Steve Levitt helpfully provides a link to the Broda and Romalis paper that Jim Surowiecki references this week, and whose findings I found so startling. After reading the paper, whose findings on inflation rates are by no means easy to pick out, I’m much less startled.
The first thing to note is that when Surowiecki says that "between 1999 and 2005 alone the inflation rate for lower-income Americans was almost seven points lower than it was for the wealthiest Americans," he isn’t referring to the inflation rate as most of us think of it, as in the annual rise in prices. Rather, he’s talking about the total rise in prices over that six-year period, during which prices rose 2.6% per year for the rich, on average, and 1.6% per year for the poor. So the difference in inflation is one percentage point per year, not seven.
But in fact it’s much lower even than that. Surowiecki’s looking only at the prices of "non-durable goods" – which are only a part of the total CPI basket. Yes, they’re a larger part of the CPI basket for the poor than for the rich. But the poor do buy other things too.
And Surowiecki’s taking his numbers from a table at the end of the paper, which doesn’t appear with a lot of explanation. If you look at what the authors actually write, they say that the inflation difference in non-durable goods is smaller still:
While the richest group in the ACNielsen had a non-
durable inflation rate of around 9.5 percent over the 1999 – 2005 period, or 1.5 percent per year,
the four poorest groups combined had an non-durable inflation of 6.2 percent, or 1.0 percent per
year. The non-durable inflation rate of the poorest income group was 0.5 percentage points
smaller than that of the richest group over the 1999 – 2005 period.
What about the overall inflation rate, not just inflation in non-durables? Broda and Romalis do address this question directly, although they use the period from 1994 to 2005 rather than the period from 1999 to 2005.
We find that the poor’s common-goods inflation rate over the 1994 – 2005 period has
been 2.1 percentage points smaller than that of the rich… simply because the
conventional CPI measures do not take into account the differences in expenditure shares by
consumption category across groups. This effect is mostly coming from the fact that the poor
consume more non-durable goods…
Column (3) in Table 7B shows the additional effect of allowing for different non-durable
“common goods” inflation rates across income groups. By using the income specific common-goods inflation rates, the inflation differential across income groups grows to 3.7 percent. When one takes into account the fact that the proportion of new goods purchased by the poor is larger
than for the rich (column (4)), the inflation differentials between rich and poor over the 1994 –
2005 period rise to 5.5 percent.
Let’s take the biggest number here: an inflation differential between rich and poor of 5.5% over 11 years. That’s not 5.5% per year, remember, it’s 5.5% total. Unfortunately, Broda and Romalis only give the difference in total inflation, not the actual inflation rates. But we know that the CPI rose by 32% between 1994 and 2005. So let’s say take a couple of numbers on either side of that 32% figure which are 5.5% apart, and assume that inflation for the poor was 30%, while inflation for the rich was 35.5%.
Those figures work out to an annual inflation rate for the poor of 2.4%, and an annual inflation rate for the rich of 2.8%. Which means that the inflation rate for the poor has only been 0.4 percentage points lower than it was for the rich – and that’s for the period of time in which the Chinese export boom really changed patterns of consumption in the US. It’s entirely possible that trend won’t continue, especially now that the dollar’s weakening and food and energy prices are soaring. (The poor spend a much higher proportion of their income on food and energy than the rich do.)
Contra Surowiecki and Levitt, then, I think it’s hard to read the Broda and Romalis paper and conclude that, in Levitt’s words, "the prices of goods that poor people tend to consume have fallen sharply relative to the prices of goods that rich people consume". They might have fallen a very little bit – about 0.4% a year, tops. But in any case that fall might well now be over.
Update: Zubin, of course, got here first.