more on the CPI–Laspeyres vs. Paasche

the current CPI

As I wrote yesterday, the CPI is a Laspeyres index.  That is, it takes a fixed basket of goods and services from a base year and calculates the changes in the total cost of that basket over time. A Laspeyres index always overstates the effect of inflation, however.  That’s because it doesn’t take into account the fact that people respond to price changes in some things by switching to less expensive substitutes that let them maintain the same lifestyle at lower cost.

The government thinks this overstatement has historically been about 0.2% per year.  The Social Security Administration estimates possible future overstatement to be about 0.3% annually.

Actually, the situation with the CPI is slightly more complicated.  Since 1998, the Bureau of Labor Statistics has been using a method that allows it to adjust the CPI if it sees, say, the price of Fuji apples rise and people switch to cheaper Honeycrisp instead (the government collects price data on over 80,000 items and includes regional price variations as well).  But it won’t adjust if people stop buying apples and switch to bananas.  So the practical effects of this index improvement have been small.

The BLS also computes a CPI for the elderly, using the same methodology but using a basket of goods and services that senior citizens surveyed say they use.  Those prices appear to be rising at 0.1% a year faster than the CPI itself.

a Paasche index is a non-starter

Paasche indices start with the current basket of goods and services that people use.  It works from the present back to construct an index of past price changes in the current basket of items.

A Paasche index tends to understate the effects of inflation because it assumes that on day one of the period being considered consumers instantaneously respond to all price changes that will occur over the period.  Hard to do if the price changes haven’t happened yet.  Also, not much practical usefulness, since the index data are only available after the fact.

a chain-weighted CPI

That’s what the current discussion in Washington is about.  It enables the BLS to adjust the CPI on the fly to shifts in consumer buying patterns based on price-driven substitutions.  The BLS has had a chain-weighted CPI up and running since 2002.  But until now the possible savings in Social Security payments hasn’t seemed to Washington worth taking on a powerful lobbying group like the AARP or the (misplaced, in my view) ire of senior citizens.