We do, however, have an empirical estimate of a major, known bias in the CPI. Prior to
1983 for the Consumer Price Index for all Urban Consumers (CPI-U) and to 1985 for the
Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W), the
homeownership component was based—very roughly—on the costs incurred in purchasing,
financing and maintaining a house. More specifically, the home-purchase expenditure weight
was the net purchase of owner-occupied houses in the survey period, and the mortgage-interest
expenditure weight was the total interest (undiscounted) that would be paid over half the term
on all mortgages incurred during the survey period. These expenditure weights were indexed
over time by a home-purchase price index and the product of a home-purchase price index and
a mortgage-interest rate index, respectively. This methodology was without conceptual
foundation (see Gillingham (1980, 1983)), and its use resulted in a substantial upward bias in
the CPI.