Abstract
Physician services comprise a substantial share of total health care spending, and the price of health care services has been cited as a key contributor to the disproportionately high rate of health care spending in the US. However, despite a large literature analyzing market power in the hospital and insurance industries, less is known about the extent to which physicians exercise market power. In this study we make use of a private health insurance claims data set to analyze physician market power for two specialties within three mid-sized US metropolitan areas. Using a method developed for hospital competition analysis, we estimate measures of consumer willingness-to-pay for physician practices within each of these markets and relate these to the prices paid to these practices for a set of physician services. Our results are suggestive of the presence of market power in the markets that we analyze. We simulate physician practice mergers for the two largest practices in each market for each specialty analyzed. Results suggest that practice mergers could result in price increases deemed significant by antitrust authorities in some markets but not in others.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
The Herfindahl–Hirschman Index (HHI) is defined as the sum of the squared market shares times 10,000.
ABA (2003) notes distance is a key factor for patient’s choice of physician, with patients preferring to obtain physician services close to where they work or live.
This assumption has implications for patient substitution patterns in that within a conditional logit model framework, the probability ratio of any two alternatives depends only on those two alternatives. This property is referred to as independence from irrelevant alternatives (IIA). Gaynor et al. (2013) note that the use of detailed consumer-level micro data mitigates the IIA property at the level of a market, since the inflexible substitution patterns generated by the IIA assumption will apply only to consumers with identical characteristics. Thus, in our application, we include in a consumer’s utility specification consumer-level characteristics such as the distance from a patient to a provider (which is essentially an interaction of the consumer’s location with the provider location), as well as interactions between observable consumer characteristics and distance.
Consistent with this assumption, we estimate our patient choice model using fee-for-service Medicare patients whose rates are set administratively and mostly have constant coinsurance rates.
As in Capps et al. (2003) we define the set of practices \(M\) as those practices that are located within the metro area analyzed. Most patients within the markets that we analyze receive care within the metro area. Specifically, for cardiology visits, 90.3 % of patients from Market 1 receive care within Market 1, 95.6 % of patients from Market 2 receive care within Market 2 and 95.7 % of patients from Market 3 receive care within Market 3. For ophthalmology visits, 94 % of patients from Market 1 receive care within Market 1, 93.4 % of patients from Market 2 receive care within Market 2 and 95.6 % of patients from Market 3 receive care within Market 3.
For practices with multiple locations, separate practice-location fixed effects are included.
The use of this specification necessitates the inclusion of only those physician visits in which we can match all procedures performed during that visit to a valid procedure code and at least one procedure to a RVU weight. For procedures with modifiers (two digit codes that indicate that a service or procedure has been altered and affect payment of service), we weight these procedures by the appropriate conversion factor given by Medicare.
Medicare RVUs are constructed to reimburse physicians based on the level of time, skill, training, and intensity required of a physician to provide a given service. For each procedure, Medicare assigns a work RVU, which proxies for the level of time, skill, and training to provide a service, a practice expense RVU which addresses the costs of maintaining a practice, and a malpractice RVU which addresses the cost of maintaining professional liability insurance. Payment is calculated based on each of these separate components, multiplied by a geographic practice cost index (GPCI) which accounts for location-specific differences in maintaining a practice. Payment is calculated according to the formula [(work RVU) * (work GPCI) + (Practice Expense RVU) * (Practice Expense GPCI) + (Malpractice RVU) * (Malpractice GPCI)] * [Conversion factor] where the conversion factor is a dollar amount used to convert RVUs to physician payment.
Furthermore, we note that this assumption is consistent with subsequent studies that make us of the Capps et al. (2003) model (Fournier and Gai 2007; Farrell et al. 2011) and that Gaynor et al. (2013) find that mergers estimated using the assumptions of Capps et al. (2003) produce estimates of price increases due to merger that are close to those that would be implied using other economic models of hospital competition.
It is important to note that a limitation of this approach is that it assumes that the WTP-per-quantity for Medicare patients is a reasonable proxy for the WTP-per-quantity unit for private patients. Depending on the source of payment for a patient, physician practices may vary in their treatment intensity and frequency with which they see patients that could generate differences in treatment patterns across insurers (McGuire and Pauly 1991; Frank and Zeckhauser 2007). While the standardization of treatment episodes using Diagnosis Related Groups (DRGs) has served to mitigate this possibility in the context of the measurement of hospital output, because we use RVUs to measure the quantity of services rendered, this quantity measure is subject to potential biases created by variation in physician treatment intensity and practice style.
Procedure codes are classified using the Healthcare Common Procedure Coding System (HCPCS), or the American Medical Association’s Current Procedural Terminology (CPT\(^{\small \circledR }\)) codes.
FAIR Health self-reports that their data include information on 120–150 million covered lives, depending on the participants in each plan and their enrollments during any given year. DaNavas-Walt et al. (2010) report that 172 million non-elderly individuals had private health insurance in 2009, implying that these data cover approximately 70–87% of the privately insured population.
Note that not all insurers in a market necessarily contribute to the FAIR Health database, and some health insurers do not provide the transaction price for all of the claims they submit to FAIR Health (presumably to protect potentially proprietary information). In all of the selected markets, for the specialties that we analyze, this step results in the elimination of between 4 and 29% of the claims in the data, and necessarily requires us to calculate price based only on insurers who contribute these fields for the markets that we study.
The insurer that meets our selection criteria is not necessarily the same insurer in each market. However, in each of the three markets that we examine, the insurers that we use comprise 95–99 % of the usable claims according to our criteria (i.e. claims with valid TINs and transaction prices) for the specialties that we analyze. For confidentially reasons we do not report either the insurer or market names.
The 20 % Medicare random sample uses a sampling methodology that selects unique Medicare beneficiaries based on the last digit of the Social Security Number of the wage earner who is eligible for benefits. If the sample is selected randomly, we should expect to have a representative sample of physician practices, with the patient volume in our sample proportional to that of the volume in the Medicare FFS population. To the extent that this sample systematically excludes smaller practices and these unobserved practices exhibit a weaker relationship between normalized WTP and price than do larger practices, our estimates could potentially overestimate the relationship between normalized WTP and price.
While the master NPI file includes a field for physician specialty, we found that in some cases, physicians who list a given specialty in the NPI file will be coded as a different specialty in the line-item file that we use to define physician visits. Because we are interested in identifying practices that include physicians whose primary specialty is cardiology or ophthalmology, we adopt the approach above which defines specialty based on physician activity in the Medicare population.
A 2012 Survey by SK&A indicates that over 95 % of cardiologist practices accept Medicare. See: (http://www.cardiovascularbusiness.com/topics/healthcare-economics/cardiology-near-top-accepting-medicare) [Accessed December 10, 2013]. Preece (2010) indicates that many ophthalmology practices derive around 50 % of their revenue from Medicare patients, and for some retina procedures, this percentage is significantly higher. These are also the specialties with the highest share of practice income that comes from Medicare (United States Government Accountability Office 2005).
While we are not aware of any work that reports the rate of multiple practice affiliations for cardiologists and ophthalmologists, Nyweide et al. (2009) finds that nearly 20 % of primary care physicians are associated with multiple practices.
In our demand specification, for practices with multiple locations, we treat each location separately from the perspective of the patient, but aggregate our WTP measure at the level of a practice. We also include in our demand specification only practices that treat at least 30 visits during the year in our 20 % sample.
It is important to note that we can include in our model only practices that treat both Medicare patients and that are included in our private-payer data set.
We encounter in our data some procedure codes that could not be matched to the Medicare payment file, and the Medicare files also include codes that do not have associated with them a Medicare payment weight. Because our utility specification requires a measurement of the Medicare payment weight, we drop from our sample visits for which we cannot match all procedure codes to a valid code in the Medicare payment file.
Diagnosis codes for symptoms involving cardiovascular system and chest pain (3-digit codes 785 and 786) were also included.
In a small number of cases, our visit construction algorithm results in the creation of visits with very large and very small RVU quantities. Because of this, we eliminate the top and bottom 1 % of visits as measured by the number of RVUs provided per visit.
For our sample of Medicare patients age 65–75, our sample includes 8,944 cardiology visits for Market 1, 26,801 cardiology visits for Market 2, and 11,740 cardiology visits for Market 3. Our sample includes 8,291 ophthalmology visits for Market 1, 15,568 ophthalmology visits for Market 2, and 6,397 ophthalmology visits for Market 3.
More specifically, using claims in the FAIR Health data, we include for the cardiology price sample those claims where the primary diagnosis field indicates the presence of a cardiac condition (indicated using 3-digit ICD9 codes 390-459) and where the claim includes either an evaluation and management procedure (HCPCS codes 99201-99499) or cardiac procedure (HCPCS codes 33010-37799 and 92950-93799). We include for the ophthalmology price sample those claims whose primary diagnosis field indicates the presence of an eye condition, (3-digit ICD9 codes 360-379) and whose visit includes either an evaluation and management procedure (HCPCS codes 99201-99499) or ophthalmologic procedure (HCPCS codes 92002-92499). In order to be consistent with the criteria used to construct our Medicare sample, the claims included when calculating price are those in which the claims can be identified as having been performed in a setting classified by Medicare as a “non-facility setting.”
We find substantial overlap in each market across both datasets. Specifically, we find that of the cardiology practices that we identify in the Medicare data, over 90 % appear in the private claims data, accounting for 84–94 % of the claims in the data and 74–94 % of the economic activity in the data (as measured by gross charges for claims in which cardiologists provided services). Of the ophthalmology practices that we identify in the Medicare data, over 90 % appear in the private claims data, and these practices account for 94–97 % of the claims and 94–97 % of the economic activity in the private claims data (as measured by gross charges for claims in which ophthalmologists provide services). Across both specialties, the practices that we analyze account for 92–99 % of the economic activity in the Medicare data (as measured by Medicare allowed amounts).
Because physicians are typically reimbursed based on the quantity of services that they provide, we do not include additional controls for disease severity in our price regressions, as this should not directly affect the price per RVU negotiated by a physician practice. In a survey of 33 health plans, Dyckman and Hess (2003) note that negotiated fee schedules for all of the plans they study are influenced by the Medicare RVU methodology, with deviations from this methodology involving mainly different conversion factors assigned to specific code ranges, as well as the frequency with which the fee schedules are updated. We also include in this specification controls for plan type (e.g. HMO, PPO, POS).
This allows the average markup for specialty codes to differ from non-specialized procedures. For example, when estimating price for cardiology practices, our indicator variable is equal to one for HCPCS codes 33010-37799 and 92950-93799. The inclusion of this indicator is consistent with the findings of Dyckman and Hess (2003) who show that markups can differ based on procedure code ranges.
Gowrisankaran et al. (2013) note that an alternative method of constructing prices would be to directly use the contracts between physicians and insurers. Since we do not observe these contracts, we use the FAIR Health claims data to formulate the price measures as described above. The results of our price regressions are presented in Table 9 in the Appendix.
For ease of exposition in Table 1, our calculations exclude claims with modifiers.
As noted in Tables 2 and 3, HHIs are calculated using the payment-weighted Medicare visits and practices used in our analysis sample. Because a small number of practices identified in the Medicare data do not appear in the FH data, we also calculated these values using all identified Medicare practices, as well as weighting by gross charges (rather than Medicare payments). Regardless of which of these measures we used, for both specialties, the HHI is consistently lowest in Market 2.
Across all markets, the physician practice sizes are calculated using physicians that practice in both facility (e.g. hospitals, outpatient hospitals) and non-facility settings.
Our study makes use of the visits detailed in row 4 of each panel.
Because the preferences of elderly patients may differ from that of the younger, privately insured population, we explore the likely effects of this assumption on our results by estimating our specification using younger Medicare patients, aged 65–75 in Table 7.
If we examine the percent of observations that are assigned either the highest predicted probability or second highest predicted probability, the models predict the actual choice for 58, 48, and 42 % of cardiology patients in Markets 1, 2 and 3 respectively, while respectively predicting actual choice for 45, 35, and 46 % of ophthalmology patients in Markets 1, 2, and 3.
Our results that make use of young Medicare patients are estimated using fewer practices than those that make use of the full sample due to our application of the criteria specified in “Estimation” section, namely the exclusion of practices with fewer than 30 young Medicare patients.
A limitation of this approach in this study is that merger effects are using an estimated WTP measure which itself has a standard error. Thus the confidence intervals associated with our estimates of \(\hat{\beta }\) likely represent a lower bound.
The Market 1 mergers would result in a practice with 41 % of the cardiology visits and 21 % of the ophthalmology visits. The Market 2 mergers would result in a practice with 21 % of the cardiology visits and 18 % of the ophthalmology visits.
While these hypothetical mergers are large, the mergers that we consider result in values for normalized WTP and price that are still mostly within our estimation sample. Specifically, for Market 1, the mergers that we consider result in post-merger values of normalized WTP that fall at approximately the 23rd and 90th percentile of the distribution for ophthalmology and cardiology respectively, and values of price with values at the 66th and 92nd percentile for ophthalmology and cardiology respectively. For Market 2, the mergers examined result in values of normalized WTP that are at 39th and 89th percentile of the distribution for ophthalmology and cardiology respectively, and values of price that are at the 37th and 89th percentile for ophthalmology and cardiology respectively. For Market 3, the resultant ophthalmology merger has a value of normalized WTP that is at the 48th percentile, while for cardiology, the resultant merger has a value of normalized WTP that is 4 % larger than the maximum value that is observed in the analysis sample. For this same market, the mergers that we analyze result in prices that are at the 22nd and 90th percentile for ophthalmology and cardiology respectively.
As noted by Brand (2013), \(\delta \) can be conceptualized as the ratio of two parameters that are separately identified in Gowrisankaran et al. (2013), the first of which is the price sensitivity parameter denoted in Eq. (6) by the term \(\gamma \), and an insurer preference parameter that scales consumer surplus into value for the insurer. We are unable to identify these parameters in our setting due to the absence of variation in the coinsurance rate for patients in our sample.
We note that this framework ignores diversion from physician group \(k\) to physician group \(j\in M\backslash k\) and will only strictly hold if \(\sum \nolimits _{j\in M\backslash k} q_{j(k)} p_j =\sum \nolimits _{j\in M} q_j p_j \) where \(q_{j(k)}\) corresponds to demand at physician practice \(j\) if practice \(k\) is removed from the network (i.e. the cost of treating patients in network \(M\) is identical to the price of treating patients in network \(M/j\)). In this case the incentive to add a physician practice to a network is generated only through the incremental value generated by the addition of this practice, and not through any cost advantages. Farrell et al. (2011) note that models that ignore these diversions limit the focus of the analysis to the first-order price effects that result directly from the potential elimination of competition between the merging firms, and therefore generally produce smaller merger effects.
References
Akosa Antwi, Y., Gaynor, M. S., & Vogt, W. B. (2009). Evaluating the Performance of Merger Simulation: Evidence from the Hospital Market in California, Carnegie Mellon working paper. Pittsburgh, PA: Carnegie Mellon University.
American Bar Association. (2003). Health care mergers and acquisitions handbook. Chicago: ABA Publishing.
American Medical Association. (2011). Competition in Health Insurance: A Comprehensive Study of US Markets: 2011 Update. Chicago: American Medical Association.
American Medical Association. (2012). Competition in Health Insurance: A Comprehensive Study of US Markets: 2012 Update. Chicago: American Medical Association.
Baker, L., Bundorf, K., Royalty, A., & Levin, Z. (2014). Physician practice competition and prices paid by private insurers for office visits. Journal of the American Medical Association, 312(16), 1653–1662.
Balan, D. J., & Brand, K. J. (2009). Simulating hospital merger simulations. Retrieved December 26, 2013, from http://works.bepress.com/cgi/viewcontent.cgi?article=1006&context=david_balan.
Boukus, E., Alwyn, C., & O’Malley, A. S. (2009). A snapshot of US physicians: key findings from the 2008 Health Tracking Physician Survey. Data bulletin (Center for Studying Health System Change), 35, 1–11.
Brand, K. (2013). Price equilibrium in empirical models of hospital competition. Unpublished Manuscript.
Capps, C. S. (May 4, 2012). Economic Analysis of Hospital Mergers in the 21st Century, A New Economic Toolkit for Assessing Hospital Mergers, Antitrust in Healthcare Conference, ABA and American Health Lawyers Association. Retrieved August 8, 2012 from http://html.documation.com/cds/health12/support/pdfs/12-1.pdf.
Capps, C., Dranove, D., & Satterthwaite, M. (2003). Competition and market power in option demand markets. RAND Journal of Economics, 34, 737–763.
Capps, C., Dranove, D., & Lindrooth, R. C. (2010). Hospital closure and economic efficiency. Journal of Health Economics, 29(1), 87–109.
Casalino, L. P. (2006). The Federal Trade Commission, clinical integration, and the organization of physician practice. Journal of Health Politics Policy and Law, 31(3), 569–585.
Clemens, J., & Gottlieb, J. D. (2013). Bargaining in the Shadow of a Giant: Medicare’s Influence on Private Payment Systems. NBER Working Paper No. w19503. National Bureau of Economic Research.
CMS. (2010). Fee-For-Service Medicare Quality And Resource Use Report. Retrieved April 4, 2012 from http://www.google.com/url?sa=t&rct=j&q=fee-for-service%20medicare%20quality%20and%20resource%20use%20report&source=web&cd=1&ved=0CC0QFjAA&url=http%3A%2F%2Fwww.cms.gov%2FPhysicianFeedbackProgram%2FDownloads%2FQRUR_Physicians.pdf&ei=AQh9T-3ONYX88gSVkdmSDQ&usg=AFQjCNF__ih5pCxU7CMw209Tg_0c1WUuYg&cad=rja.
CMS. (2011). Physician Group Practice Transition Demonstration Design Overview. July 14. Retrieved April 4, 2012 from https://www.cms.gov/Medicare/Demonstration-Projects/DemoProjectsEvalRpts/downloads//PGP_Transition_Design_Summary.pdf.
Dafny, L. S. (2010). Are health insurance markets competitive? The American Economic Review, 100, 1399–1431.
Dafny, L., Duggan, M., & Ramanarayanan, S. (2012). Paying a premium on your premium? Consolidation in the US Health Insurance Industry. The American Economic Review, 102(2), 1161–1185.
Davis, K., & Carper, K. (2012). Use and Expenses for Office-Based Physician Visits by Specialty, 2009: Estimates for the U.S. Civilian Noninstitutionalized Population. Statistical Brief #381. August 2012. Agency for Healthcare Research and Quality, Rockville, MD. Retrieved June 2, 2014 from http://www.meps.ahrq.gov/mepsweb/data_files/publications/st381/stat381.pdf.
DaNavas-Walt, C., Proctor, B. D., & Smith, J. C. (September 2010). Income, Poverty, and Health Insurance Coverage: 2009. Current Population Reports, Consumer Income, P60–238. Washington, DC: US Census Bureau.
Dunn, A. & Shapiro, A. H. (April 2012). Physician Market Power and Medical-Care Expenditures, Working Paper No. 84, Bureau of Economic Analysis.
Dyckman, Z., & Hess, P. (2003). Survey of Health Plans Concerning Physician Fees and Payment Methodology. Prepared for Medicare Payment Advisory Committee. Washington, DC: Dyckman & Associates.
Elzinga, K. G., & Swisher, A. W. (2011). Limits of the Elzinga–Hogarty Test in hospital mergers: The Evanston case. International Journal of the Economics of Business, 18(1), 133–146.
Ericson, K. M. M., & Starc, A. (2012). Pricing Regulation and Imperfect Competition on the Massachusetts Health Insurance Exchange. NBER Working Paper No. w18089. National Bureau of Economic Research.
Farrell, J., Balan, D. J., Brand, K., & Wendling, B. W. (2011). Economics at the FTC: Hospital mergers, authorized generic drugs, and consumer credit markets. Review of Industrial Organization, 39(4), 271–296.
Federal Trade Commission & Department of Justice. (2011). Statement of antitrust enforcement policy regarding accountable care organizations participating in the Medicare shared savings program. Fed Regist, 76(209), 67026–67032.
Fournier, G. M., & Gai, Y. (2007). What Does Willingness-to-Pay Reveal about Hospital Market Power in Merger Cases? (April, 2007). iHEA 2007 6th World Congress: Explorations in Health Economics Paper. Retrieved Accessed 12 August, 2011 from SSRN: http://ssrn.com/abstract=993213.
Frank, R. G., & Zeckhauser, R. J. (2007). Custom-made versus ready-to-wear treatments: Behavioral propensities in physicians’ choices. Journal of Health Economics, 26(6), 1101–1127.
Gaynor, M., & Vogt, W. B. (2000). Antitrust and competition in health care markets. In A. J. Culyer & J. P. Newhouse (Eds.), Handbook of health economics, Ch.27 (Vol. 1B). New York: North-Holland.
Gaynor, M., & Vogt, W. B. (2003). Competition among hospitals. RAND Journal of Economics, 34, 764–785.
Gaynor, M., & Town, R. (2012). Provider competition. In M. Pauly, T. McGuire, & P. Barros (Eds.), Handbook of health economics (Vol. 2, pp. 499–637). New York: Elsevier.
Gaynor, M. S., Kleiner, S. A., & Vogt, W. B. (2013). A structural approach to market definition with an application to the hospital industry. The Journal of Industrial Economics, 61(2), 243–289.
Gerardi, K. S., & Shapiro, A. H. (2009). Does competition reduce price dispersion? New evidence from the airline industry. Journal of Political Economy, 117(1), 1–37.
Ginsburg, P. B. (2010). Wide Variation in Hospital and Physician Payment Rates Evidence of Provider Market Power. Research Brief No. 16. Washington, DC: Center for Studying Health System Change.
Gowrisankaran, G., Nevo, A., & Town, R. (2013). Mergers When Prices Are Negotiated: Evidence from the Hospital Industry. No. w18875. National Bureau of Economic Research.
Hartman, M., Martin, A. B., Benson, J., & Catlin, A. (2013). National Health Spending In 2011: Overall growth remains low, but some payers and services show signs of acceleration. Health Affairs, 32(1), 87–99.
Health Care Cost Institute. (2012). Health Care Cost and Utilization Report: 2010. Retrieved December 25, 2013 from http://www.healthcostinstitute.org/files/HCCI_HCCUR2010.pdf.
Ho, K. (2009). Insurer–provider networks in the medical care market. The American Economic Review, 1, 393–430.
Ho, K., & Lee, R. S. (2013). Insurer Competition and Negotiated Hospital Prices. NBER Working Paper, w19401.
Iglehart, J. K. (2011). Assessing an ACO prototype–Medicare’s physician group practice demonstration. New England Journal of Medicine, 364(3), 198–200.
Irving Levin Associates Inc. (2007). Deals and Dealmakers: The Health Care M&A Year in Review. Norwalk, CT, 12th Edn. Retrieved July 16, 2010 from http://www.levinassociates.com/compallconfirm?sid=6050.
Irving Levin Associates Inc. (2008). Deals and Dealmakers: The Health Care M&A Year in Review. Norwalk, CT, 13th Edn. Retrieved July 16, 2010 from http://www.levinassociates.com/compallconfirm?sid=6050.
Irving Levin Associates Inc. (2009). Deals and Dealmakers: The Health Care M&A Year in Review. Norwalk, CT, 14th Edn. Retrieved July 16, 2010 from http://www.levinassociates.com/compallconfirm?sid=6050.
Irving Levin Associates Inc. (2010). Deals and Dealmakers: The Health Care M&A Year in Review. Norwalk, CT, 15th Edn. Retrieved July 16, 2010 from http://www.levinassociates.com/compallconfirm?sid=6050.
Kaiser Family Foundation. (2005). Trends and Indicators in the Changing Health Care Marketplace, publication number 7031. Retrieved July 16, 2010 from http://www.kff.org/insurance/7031/print-sec5.cfm.
Kleiner, S. A., Lyons, S., & White, W. D. (2012). Provider concentration in markets for physician services for patients with traditional medicare. Health Management, Policy and Innovation, 1(1), 3–18.
Krishnan, R. (2001). Market restructuring and pricing in the hospital industry. Journal of Health Economics, 20(2), 213–237.
Laugesen, M. J., & Glied, S. A. (2011). Higher fees paid to US physicians drive higher spending for physician services compared to other countries. Health Affairs, 30(9), 1647–1656.
Lewis, M. S., & Pflum, K. E. (2013). Diagnosing Hospital System Bargaining Power In Managed Care Networks. Working Paper.
McGuire, T. G., & Pauly, M. V. (1991). Physician response to fee changes with multiple payers. Journal of Health Economics, 10(4), 385–410.
Nicholson, S. (2012). Research opportunities of a new private health insurance claims data set. Health Management, Policy and Innovation, 1(1), 37–41.
Nyweide, D. J., Weeks, W. B., Gottlieb, D. J., Casalino, L. P., & Fisher, E. S. (2009). Relationship of primary care physicians’ patient caseload with measurement of quality and cost performance. JAMA, 302(22), 2444–2450.
Pham, H. H., Schrag, D., O’Malley, A. S., Wu, B., & Bach, P. B. (2007). Care patterns in medicare and their implications for pay for performance. New England Journal of Medicine, 356(11), 1130–1139.
Preece, D. (2010). Six tips for surviving a Medicare fee cut. Retrieved June 2, 2014 from http://www.aao.org/yo/newsletter/201010/article01.cfm.
Schneider, J., Li, P., Klepser, D., Peterson, N. A., Brown, T., & Scheffler, R. (2008). The effect of physician and health plan market concentration on prices in commercial health insurance markets. International Journal of Health Care Finance and Economics, 8, 13–26.
Starc, Amanda. (2010). Insurer pricing and consumer welfare: Evidence from medigap. Manuscript. Harvard University.
Town, R., & Vistnes, G. (2001). Hospital competition in HMO networks. Journal of Health Economics, 20(5), 733–753.
United States Government Accountability Office (GAO). (March 2005). Medicare Physician Fees: Geographic Adjustment Indices are Valid in Design, but Data and Methods Need Refinement’. Retrieved October 27, 2014 from http://www.gao.gov/products/GAO-05-119.
Vita, M. G., & Sacher, S. (2001). The competitive effects of not-for-profit hospital mergers: A case study. Journal of Industrial Economics, 49(1), 63–84.
Wilensky, G. R. (2011). Lessons from the Physician Group Practice Demonstration—a sobering reflection. New England Journal of Medicine, 365(18), 1659–1661.
Acknowledgments
We are grateful to Keith Brand, Cory Capps, Matthew Eisenberg, Martin Gaynor, Katherine Hempstead, Sean Nicholson, Carol Simon, Amanda Starc, Robert Town, two anonymous reviewers, and seminar participants at the American Economic Association Annual Meetings and the American Society of Health Economists Conference for helpful comments. Daniel Ludwinski provided excellent research assistance. We gratefully acknowledge support from the Robert Wood Johnson Foundation and FAIR Health Inc. FAIR Health, Inc. is not responsible for the research conducted, nor for the opinions expressed in this article
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
Bargaining between physician groups and insurers
Following Farrell et al. (2011) and Brand (2013), and defining (as in Eq. (4)) the Willingness-to-Pay for physician practice \(k\) as the incremental utility value of physician practice \(k\) to a given network, \(M\), as
we characterize an insurer’s expected pay-off for reaching an agreement to include all practices in their network as
where \(\delta \) is defined as a payoff parameter for an insurer, \(q_j \) is the quantity of care at physician practice \(j \)and \(p_j \) is the price of care at practice \(j\).Footnote 42 Using this notation, an insurer’s expected payoff for failing to reach an agreement with practice \(k, \)but reaching an agreement with the other \(M-1\) practices can be represented asFootnote 43:
where \(q_{j(k)} \) denotes the expected volume of the insurers enrollees at physician group \(j\) if the insurer fails to reach an agreement with physician group \(k\). Assuming a zero disagreement pay-off to a physician group if no agreement is reached, the objective function in a Nash bargaining game for each physician group-insurer combination can be written as
where \(c_k\) is firm \(k\)’s cost. Maximizing this expression with respect to \(p_k \) yields
that is estimable using the equation given in (7) for a fixed value of \(b_k \). Given the terms in Eq. (15), in equilibrium, the practice receives a fraction \(b_k \) of the equilibrium per-patient incremental value that the physician practice brings to the network. Note that given the context of this model, for an insurer that has considerable bargaining power (i.e. \(b_k \rightarrow 0\)) price will approach marginal cost whereas if a physician practice has full leverage over the insurer (i.e. \(b_k \rightarrow 1\)) this practice will fully capture its incremental value to the network in the form of higher prices.
We are careful to note a number of limitations to this model. First, as noted by Gaynor and Town (2012) and Farrell et al. (2011) this relationship is best conceptualized as an approximation to a bargaining game, where the \(\textit{WTP}_{k}\) measure serves as a proxy to the incremental gross payoff to the insurer generated by the inclusion of physician practice \(k\) by its inclusion in a network. Secondly, the residuals from this regression capture all factors that affect price but are not included in the observables which include bargaining ability and unobserved costs. Finally, the use of this framework does not account for spillover effects to other practices due to merger and hence using this approach is likely to underestimate merger effects.
Rights and permissions
About this article
Cite this article
Kleiner, S.A., White, W.D. & Lyons, S. Market power and provider consolidation in physician markets. Int J Health Econ Manag. 15, 99–126 (2015). https://doi.org/10.1007/s10754-014-9160-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10754-014-9160-y