New Physicians and the Role of Medical Malpractice Rates

Jonathan Heim

(2007)


Abstract

This study examines the role that medical malpractice rates have on newly graduated physicians entering into the labor force as well as physicians who have chosen to relocate to a new state.  Medical malpractice rates may have a negative influence on the number of physicians entering into the labor market for any geographic area in a year.  If a negative relationship is found between medical malpractice rates and the amount of licenses issued, the physician labor supply could be influenced by policies aimed at lowering health care practitioner costs.  Using statewide data on both medical malpractice rates and number of licenses issued, this paper finds that for a $1000 increase in medical malpractice rates, the percentage of licenses issued to relocating physicians decreases by 1%-7%.  The results also indicate that for a $1000 increase in the malpractice rates of internal medicine specialists, the number of newly graduating physicians in a given state will decrease by 6%.
Introduction

            Physicians in the United States have been facing rising medical malpractice premiums throughout the past few decades to the point where the American Medical Association (AMA) has declared a state of crises in 21 states in the union[1].  Some states such as California have passed legislation that puts a maximum limit on how much medical malpractice premiums can cost.  Many other states have passed the same legislation in an attempt to curb the cost of overall medical malpractice rates.

            Economic theory and previous studies have concluded that there are four contributing factors to the rise in medical malpractice premiums: declines in insurers’ investment income, a less competitive insurance market, climbing reinsurance rates, and insurer losses from a growing number of malpractice claims and awards to plaintiffs (Mello, Studdert, and Brennan, 2003). 

            High medical malpractice rates can have several effects on the physician labor force. First, research suggests that when malpractice premiums rise, physicians begin to practice defensive medicine (James Thornton, 1999).  This translates to physicians ordering more tests and procedures for every patient in order to ensure there is no room for any malpractice claims.  Second, research also suggests that the number of hours per week that the average physician works is positively affected by rising malpractice premiums (Thornton, 1999).  Third, the increased costs that physicians face in the form of fees is passed onto patients or insurance companies in excess of their increased costs (Greenwald and Mueller, 1978; Sloan, 1981; Reynolds et al., 1987; and Danzon et al., 1990).  Finally, research has found a connection between previous claim records and the probability that a physician will be sued a second time (Weycker and Jensen, 2000).  This pattern could lead to physicians leaving the labor force due to surmounting legal fees and rising malpractice premiums.

            Baicker and Chandra (2004) conclude that high medical malpractice costs do not seem to have any effect on the existing physician workforce aside from marginal entry and exit, though their research does not focus on the influence of medical malpractice rates have on new entrants into the labor market for physicians.  In order to overcome this gap, this study will analyze this by looking at average state wide medical malpractice rates and comparing them to the number of medical licenses issued to physicians for the first time in their career.  This study would shed some light on the overall labor market effects of medical malpractice rates from the perspective of newly graduated physicians from residency training. 

            The results of this study could have a large impact on future policy dealing with medical malpractice rates.  Many states have already passed legislation that puts a maximum limit on how much medical malpractice premiums can cost any individual physician.  As of yet, none of the research used to influence these laws have looked at the market for newly graduated physicians.  This could potentially give legislators an edge in curbing medical malpractice rates from the very beginning of a physician’s career in order to keep them in their state for an extended period of time.

            The outline of this paper is as follows. Section II introduces the empirical model that will be used to estimate the effects of medical malpractice rates on the number of initial licenses issued per state in 2004.  In section III, data will be introduced and explained. Section IV details the results of the empirical model and Section V will provide conclusions and limitations of the study.

II.  Empirical Framework

         This study will explore whether states with higher medical malpractice rates see a greater decline in number of initial licenses issued on a statewide level.  Implicit in this analysis is the concept that higher medical malpractice rates will increase the total cost of doing business and thus will decrease the overall profitability of physicians.  Since most physicians are profit maximizers (Thornton, 1999), they will seek to set up their offices in a geographic area that will give them the greatest potential to maximize their income.

            Similarly to Baiker and Chandra (2004), this paper assumes that the number of physicians entering into the labor market for the first time is influenced by the total medical malpractice rate:

               (1)

 

            Initial licenses refer to the total number of licenses issued to physicians for the first time in state “i”.  These licenses are being issued to these physicians for the first time in their careers.  There are three different rates that are used for average malpractice insurance premiums paid in a given state: Internal medicine rates, General surgery rates, and the rates of OB/Gyn practitioners.  This is done in an attempt to see whether physicians pay attention to the high-end rates (OB/Gyn), the medium rates (General surgery), or the low rates (Internal medicine).  In order to control for time varying factors, the variable, X, will include the unemployment rate, the mortality rate per 100,000 capita, per capita income, and population.  These factors are included to account for the economic situation in each state.

            In an attempt to examine only the percentage change that is felt from year to year in the number of initial licenses, the difference in logs will be taken of the variable Initial Licenses.  One major problem that exists with equation #1 is that of multicollinearity.  In order to control for the correlation between variables, each medical malpractice variable is examined separately in equations #2-4.

            Equation #2 will look at the Medical Malpractice rates of Internal Medicine specialists on the percentage change in the number of initial licenses issued.

                     (2)

            Equation #3 will be responsible for the rates of General Surgeons and their effect on the percentage change in the number of initial licenses issued.

                       (3)

            Equation #4 will examine the effect that the rates of Obstetrics and Gynecology specialists have on the percentage change in the number of initial licenses.

                                   (4)

            A second set of equations will be examined to see if medical malpractice rates have an effect on the number of physicians that relocate to a different state in a given year.

              (5)

            Equation #5 will examine the effect that the rates of Internal Medicine specialists have on the percentage change in the number of relocating physicians.

                (6)

            Equation #6 will account for the effect that the medical malpractice rates of General Surgeons have on the percentage change in the number of relocating physicians.

                          (7)

            Finally, equation #7 will look at the medical malpractice rates of Obstetrics and Gynecology specialists and their effect on the percentage change in the number physicians that relocate to a different state in a given year.

            Each of the above models will be estimated using the fixed effects method in order to determine the actual effect that each covariate will have on the dependant variable.  This method is much more powerful than the OLS method because it allows the use of panel data.  This allows us to see the effects that medical malpractice rates have on the percentage change in the number of licenses issued to both newly graduated physicians and physicians that have decided to relocate to a different state from one year to the next.  In order to ensure that the fixed effect method should be used in this particular analysis, the F-Test for fixed effects will be used as well as the Log Likelihood Ratio Test and the statistics for both tests will be included in the table of results for each equation. 

            The dependant variable is measured as a percentage change in order to allow us to examine the effect that each covariate has on the change in the number of licenses issued as opposed to looking at absolute figures. 

III. Data

            Several data sources are used to compile the appropriate data set for use in the estimated equations.  All observations for all variables are measured at the statewide level (does not include Washington D.C. or any other U.S. provinces) for the years 2002 and 2003.  Summary statistics are included in table 1.

            The number of physicians entering into the labor market is measured in terms of initial licenses issued.  An initial license is issued when a physician without any previous licensure receives the first license of his/her professional career.  This data are obtained from State Medical Licensure Requirements and Statistics.  Their data is taken from the American Medical Association’s Physician Masterfile, which tracks physicians’ entire educational and professional careers.  The data only includes full and unrestricted licenses.  All data that are reported is on a statewide scale.

            The number of physicians that are relocating to a different state will be derived from the State Medical Licensure Requirements and Statistics book.  Two variables are reported in this series of publications; Initial licenses and new licenses.  New licenses are defined as licenses that are issued to physicians for the first time in that particular state.  The difference between New and Initial licenses is that initial licenses are only issued to a physician if it is the first licenses that they will have ever received.  Thus, Initial licenses are a subset of New licenses.  In order to single out only the physicians that have relocated in a given year, the number of initial licenses will be subtracted from the number of new licenses. 

            Medical malpractice rates are taken from an annual survey conducted by the Medical Liability Monitor (MLM).  Every year since 1991 the MLM has conducted an extensive nationwide survey of physician malpractice insurance premiums for policies offering $1 million in coverage for a claim and $3 million in total coverage for a year.  This survey issue breaks down the rates that physicians pay in three specified fields of medicine: Internal Medicine, General Surgery, and Obstetrics and Gynecology.  They report the findings on a countywide level for all states as well as reporting different rates for different insurance companies that service the same geographic area.  All countywide data has been averaged in order to compile a statewide rate.

            All other variables, including unemployment rate, mortality rate, per capita income, and population are measured on a statewide level and are reported by the Census Bureau in their statewide data profiles.  Information is reported per state on a yearly basis.  Of these variables, only the unemployment rate is expected to have a negative relationship with the dependant variable.  All of the other control variables are expected to have a positive influence on the percentage increase in the number of licenses issued to physicians in a given state and year.

IV. Empirical Results

            When the model is estimated using the fixed effects method, the hypothesis that medical malpractice rates have no effect on the number of initial licenses issued cannot be rejected when using equation #1, however, as discussed above, high levels of multicoliniarity are found to exist when a correlation matrix is produced.  This drastically reduces the explanatory power of equation #1.  Due to this multicoliniarity, the parameter estimates for all of the other variables are biased.  Full results for equation #1 can be found on table #2.

            The hypothesis that the rates of Internal Medicine specialists have no effect on the number of initial licenses issued is rejected at the 95% significance level.  For a $1000 increase in the rates of Internal Medicine specialists, the number of initial licenses issued decreases by 6%.   Only two of the control statistics were found to be significant at any level; population, and the unemployment rate.  Both of these variables were found to have a positive relationship with the number of initial licenses issued per state.  The complete results can be viewed on table #3.

            Equations #3 and #4 yielded the same results that the medical malpractice rates of General Surgeons and Obstetrics and Gynecology specialists do not have a statistically significant effect on the percentage change in initial licenses issued in a given state.  Similarly to equation #2, however, the variable for population and the unemployment rate are found to have statistically significant positive effects on the dependant variable.  See tables #4 and #5 for complete results.

            Equation #5 examined the effect that the medical malpractice rates of Internal Medicine specialists have on the percentage change in the number of relocating physicians.  This model was able to explain 57% of the variation in the percentage change in the number of relocating physicians in a given state.  The rates of Internal Medicine specialists are significant at the 95% significance level.  For a $1000 increase in the rates of Internal Medicine specialists, the percentage of licenses issued to relocating physicians decreases by 7%.  Three of the control variables in equation #5 are significant at the 99% level, population, the unemployment rate, and the mortality rate.  Each having a positive effect on the percentage of licenses issued to relocating physicians. See table #6 for complete results.

            The model for equation #6 yielded similar results to Equation #5.  The malpractice rates of General Surgeons are significant at the 95% significance level.  For a $1000 increase in the rates of General Surgeons, the percentage of licenses issued to relocating physicians decreases by 2%.  Similarly to Equation #5, the control variables for population, the unemployment rate, and the mortality rate have a positive relationship with the dependant variable and are significant at the 99% level.  Equation #6 was able to explain 56% of the variation in the percentage change in the number of licenses issued to relocating physicians.  See table #7 for complete results.

            The Estimation of equation #7 reinforces the previous two equations in that the rates of Obstetrics and Gynecology specialist have a negative relationship with the dependant variable and is significant at the 95% confidence level.  For a $1000 increase in the medical malpractice rates of OB/Gyn specialists, the percentage of licenses that are issued to relocating physicians decreases by 1%.  The same three control variables that were significant for equations #5 and #6 remain significant in equation #7 at the 99% level.  For a full table of results of equation #7, refer to table #8.

            One interesting observation is that for all of the equations, the unemployment rate has a positive relationship with the percentage increase or decrease in the number of licenses issued to physicians in one facet or another.  For a 1% increase in the state unemployment rate, the percentage of licenses issued to both new physicians and relocating physicians increases by .56%-.88%.  The reason for this positive relationship is unknown though with more years worth of information this relationship could change in intensity or could possibly become negative.  As expected, the variables for population and mortality rate both had positive and significant relationship with the dependant variable.

            For all equations, there is a large discrepancy between the F-test for Fixed Effects and the Log Likelihood Ratio test.  The two tests continually conflict with one another for this set of equations with the Log Likelihood Ratio test always finding fixed effects to be significant at least in the 90% significance range.  The F-test for Fixed Effects finds for all equations that the fixed effects method is not necessary.  Since there is only two years included in the analysis at this time, this paper assumes that with more years of data, the two tests for fixed effects will both become significant.  Due to this assumption, the fixed effects method will still be carried out at this time in order to offer the most explanatory power.

V. Conclusions and Limitations

            This study was aimed at trying to find a correlation between the statewide medical malpractice rates of physicians and the total number of licenses issued in each state both for newly graduated physicians entering into the labor force for the first time as well as physicians who have relocated from one state to another.  A negative relationship was found between the rates of Internal Medicine specialists and the percentage of initial licenses issued in a given state though the rates of General Surgeons and OB/Gyn specialists were not statistically significant.  However, in the case of relocating physicians, medical malpractice rates of all three of the specified fields were found to have a negative relationship with the dependant variable.  This discrepancy may be due to the fact that newly graduating physicians generally have never operated a business before and due to this they do not think about the costs that are associated with practicing medicine.  Another reason may be because newly graduated physicians do not care where they begin their careers, so long as they are able to start working in the medical profession as soon as possible.  In this latter case, the price of medical malpractice insurance would not have an effect on the decision of where to locate their practice.  It may also be the case that the majority of physicians who attain an initial license may be Internal Medicine specialists.  In this case, the rates of General Surgeons and OB/Gyn specialists are null and void because they do not apply to Internal Medicine specialists.

            These results indicate that higher medical malpractice rates for Internal Medicine specialists do have a negative effect on physicians entering into the labor market for the first time.  However, the results indicate that the rates of other specialties do not have an effect on the growth of the physician labor force.  In contrast, seemingly all medical malpractice rates have a negative influence on the number of physicians that choose to relocate to a different state.  Medical malpractice rates may not have the strongest influence on the number of physicians entering into any particular state, however they do have a significant influence and thus should be considered when drafting legislation that deals with medical malpractice rates.

            These findings contradict previous research in that a negative effect was found and was statistically significant.  Previous studies have concluded that physicians do not close their doors because of rising malpractice rates, aside from marginal entry and exit; however, physicians who have decided to move to a different state are affected by medical malpractice rates.  Previous literature has also failed to look at the two sides of any labor force, being entry and exit.  The overall size of the physician labor force may not be affected by medical malpractice rates though this study has found that the influx of physicians into any particular state is affected by medical malpractice rates. 

            Legislators should use this information to write legislation that will bring more new physicians into their state in hopes of driving end-user health care costs down.  Drafting such legislation may also aide in preventing the practice of “defensive medicine” that plagues the health care industry by running extraneous tests and procedures, which only purpose is to protect the physicians from lawsuits.

            Though the results of this study are somewhat robust, there are some limitations that must be considered.  First, only two years of data was included in the fixed effects model.  The inclusion of at least one more year of data would help to further solidify the estimation of the model.  Second, statewide medical malpractice rates have been averaged from countywide data, which somewhat reduces the accuracy of that particular variable.  If statewide data was collected and reported, the results of the study may be positively influenced.  Future research should aim their studies at finding a relationship between the number of practicing physicians in a particular state and medical malpractice premium caps that many states in the union are starting to enforce on the insurance market.

VI References

American Medical Association, 6-19-06, Medical Liability Reform-NOW!, (Accessed 2-14-07 from http://www.ama-assn.org/ama1/pub/upload/mm/-1/mlrnow.pdf).

American Medical Association, 2004, “State Medical Licensure Requirements and Statistics, 2004.” AMA Press, United States, pp. 65-66.

American Medical Association, 2005, “State Medical Licensure Requirements and Statistics, 2005.” AMA Press, United States, pp. 65-66.

American Medical Association, 2006, “State Medical Licensure Requirements and Statistics, 2006.” AMA Press, United States, pp. 65-66.

Baicker, Katherine., Chandra, Amitabh (2004), “The Effect of Malpractice Liability on the Delivery of Health Care”, NBER Working Paper Series, Working Paper 10709.

Danzon, P., Pauly, M. and Kingston, R. (1990). “The Effects of Litigation on Health Care Costs”, American Economic Review, Volume 80, pp 122-127.

Greenwald, B. and Mueller, M. (1978). “Medical Malpractice and Medical Costs”, The Economics of Medical Malpractice, ed. S. Rottenberg (American Enterprise Institutes, Washington DC), pp. 65-86.

Medical Liability Monitor, “Rate Survey Issue”, Oct. 2002.

Medical Liability Monitor, “Rate Survey Issue”, Oct. 2003.

Medical Liability Monitor, “Rate Survey Issue”, Oct. 2004.

Mello MM, Studdert DM, Brennan TA. “The New Medical Malpractice Crisis”. N Engl J   Med. June 5, 2003; 348 (23): 2281-2284.

Reynolds, R., Rizzo, J. and Gonzalez, M. (1987). “The Cost of Medical Professional Liability”. Journal of Health Economics, Vol. 257. Pp. 2776-2779.

Sloan, F. (1981). “Physician Fees and Length of Visit”. Analysis of Survey Data on Physicians’ Practice Costs and Incomes (Vanderbilt Institute of Policy Studies, Nashville), pp. 79-121.

Thornton, James (1999), “The Impact of Medical Malpractice Insurance Cost on Physicians Behaviour: The Role of Income and Tort Signal Effects”, Applied Economics, Vol. 31, 779-794.

Weycker, D., Jensen, G. (2000). “Medical Malpractice among Physicians: Who will be Sued and Who will Pay?”. Health Care Management Science, Vol. 3, no. 4, pp. 269-277.

U.S. Census Bureau. Data profiles 2002. http://factfinder.census.gov/servlet/DatasetMainPageServlet?_program=ACS&_submenu            Id=datasets_1&_lang=en&_ts= .  Accessed 2-07

U.S. Census Bureau. Data profiles 2003. http://factfinder.census.gov/servlet/DatasetMainPageServlet?_program=ACS&_submenu            Id=datasets_1&_lang=en&_ts= .  Accessed 2-07

U.S. Census Bureau. Data profiles 2004. http://factfinder.census.gov/servlet/DatasetMainPageServlet?_program=ACS&_submenu            Id=datasets_1&_lang=en&_ts= .  Accessed 2-07
Table #1 Simple Statistics

Variable

Definition

Minimum

Maximum

Mean

Standard Deviation

Internal Medicine       (IM)

Total Rate paid per year by physicians that specialize in Internal Medicine (Measured in Dollars).

3055

42272.5

10521.55

6595.68

General Surgery       (GS)

Total Rate paid per year by physicians who specialize in General Surgery (Measured in Dollars).

9301.33

132900.75

36586.33

20441.1

OB/Gyn      (OB/Gyn)

Total Rate paid per year by physicians who specialize in OB/Gyn (Measured in Dollars).

16082

181509.75

56159.29

27751.61

Relocating Physicians     (RI)

Total number of licenses issued to physicians that have relocated to a different state.

73

2288

633.33

478.24

Initial Licenses   (IL)

Total number of licenses issued to doctors for the first time only.

5

2822

405.78

515.26

Unemployment Rate         (URate)

State level Unemployment Rate.

3.3

8.1

5.44

1.03

Mortality Rate per 100,000 people       (Mrate)

State level Mortality Rate.

470.7

1176.9

863.94

128.21

Population      (Pop)

Total number of people in each state.

484833

34650690

5623858.24

6225210.01

Per Capita Income       (PCInc)

Dollar Earnings per person per state

16398

31474

22217.17

3192.92

 

 


Table #2 Results For Equation #1

 

Variable

Parameter Estimate

Corrected Standard Error

Corrected T-Value

IM

-.29

0.12

-2.37

GS

.02

0.02

0.91

OB/Gyn

.04

0.03

1.49

URate

.88***

0.24

3.69

MRate

.007

0.005

1.58

PCInc

-.0002

0.0002

-1.14

Pop

 

R^2

 

Likelihood Ratio Statistic

 

F-Statistic

 

N

.000002***

 

.5430

 

69.353

 

 

 

.858

 

100

0.0000006

 

 

 

 

2.82

 

 

 

 

 

* denotes 90% level of significance, ** represents the 95% level of significance, and *** represents the 99% level of significance.

 


Table #3 Results For Equation #2

 

Variable

Parameter Estimate

Corrected Standard Error

Corrected T-Value

IM

-.06**

0.03

-1.99

URate

.83***

0.24

3.38

MRate

.007

0.005

1.5

PCInc

-.0002

0.0002

-1.17

Pop

 

R^2

 

Likelihood Ratio Statistic

 

F-Statistic

 

N

.000002***

 

.5305

 

67.533

 

 

 

.866

 

100

0.0000006

 

 

 

 

2.73

 

 

 

 

 

* denotes 90% level of significance, ** represents the 95% level of significance, and *** represents the 99% level of significance.

 


Table #4 Results For Equation #3

 

Variable

Parameter Estimate

Corrected Standard Error

Corrected T-Value

GS

-.01

-0.01

-1.12

URate

.81***

0.25

3.29

MRate

.008

0.005

1.57

PCInc

-.0002

0.0002

-1.08

Pop

 

R^2

 

Likelihood Ratio Statistic

 

F-Statistic

 

N

.000001**

 

.5215

 

65.456

 

 

 

.830

 

100

0.0000007

 

 

 

 

2.28

 

 

 

 

 

* denotes 90% level of significance, ** represents the 95% level of significance, and *** represents the 99% level of significance.

 


Table #5 Results For Equation #4

 

Variable

Parameter Estimate

Corrected Standard Error

Corrected T-Value

OB/GYN

-.01

-0.01

-1.42

URate

.82

0.25

3.31

MRate

.007

0.005

1.49

PCInc

-.0002

0.0002

-1.15

Pop

 

R^2

 

Likelihood Ratio Statistic

 

F-Statistic

 

N

.000002

 

.5245

 

65.977

 

 

 

.839

 

100

0.0000007

 

 

 

 

2.4

 

 

 

 

 

* denotes 90% level of significance, ** represents the 95% level of significance, and *** represents the 99% level of significance.

 


Table #6 Results For Equation #5

 

Variable

Parameter Estimate

Corrected Standard Error

Corrected T-Value

IM

-.07**

0.03

-2.1

URate

.58***

0.18

3.26

MRate

.01***

0.003

2.97

PCInc

-.0001

0.0001

-1.01

Pop

 

R^2

 

Likelihood Ratio Statistic

 

F-Statistic

 

N

.000001***

 

.5670

 

76.845

 

 

 

1.038

 

100

0.0000005

 

 

 

 

3.15

 

 

 

 

 

* denotes 90% level of significance, ** represents the 95% level of significance, and *** represents the 99% level of significance.

 


Table #7 Results For Equation #6

 

Variable

Parameter Estimate

Corrected Standard Error

Corrected T-Value

GS

-.02**

0.1

-1.87

URate

.56***

0.18

3.11

MRate

.01***

0.003

3.1

PCInc

-.0001

0.0001

-0.87

Pop

 

R^2

 

Likelihood Ratio Statistic

 

F-Statistic

 

N

.000001***

 

.5643

 

75.437

 

 

 

1.011

 

100

0.0000005

 

 

 

 

2.94

 

 

 

 

*denotes 90% level of significance, ** represents the 95% level of significance, and *** represents the 99% level of significance.

 


Table #8 Results For Equation #7

 

Variable

Parameter Estimate

Corrected Standard Error

Corrected T-Value

OB/Gyn

-.01**

0.007

-1.99

URate

.57***

0.18

3.18

MRate

.01***

0.003

2.9

PCInc

-.0001

0.0001

-1.02

Pop

 

R^2

 

Likelihood Ratio Statistic

 

F-Statistic

 

N

.000001***

 

.5670

 

75.989

 

 

 

1.022

 

100

0.0000005

3.04

 

* denotes 90% level of significance, ** represents the 95% level of significance, and *** represents the 99% level of significance.

 



[1] American Medical Association, 6-19-06, Medical Liability Reform-NOW!, (Accessed 2-14-07 from http://www.ama-assn.org/ama1/pub/upload/mm/-1/mlrnow.pdf)