Physician Compensation

Physician Compensation in Canada

Does physician compensation affect the quantity of medical care provided?  My paper “Operating on Commission” claims that the answer is yes.  I find that surgery rates increase 78% when patients switch from capitation to fee-for-service (FFS) specialists.

A paper by Devlin and Sarma (2008) examines a similar question for Canadian family physicians.  Since the inception of Canadian Medicare, 89% of family physicians have been paid on a fee-for-service basis.  The authors aim to estimate how fee-for-service compensation affects the quantity of medical care controlling for the fact that physicians who favor more aggressive treatments likely will sort into fee-for-service compensation schemes.  

The authors control for the endogeneity problem with using 2 econometric specifications.  The first uses an instrumental variables (IV) specification with 4 instruments.  The first three instruments are physician preferences for research, teaching, and non-work interests.  Physicians who enjoy teaching and research are more likely to prefer salaried compensation schemes.  The final instrument is the physician’s response to the compensation scheme they prefer.  Each of these four instruments is likely correlated with the actual way the physician is compensated, but the instrument must be uncorrelated with unobserved factors which effect physician quality.  

The second econometric specification is the treatment effects estimator (the restricted control function approach).  The treatment effects estimator assumes the following econometric structure:

  • ln(qi) = Xiα + βRi + εi
  • ΔVi = Ziγ + ui

The first equation shows how the physician remuneration scheme (Ri) affects the log quantity of medical care [ln(qi)] after controlling for covariates (Xi).  The second equation gives physician’s latent utility (ΔVi) of choosing one remuneration scheme over another.  The treatment effects specification estimates the coefficients based on functional form; the first equation is estimated with OLS and the second equation is estimated with a probit model.

With these two specifications, the authors find strong evidence that physicians select into different compensation schemes based on their practice styles.  “…those who choose a non-FFS environment engaged in more patient visits per week than those who choose the FFS scheme.”  After controlling for the selection effect, the authors found that that the direct incentive effect of physician compensation was strong.  “FFS schemes appear to strongly encourage physicians to see many more patients relative to alternative remuneration schemes.”

As I find in my “Operation on Commission” paper, financial incentives do matter.