Abstract
Particularly since the 1990s, federal statistical agencies have worked to improve the ability of
various price indexes to measure changes in the cost of living. However, in recent years, some
have sent mixed signals to researchers about the relative merits of different measures. As a
result, academic and policy researchers routinely use theoretically and empirically inferior price
indexes in their analyses of real income changes.
Moreover, a cumulative body of research has identified and estimated the magnitude of a number
of biases that affect all of today’s widely used price indexes. These biases remain due to data
inadequacy, methodological uncertainty, bureaucratic inertia, and political considerations.
Nevertheless, there is little dispute about the direction of the overall bias in these indexes, and
considerable consensus as to the magnitudes of the individual biases. It is very likely that all the
most commonly used price indexes overstate the rise in the cost of living by a substantively
important amount.
This paper summarizes the evidence on these biases and translates it into a new “More Accurate
Consumer Price Index” (MACPI). It provides annual index values from 1973 to 2023 and
illustrates the importance of bias correction by showing a number of long-term trends in wages,
earnings, income, and wealth. While the most widely cited inflation measure, the Consumer
Price Index for All Urban Consumers (CPI-U) suggests that the average hourly wage of
production and nonsupervisory workers rose by 2 percent from 1973 to 2023, and the superior
Personal Consumption Expenditures price index (PCEPI) indicates a rise of 30 percent, using the
MACPI, wages rose by 61.5 percent. The median wage of prime-age male workers fell by 15
percent using the CPI-U and rose by 9 percent using the PCEPI, but it rose by 35 percent using
the MACPI. Other comparisons are similarly striking.