NBER WORKING PAPER SERIES
PHYSICIAN FINANCIAL INCENTIVES
AND CESAREAN SECTION DELIVERY
Jonathan Gruber
Maria Owings
Working
Paper No. 4933
NATIONAL
BUREAU OF ECONOMiC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
November 1994
We are grateful to Ed Bacon, Peter Diamond, Victor Fuchs, Marty Gaynor, Tom McGuire. Joe
Newhouse, Robert Pokras, Jim Poterba, Richani Zeckhauser, Jon Skinner, seminar participants
at MIT, Harvani, and the NBER, and to the editor and two anonymous referees for helpful
comments. This paper is part of NBER's research programs in Health Care and Public
Economics. Any opinions expressed axe those of the authors and not those of the National
Center for Health Statistics or the National Bureau of Economic Research.
1994 by Jonathan Gruber and Maria Owings. All rights reserved. Short sections of text, not
to exceed two paragraphs, may be quoted without explicit permission provided that full credit,
including
notice, is given to the source.
NBER Working Paper #4933
November 1994
PHYSICIAN FINANCIAL INCENTIVES
AND CESAREAN SECTION DELIVERY
ABSTRACT
The "induced demand" model states that in the fact of negative income shocks physicians
may exploit theft agency relationship with patients by providing excessive care in order to
maintain their incomes. We test this model by exploiting an exogenous change in the financial
environment facing obstetrician/gynecologists during the 1910s: declining fertility in the U.S.
We argue that the 13.5% fall in fertility over the 1970-1982 period increased the income pressure
on ob/gyns, and led them to substitute from normal childbirth towards a more highly reimbursed
alternative, cesarean delivery. Using a nationally representative micro-data set for this period,
we show that there is a strong correlation between within state declines in fertility and within
state increases in cesarean utilization. This correlation is robust to consideration of a variety of
alternative hypotheses, and appears to be symmetric with respect to periods of fertility decline
and fertility increase.
Jonathan Oruber
Maria Owings
Department of Economics, E52-274c
National Center for Health Statistics
M.I.T.
3700 East West Highway
50 Memorial Drive
Hyattsvilte, MD 20782
Cambridge, MA 02139
and NBER
A standard model of physician behavior in the health economics literature is that of "induced
demand'. This model states that in the face of negative income shocks physicians may exploit their
agency relationship with patients by providing excessive care in order to maintain their incomes.
Such a model has traditionally been tested by assessing the effects on the utilization of medical
procedures of two alternative changes in the environment facing physicians. The first is reductions
in the fees paid to physicians, generally by government payers. The second is variations., in the
physician/population density across areas; increased density lowers the income of the existing stock
of physicians, and will lead to increased utilization of medical procedures in an inducement-type
model.
Each of these approaches faces important problems, however. Fee changes cannot identify
supply responses because there may be a contemporaneous demand response to changing prices; that
is, if quantity rises when fees fall, this could reflect either induced or true increases in demand.
Using fee changes to identify inducement also raises the problem that substitution and income effects
go in opposite directions; even if there is no measured utilization response, there may be inducement,
but substitution and income effects may simply be cancelling. Studies that use physician density
differences to proxy for income shocks also face a fundamental identification problem. If there is
some unobserved area characteristic which is correlated with taste for medical interventions, an area
will feature both higher procedure utilization and more physicians, regardless of the extent of demand
inducement An obvious candidate for such sit omitted variable is the average coinsurance rate in
the area.
In this paper, we propose a new means of identifying the effect of induced demand on
procedure utilization. We exploit a plausibly exogenous change in the financial environment facing
2
obstethcianlgyttcologists
during the 1970s: declining fertility in the U.S. Fertility (births/100
population)
fell by 13.5% during
the 1970-1982
period. We argue that declining fertility increased
the income pressure on ob/gyns, and led them to substitute from normal childbirth towards a more
highly reimbursed alternative: cesarean delivery. Used in only 5.5% of births in 1970, cesarean
delivery
rose by over 240% over the
subsequent 12 years, and it is now the second most frequently
performed major surgical procedure in the U.S.,
with a rate of
23.5 cesarean
deliveries per 100
births.' This time series correlation between the fertility declinc and the cesarean utilization increase
raises the possibility that inducement played an important role in the substitution of cesarean delivery
for normal thildbirth.
The primary purpose of
this
paper
is to test for
a causal role of financial incentives in the use
of
cesarean delivery over the 1970-1982 period. We do so by exploiting
the dramatic change in
fertility patterns across the U.S.
states during this era.
Conditional on
the characteristics
of the
mother and of the birth, statewide fertility changes should provide exogenous measures of the
financial pressure on ob/gyns in that state. Thus, we ask: in states where fertility was falling the
most, did cesarean delivery rise the most?
Our teat is carried out using a rich dataset of individual birth records from the 1970-1982
period. The National Hospital Discharge Survey (NHDS) is a nationally representative survey of
hospitals which collects information, for a sample of discharges, on the nature of the hospital stay,
procedures performed during that stay, and the demographic characteristics of the patient. There is
data on approximately 20,000 deliveries annually from a wide cross-section of hospitals. We match
this data on individual births to information on changes in fertility levels by state and year, and
examine the effect of within state changes in fertility on the use of cesarean delivery. We use both
1Unpublished tabulations, National Center for Health Statistics.
3
individual lent and state level data to control for a number
ofother characteristics of mothers and
births, such as age, birth order, and infant health, which may be spuriously correlated with both the
fertility rate and the use of cesarean delivery.
To summarize, we find a strong correlation between within-state fertility declirts and within-
state increases in cesarean section delivery rates: a 10% deciS in the fertility rate is associated with
a 0.97 percentage point increase in the cesarean delivery rate. This finding is robust to a wide
variety of specification checks. It implies that financial incentives do play an important role in the
substitution of cesarean delivery for normal childbirth. This role was a small one relative to the
rapid growth in cesarean delivery, however; increased ob/gyn income pressure can only explain
between 16% and 32% of the substitution of cesarean delivery for normal childbirth during this era.
We begin,
in Part
1,
by
discussing the causes and consequences of the diffusion of cesarean
childbirth in the l970s. In Pan
II, we
summarize the literature on the induced demand hypothesis,
and present a brief model which notes the implications of this hypothesis for the use of cesarean
delivery when the fertility rate falls. Part III
describes
the data sources and the empirical framework
used. Paxt IV presents the basic results and specification checks. Part V
assesses
whether cesarean
delivery responded asymmetrically to fertility rises and falls. Part VI concludes.
Part I: Background on Cesarean Delivery
Cesarean Delivery in the United Suites
The upper IS in Figure 1 charts the growt in the cesarean section rate (c-sections per 100
births) nationally over the 1970 to 1991 period. The rate doubled from 1970 to 1975, and doubled
again by 1984. Utilization of the procedure flattened out in the mid-1980s, and fell somewhat at
the
end of the decad& This rapid growth in c-section utilization was not restricted to any particular
