" adjustments, except to say that the aggregate effect of quality adjustments is so small as to be hardly worth considering. Hedonic adjustment lowers the CPI for a number of categories whose weight total to around 0.7% of the index, but increases the CPI for Rent, Owners' Equivalent Rent, and Apparel, plus some smaller categories, totaling about 32% of the index. So large downward adjustments on lightweight categories are balanced by very small upward adjustments on large categories. The net effect is thought to be less than 0.01% per year (see Johnson, Reed, and Stewart, "Price Measurement in the United States: a Decade after the Boskin Report", 2006).