This study develops two sets of price indices for Medicare physician s
ervices. The first measures price changes, and the second measures geo
graphic price differentials. The indices can be used to adjust Medicar
e physician spending data to examine growth or variations in the volum
e and intensity of services. In both instances, it is necessary to app
ly an index form that reflects both the rapid changes and variability
in the mix of physician services received by Medicare beneficiaries an
d their relative importance. This suggests that an index based on a fi
xed basket of services (e.g., a Laspeyres index) can produce a biased
measure of price. An alternative methodology based on the Fisher's Ide
al Index form was used. This index allows service weights to vary over
time and across areas. In the case of price change, the index is ''ch
ained'' to acknowledge the introduction of new services over several y
ears. It is concluded that the Fisher's Ideal approach is essential fo
r cross-sectional price comparisons, in light of the high variability
in service mix across areas. In measuring price changes, it was found
that ''chaining'' was more important empirically than allowing the rel
ative importance of services to change between years. During the 1985-
1989 study period, Medicare payment rates grew, on average, by 3.5% an
nually. This rate varied across both time and types of services as a r
esult of differential fee updates and explicit pricing policies implem
ented by Medicare (e.g., reductions in payments for ''overvalued'' pro
cedures). Cross-sectionally, our results show that 1988 fees in the hi
ghest-priced areas were more than 1.5 times those in the lowest-priced
areas.