Health care is a current media buzz word. The focus ranges from the Affordable Care Act
to restructuring Medicare, but there are other important aspects of the medical care system that
need to be examined. One of the controversial areas that researchers have started studying with
inconclusive results is the effect of liability pressure on the supply of physicians.
The concern for liability pressure on the supply of physicians has grown since the price
of premiums has grown 8.1% annually from 1994 to 2002,i and continues to increase at a high
rate now. Malpractice premiums have increased due to the increase in malpractice suits and the
rising medical and health care costs. If nothing stops a patient they can sue a doctor for any
amount and since going to court is more of a cost in terms of time and money the cases are often
settled outside of court. This means that the physicians’ insurance will pay whatever the lawyers
come up with. As malpractice suits increase, insurers must respond by raising their premiums to
remain profitable. With these rising costs from liability pressure physicians may choose to stop
practicing or to relocate. Currently, the policy response to deterring liability pressure is to pass
tort reform laws. The scopes of these laws vary by state from nothing to severe economic cap.
Do states with more stringent malpractice tort reform laws have an increase in supply of
physicians or conversely do states without malpractice tort reform laws see a decrease in the
number of physicians? More importantly do the rising costs of medical malpractice premiums
drive physicians to locate in “cheaper” places to practice.
Previous studies have looked into these questions. When this problem came to the
attention of medical researchers their first response was to survey physicians. Although many
survey studies I disregarded because it is more difficult to make economically significant
conclusions from them, there is one survey study that I found highly relevant to the question and
what I want to research. The study, by Mello and Studdert, that uses a survey focuses on the
south eastern area of Pennsylvania and randomly surveys doctors in that area who are registered
with the AMA.ii What I found beneficial about this method is that the researchers are trying to
find out if the doctors change their practice, not just if they stop practicing or relocate. The
researchers do find that physicians are cutting down in their high risk practices and that some are
willing to retire early. “Forty-three percent had already personally reduced or eliminated high-
risk aspects of their practice, and 50% said they would likely (continue to) do so over the next 2
years”iii The authors also look into how patient experience has changed. Again I do not believe
survey results can be considered for thorough statistical analysis, but it is insight into the way
physicians think about malpractice premiums. Also another study I looked at highlighted that
physicians might not relocate, but may just cut out the highest risk aspects of their practice. By
just counting numbers it would be hard to account for that change in supply.
The other studies I looked at do use raw data from the American Medical Association,
(AMA), although two of them use differences in differences methods, while the others do run
regressions. I will talk about these in order that they were published since these articles have in
fact built off each other and set up the framework that I hope to build off of. Kessler et. al.
published Impact of Malpractice Reforms on the Supply of Physician Services in 2005 and he
uses a differences in differences model to look at the effect of direct tort reform laws, (economic
caps on damages), on the supply of physicians in group practices verse solo practices.iv The
authors find that physician supply increased by 3.3% after three years. They also find that the
effect is larger on non-group physicians. I am critical of the differences in differences method
and think that a standard OLS regression with an indicator variable for group practice or non-
group would have been a more sufficient way to study this, but I do like that the authors identify
the difference. I dislike that they look at doctors with greater than twenty years’ experience and
less than 20 years. I think the more important groups of physicians to look at are those just
entering the market and those close to retiring since they will be the ones most likely to choose to
practice or to stay in practice in areas where malpractice premiums are low.
In 2007 Klick builds off the Kessler study by analyzing AMA data with a triple
differences model.v Klick et. al. introduces the idea of looking at high risk specialists since they
face the higher premiums and therefore are more likely to react to an increase in premiums.
Klick uses a triple differences model looking at the difference between states with verse without
tort reform laws on high risk specialists vs. low risk specialist within each state. Basically the
low risk specialists serve as a control group and the high risk as the treatment group. Again I
would have preferred a regression with an indicator variable for high or low risk physicians.
Klick et. al. found that caps on noneconomic damages have a significant effect on increasing
physician supply and that this effect is concentrated on those physicians with the highest risk.vi
The authors also do an interesting analysis of the effect of tort reform on health outcomes.
Although I do not plan to focus on this for my capstone I believe it is another area that should be
The last two articles I look at move into some higher economic analysis, but still lack the
rigor I hope to use with my study. In 2008 Yang et. al. focus more on the idea of high risk
specialties, by only looking at the supply of OB/GYN in relationship to malpractice premiums
and tort reforms.vii The authors use panel data form the AMA from 1994 to 2002 for information
OB/GYN for all 50 states and Washington D.C., along with data on number of births and number
of birthing age women to find the effect or premiums on relative supply. The authors run a mixed
model, which basically means they control for time through trend variables, but they do not
control for states. Instead they include many variables they are supposed to prevent omitted
variable bias, but that allow for with in state variation. This is one of the biggest issues I have
with the study. A fixed effects model should have been run, even if it was not their base
regression. Now I did like that the authors introduce the idea of lagging the data from the AMA.
I thought this was a good idea because physicians cannot immediately react to changes in
malpractice premiums there are too many hindrances that come with relocating to allow that. The
authors find that their no significant relationship between malpractice reform or malpractice
premiums with the supply of OB/GYN. They argue one of the reasons for this may be that
OB/GYN can cut out the highest risk aspects of their practice in response to increase in
premiums because it is a lower cost than moving.
Finally a study in 2009 by Chou et. al. looks at how new physicians choose to locate
based on malpractice premiums.viii Chou only looks at residents who graduate from New York
Medical schools and in three specialties: surgery, primary care, and OB/GYNs. The authors find
that there is a significant negative association between malpractice premiums and surgeons, but
neither of the other specialties are significantly related. This does not surprise me that primary
care physicians are not related significantly since PCP are low risk. The OB/GYN find is similar
to that of what Yang et. al reported in 2008. What I especially like about this study is that the
authors are looking at what residents chose to do. This is the right question to ask since it looks
at the marginal decision not the cumulative, which is the smarter economic question to ask. Also
physicians finishing their residency are more likely to choose where they practice based on
premiums because they have more choice then physicians who are settled with a family. My
ultimate goal is to ask a similar question as Chou et al, but with general panel data from AMA.
To build on this previous work and to hopefully clear up the answer I plan to design my
own project. My study will not be based solely on AMA data, which some of the previous
studies have used. Similar to the studies by Mello and Chou I will be focusing on specific states
first to lay the ground work. The two states I will look at have opposite reactions to tort reform
laws. Texas, which leading up to its tort reform had really high malpractice premiums, but in
2003 passed a stringent malpractice law to drive down malpractice premiums. From Texas
Department of State Health Services I have data from 2000-2010 that account for all physicians
in direct patient care.ix The data set includes where the physicians went to medical school, where
they currently practice, where they are from, their specialty, whether the area is urban or rural,
and much more. From this data I plan to study if and how supply has changed leading up to the
2003 law and after in Texas. My dependent variable will be the number of physicians choosing
to start practice in Texas that year and I will use an indicator variable for before tort reform or
after as my focal independent variable. Other variables I will control for are population, an
indicator for urban or rural, high risk or low risk specialty, and possibly cost of living. I plan to
explore different variables that need to be controlled for. I would also like the run the same
regression with malpractice premium data. Although there are flaws in this model, (for example
it cannot be generalized), I believe it will give me insight into how a state with tort reform effects
the choice of residents. I will be expanding on the study by Chou because I will look at several
different specialties. I will also be looking at doctors who come from many different medical
schools and areas who choose to practice in Texas.
The state that has an opposite approach to tort reform is Pennsylvania. Now
Pennsylvania has some of the highest malpractice premiums and has had no direct malpractice
tort reforms. One of the studies I mentioned earlier was the study that used the survey of
Pennsylvania physicians, focused primarily on the Philadelphia area. I want to look at how the
Pennsylvania physician supply changes. I do not have the same data set as the Texas data set
hence any real comparison would be difficult to do, but I think an analysis of both side by side
yield interesting results. For Pennsylvania my data comes from Jefferson Medical school and it
follows their alums back from 1968.x The data accounts for the year they graduated from medical
school, their age, their sex, where they are from, where they practiced their first year of
residency, where they practice now, their specialty, and their type of practice. I would like to
either run a probit model that looks at the probability of practicing in Pennsylvania and uses
control variables from above or I would like to look at how the number of physicians change
running a similar regression as the one for Texas.
Finally if possible I would like to get my hands on the American Medical Association
Physician Master file.xi This data costs money and has been elusive to get, but if I can get it I
would like to run a regression similar to the one I described using the Texas data. What I would
do is regress malpractice premiums in a given year, fixed effects for states and trends, and other
control variables against the number of physicians finishing resident that choose to practice in a
given state in a given year. If possible I will break this down into specific high risk specialties. I
may also do a similar regression, but input tort reform law for each state for malpractice
premiums. The malpractice premium data will come from Medical Liability Monitor.xii This data
does cost money and thus will cause a roadblock in my work.
What I hope to get from my research is a thorough, rigorous, economic analysis of the
relationship between liability pressure and physician supply. As I highlighted in my introduction
this is an extremely important topic because of the rapid grown in health care costs and the need
for physicians. I hope to be able to conclude if they are significantly related and to feel as though
my analysis puts an end to the inconclusiveness that still remains in the topic area. If you’ll
notice none of the studies I mention above are published in economic journals in fact using econ
lit, I was only able to find three studies listed, only one of which was published in a journal.
This topic area lacks the analysis of economists and I hope that my capstone will be a step,
(albeit very small), towards filling that void.
A Longitudinal Analysis of the Impact of Liability Pressure on the Supply of Obstetrician-Gynecologists Author:
Yang, YT. Published in: journal of empirical legal studies, v. 5 no. 1, pp. 21 Date: 2008
Effects of a Malpractice Crisis on Specialist Supply and Patient Access to Care, MM Mello, DM Studdert… -
Annals of Surgery, 2005
Mello et. al.
Impact of malpractice reforms on the supply of physician services Author: Kessler, DP. Published in: jama, v.
293 no. 21, pp. 2618 Date: 2005
Medical malpractice reform and physicians in high-risk specialties Author: Klick, J. Published in: journal of legal
Klick et. al.
Yang et. al.
Practice Location Choice by New Physicians: The Importance of Malpractice Premiums, Damage Caps, and
Health Professional Shortage Area Designation Author: Chou, CF. Published in: health services research, v. 44 no.
4, pp. 1271 Date: 2009