Ownership and rent-seeking behavior in specialty health care practices. Abstract

Transcription

1 Ownershi and rent-seeking behavior in secialty health care ractices Dan Friesner onzaga University Chris Stevens Ohio University, Eastern Camus Abstract Secialty health care ractices are unique in that they ehibit a wide range o ownershi tyes, rom large cororations controlled by third arties to those directly owned by ractitioners hysicians, theraists, etc.. Many o these ractices also emloy licensed assistants whose labor is artially substitutable with those o the ractitioners. This aer resents a theoretical model that eamines the imact that dierent levels o ownershi have on rent-seeking behavior and eiciency within secialty ractices. Our rimary ocus is on whether lower levels o ownershi induce ractitioners to etract larger economic rents by substituting their services or those o their assistants. e ind that i the ractitioners are not required to be technically eicient then they unambiguously resond to lower ownershi with rent-seeking. owever, requiring the irm to be technically but not allocatively eicient, may be suicient to mitigate this incentive. e thank John Conley, an anonymous reviewer, Robert Rosenman and Rod Fort or roviding a number o helul comments that imroved the aer. Any remaining errors are our own. Citation: Friesner, Dan and Chris Stevens, 007 "Ownershi and rent-seeking behavior in secialty health care ractices." Economics Bulletin, Vol. 9, No Submitted: November 6, 006. Acceted: June, 007. URL: htt://economicsbulletin.vanderbilt.edu/007/volume9/eb-06i000a.d

2 . Introduction Rent-seeking is a common activity or economic agents. hen irms oerate in imerectly cometitive markets, the owners and/or managers o these irms can make oerating decisions that redistribute wealth rom other members o society to themselves. This roduces allocative ineiciency and reduces social welare below the otimal level. The health care industry is no ecetion to this rule. For eamle, a recent study by Town et al. 006 estimated that between 990 and 00, over $40 billion in consumer surlus was transerred rom consumers to roducers solely rom mergers in the US hosital industry. iven that hositals are only one albeit a major subset o the health care industry, and that other roducers may also ractice rent reeking, the social welare imlications o rent seeking in health care are highly signiicant. A crucial issue is to whom these otential rents accrue. Firms oerating in the health care industry are unique in that they have a wide array o ownershi orms. For eamle, a irm may be comletely controlled by an outside, third arty, or it may be at least artially owned by a grou o individuals sulying a crucial services such as insurance, inatient hositals services and/or hysician services within the roduction rocess. Since outside owners and medical ersonnel have dierent objectives, it stands to reason that the incentive to etract rents, as well as the allocation o those etracted rents, may dier deending on who owns the irm. The literature on rent seeking in health care has been emirical, ocused on hysician behavior in general, acute care hositals articularly those that own, or are owned by, a MO and has attemted to identiy whether various ownershi structures lead to dierences in rent seeking and eiciency. The elanation osited by most o these studies is that certain ownershi tyes have a vested interest in roviding a articular tye o care, and unnecessarily substitute that service or other tyes o medical care whenever ossible in order to increase the monetary gain to the individual or grou roviding that service. For eamle, Ahern et al. 996 ound that hosital-owned MOs were more likely to over-utilize inatient services than other tyes o MOs. In doing so, the hosital could garner a larger ortion o the available net revenue er atient. Unortunately, by over-utilizing inatient services to cature these rents, hosital-owned MOs were necessarily less eicient than other MOs. Other studies have ound similar results or a number o dierent hositals tyes, including MO-ailiated Siddharthan et al. 997; Rosenman et al. 997; illman et al. 999; Siddharthan et al. 000 and non-mo-ailiated hositals Kuntz and Vera 003. Two recent trends have sarked renewed interest in rent seeking activity in health care. The irst is the raid rolieration o secialty care hositals and medical ractices eneral Accounting Oice 003; Nallamothu et al These irms dier rom general, acute care hositals in that they rovide only a small range o services or eamle, coronary care, hysical theray or inatient sychiatric services, are more likely to be or-roit, and also ocus on a narrower range o atients, usually those covered by insurance olicies that reimburse generously or the services being erormed. Moreover, these irms also have disroortionately higher degrees o ractitioner-ownershi, which may increase the otential or rent seeking activity Casalino et al The second trend is an increase in the use o variable inuts that ehibit a high degree o substitutability with a traditional ractitioner s labor. For eamle, many irms articularly To the best o our knowledge, the only theoretical treatment o rent seeking in health care was conducted by Ahern et al. 996, which rovides only a simle, grahical analysis o how vertical integration allows or transer-ricing and rent-seeking within an organization.

3 those oerating in the US and other industrialized nations not only emloy traditional medical ractitioners such as hysicians, harmacists and theraists, but also licensed assistants such as hysician assistants, amily nurse ractitioners, harmacy technicians, and hysical and/or occuational theray assistants who are legally qualiied to erorm many o the same tasks as the ractitioner under whom they work. There is currently much debate in the medical literature about the role these licensed assistants should lay in the delivery o medical care Riortella- Muller et al. 995; Banham and Connelly 00. hile some degree o substitutability is certainly desirable, issues o medical ethics, the quality o care rovided to atients assuming atients, on average, eerience a higher real and/or erceived outcome when treated by the ractitioner and not the assistant and the increased otential or rent seeking activity may lead olicy makers to review and ossibly regulate the etent to which these licensed assistants are used, articularly in secialty care settings. To date, there has been little theoretical elanation about the imact o ownershi on rent-seeking within secialty care ractices that utilize both ractitioner and licensed assistant labor. Moreover, there has been no theoretical eloration o how the degree o irm ownershi imacts and otentially mitigates the incentive to rent-seek. This aer resents a theoretical model that eamines the imact that dierent levels o ractitioner ownershi have on rentseeking behavior and eiciency within secialty ractices. Our rimary ocus is on whether lower levels o ownershi induce ractitioners to etract larger economic rents by substituting their services or those o their assistants. In doing so, we rovide some olicy rescritions which can be used to curb rent seeking behavior. To the etent that the substitution o ractitioner or assistant labor maniests itsel in the quality o care whether real or erceived, our analysis also allows us to theoretically characterize the relationshi that eists between rent seeking, eiciency and the quality o medical care. Thus, our analysis rovides a theoretical rationale or why and how quality discrimination and cost adjusting might occur Dor and Farley 996; Friesner and Rosenman 004; Rosenman, Friesner and Stevens A Simle Model o 3 rd Party Ownershi To establish a useul benchmark, we begin by modeling the third arty ownershi o a tyical, or-roit secialty ractice. These owners do not suly actors o roduction, but instead urchase a quantity o ractitioner labor and a quantity o assistant labor at rices and, resectively. e note in assing that i inut markets are erectly cometitive then the inut suly unctions and are eogenously determined constants. The inuts are transormed into a single outut which we will deine as secialty health care services using a continuously at least twice-dierentiable roduction unction Q =,. e assume that satisies all usual regularity conditions. The irm has the ability to set rices in the outut market, where the rice is given by P= Q. e urther assume that Q is at Clearly, the roduction o health services, even in secialty ractices, is more comlicated than the rocess described above. Not only do these irms utilize ied inuts such as oice sace and major equiment, but also other variable inuts, including but not limited to utilities and sulies. owever, because ied inuts are not chosen by the irm and do not qualitatively imact the irst order conditions o the model, we suress these variables in the interests o arsimony. Similarly, incororating additional variable inuts will not change the signs o our comarative statics, and thus any olicy imlications derived rom out model. Lastly, in the case where a secialty irm rovides a wider array o services, one can interret our outut measure as a weighted average o these dierent services, where the weights are deined by the severity o the illness being treated or eamle, in terms o RBRVS units or case mi-adjusted admissions.

4 least twice-dierentiable, real-valued, and inite unction or all non-negative values o Q. The irst derivative o P must also be non-ositive or all allowable values or Q. 3 Based on these assumtions, the irm s roit unction can be eressed as: R, where R, is the irm s revenue unction, which can be reresented as the roduct o the inverse demand and roduction unctions: R,,, The third-arty owners are assumed to choose and to maimize roit. Substituting into, taking artial derivatives with resect to and, and setting these eressions equal to zero gives the necessary irst order conditions: i where 0 ; 0 ; and 0 or i =, i i The irst term in equations 3 and 4 is the marginal revenue roduct o inut i, or i =, while the second and third terms jointly reresent the marginal actor cost or the ractice. One can rewrite equation 3 and 4 in terms o the inut s marginal roduct: 3a 4a Let * and * denote the otimum demand or and derived rom equations 3a and 4a. Then assuming diminishing marginal returns, the smaller the value on the right-hand side o each equation, the larger the value o i * or i =,. 3. The Case o Constrained Practitioner Ownershi One interesting etension o the above roblem is the case where one o the inut suliers namely a grou o ractitioners is a artial owner o the irm. Because the ractitioners now own art o the irm, they not only receive a share o its roits, but they also have the ability to determine inut usage. Moreover, as the ractitioners alter the use o their own services, they move along the marginal revenue roduct curve or ractitioner services, thereby altering their wage. This last oint is imortant, because it imlicitly assumes that the ractitioners have the ability to set the level o use or their inut or the rice charged or their inut based on the demand curve or those services. Thus, regardless o the degree o ownershi, the ractitioners are constrained in that they must choose a technically, but not necessarily allocatively eicient level o resource usage or the irm. e deine unconstrained control as the case where the 3 In a ew cases, secialty care ractices may derive the entirety o their revenues by treating atients covered by an insurance lan such as Medicare that reimburses solely on a ied ee or service basis. In that case, the irm cannot directly control the rice it sets or its services. This secial case can easily be incororated into our model without loss o generality by restricting P to a constant. 3

5 4 ractitioners have the ability to set rices based o o the suly curve or, and thus can induce both allocative and technical ineiciency. This model is discussed in section 4. Assuming that all o the irm s ractitioners act as a cohesive unit, the ractitioners total return unction consists o a share o the roit lus inra-marginal rents. Deining this unction as, these returns are given by:,,, 0, d 5 where is the eogenously determined share o hysician ownershi in the irm which is normalized to the 0,] interval and is the suly unction or the ractitioners to rovide their services. 4 In equation 5, the irst bracket is simly the total amount o irm roit earned by the ractitioners. The second bracket reresents the inra-marginal rents the ractitioners are able to etract or their services. Unlike marginal actor costs, the marginal revenue roduct will be a unction o other inuts, since we assume that changes in the level o one o the inuts aect the marginal roducts o the other inuts; that is, the roduction unction is not searable. Under these conditions, the ractitioners will maimize their gain by setting the necessary irst order conditions equal to zero: Since the wage is now determined by the irm s demand curve, these equations do not match u with the third arty ownershi case. owever, to allow or a meaningul comarison across models, it is useul to eress 6 and 7 in terms o their marginal roducts: 6a 7a In section 5, we use comarative static analysis to determine how a small change in the share ractitioner ownershi aects inut usage, and consequently rent-seeking activity. owever, at this oint it is interesting to consider the two discrete cases: where is either one or less than one. e begin with the case where equals one; that is, where the ractitioners receive all o the irm s roit. In this case 6a and 7a simliy to: 4 Note that actually reresents the ractitioners oortunity costs. As such, there is no reason to consider the ortion o the curve above the ractitioners actual oortunity cost since the irm will decide inut usage based uon their demand curve, the.

6 ; 6b, 7b Equation 6b, when comared to 3a indicates that the irm will use more o, or ractitioner services, than in the case o third arty ownershi. Since the numerator o 6b is smaller than that o 3a, the otimal value o the marginal roduct must be smaller. iven diminishing returns, the otimal level o must also be higher. In the third arty case, the irm has monosony ower. As a result, the marginal revenue roduct o ractitioner services is greater than the oortunity cost or ractitioners at that oint. Since the ractitioners can act as erect wage discriminators or themselves, they will increase roit, as well as their rents, by increasing the use o until its marginal revenue roduct is equal to the oortunity cost or ractitioners. Equations 7b and 4a are eactly the same in orm. owever, since the otimal level o has changed, the values or 7b will likely be dierent than those o 4a. As a result, whether the otimal level o is higher or lower under third arty ownershi is an emirical issue. To the etent that greater net use o ractitioners enhances the quality o care whether real or erceived, these indings imly that ull ractitioner ownershi, in the majority o cases, leads to both more rent seeking and higher quality o care. The ractitioners unambiguously use more o their own services, and may or may not use ewer licensed assistants. I both inuts are increased, quality is made unambiguously better o. Quality o care is also increased, albeit to a lesser etent, i licensed assistant use is decreased, but the decrease is disroortionately smaller than the increase in ractitioner usage. It is only when the otential decrease in licensed assistants outweighs the increase in ractitioner use that there is a concern about the quality o care oered to atients. Now consider the case where <. ere the ractitioners may want to charge a higher rice or their services than when they ully own the irm. Although the amount o will decrease, as will the irm s roits, the ractitioners gain more rents than beore. Again, this becomes an emirical issue. For certain values o, the increased rents will oset the loss in ractitioner roits. owever, other values o lead to less rent-seeking because the gain in roitability osets the incentive to rent-seek. As or the use o inut, the denominator on the third term is larger than that o the second term. Meanwhile, both terms have the same numerator. I the two inuts are substitutes in roduction i.e., i is negative, then the irm will use more than when is one. I the two inuts are comlements in roduction i.e., i is ositive, then the irm will use less. Thus, whether there is a ositive or negative relationshi between rent seeking activity and the quality o care is a undamentally emirical issue, deending on both revenue conditions as well as the substitutability/comlementarity o the two inuts. 4. The Case o Unconstrained Practitioner Ownershi e also wish to eamine the case where the ractitioners have an ownershi stake and unconstrained control over inut usage in the sense that they can set both the rice and the quantity o the irm s inuts. The rice o the inut is now determined along the ractitioners 5

7 6 suly curve; thus we use instead o when determining the rice o. As such, the irm can be both technically and allocatively ineicient. This creates a new roit unction or the ractitioners which we deine as, as shown below:,, 0 d 8 The irst-order conditions that result are: 0 9 = 0 0 For comarison, we can rewrite these equations as: ; 9a, 0a The second term in equation 9a is unambiguously ositive as long as 0. Thus, the value o in the case o unconstrained ractitioner ownershi is smaller than the third arty case as long as the ractitioners have some ownershi stake. ith diminishing marginal returns, this means the level o is larger regardless o the magnitude o as long as is ositive. Unlike the constrained case, the wage earned by when ractitioners have unconstrained control increases as the level o increases, since the wage is determined by the suly curve. As a result, the rents actually increase or the inut-suliers o. As long as these rent-seeking gains are greater than the loss in roit or the irm, the inut-sulier will continue to increase. Meanwhile, equation 4a and 0a are identical in orm. owever, with a dierent level o being chosen, we eect the level o to dier as well. hether this is an increase or a decrease is an emirical issue. As long as the use o licensed assistants either increases, or decreases a suiciently small amount, then rent seeking will lead to equal or higher quality care. It is only when the decrease in licensed assistant hours outweighs the increase in ractitioner hours that there is any concern about a reduction in the quality o care. 5. A Comarative Static Analysis e are now in a osition to not only eamine how a change in the degree o ractitioner ownershi aects the incentive to rent-seek, but also to determine whether and how these marginal incentives dier across each o our models. Our key arameter o interest is the level o ownershi by the ractitioners,. e begin by creating comarative statics or our constrained ownershi model. As shown in the aendi, the comarative statics o interest are given by: d d d d where,,,, and are deined in the Aendi.

8 In general, both o these eressions are ambiguous in sign. owever, we can identiy some general conditions under which each o these eressions is unambiguously signed. The denominator or both and is the stability condition or a maimization roblem, which must be unambiguously ositive. and must also be unambiguously negative to ensure that the objective unction is maimized. As such, the sign o is determined by the signs and relative magnitudes o, and. The irst two terms are the marginal roitability o and, resectively, while the last term is the imact o a change in on the marginal net beneit o. As discussed earlier, the ractitioners in this roblem have an incentive to overutilize, making negative in sign. This also makes the irst set o terms in the numerator o negative. Thus, i and are oosite in sign, then is unambiguously negative, and ractitioners react to lower levels o ownershi by over-utilizing. In short, the ractitioners are rent-seeking. The same is true i and are the same sign, but the roduct o these terms is smaller in magnitude than. The only time no rent seeking occurs is when and are the same sign, and the roduct o these terms is greater in magnitude than. hether this last event occurs deends on the comlementarity or substitutability o the two inuts in the roduction rocess. I the inuts are highly substitutable, then decreasing ractitioner ownershi would likely lead to more rent seeking, as it allows the ractitioners a greater ability to over-utilize in order to recou lost roit. igher levels o comlementarity would allow or less over-utilization and thus a ositive sign or, esecially when ownershi is low or declining. The sign o also deends on the signs and relative magnitudes o, and. owever, in this case the only time when is clearly ositive is when is ositive and is negative. And when is negative and is ositive then is negative. All other ossibilities lead to ambiguous results. In both o these cases, the sign o is negative, imlying that rent seeking occurs. So whether rent seeking leads to more or less use o the other inut deends on whether the two inuts are substitutes or comlements in the roduction rocess, and also how this relationshi imacts the marginal roitability o both inuts. In general one would eect that as ownershi share increased, the value o any etracted rents would begin to disaear, as these rents become more o a dollar or dollar transer between the two sources o income. Meanwhile, the loss o roit rom the ineicient choice o inuts would become a larger loss or the ractitioner owners. Thereore, the ractitioners would increase the use o until it reached the level o use or erect cometition. Again, while a undamentally emirical issue, these indings that, unless the ractitioners and their assistants ehibit etremely high degrees o comlementarity, rent seeking is likely to result in higher real and/or erceived quality o care oered to atients. There are also comarative statics or the case o unconstrained ownershi. As shown in the Aendi, the comarative statics are given by d d 3 d d 4 where,,,, and are deined in the Aendi. As in the constrained roblem, the denominator or both 3 and 4 is the stability condition or a maimization roblem, which must be unambiguously ositive. and must also be unambiguously negative to ensure that the objective unction is maimized. Since 7

9 the ractitioners over-utilize relative to the ure roit maimization case, must also be negative. The rimary dierence between the constrained and unconstrained cases is that we can now sign. As long as is ositive, 0 guarantees that = 0. As such, the sign o 3 is unambiguously negative. That is, in the unconstrained roblem ractitioners always rent-seek when their ownershi share declines. The sign o 4 is ambiguous, and deends on conditions analogous to those or signing. The only time when 4 is clearly ositive is when is ositive and is negative. And when is negative and is ositive then 4 is negative. 6. Conclusions and Policy Imlications In this aer, we investigate whether increasing or decreasing the degree o irm ownershi increases or decreases the likelihood o rent-seeking on the art o ractitioners. In general, we ind this relationshi to be ambiguous, articularly when the ractitioners must ensure that the irm is technically eicient. On the other hand, when the irm can be both allocatively and technically eicient, the ractitioners unambiguously resond to lower ownershi levels by increased rent-seeking. Our indings resent several imlications or olicy makers and administrators. First, administrators have both a carrot and a stick to revent ractitioners rom rent-seeking. The stick involves enalties or ractitioners who waste resources. By ocusing on waste reduction olicies, ractitioners are constrained to be less technically ineicient, and thus less likely to rent-seek. The carrot is simly to give ractitioners a larger ownershi share in the irm. Our study suggests that rent-seeking is a resonse to a reduction in ractitioner ownershi. Thus, i ractitioner ownershi increases, there is less incentive to rent-seek, since rent-seeking reduces roitability. A second imlication is that the tye o medical care oered by the irm is crucial in whether or not rent-seeking occurs. Rent seeking is much less likely when inuts are not substitutable. Thus, rent-seeking is much more likely in irms roviding multile, but similar tyes o secialized care. This may be the reason, or eamle, why rent-seeking is likely to occur when irms rovide both ambulatory and inatient services Ahern et al Additionally, olicy makers may be able to mitigate or enhance rent seeking activity by limiting the tasks licensed assistants can legally erorm, thereby reducing their substitutability with ractitioners. A third imlication is that olicy makers need to be careul when designing olicies to reduce rent seeking and enhance irm eiciency. To the etent that higher use o ractitioners leads to higher real or erceived quality o care, the roit that is transerred to ractitioners and the resulting ineiciency incurred by the irm may ultimately create beneits in the orm o better atient care. owever, a crucial issue in whether rent seeking leads to higher quality deends on i what asects o quality are imortant to administrators and olicy makers and ii the comlementarity/substitutability o the ractitioners and their assistants. Future work to emirically identiy the nature o the rent-seeking/quality tradeo in seciic tyes o health care ractices would rovide valuable insights about whether and how this welare transer imacts atient care. 8

10 9 Aendi: Creating the Comarative Statics e begin with the constrained ownershi roblem, and re-deine equations 6 and 7 as ollows: A A Totally dierentiating A and A: d = d + d + d = 0 A3 d = d + d + d = 0 A4 where: A5 A6 A7 A8 A9

11 A0 In order to guarantee that the objective unction is maimized, and must both be negative. Additionally, while symmetry requires that = the signs o these eressions are unknown. aving deined A5 A0, we can eress A3 and A4 in matri orm as: d d d A Solving or the comarative statics using Cramer s rule leads to: d d d d A A3 Comarative statics or the unconstrained roblem can be created in an analogous ashion. First, we re-deine equations 9 and 0 as ollows: A4 A5 Taking the total dierential, we ind: d = d + d + d = 0 d = d + d + d = 0 where: A6 A7 A8 A9 0

12 A0 A A A3 As beore, in order to guarantee that the objective unction is maimized, and must both be negative. Additionally, while symmetry requires that = the signs o these eressions are unknown. aving deined A8 A3, we can eress A6 and A7 in matri orm as: d d d A4 Solving or the comarative statics using Cramer s rule leads to: d d A5 d d A6

13 Reerences Ahern, M., et al. 996 Predictors o MO Eiciency Alied Economics 8, Banham, L. and J. Connelly 00 Skill Mi, Doctors and Nurses: Substitution or Diversiication? Journal o Management in Medicine 6, Casalino, L., K. Devers and L. Brewster 003 Focused Factories? Physician-Owned Secialty Facilities ealth Aairs, Dor, A. and D. Farley 996 Payment Source and the Cost o osital Care: Evidence rom a Multi-roduct Cost Function with Multile Payors Journal o ealth Economics 5, -. eneral Accounting Oice 003 Secialty ositals: Inormation on National Market Share, Physician Ownershi and Patients Served OA Reort: ashington, D.C. Friesner, D. and R. Rosenman 004 Nonroit Cost Adjusting with Quality as a Private ood Alied Economics 36, illman, A. L. 987 Financial Incentives or Physicians in MOs: Is There a Conlict o Interest? New England Journal o Medicine 37, illman, A.L., M. Pauly, and J. Kerstein 989 ow Do Financial Incentives Aect Physician s Clinical Decisions and the Financial Perormance o ealth Maintenance Organizations? New England Journal o Medicine 3, Kuntz, L. and A. Vera 005 Transer Pricing in ositals and the Eiciency o Physicians: The Case o Anesthesia Services ealth Care Management Review 30, Nallamothu, B., et al. 007 Oening o Secialty Cardiac ositals and Use o Coronary Revascularization in Medicare Beneiciaries Journal o the American Medical Association 97, Riortella-Muller, R., D. Libby, and D. Kindig 005 The Substitution o Physician Assistants and Nurse Practitioners or Physician Residents in Teaching ositals ealth Aairs 4, 8-9. Rosenman, R., M. Ahern, and K. Siddharthan 997 Outut Eiciency o MOs in Florida ealth Economics 6, Rosenman, R., D. Friesner and C. Stevens 005 Do ealth Care Providers Quality Discriminate? Emirical Evidence rom Primary Care Outatient Clinics Eastern Economic Journal 3,

14 Siddharthan, K., M. Ahern, and R. Rosenman 997 The Imact o Ownershi on ealth Care Services in MOs ealth Manower Management 3, 6-. Siddharthan, K., M. Ahern, and R. Rosenman 000 Data Enveloment Analysis to Determine Eiciencies o ealth Maintenance Organizations ealth Care Management Science 3, 3-9. Town, R., et al. 006 The elare Consequences o osital Mergers NBER orking Paer No