The Property Rights Theory of the Firm: Some Evidence from the U.S. Nursing Home Industry Author(s): H.E. Frech III Reviewed work(s): Source: Zeitschrift für die gesamte Staatswissenschaft / Journal of Institutional and Theoretical Economics, Bd. 141, H. 1., 2nd Symposium on The New Institutional Economics (März 1985), pp. 146-166 Published by: Mohr Siebeck GmbH & Co. KG Stable URL: http://www.jstor.org/stable/40750790 . Accessed: 02/01/2013 12:57 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. . Mohr Siebeck GmbH & Co. KG is collaborating with JSTOR to digitize, preserve and extend access to Zeitschrift für die gesamte Staatswissenschaft / Journal of Institutional and Theoretical Economics. http://www.jstor.org This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions fürdiegesamteStaatswissenschaft Zeitschrift (ZgS) 141(1985),146-166 Journal ofInstitutional andTheoretical Economics TheProperty RightsTheoryoftheFirm: SomeEvidencefromtheU.S. NursingHomeIndustry by H.E. Frech III* /. Introduction Oskar Lange is reportedto have said thattheincentives facingthemanagers of his socialistfirmswereso unimportant thattheycould be leftto thesociologists.But,manyeconomistssinceLange have disagreedand have pursued theincentiveproblem.Theymayproperlybe called new institutional economists. An important economicsis theproperty rights partof thenewinstitutional with ArmenAlchianand Oliver of the associated most firm, theory closely Williamson.It analyzesthe externalconstraints faced by the top decisionof the classical textbookfirm makerof the firm.The top decision-maker is regardedas havingfullprivatepropertyrights.The theorypredictsthat if the propertyrightsof that decision-maker are attenuated,theseweaker to thetop decisionmaker. the on amenities reduce rights price nonpecuniary In turn,thisleads to greaterconsumption of amenitiesand lowerfirmwealth, thusreducedmanagerialefficiency (Alchian [1961],Furubotn and Pejovich [1972],Frech [1980]). Other,implicationsof the theorycan be developedwiththe additionof auxiliaryhypotheses(Clarkson [1972], Borjas, Frech and Ginsburg [1981a]).This theoryhas formedthebasis of numerousempiricalstudies(De firmbehavior. foundto be usefulaid in explaining Alessi [1980])and generally At the same time,thereis a greatdeal of policyinterestin the issues of In particular, someobof thehealthcare industry. theinstitutional structure * Thanksforsupportare due to theNationalCenterforHealthServicesResearch and Developement,U.S. Departmentof Health and Human Service,underGrant of California,Santa Barbara,Committee No. ROI HS02674-02and to theUniversity on ResearchoftheAcademicSenate.Thispaperdrawson Frech and Ginsburg[1981], Frech and Ginsburg[1982]and Borjas, Frech and Ginsburg[1983].The paperwas and Economicsat Mettlach/Saar presentedat the Seminaron the New Institutional at a seminarat the International ManagementInstitutein Berlin.Helpfulcomments fromGérardGäfgen at both meetings, werereceivedfromparticipants particularly and JanKmenta. This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S.Nursing 147 servershave suggestedthat the traditionalnonprofitformof ownershipin U.S. hospitalsis inappropriate and thattheprofit-seeking formmightbe superior(Clark [1980]).Policymakershavealso beenconcernedwiththedesign ofsystems ofpaymentforhealthcareproviders(calledreimbursement systems in theU.S.). In thispaperI willshowthattheownershipand reimbursement issuesarepartofthegenericproblemofproperty rights.I willuse theproperty of a largeempiricalstudy rightstheoryof thefirmto guidetheinterpretation of theAmericannursinghomeindustry. 2. TheProperty RightsTheoryoftheFirm The propertyrightstheoryof the firmapplies standardeconomicanalysis to theincentive effects ofvariousconstraints on thefirm.Originally, thestress was on the particularlaw of propertythat applies to the firm,hence the name.Most of theworkon thetheoryhas stressedtheeffectof theproperty on managerialefficiency or cost. In thissection,I willsketch rightsstructure the theory,thenextendit to consideroutsideconstraintsarisingfromthe reimbursement government systemand effectson both costs and on wages paid. Considerthetop decision-maker of thefirmunderprivateproperty rights. The ownerwill maximizehis utilityby choosingthe optimalcombination of nonpecuniary benefitsand monetarywealthof the firm,as in point Op in Figure1. Supposenowwe alterthelegalconstraints on thefirmbyconvertfirm.Althoughthesetwo types ing it to a nonprofit privateor government of attenuatedproperty in degreeand in detail,theessential rightsfirmsdiffer natureof the constraintis the same. The top decision-maker can no longer keep the residualincomeof the firmas his own incomeand wealth.He is limitedto, in wealthterms,thepresentvalue of a "reasonable"salary,which I willdenoteas S. The additionof thisnewnonprofit constraint wouldcause the decision-maker to choose a new point to maximizeutility,point 0a in Figure 1. At thispoint,he has reducedthe firm'swealthand increasedhis consumptionof nonpecuniary goods. As long as the firmearns enoughto him the value these S, pay present nonpecuniary goodsare freeto thedecisionmaker.In choosingto increasehis consumptionof nonpecuniary goods at the expenseof firmwealth,the decision-maker has loweredthe productive ofthefirm.Thisis easyto seewhenone considersthatthenonpecunefficiency iarygoods includeleisure,thequietlife,prestigein thecommunity, pleasant colleaguesand so on. Thus, the model predictsthat firmswithattenuated thanpriproperty rightswould have highercostsand produceless efficiency vate propertyrightfirms.As I mentionedabove, thispredictionhas been borneout in manystudies,ablyreportedby De Alessi [1980]and Furubotn and Pejovich [1972]. 10* This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 148 H.E. FrechIII ZgS The Substitution fromwealthtowardsnonpecuniary goods by thetop decision-maker has manyimplications forthefirm'swage settingprocess,as well as forcosts and efficiency. to First,the shiftwill lead the decision-maker hireworkersand managerswithwhichhe feelsmorecomfortable and towards whichhe feelsmoresympathetic. Thus,dependingon theowner'stastes,the firmmay hire individualsof particularethnic,religiousand racial groups and withparticular physicalor mentalattributes (Alchian and Kessel [1962]). More generally,the ownerwill hireworkerswho are more cooperativein pursuingthe nonpecuniary goods that he favors.What I have said so far ones, but may indicatethatthe workershiredare not the cost minimizing I have not yetdevelopedan argumentforwagesto be higherin non private propertyfirms.However,such a conclusioncan be reachedby severallines of argument. First,George Borjas [1980]has arguedthatone way to purchaseworker cooperationis simplyto pay themmore. Further,he has arguedthat this occursin theU.S. federalgovernment. This,of course,impliesthatworkers in attenuatedproperty be paid moreto securetheirgreater firms would right than would be ordinarily sought.Second,MartinFeldsteinargues cooperation firmsare partly"spent"by overthattheeconomicrentsearnedbynonprofit This is calledphilanthropic workers as a charitable act the owners. paying by wage setting.Finally,attenuatedpropertyrightswill lead the ownerto be to thelabormarkets.Thus,he willmakemoreerrorsin offering lessattentive wagesthatare eithertoo highor too low, relativeto themarketwage. Even thedistribuiftheoffers too high,themarkettruncates are notsystematically tionofoffers sinceonlyoffers thatexceedthepotentialemployee'sreservation wagewillbe accepted.Of course,thesehigherwagesforemployeesof governwithfullcompetitive mentand nonprofitfirmsis inconsistent equilibrium in the labor market.There mustbe an implicitqueue or excess supplyof workerswillingto workin thesehighwagefirms. 3. TheAmerican NursingHomeIndustry U.S. nursinghomesprovidehousing,nursingcare and low levelmedicalcare to manyelderlypeople and a small numberof youngerpeople. About five percentof the elderlypopulationlivesin nursinghomes,but thereis great variationacrossAmericanstatesfromabouttwoto nineor tenpercent(Levit homecareamountedto $ 27.3 billion, on nursing [1982]).In 1982,expenditures Theseexpendiwhichwas about eightpercentof all healthcare expenditures. tureshave been growingat about one and one halftimesthe growthrate ofGrossNationalProductduringthepastdecade(Gibson,Waldo and Levit Medimostlythroughthestateadministered [1983]).Government payments, caid programforthepoor,accountedforabout 55 percentof totalexpendiThe is large,it is dominatedby smallinstitutions. tures.Thoughtheindustry This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) HomeIndustry SomeEvidence fromtheU.S. Nursing 149 Wealth s __ΐ^Λ2* ' ρ Nonpecuniary Benefits Fig.1 Property Rightsand Choice averagenursinghome has only74 beds and onlyfivepercentof the homes containover200 beds. The nursinghomeindustry has historically beendominatedbyprivatenonprofitfirms,but recentgrowthhas been largelyin the ranksof the private firmsso thatby 1976,69 percentof the homeswerein that profit-seeking group,21 percentwereprivatenonprofitand 10 percentwere government owned. Most of the latterare owned by state and local government. The propertyrightstheorywould predictlower costs and lower wages for the firms. profit-seeking the Medicaid, joint federal/state programthat financesmedicalcare for thepoor playsa majorrolein theindustry throughitsmassivereimbursement fornursinghome care. In particular,variationsacross statesin the method Medicaid uses forreimbursement createsdifferences in the incentivesfacing homes withingiven ownershipgroups. In this system,both the level and methodofpaymentformedicalcareis setbythestate.Thishas led to substantial variation.In 1972,the yearof the data forthisstudy,some statespaid a flatrate,regardlessof costs,whileat theotherextreme, some statessimply reimbursed the home forcosts incurred.Betweentheseextremesare subtle variationson themethodsand also theprospective thatpaysan amount system This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 150 H.E. FrechIII ZgS on thepast costsof theindividualhomeor a groupof similarhomes.Table1 schemesfor shows the distribution across statesof nursingreimbursement 1972.Since 1972,partlyin responseto new federalrules,theuse of flatrate exist systemshas declined.Since the best incentivesfor cost minimization in the flatrate states,the propertyrightstheorywould predictboth lower costsand lowerwagesforhomesunderthatsystem. 4. EmpiricalResultson BehavioralDifferences In thissectionI surveytheempiricalresultsfoundon the two main focuses and wages. But, I also considerthe effects of research,costs and efficiency of servicesprovidedto conof theproperty on theintensity rightsstructure is of interestbecause some have and selection. The latter sumers on patient maintain lowercoststhannonprofit homes assertedthatprofit-seeking nursing ones simplyby selectingthe healthiestconsumers,who are least likelyto need costlyservices,as wellas simplyprovidingfewerservices.This is often draws on called "cream skimming".The materialon costs and efficiency Frech and Ginsburg [1981],whilethaton wages drawson Borjas, Frech and Ginsburg[1983]. 4.1. Data The data on nursinghomes,employeesand patientsare fromthe 1973-1974 and state NationalNursingHome Survey1.Data on Medicaidreimbursement regulationare takenfromunpublishedMedicaid documentsand an Urban thesample For thestudyof costsand efficiency, Institute surveyrespectively. was limitedto thosenursinghomesthatprovidednursingcare, ratherthan thatwerecertified simplecustodialcare,to at least halfof theircustomers, of theneeligibleforpaymentunderMedicaid,thathad completereporting StatisticalAreas cessarydata and thatwerelocatedin StandardMetropolitan (SMSAs). For the wage survey,we includeemployeesof homes thatwere not in SMSAs. This editingreducedthenumberof nursinghomesfromthe totalsampleof 2100 to about 600. The numberof patientswas about 8000, whilethe sampleof nonmanagerial employeeswas about 11000. Since the of are based on a data sample 10foreachhome,thereis somesampling patient errorin thesedata whenthe nursinghome,ratherthanthe patient,is used as theobservation. 1 A discussionof thedata is containedin National Center for Health Statistics variablesand data editing of the reimbursement [1979].Details on the construction generallyare containedin Frech and Ginsburg[1981]and especiallyin Frech and Ginsburg[1982]or fromtheauthor. This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S. Nursing 151 Table1 StateNursing HomeReimbursement SystemsUnderMedicaid,1972 (SkilledNursing Homes) TypeofSystem Flat Rate Flat Rate,VariesbyLevelofCare Rate Setting, no Ceiling Prospective Rate Setting, withCeiling Prospective Rate Setting,Based on GroupPerformance Prospectice Cost Reimbursement, withCeiling Cost Reimbursement, no Ceiling NumberofStates 7 2 4 2 6 11 18 Source: Classifiedbyauthorson basisofunpublished surveyconductedbytheSocial and Rehabilitation ofHealth,Educationand Welfare. Service,U.S. Department 4.2. CostsandEfficiency 4.2.1. The EmpiricalModel In thestudyofcostsand efficiency, we useda translogcostfunction to explain the variationin total operationcosts,excludingcapital costs. Capital costs are problematical sincetheyare systematically forgovernment underreported firmsand, underinflation,by profit-seeking firmsin and privatenonprofit flatratestates.Resultsare not verysensitiveto thefocuson operatingcosts alone,whichmakeup 80 percentof totalobservedcosts. variablesincludefirmownership, typeof MedicaidreimburseIndependent ment,inputprices,outputand variablesto standardizetheserviceprovided. These latterincludean indexforpatientdebility,forservicesavailable and forservicesactuallyprovided.These serviceand debilityvariablesallow us to controlfortypeof serviceand qualitymorecompletely thanpast studies. Variablesto capturethe interaction betweenownershipand reimbursement werealso included.All variablesare listedin Appendix1. The functionalformis a hybridof translogand linear.It is translogin outputand inputprices,includingall second orderterms,whilethe output are simplylinear. variables,mostlydichotomous, standardizing lnC=ao + a1lnr+XailnPi + ixZaolnPimP/+Zaiym/>im:r l i i j + Σ*Λ+ Σ^9ι + I^J > Hißtet i i i i This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 152 H.E. FrechIII ZgS where: C=the totaloperatingexpenses, Y= output,inpatient days, Pi= theithinputprice, variable, Prt= the/thproperty rightstructure = the ι th reimbursement variable, Ri variable, Rqi= theιthstateregulatory = the ι th variable. St outputstandardizing The value of the translog'sflexibility was shownin our sensitivity testing, as thedata strongly of theCobb-DouglasForm. rejectedtherestrictions The equationwas estimatedwithordinaryleastsquares(OLS). The treatmentof mostvariablesas exogenousis natural.However,the treatment of (Nerlove [1965]).Althoughquite outputas exogenousis moreproblematical due to location,medicaland competitive, nursinghomesare heterogeneous Hence,each homehas a smallamountof marketpower. amenitydifferences. Marketconditions, an optimalplannedoutputforeach home, thus,determine whichwe take as exogenous.In the regressionwe enterpatientdays, but holdconstanttheoccupancyrate.Thus,patientdaysareequivalentto planned output. a moreseriousproblem. Takingoccupancyrateas exogenousis potentially and Yett [1981] shows that the work of Marshall However, Greenlees, thereis a largerandomcomponentto arrivalsat nursinghomes.Thus,even if plannedoccupancyrate is endogenous,the measuredoccupancyrate includes a large random,exogenouscomponent.In any case, excludingthat variablehinderedthe fitof the equation a bit, but did not alterthe main on theownershipand reimbursement variables.It would have been findings valuableto testthespecification. However,thatrequiresa moregeneralmodel (Krameret al. [forthcoming]). functions withOLS aredifTheassumptions toestimate necessary production ferent (Zellner, Kmentaand Dreze [1966])and lessplausibleforthisindusa production analysis,we performed try.However,as a partof our sensitivity functionanalysis.These resultsalso showedthe same qualitativeeffectsof the propertyrightsvariables,but slightlysmallermagnitudes.This is what of theattenuatedproperty one wouldexpect,sincetheinefficiency rightfirms bundleof inputs. is composedof two parts.First,theyselectan inefficient Second, theyuse the inputsthat theyhave in suboptimalways. The cost functionresultscapturethe total effecton costs of both sourcesof waste, measuresonlythelattersourceon inefficiency. whiletheproductionfunction 4.2.2. Results The modeloverallexplainedabout 91 percentof thecrosssectionalvariation theresultsprovidereasonably in totaloperatingcosts.Much moreimportant, This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S. Nursing 153 Table2 HomeCostFunction:TheEffects andReimbursement Method Nursing ofOwnership on Log TotalOperating Expenses Β Standard Independent Variable Value Deviation Variables* NONPROF ÕÕ5 ÕÕ4 Ownership GOVT 034 0.07 TPMED2 Reimbursement Variablesa -0.02 0.09 TPMED3 0.56 0.32 TPMED4 0.06 0.13 TPMED5 0.03 0.12 TPMED6 0.04 0.10 TPMED7 041 0.14 NPMED 0.24 0.12 Ownership/Reimbursement Interactions'» GMED 027 0.16 a Reimbursement variablesand ownership/reimbursement interactions are all interacted withPCAID, the percentageof patientssupportedprimarily by Medicaid.To obtaintheeffectof any thesevariablesforhomeswiththemean Medicaidpercentage forhomesincludedin theregression, thecoefficient by0.615. * Key to Variables: NONPROF multiply =1 if nursinghome is nonprofit, 0 otherwise; GOVT=1 if nursinghome is government operated,0 otherwise;TPMEDi = l times theproportionof patientsreimbursed by Medicaidif nursinghomeis reimbursed by theithmethod,0 otherwise, TPMED1 = flatrate,TPMED2 = flatrate,variesbypatient, = TPMED3 prospectiverates,base periodcosts,subjectto ceiling,TPMED4 = same as TPMED3, exceptno ceiling,TPMED5 = prospective rates,based on groupperformance, TPMED6 = cost reimbursement, subjectto ceiling,TPMED7 = cost reimbursement,no ceiling;TPMED = interaction amongNONPROF, PCAID (percentofpatients supported primarily byMedicaid)and TPMED 1 or TPMED2 (flatratereimbursement); GMED = sameas above,onlyforGOVT ratherthanNONPROF. The meanIn (operatingcosts)is 13.315. of the effectsof ownershipand reimbursement precisemeasurements type. The coefficient estimates forownership and reimbursement are shownin Table 2. The fullregression is availablefromtheauthor. a. Ownership As one would expect,nonprofit and government firmshave generallyhigher costs than privateproperty, firms. The most costlyfirmsare profit-seeking the government owned homes.Further,the impactof ownershipis highest in thestateswithflatratesreimbursement. Privatenonprofit firmsappear to be muchmorecostlythanprofit-seeking firmsin flatrate environments. Our point estimatefor this is 29 percent, the sum of thecoefficient on NONPROF and NPMED, thelatteran interaction on nonprofitstatusand locationin a flatrate state.In stateswith thewasteful costbased systems, theefficiency shortfall oftheprivatenonprofit firmsis less, witha point estimateof only fivepercentexcess costs. This This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions H.E. FrechIII 154 ZgS is significant at about the 10 percentlevel on a one-tailedtest.This result firmsto obtaindooccursin spiteof theabilityof some religiousnonprofit natedlaborat lessthanmarketwages. farworsethanevenprivatenonprofit Government ownedhomesperform ones. Our estimatesshow thattheircosts are 51 percenthigherin flatrate statesand fully34 percenthigherevenin thecostbasedstates.One explanation forthismightbe thatwiththeiraccessto tax financedsubsidies,theincentives forefficiency ofthegovernment homesmaybe muchweakerthanthenonprowe findlaterthatgovernment homesoverpay fitprivatefirms. Further, nursing firmsdo. theiremployeesbysubstantially morethantheprivatenonprofit the reimbursement variables are deflated the by percentageof the (All home's patientswho are financedby Medicaid. This variablehas a mean mustbe interpreted thecoefficients of0.615in oursample.Therefore, reported as theeffectof thereimbursement schemeon a homewith100 percentMedicaid residents. To gettheeffect on a homewiththesampleaverageproportion of Medicaidresidents, thecoefficient by0.615.) multiply In theninestudies withtheempiricalliterature. Our resultsare consistent firmswere wellsurveyedby ChristineBishop[1979],all foundthatnonprofit more costlythan profit-seeking homes. Some of the studiescontrolledfor outputby includingvariousqualityand serviceproxies,but none were as completein thisrespectthanours.Anotherweaknessofthesepreviousstudies As we see by is that theydid not controlfor the typeof reimbursement. interacon NPMED and GMED, thereareimportant theestimatecoefficients We now turnto thereimbursetionsbetweenownership and reimbursement. mentvariables. b. Reimbursement are assoThe purecostreimbursement systems (withoutceilingson payments) with the when For ciatedwiththe highestcosts. simple compared example, leads to costs thatare flatratesystem(TPMED1) purecost reimbursement or 25 percenthighercomparing 41 percenthigherforan all Medicaidfacility home withthe mean proportionof Medicaid residents.This is the easiest to interpret, sincetheincentives reimbursement coefficient generatedare the clearest.Also,thisresultis quiteprecise. withceilingson theamountpaid by thestate Note thatcostbased systems forMedicaid patients(TPMED6) lead to muchlowercosts than the open ended cost systems.In fact,costs are onlyfourpercentabove the flatrate systems.This indicatesthat the stateshave set the ceilingslow enoughto formanyof the nursinghomes.Thus, mostnursing be bindingconstraints homesfacea de factoflatratesystemin thesestates. We findmixedresultsforthe prospectivesystems.Two of themlead to costswithina fewpercentof thosefortheflatratesystem,whileTPMED3 leads to veryhighcosts,more like the cost based systems.This anomaly This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S.Nursing 155 is probablyconnectedto the factthat New York is one of the few states withthe TPMED3 systemand it has a historyof cost reimbursement and costs. high flatratesystemswheretheratedependson Lastly,themoresophisticated thetypeof patient(TPMED2) resultedin costssimilarto thoseof thesimple sincetheincentives forcostminimizaflatratesystem.This is not surprising, thenursinghomecan affectitspayment tionare identical.Unterthissystem, of the resident,not by allowingcosts to onlyby changingthe classification rise. Interactions c. Ownership/Reimbursement The resultson theinteraction termsNPMED and GMED are veryinteresting. I interpret theresultsas follows.In flatratestates,thereimbursement system market.That is, if the decision-maker allows costs emulatesthecompetitive willnotpay to riseby consumingmorenonpecuniary goods,thegovernment any part of the highercosts. At the otherextreme,in a cost based state, in effect, thegovernment, pays forall or partof anycost increasesby higher reimbursement payments. firmin a cost based state of a profit-seeking Thus,thetop decision-maker would findthat he faces a similarconstraintto that of a nonprofitfirm. The Medicaidsystemwould,in effect, pay forhimto consumemorenonpecuThe constraints of all the firmsin thecost reimbursement states niarygoods. would be describedas attenuated.The constraintsfor the nonprofitfirms wouldbe twiceattenuated. The data tellsus thatnonprofit and profit-seeking firmsbehave similarly.Thus, giventhatincentiveshave been dulledby the costreimbursement thefurther attenuation ofproperty system, rightsinherent in privatenonprofitownershiphas littleeffecton raisingcosts (estimated at onlyfivepercent).It may be thatthe limitedmonitoring of costs by the state reimbursement agencyis all that constrainsthe costs of any firmin theseareas, so that the ownershipvariableshave no role. The same story is true for the comparisonof governmental and profit-seeking firms,but ownedfirmsare so costlythatevenin costreimbursement government states, theircostsare still34 percentabove thoseofprofit-seekers. It is in the flatrate states,wherethereis no outsidemonitoring of the firms'coststo complicatethepicture,thattheownershipvariablesplay such a powerfuland obvious role. In theseareas, privatenonprofitfirms'costs are 29 percenthigherthan the profit-seekers, while governmental nursing homes'costsare estimatedat 51 percenthigher. d. Sensitivity Testing Withinthelimitsof our data themajorresultson theownership and reimbursementvariableswerenot sensitiveto the specification used. First,reported This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions H.E. FrechIII 156 ZgS totalexpensewas substituted foroperating variable. expensesas thedependent The qualitativeresultswerenot altered.As expected,the magnitudeof the effectwas reduced.Also as expected,thereimbursement variables ownership had a greatereffect. Second,variousblocksof variablesweredroppedfromtheequation.With theownership variablesexcluded,thereimbursement variablesshowedalmost the identicalcoefficients. When the reimbursement variableswere leftout, theownershipvariableswereestimateda bitless preciselyand thepointestimatesfellsomewhat.Droppingtheoccupancyrateslightly enhancedtherelativeperformance oftheprivatenonprofit butdid notalterthecomparifirms, son betweenthegovernment and profit-seeking firms.Surprisingly, dropping the outputstandardizing variablesblock by block and in toto harmedthe fitof theequation,but led to almostno changein thecoefficients of either the ownershipor the reimbursement coefficients. This indicatesthatall the firmsprovidecomparablecareto comparableconsumers. e. Is it Inefficiency? The measuredexcesscostsforfirmswithproperty rightsthatare attenuated or reimbursement are large.The theoryleads us to predictthat by ownership thedifferences in managerialefficienwouldreflect, at leastin part,differences thereare at leasttwocounterdifferences, cy.Butespeciallyfortheownership thatcompeteforour belief.The firstis thatthecost differences hypotheses in patientdebilityor thequalityof care (The reflectunmeasureddifferences firms.No one would qualityargumentwould be only made for nonprofit assertthequalityis above averagein government homes).This possiblycan neverbe ruledout completely because one can alwaysimaginesome subtle, in output.However,the patientdebilityand difficult to measuredifference servicevariablesused hereas proxiesforqualityand typeof outputare quite variableshad littleeffect extensive. And droppingtheseoutputstandardizing of patientand quality on the ownershipvariables.Thus, forthe hypothesis to hold,thesedifferences mustnotonlybe unmeasured, differences theymust be uncorrelated withthe measuredoutputvariables.This strainscredulity. firmsselect Further,otherworkreportedbelowindicatesthatprofit-seeking than and more debilitated government nursing patients nonprofit slightly homesand thattheyprovideserviceof slightlygreaterintensity, giventhe debilitystatusof the resident.Also, recentworkby Burton Weisbrodand Mark Schlesinger [1983],based on regulatory violationsfindsthat,at least ones. forsomeoutputs,profit-seeking nonprofit outperform holdsthatthehighercostssimply The othercompeting counter-hypothesis fromnursinghomepatientsto employeesof attenuatedproreflect a transfer pertyrightnursinghomes.As discussedabove, thereare severalreasonsto firmswouldindeedoverpaytheiremployees. suspectthatnonprivate property This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S.Nursing 157 We havefoundthatnonprofit and government nursinghomesindeeddo this. is not nearlyenoughto explain However,themagnitudeof thisoverpayment theentiremeasuredcostdifference. firms Government overpaytheiremployees an averageofsevenpercent, buttheirexcesscostsrangefrom34 to 51 percent. On average,privatenonprofit firmsactuallyunderpay theiremployeesrelative to profit-seeking homesbecauseoftheavailability ofdonatedlaborto religious nursinghomes. Theirestimatedexcesscosts rangefromfivepercentto 29 observedin thisdata are mostly percent.I believethatthe costs differences reflective ofinefficient productionbytheattenuatedproperty rightfirms. 4.3. Wages 4.3.1. The EmpiricalModel To developan empirically usefulmethodformeasuringtheeffectof thepropertyrightsvariableson wage setting,assume that the naturallog of the wage rate of the individualcan be decomposedinto an individual-specific If we approximate fixedeffect and a firm-specific fixedeffect2. theindividuala of socioeconomic effect vector individual variables,the wage specific by to be estimated can be written as determining equation (3) = 'nwij XijB^Ij + ei whereetis a statistical residualuncorrelated withthesocioeconomicvariables effectIjmThe model containsboth individual Xij nor withthe firm-specific and firmcharacteristics. The individualcharacteristics includea greatdeal of detailon experienceand schooling,as well as demographicinformation. The firmspecificvariablewas eitherownershipor reimbursement, depending on theequation.Earlyexperimentation showeda surprisingly negativeeffect of privatenonprofit relativeto profit-seeking. This led to our disownership, church related homes that below market tinguishing nursing pay wagesfrom othernonprofit firmsthatpaytheabovemarketwagespredicted bythetheory. Location variableswere includedto controlfor regionalwage differences. The equationwas estimatedusingOLS. The dependentvariableis thenatural semiskilled log of thewage forall employeesin threeoccupationalcategories, licensedpracticalnurses,skilledregistered nursesand unskilledaides and orderlies.Managerialpersonnelare excludedforfearofconfounding ownership returns withwages. 2 The formis motivatedby theoretical considerations fromthe semilogfunctional humancapitalliterature (Mincer [1974]). This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions H.E. FrechIII 158 ZgS 4.3.2. Results a. Ownershipand Wages The resultson the effectof ownershipon wages are reportedin Table 3. The equationexplains44 percentof thevariationacrossemployees.The socioeconomiccoefficients are generallyreasonableand indicatea highvalue to nursing educationand experience. is in theownership But,themaininterest variables.Generally,the data show thatattenuatedpropertyrightfirmsdo firmsdo. The greatestexcess pay theirworkersmorethantheprofit-seeking pay comesfromgovernment nursinghomes.Theypay about6.6 percentmore to theirworkersthando privateprofit-seeking firms, accordingto thecoefficienton GOVT. This is estimatedwithgood precision,as thestandarderror is lessthanone fifth withthegeneraloverpay oftheestimate.Thisis consistent Smith[1977].As of stateand local government found Sharon employees by as this it is not is, nearlylargeenoughto account large wage overpayment forthe27 to 51 percentexcesscostsof thegovernment firms. In analyzingnonprofit we found it between firms, necessaryto distinguish ones because their church-related homes and secular wage setting nursing on ONONPROF showsthatsecubehavioris clearlydifferent. The coefficient lar nonprofit homespay theiremployeesmorethantheprofit-seeking firms, at but the pointestimateis only 1.6 percent.This is statistically significant onlyabout the 10 percentlevel,based on a one-tailedtest.Needlessto say, thiscannotaccountforthe nonprofit firms'costs beingfiveto 29 percent homes'.The coefficient on CHURCH indihigherthantheprivateproperty cates that the church-related nonprofitnursinghomes underpaytheir employeesby an averageof 3.9 percent.We believethat labor is donated motivatedworkers.This to thereligioushomesby spiritually and charitably Since the secularnursing can be consideredan earlyresultin Theometrics. homesoutnumber thereligiousonesbyonlya factorof about 1.7,theconclufirmslightly sion is thattheaveragenonprofit underpaysits workers.Thus, firmscannotbe attributed thefiveto 29 percentexcesscostsof thenonprofit to employees, notevenin part.We also splitthesamplevarious to a transfer The equationsare reportedin ways to providemore detailedinformation. Frech and Ginsburg Borjas, [1982]. b. Occupationand Reimbursement the we disaggregated As a controland also to providemore information, data by occupationand ran separateequations.The resultsindicatedthat firmsoverpaidtheiremployeesin everycategory,but with the government a decidedegalitariantwist- aids and orderlieswerepaid 16 percentmore thanmarket,licensedpracticalnursessevenpercentin excessand registered This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S. Nursing 159 Table3 on WageRates Effects of TypeofOwnership = In ( WageRate) Dependent Pooled Sample Variable* Constant ED NED EXP EXP2 FEXP NEXP HEXP TRCS WMALE NMALE NFMALE NEAST NCENT SOUTH SMSA CHURCH ONONPROF GOVT R2 No. ofObservations Õ) χ 11.84 1.71 22.23 669.04 2.08 .78 1.12 .47 .07 .01 .14 .28 .33 .23 .64 .07 .12 .15 .441 11542 (2) Coeff. -.0355 .0552 .1736 .0128 -.0002 .0070 .0046 .0076 .0767 .4139 .1282 -.0148 .0728 -.0222 .0831 .1187 -.0392 .0160 .0661 (3) f-ratio (-.40) (7.46) (61.25) (11.77) ("8.55) (4.99) (1.59) (5.23) (9.46) (26.28) (3.21) ("1.23) (5.76) (-1-79) (-6.37) (14.02) (-2.52) (1.29) (5.68) * Key to Variables: ED = yearsof completededucation;NED = yearsof nursing education;EXP = Age-Education -6; FEXP = yearsof experiencein firm;NEXP = yearsof nursingexperience;HEXP = yearsof hospitalexperience;RCS= nondegree coursewithinthepastyear;WMALE = 1 ifwhitemale,0 otherwise; NMALE = training 1 if black male,0 otherwise;NFMALE = 1 if black female,0 otherwise;NEAST = 1 ifresidesinNortheast, 0 otherwise; NCENT = 1 ifresidesinNorth-Central, 0 otherwise; SOUTH = 1 if residesin South,0 otherwise;SMSA= 1 if nursinghomeis locatedin an SMSA, 0 otherwise; CHURCH = 1 ifnursinghomeis church-affiliated, 0 otherwise; ONONPROF = 1 ifnursing homeis secularnonprofit, 0 otherwise; GOVT = 1 ifnursing homeis government The mean In (wage) in thepooled sample operated,0 otherwise. is 1.241. nursesonly2.5 percentmorethanprofit-seeking firms.These estimateswere all quite precise,with/ values of 9.1, 4.1 and 1.8 respectively. A charitable calculus may be behindthis pattern.A impulseand/ora vote-maximizing similar,but weaker,patternwas observedforthesecularand church-related nonprofits. To examinetheimpactoftheotherproperty rightsvariables,thereimbursementvariables,wage equationswereestimatedforeach ownershipclass with thereimbursement variablesentered.The resultsforprofit-seeking firmswere Cost based reimbursement led to wagesabout 10 percenthigherthan striking. flatratereimbursement. These estimateswereprecise,witht statisticsof 3.4 and 4.2 forthe two typesof cost reimbursement. For the otherownership and wage rates. types,therewas almostno relationbetweenreimbursement This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions H.E. FrechIII 160 ZgS Coefficient pointestimatesweremostlylower,some signswerenegativeand thisas indicatingthatfirmswith We interpret estimateswereveryimprecise. attenuatedproperty rightsin legal ownershipare less responsiveto thereimthatthereimbursement withour finding bursement system.This is consistent in costsforprofit-seeking firmsthan systemmade a muchgreaterdifference firms.The reimbursement fornonprofit and government systemalso makes firms. in wagesfortheprofit-seeking a greaterand morepredictable difference disOnly theprofit-seeking subsamplehad enoughobservationsforfurther agregation. thesampleshowthe same egalitariantwist The resultsof further splitting as theresultsof ownershipin the different occupations.Aides and orderlies in the cost reimbursement are overpaidby about 10 percent states,but the nursesare onlyoverpaidby about two moreskilledand educatedregistered percent. 4.4. ServiceIntensity The issueofqualityand ofproperly holdingconstantthenatureoftheoutput in in is alwayslurking thesidelines anyempiricalanalysisof costs.This study laid theissue to rest,but we do have data thatallow maynot have entirely The cost regressions us to narrowthe scope of disagreement substantially. hold constantpatientdebility,and the servicesofferedand used. However, firmssimplyprovide one mayobjectthatperhapsthelowercostprofit-seeking to theirpatients,thusachievinglowercosts without lowerserviceintensity In thispart of the analysis,we model and estimatethe greaterefficiency. in service We findonlyverysmalldifferences ofserviceintensity. determinants notless, more firms the services, provide intensity. Surprisingly, profit-seeking ones. thanthenonprofit 4.1.1. EmpiricalModel We studythe serviceintensity problemwitha two-stageprocedure.First, we createa hedonicindexforservicesthatgiveus theimplicitpricesof the servicesreceived.That is, it givesus an expectedcost of providingeach of the 20 serviceson whichwe have data. To get as close as possibleto true theindexis estimatedfortheprivateresidents marketweightson theservices, homesonly,butcalprograms)ofprofit-seeking (notreimbursed government culatedforall residents.The equationis reportedin Frech and Ginsburg [1982ch.3, p. 3a]. The indexallowsus to aggregateservicesusingtheimplicit This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S.Nursing 161 marketweightsto get a sealermeasureof the totalservicessuppliedto each resident. 4.4.2. Results In the finalstage,we regressthe measureof total servicesHEDEX1 on the forpatientdebilityand mentalstate,reimburvariables,controlling ownership sementand many otherstandardizing variables.The resultsare shownin Table4 ServiceIntensity Regression: DependentVariableis an IndexofServiceIntensity Parameter Standard Variable* DF Estimate Error tratio Prob>|t| - 79.083339" Î INTERCEPT 76.944625 -1.0278 0.3041 NPMED Ï -1.822803 4.258421 -0.4280 0.6686 1 GMED 3.543425 4.350992 0.8144 0.4155 NONPROF 1 -3.365793 1.316036 -2.5575 0.0106 1 GOVT -0.147356 2.053820 -0.0717 0.9428 1 FAMINC 0.0006555807 0.0007614285 0.8610 0.3893 1 ELG -1.035013 1.967233 -0.5261 0.5988 MALE 1 -0.885631 0.954120 -0.9282 0.3533 WHITE 1 0.229295 1.636940 0.1401 0.8886 Ï LCONSDEX -3.015082 4.176162 -0.7220 0.4703 LRNEARN 1 22.953074 12.822421 1.7901 0.0735 LLPEARN Î -21.812325 9.745216 -2.2383 0.0252 LNAEARN 1 38.298062 7.664554 4.9968 0.0001 1 LCSEARN -29.726516 7.709798 -3.8557 0.0001 TR1C 1 0.414457 5.293541 0.0783 0.9376 1 TR3C -27.500397 5.190387 -5.2983 0.0001 TR4F 1 -5.562038 20.463671 1 TR6F 9.370556 3.512063 2.6681 0.0076 TR7F 1 -1.526452 3.055132 -0.4996 0.6174 - 3.928936 TPREM2 1 2.014008 -1.9508 0.0511 Ï TPREM3 10.047451 4.541629 2.2123 0.0270 TPREM5 1 5.404015 1.970902 2.7419 0.0061 - 5.662692 TPREM6 1 1.896633 -2.9857 0.0028 TPREM7 1 2.944037 2.423492 1.2148 0.2245 KATZ1 1 1273.88 154.790527 8.2297 0.0001 KATZ2 1 1420.05 176.280595 8.0556 0.0001 KATZ3 1 1222.906 214.947224 5.6893 0.0001 KATZ4 1 2091.349 145.995638 14.3247 0.0001 1 KATZ5 3089.556 147.127266 20.9992 0.0001 KATZ6 Ï 4403.239 149.072497 29.5376 0.0001 KATZ7 1 1671.788 186.839355 8.9477 0.0001 MENTAL1 1 4.212444 0.910502 4.6265 0.0001 MENTAL2 1 1.041522 0.941107 1.1067 0.2685 11 ZgS 141/1 This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 162 H.E. FrechIII Table4 (Continued) Standard Parameter Error Estimate DF Variable* 1.188605 -2.017055 Í MENTAL3 0.922596 1 -5.833174 MENTAL4 1.158320 1 5.754409 MENTAL5 5.683980 4.838800 1 ADMP 6.057627 Ï -7.972128 ADMS 5.829492 -1.508673 1 ADMB 2.941528 -6.974891 1 EPROXY 0.854583 -2.935799 1 STF 1.753249 2.047153 1 STFO F ratio 6156999 SSE Prob> F 5877 DFE 1047.643 MSE R-square Ν 38.08 MEANDEPVAR ZgS tratio -1.6970 -6.3226 4.9679 0.8513 -1.3160 -0.2588 -2.3712 -3.4354 1.1676 36.90 0.0001 0.2047 5919 Prob>|t| 0.0898 0.0001 0.0001 0.3946 0.1882 0.7958 0.0178 0.0006 0.2430 * Key to Variables:NPMED = interaction amongNONPROF, PCAID (percentof by Medicaid)and TPMED1 or TPMED2 (flatratereimpatientssupportedprimarily GMED = sameas above,onlyforGOVT ratherthanNONPROF; NONbursement); 0 otherwise;GOVT=1 if nursinghome is PROF =1 if nursinghome is nonprofit, = medianfamilyincomein SMSA, 1969; FAMINC 0 otherwise; operated; government in thatstate; MALE = 1 if patientis Medicaid for ELG = medicallyindigent, eligible = = is 1 if WHITE 0 male, otherwise; patient white,0 otherwise;LCONSDEX natural log of Dodge BuildingCost Index for U.S. and Canadian cities,September1972; nursesfor LRNEARN = naturallog of medianearningsof female,fulltime registered the SMSA, 1969; LLPEARN = same as above for licensed practical nurses; LNAEARN = same as above fornursesaides,orderliesand attendants;LCSEARN = same as above forcleaningservicepersonnel;TriC= 1 timesPCAID if thereimbursementsystemwas recentlychangedfromcost based to the ith system,0 otherwise; TRiF = 1 timesPCAID if the reimbursement changedfromflat systemwas recently rate to the ith system,0 otherwise;TPREMi 1 if nursinghome is reimbursed by TPREMI = flatrate,TPREM2 = flatrate,variesbypatient, theithmethod,0 otherwise, TPREM 3 = prospectiverates,base periodcosts,subjectto ceiling,TPREM4 = same as TPREM3, exceptno ceiling,TPREM5 = prospective rates,based on groupperforreimbursesubjectto ceiling,TPREM7 = cost mance,TPREM6 = cost reimbursement, = ment,no ceiling;Κ ATZi= versionof the Katz debilityindex,Κ ΑΤΖ0 independent KATZ1 throughKATZ7 increasingdependenceon help in all dailylivingactivities, behaviors,0 otherfordailyliving;MENTALi= 1 ifthepatientexhibitsthefollowing = wise,MENTAL 1= depressed,MENTAL2 ==agitated,MENTAL3 abusive,aggressive, MENTAL4 = confused,senile,MENTAL5 disturbedsleep; ADMP=1 ifpatientadmittedforphysicalreasons,0 otherwise;ADMS = 1 ifpatientadmittedforsocial reasons,0 otherwise;ADMB = 1 ifpatientadmittedforbehavioralreasons,0 otherwise; EPROXY = 1 if earningsfroma nearbySMSA was used fora smallSMSA; STF = nursingstaffrequirement, nursinghoursperbed day; STFO = 1 ifstatehas a minimum 0 otherwise. Table 4. The equationexplains20 percentof thevariationin serviceintensity measures thatthisis microdata withimperfect acrossindividuals. Considering of the patient'shealthstatus,thisfitis surprisingly good. One can see that variableshave onlyverysmalleffects. But,thepointestimates theownership levelof serviceintensity a lower is associatedwith is thatnonprofit ownership This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S.Nursing 163 than profit-seeking nursinghomes. What is more important,the standard erroris smallenoughthatone can say witha greatdeal of confidencethat theeffectof ownershipis verysmall.The effectof reimbursement is likewise sinceexperience withhospitalinsurancewould quitesmall.This is surprising, lead one to expectgenerousreimbursement to lead to manymore services. The majordeterminants of theserviceintensity are,as one wouldexpectand hope,thedebilityof thepatients.Clearly,one cannotexplainthehighercosts ofnonprofit and government firmsby serviceintensity. 4.5. PatientSelection Anotherpossiblenon-effiqency explanationforthehighercostsof theattenis uatedproperty firms that theprofit-seeking firmssystematically select rights healthier who will be less for. to care This kind of selection patients expensive is oftencalled"creamskimming" in theAmericanliterature. Greenlees,Marshall and Yett [1982]have shownthatnursinghomesexercisein selection patientsand thatwhenoccupancyratesare high,theyavoid Medicaidpatients where Medicaid patientsare poorly reimbursed.Doubtlessly,all nursing homesselectthepatientsthattheypreferforwhateverreason.The question is, does this selectionfavorthe location of relativelyhealthypatientsinto the profit-seeking firmsso that thisexplainsthe apparentlylowercosts of thesefirms. The factthattheprofit-seeking firms havelowercostswhenpatient debilityis controlledforgivesone pieceofevidencethatthisis notthecase. But directlyexaminingpatient selection provides another approach. in its own right.It rounds Further,theissueof patientselectionis interesting out ourunderstanding ofthedifferences in behavioramongthenursinghomes withdifferent legalownership. Examinationof the simplecorrelationsbetweenthe debilitylevel of the or notthepatientwas selectedbya profit-seeking patientand whether nursing home,Table 5, showsthatforall of themoredisabledcategories,theprofitnotless thantheattenuatedproperty seekingfirmshad moreresidents, rights firms.The privateproperty rightsfirmshave fewerof thosewithno disability and of thosewiththe lowestlevel of disability.This is exactlycontraryto theconventional wisdomof "creamskimming" and providesadditionalsupfor an of the observation thatcosts are much port efficiency interpretation lowerforprofit-seeking firms.Given thatwe know of selectionby nursing homesfromthe Marshall,Yett and Greenlees,how can we explainthis?I believethattheanswerlies not in the selectionof patientsby profit-seeking firms,but in the selectionof theirnursinghomesby consumers.Those who choosea nursinghomeforsocial,ratherthanmedical,reasonsare mostlikely to choosea religiousone. Othernonprofit homesalso caterto naturalaffinity groups.Anotherclass ofconsumerswho are notill,butmaychoosea nursing 11* This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 164 H.E. FrechIII ZgS Table5 Selectionbya Profit-SeekingFirmandPatientDebilityLevel SimpleCorrelations: DebilityLevel* KATZ1 KATZ2 KATZ3 KATZ4 KATZ5 KATZ6 KATZ7 Ν = 7,992 Correlation -0.00133 0.01466 0.01124 0.01164 0.05389 0.00519 0.01294 * Key to Variables:Κ ATZi= versionof theKatz debilityindex,Κ ΑΤΖ0 = independentin all dailylivingactivities, KATZ1 throughKATZ7 increasingdependenceon helpfordailylivingactivities. home are thosewho are destituteand have nowhereelse to go. These consumers,I suspect,are mostlikelyto choosea government nursinghome. 5. Conclusion the The property rightstheoryof thefirmhas provedusefulin illuminating economicbehaviorof the nursinghome industry.Its predictionof higher costs forthe attenuatedpropertyrights,privatenonprofit and government firmsis borneout. Government nursinghomescost from34 to 51 percent firms'costs are fromfive more than profit-seeking ones. Privatenonprofit to 29 percenthigher.These differences seemverylikelyto reflectefficiency not variationsin output,as manydimensionsof patientdebility differences, and servicesprovided,proxiesforqualityand typeof output,are held conis not necessary,as profit-seeking stant.Further,a regression interpretation moreservicesto a moredisabledpopulation firmsactuallyprovideslightly firms.Partof thecostdifference of residents thando thenonprivateproperty reflects a transfer to the employeesof the attenuatedpropertyrightsfirms. Government firms,in particular,overpaytheiremployeesby sevenpercent, but thiscannotcome close to accountingforthelargeobservedcost differences. programsforthe Turningto the effectof the stateMedicaid government leads to 41 percenthigher poor, we see thatunlimitedcost reimbursement marketincentives costs than the flatrate systemthatemulatescompetitive withownership. thereimbursement forcostcontrol.Further, systeminteracts makesmuchmoredifference Wheretheflatratesystemis used,theownership forall nursinghomeshave alreadybeendulled thanin stateswhereincentive This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 141/1(1985) SomeEvidence HomeIndustry fromtheU.S.Nursing 165 OskarLangewas wrong.The issueofincentives should bycostreimbursement. notbe leftto thesociologists. References Rand Corp.P-2316,SantaMonica. Alchian, A. A. [1961],SomeEconomicsofProperty, , and Kessel, R. [1962],"Competition,Monopoly,and the Pursuitof Money", in: Aspectsof Labor Economics,NationalBureauof EconomicResearch,Special Conference N.J. Series,Vol. 14,Princeton, Information Costsand EconomicOrganiza, and Demsetz,H. [1972],"Production, EconomicReview.62. 777-795. tion".American Bishop,C. E. [1979],"NursingHome BehaviorUnderCost-RelatedReimbursement", DiscussionPaper,University HealthPolicyConsortium, BrandeisUniversity. in the FederalGovernment: The Role of Borjas, G.J. [1980],"Wage Determination Constituents and Bureaucrats", JournalofPoliticalEconomy, 88, 1110-1147. , Frech III, H.E., and Ginsburg,Β. [1983],PropertyRightsand Wages: The Case ofNursingHomes",JournalofHumanResources,17,231-246. Clark, R.C. [1980]"Does theNonprofitForm Fit theHospitalIndustry?"Harvard Law Review,93, 1416-1489. of PropertyRightsin HospitalManageClarkson, K.W. [1972],"Some Implications ment",Journal ofLaw andEconomics,15,363-384. De Alessi, L. [1980],"The Economicsof Property Rights:A Reviewof theEvidence", ResearchinLaw andEconomics, 2, 1-47. D.C. Feldstein,M. S. [1971],TheRisingCostofHospitalCare,Washington, Frech, H. E., Ill [1976],"The Property RightsTheoryof theFirm:EmpiricalResults froma NaturalExperiment", JournalofPoliticalEconomy, 84, 143-142. Marketsfor [1980],"PropertyRights,the Theoryof the Firmand Competitive ResearchinLaw andEconomics Top Decision-Makers", , 2, 49-63. , and Ginsburg,P. B. [1981a],"The Cost of NursingHome Care in the United States:Government Financing,Ownershipand Efficiency", pp.67-81, in: J. van der Gaag and M. Perlman(eds.), Health,Economicsand HealthEconomics:Proceedingsof the WorldCongresson Health Economics,Leiden,The Netherlands, 1980. September HealthInsurance:MultipleObjectives [1981b],"PropertyRightsand Competition forNonprofit Firms",ResearchinLaw andEconomics, 3, 155-172. and Financing,Final [1982],The Economicsof NursingHome Care: Ownership Report,Grant 1RO1 HS 02674, National Centerfor Health ServicesResearch, of Healthand HumanServices,Springfield, Department Virginia. Furubotn, E. and Pejovich, S. [1972],"PropertyRightsand EconomicTheory:a JournalofEconomicLiterature. Surveyof RecentLiterature", 10, 1137-1162. Gibson,R. M., Waldo, D. R. and Levit,K. R. [1983],"NationalHealthExpenditures, 1982",HealthCareFinancing: Review,5, 1-32. Greenlees,J.S., Marshall, J.M. and Yett, D. E. [1982],"NursingHome Admissions PoliciesUnderReimbursement", Bell JournalofEconomics,13,93-106. Levit,K. [1982],"PersonalHealthCare Expenditures byState",HealthCare Financing Review,4, 1-46. andEarnings, New York. Mincer,J.[1974],Schooling, Experience Kramer,W., Havlick, P., Maurer, J.and Sonnberger,H. [1985],"DiagnosticCheckingin Practice",ReviewofEconomicsandStatistics (forthcoming). National Center for Health Statistics [1979],TheNationalNursingHome Survey: 1977Summary ofHealth, fortheUnitedStates,Series13,No. 43, U.S. Department Educationand Welfare(PHS)79-1794. This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions 166 H.E. FrechIII ZgS Amsterdam. Nerlove, M. (19651,Identification ofCostandProduction Functions, N.J. Smith,S. [19711,Equal Pay inthePublicSector,Fact orFancy?Princeton, Weisbrod,B.A. and Schlesinger,M. [1983],"Public,Private,NonprofitOwnership and theResponseto Asymmetric DiscussionPaper No.209, Center Information", Madison. forHealthEconomicsandLaw, University of Wisconsinand Estimation of CobbZellner, Α., Kmenta,J.and Dreze, J.[1966],"Specification 34, 784-795. Douglas ProductionFunctionModels",Econométrica, Professor HarryE. FreehHI University ofCalifornia Department ofEconomics Santa Barbara,CA 93106 U.S.A. This content downloaded on Wed, 2 Jan 2013 12:57:52 PM All use subject to JSTOR Terms and Conditions