Dipartimento di Scienze Economiche Università degli Studi di Brescia Via San Faustino 74/B – 25122 Brescia – Italy Tel: +39 0302988839/840/848, Fax: +39 0302988837 e-mail: [email protected] – www.eco.unibs.it WELFARE PROPERTIES OF RESTRICTIONS TO HEALTH CARE SERVICES BASED ON COST EFFECTIVENESS By Laura Levaggi Rosella Levaggi Discussion Paper n. 0902 These Discussion Papers often represent preliminary or incomplete work, circulated to encourage discussion and comments. Citation and use of such a paper should take account of its provisional character. A revised version may be available directly from the author(s). Any opinions expressed here are those of the author(s) and not those of the Dipartimento di Scienze Economiche, Università degli Studi di Brescia. Research disseminated by the Department may include views on policy, but the Department itself takes no institutional policy position. Welfare properties of restrictions to health care services based on cost e¤ectiveness Laura Levaggi and Rosella Levaggiy March 11, 2009 Abstract In this note we explore the welfare properties of the cost e¤ectiveness criterion as an instrument to improve the appropriateness of health care. We show that such instrument is optimal only under speci…c assumptions relating to the shape of the welfare function and the utility of health care. JEL Classi…cation: I11,I18 Key Words: Cost e¤ectiveness, welfare analysis Department of Mathematics, University of Genova, Via Dodecaneso 35 - 16146 Genova (Italy). E-mail: [email protected]. y Department of Economics, University of Brescia, Via S. Faustino, 74b, 25122 Brescia (Italy). E-mail: [email protected]. 1 1 Introduction In most countries, national authorities implement controls (e.g. budget impact limitations) and incentives (e.g. prescription limitations to be followed by physicians) to in‡uence the use of health care. These systems, widely used for drugs, are becoming more common for other types of treatments (Jacobzone, 2000). Restrictions on health care access are powerful instruments to reduce demand. In their direct form (restriction on access or reimbursement) the number of patients receiving a speci…c treatment is restricted to those that are expected to derive the highest bene…ts. The aim of this note is to study the welfare properties of direct restrictions based on cost-e¤ectiveness measurement from a theoretical point of view, as a policy instrument that can be used to improve equity in access to and …nancing of public health care systems. Its use increases the average e¤ectiveness of care in a pure public health care system; the comparision with a pure private health care system depends on the shape of utility. Such criterion is welfare maximising only if the function representing utility and the distribution of weights have speci…c properties. In general we can therefore say that such criterion may increase average e¤ectiveness, but not necessarily welfare. 2 The model The community consists of N individuals, normalised to one. Each individual has a …xed exogenous income Y in the support (0; Q) and an endowment H of health which produces utility according to a strictly increasing function (Y ) with 0 (Y ) > 0. This parameter depends upon individual characteristics such as education, income, social status and lifestyle. The utility function is additive and separable in income and health; there is a …xed probability of being ill, if so the health stock reduces to H. The utility function for the representative consumer is: ( y + (1 ) (Y )H when healthy, U= (1) y + (Y ) H; if ill where y and Y are the gross and net income respectively. Here captures the greater concern of people with higher incomes with their health, due to several characteristics: a greater opportunity cost of leisure and work time spent in bad health, a greater concern for their physcal strength. 2 If ill, the patient can partially restore his health using a speci…c treatment that allows H units to be gained and has a cost equal to p1 . is assumed to follow a known distribution on the support [0; 1 ] with 2 g( ) > 0 representing its density. 3 Private provision In a private health care market with no insurance, the patient buys the treatment if he can a¤ord it (Y p) and if the utility received is greater p than the cost. From (1) we can write (Y ) H p, i.e. (Y )H . Since p 2 [0; 1 ] the condition (Y ) H(1 ) is also necessary in order to have a non-void interval of su¢ ciently high -s for convenience. To simplify matters we assume that p (p) ; (2) H(1 ) so that any Y p will satisfy this requirement, since is increasing. Therefore possibility/convenience conditions are p Y p; : (3) (Y )H RQ Let’s denote by E(Y ) = 0 Y f (Y ) dY the average income and by E( ) = RQ (Y ) f (Y ) dY the average . 0 In the private market the average e¤ectiveness of the drug will be equal to RQR1 g( ) d f (Y ) dY p p (Y )H pr = : R R m Q 1 g( ) d f (Y ) dY p p (Y )H For uniformly distributed pr m and Y we get RQh )2 1 p (1 h = 2 R Q (1 ) p p2 )H 2 2 (Y p (Y )H i i dY : dY Such criterion may not be optimal both for e¢ ciency and equity reasons. Patients buy the treatment on the basis of their income and e¤ectiveness. Since the latter is just one of the determinants, productivity in terms of health will never be maximised under this system. 1 To simplify the model, we assume that one unit of treatment is su¢ cient to treat the patient. 2 For example, an active principle that reduces blood pressure might be less e¤ective if used with other drugs, i.e. for a class of patients with multiple diseases. 3 4 The cost e¤ectiveness criterion In a pure public health care system the treatment should in principle be o¤ered to all those needing it. In this case, even if we assume that only appropriate treatments are Rsupplied, the average e¤ectiveness of the new 1 treatment will be equal to 0 g( ) d . In order to improve the average e¤ectiveness, most regulators use a simple cost e¤ectiveness criterion for reimbursement: p p <1 (4) > ; H H which in turn implies an expected average e¤ectiveness equal to: R1 g( ) d p E = RH1 ; g( ) d p H which is always greater than the total average e¤ectiveness. For example in the uniformly distributed case we obtain E 1 = 2 p2 H2 p H (1 (1 )2 1hp = + (1 ) 2 H i ) pr and E > 1 2 . In contrast, the sign of E m depends on the shape of (Y ). From a policy point of view, what is even more important is that we do not know the welfare properties of such a criterion. 5 Welfare properties Let’s examine the behaviour of a benevolent regulator who in this environment wishes to maximise social welfare. The derivation of social welfare from individual preferences depends on the objectives pursued by the regulator. In general utility is assumed to be the sum, possibly weighted, of individual utility3 . We use this approach here and de…ne total welfare as follows: Z QZ 1 V = [y + (Y )H(1 + + (Y; ))] (Y )g( )d f (Y )dY: 0 0 The function (Y; ) 2 f0; 1g depends on the possibility/bene…t of the individual in being cured and re‡ects the choices of the regulator’s admission 3 For a discussion of the use of weights with a linear utility function see Fossati and Levaggi (2007) 4 to the public market. The term 0 represents the weight the regulator attaches to the utility of each individual. Such weights are positive, but decreasing in income, i.e. 0 (Y ) 0. From the above equation it is clear that, without loss of generality, we can assume Z Q (Y ) f (Y ) dY = 1; (5) 0 that is the function (Y ) := (Y )f (Y ) is a distribution on [0; Q]. 5.1 Public provision Let’s assume that the good can be supplied only by the public provider and that the cost is …nanced through a linear income tax. The regulator de…nes the minimum e¢ ciency level c over which the treatment is supplied, free of charge. The optimal level has to be found to maximise the following welfare function: Z Q Z Q V = (1 t) Y (Y )f (Y )dY + H (1 + ) (Y ) (Y )f (Y )dY 0 + H Z 0 Q (Y ) (Y )f (Y )dY Z (6) 1 g( )d ; c 0 The budget constraint is Z Q Z t Y f (Y )dY = p 1 g( )d : (7) c 0 We introduce the notation Z Q E (Y ) := Y f (Y ) (Y ) dY; E ( ) := 0 Z Q (Y )f (Y ) (Y ) dY; 0 i.e. E (Y ) and E ( ) are the averages of income and w.r.t. the new distribution . As shown in Appendix A, the optimal value is: c = For pE (Y ) : HE ( )E(Y ) (8) p : HE( ) (9) = 1 we get c = From the above analysis we can derive the following propositions: 5 Proposition 1 For an utilitarian welfare function ( = 1) the cost e¤ ectiveness criterion (4) is optimal only if E( ) = 1. If this is not the case, the average e¤ ectiveness is higher than E if E( ) < 1, lower when E( ) > 1. Proposition 2 In a pure public health care system with weights on utility, the cost e¤ ectiveness criterion is optimal if and only if Z 0 Q Y f (Y ) (Y ) dY = E(Y ) Z Q (Y )f (Y ) (Y ) dY: (10) 0 Even in a context where there is no private market and the utility function is linear in income, the cost e¤ectiveness criterion is optimal only under special circumstances. For a system without weights on utility, using the cost e¤ectiveness rule is equivalent to assuming that the average utility derived from health care is equal to one; in a system where utility is weighted, the Y latter must have a particular shape. For a linear tax system E(Y ) is the tax incidence function (Gouveia, 1997), i.e. the relative burden of each individual. This allows interpretation of the equation (10) as follows: for the cost e¤ectiveness criterion to be e¢ cient, the function of the weights must equalise the average weighted “willingness to pay” (E ( )) with the average weighted incidence of the …scal burden. If this is not the case, the cost e¤ectiveness criterion is not optimal from a welfare point of view. 5.2 A mixed market In most public health care systems the approval to commercialise the drug is a process independent of reimbursement. In the presence of a prescription limitation, patients whose productivity level is below c can still buy it on the private market, provided that Y p and p (Y )H and c. Therefore buyers on the private market are identi…ed by this combination of Y and p (Y )H c and p Y Q: The number of buyers on the private market depends on c and we can identify a minimum income Y (c) (see (18) in Appendix B) such that consumers satisfying p Y (p) Y Q; c (Y )H receive a utility surplus when receiving treatment. Using the notation of Section 5.1, for a mixed health care market, the welfare function should be 6 written as: W = (1 t)E (Y ) p Z Q f (Y ) (Y ) Y (c) +(1 + H + Z Q Y (c) Z c p (Y )H Z )HE ( ) + HE ( ) ! Z c (Y ) p (Y )H g( ) d ! dY 1 (11) g( )d c g( )d (Y )f (Y )dY The budget constraint does not change, thus (7) is still valid. In Appendix B we show that a necessary and su¢ cient condition for the existence of a maximum is Z Q Z p E (Y ) p (Y ) dY < (1 )H (Y ) (Y ) dY: (12) E(Y ) p 0 If this condition is satis…ed, from the F.O.C. ! Z Q Z Y (c) p E (Y ) (Y ) dY = c (Y ) (Y ) dY; H E(Y ) Y (c) 0 (13) by (5) we can write the optimal marginal reimbursed e¤ectiveness as: ! R Y (c) (Y ) dY p E(Y ) E (Y ) 0 c= 1 : (14) R R Y (c) H Y (c) (Y ) (Y ) dY E(Y ) (Y ) dY 0 0 The term E ( ;Y Y (c)) = R Y (c) 0 (Y ) (Y ) dY R Y (c) (Y ) dY 0 is the average of under the distribution for individuals having income Y Y (c). The expression (14) does not have a clear interpretation in its general formulation. To gain some insights, let’s examine the case of a weight , for which the averages E(Y ) and E (Y ) are equal 4 . We show in Appendix B that in this case the optimal value is given by p HE ( ; Y c = max 4 This condition is satis…ed whenever uniformly distributed case f (Y ) = 1 Q RQ 0 Y (1 p) ;1 (Y ))f (Y ) dY = 0; for example in the a feasible choice is (Y ) = 7 : 6 Q Y Y2 Q . The expression E ( ; Y p) is the average weighted utility of consumers having insu¢ cient income to buy on the private market. This result is in fact comparable to (9). In a pure public health care system, the regulator chooses the marginal e¢ ciency level using the average level of , while in this case only the utility loss of non-buyers is taken into account. If the regulator gives higher weight to low income levels Y , then obviously E (Y ) < E(Y ) and the optimal c is lower. Let’s now compare the cost e¤ectiveness criterion (4) with welfare maximisation in a mixed market. From (13) a minimum level Hp would be optimal if and only if Z Q Z Y( p ) H E (Y ) (Y ) dY = (Y ) (Y ) dY p E(Y ) Y( ) 0 H Rearranging the terms, since by (18) (Y ( Hp )) = 1 we can show the following: Proposition 3 In a mixed market case with weights on utility the cost effectiveness criterion is optimal if and only if Z Q Z Q Y (Y ) f (Y ) dY = minf (Y ); 1g (Y ) f (Y )dY (15) 0 E(Y ) 0 Note that if (Y ) = 0 for Y Y ( Hp ) conditions (15) and (10) coincide. Observe also that the right-hand side in (15) is lower than E ( ). Thus, if we compare a weight function ful…lling (15) with one satisfying (10), the …rst has to give higher weight to incomes Y E(Y ) than the second. 6 Discussion and conclusions Cost e¤ectiveness measures are becoming increasingly popular among regulators as a tool to improve the appropriateness of health care services. Prescription limitations are an instrument used at various degrees by national regulators. They may be e¤ective in increasing the expected e¤ectiveness of a drug, but their welfare properties have not been explored so far. In this paper we use a linear function very similar to the one proposed by the literature (Jelovac, 2002) to show to what extent the cost e¢ cacy criterion is also welfare maximising. We show that in a pure public health care system where the drug is available only through the public provider, cost e¢ cacy is also welfare maximising either if the utility derived from health has special properties or if 8 the set of weights is chosen so as to equalise average utility with average tax incidence. In a mixed market for health care, a straightforward comparison is more di¢ cult because the marginal e¤ectiveness level that maximises welfare depends on and . In this case we might expect the cost e¤ectiveness criterion to be welfare improving only if the distribution of weights is skewed towards low income brackets. In general, we can say that the cost e¤ectiveness criterion is welfare maximising only under speci…c conditions relating to the shape of these two functions. Cost e¤ectiveness may be an easy to understand - easy to implement criterion, but it does not necessarily imply that it is also welfare maximising. References [1] Breyer, F. and A Hau‡er (2000) Health Care Reform: Separating Insurance from Income Redistribution, International Tax and Public Finance, 7( 4-5), pp. 445-61 [2] Fossati, A. and Levaggi, R. (2008) ,Delay is Not the Answer: Waiting Time in Health Care & Income Redistribution, http://ssrn.com/abstract=1081928 [3] Jacobzone S (2000), Pharmaceutical policies in OECD: reconciling social and industrial goals, OECD Labour market and social policy, occasional papers. [4] Jelovac, I (2002) On the relationship between negotiated prices of pharmaceuticals and patients’copayment, paper presented at the VI EHEW meeting, Marseille, 8-9 November. [5] Levaggi, L. and R. Levaggi (2006) Optimal copayment strategies in a public health care system, http://ssrn.com/abstract=734923 [6] Smith, P (2005) User charges and priority setting in health care: balancing equity and e¢ ciency, Journal of Health Economics, 2005;24:10181029. 9 A No private market By de…ning the average level of from (7) we have t = t(c) = as E( ) and the average income as E(Y ), p E(Y ) Z 1 g( )d : (16) c Using the notation E (Y ) and E ( ) for the average of income and w.r.t. the weighted distribution = f , we can write the objective function in equation (6) as: Z 1 V (c) = (1 t(c))E (Y ) + H (1 + ) E ( ) + HE ( ) g( )d : c (17) The optimization problem is the maximization of V on [0; 1 ]. Since p dt dc = E(Y ) g(c), the …rst order condition for the optimization problem is dV E (Y ) = p g(c) dc E(Y ) HE ( )cg(c) = 0 and the optimum value is c = pE (Y ) ; HE ( )E(Y ) which is independent of g. B Mixed market Buyers on the private market are identi…ed by the conditions p (Y )H c and p Y Q: p 1 Inequality (Yp)H c is satis…ed if Y cH : this lower bound is greater p p than p when c (p)H and greater than Q when c (Q)H . From (2) we p always have (p)H 1 and we can de…ne Y (c) := 8 > > <Q > > :p 0 1 p cH p (Q)H p (p)H 10 p (Q)H c p (p)H c c 1 : (18) Let’s write the …rst order optimality conditions for the problem of maximizing the welfare function W ; to simplify notations write (Y ) = (Y )f (Y ); we have Z Q dt dW = E (Y ) p g(c) (Y )dY HE ( ) c g(c) dc dc Y (c) Z Q + H cg(c) (Y ) (Y ) dY Y (c) pE (Y ) = g(c) HE ( ) c g(c) + H E(Y ) Z Q p g(c) (Y )dY: Z Q cg(c) (Y ) (Y ) dY Y (c) Y (c) Thus, using (5), we obtain 1 dW =p g(c) dc E (Y ) E(Y ) Z =p E (Y ) E(Y ) 1 + Q (Y ) dY Y (c) Z ! Z Y (c) cH (Y ) (Y ) dY 0 Y (c) (p cH (Y )) (Y ) dY (19) 0 We are interested in maximizing W on the interval [0; 1 ], therefore we have to analyze the sign and zeros of the above expression. The derivative w.r.t. c of the right-hand side is H Z Y (c) (Y ) (Y ) dY; 0 thus W is concave. Because the right-hand side of (19) is positive for c = 0, an optimum exists if and only if Z Q Z p E (Y ) p (Y ) dY (1 )H (Y ) (Y ) dY < 0 (20) E(Y ) p 0 and in this case it satis…es the F.O.C. ! Z Q Z Y (c) p E (Y ) (Y ) (Y ) dY: (Y ) dY = c H E(Y ) Y (c) 0 (21) If the introduction of the weight does not in‡uence h the averagei income, the p p sign of the …rst derivative of W on the interval (Q)H ; (p)H is positive. 11 In fact (19) is increasing, thus if Y Z dW = dc Y (c) then by (18) (Y ) p cH , thus by Y (c) (p cH (Y )) (Y ) dY > 0 0 and the optimal value can be found only if p c = HE ( ; Y h lies in the interval under the distribution is increasing we have p (p)H ; 1 of Z i Rp (Y ) dY p Rp 0 = p) H 0 (Y ) (Y ) dY . The term E ( ; Y p) is the average for individuals having income Y p. Since p [ (p) (Y )] (Y ) dY 0; 0 thus (p) E ( ; Y value is given by p) and c c = max the maximum point being 1 p (p)H . p HE ( ; Y Therefore in this case the optimal p) ;1 if (20) is not satis…ed. 12 ; Discussion Papers recentemente pubblicati Anno 2006 0601 – Francesco MENONCIN “The role of longevity bonds in optimal portfolios” (gennaio) 0602 – Carmine TRECROCI, Matilde VASSALLI “Monetary Policy Regime Shifts: New Evidence from Time-Varying Interest-Rate Rules” (gennaio) 0603 – Roberto CASARIN, Carmine TRECROCI “Business Cycle and Stock Market Volatility: A Particle Filter Approach” (febbraio) 0604 – Chiara DALLE NOGARE, Matilde VASSALLI “A Pressure-Augmented Taylor Rule for Italy” (marzo) 0605 – Alessandro BUCCIOL, Raffaele MINIACI “Optimal Asset Allocation Based on Utility Maximization in the Presence of Market Frictions” (marzo) 0606 – Paolo M. PANTEGHINI “The Capital Structure of Multinational Companies under Tax Competition” (marzo) 0607 – Enrico MINELLI, Salvatore MODICA “Credit Market Failures and Policy” (gennaio) 0608 – J.H. DRÈZE, E. MINELLI, M. TIRELLI “Production and Financial Policies Under Asymmetric Information” (febbraio) 0609 – Françoise FORGES, Enrico MINELLI “Afriat’s Theorem for General Budget Sets” (marzo) 0610 – Aviad HEIFETZ, Enrico MINELLI “Aspiration Traps” (marzo) 0611 – Michele MORETTO, Paolo M. PANTEGHINI, Carlo SCARPA “Profit Sharing and Investment by Regulated Utilities: a Welfare Analysis” (aprile) 0612 – Giulio PALERMO “Il potere come relazione sociale. Il caso dell’università baronale italiana” (giugno) 0613 – Sergio VERGALLI “Dynamics in Immigration Community” (luglio) 0614 – Franco SPINELLI, Carmine TRECROCI “Maastricht: New and Old Rules” (luglio) 0615 – Giulio PALERMO “La valutazione dei titoli scientifici dei docenti del Dipartimento di Scienze Economiche dell’Università di Brescia” (settembre) 0616 – Rosella LEVAGGI “Tax evasion and the cost of public sector activities” (settembre) 0617 – Federico BOFFA, Carlo SCARPA “Exporting Collusion under Capacity Constraints: an Anti-Competitive Effect of Market Integration” (ottobre) 0618 – Monica BILLIO, Roberto CASARIN “Stochastic Optimisation for Allocation Problems with Shortfall Risk Constraints” (ottobre) Anno 2007 0701 – Sergio VERGALLI “Entry and Exit Strategies in Migration Dynamics” (gennaio) 0702 – Rosella LEVAGGI, Francesco MENONCIN “A note on optimal tax evasion in the presence of merit goods” (marzo) 0703 – Roberto CASARIN, Jean-Michel MARIN “Online data processing: comparison of Bayesian regularized particle filters” (aprile) 0704 – Gianni AMISANO, Oreste TRISTANI “Euro area inflation persistence in an estimated nonlinear DSGE model” (maggio) 0705 – John GEWEKE, Gianni AMISANO “Hierarchical Markov Normal Mixture Models with Applications to Financial Asset Returns” (luglio) 0706 – Gianni AMISANO, Roberto SAVONA “Imperfect Predictability and Mutual Fund Dynamics: How Managers Use Predictors in Changing Systematic Risk” (settembre) 0707 – Laura LEVAGGI, Rosella LEVAGGI “Regulation strategies for public service provision” (ottobre) Anno 2008 0801 – Amedeo FOSSATI, Rosella LEVAGGI “Delay is not the answer: waiting time in health care & income redistribution” (gennaio) 0802 - Mauro GHINAMO, Paolo PANTEGHINI, Federico REVELLI " FDI determination and corporate tax competition in a volatile world" (aprile) 0803 – Vesa KANNIAINEN, Paolo PANTEGHINI “Tax neutrality: Illusion or reality? The case of Entrepreneurship” (maggio) 0804 – Paolo PANTEGHINI “Corporate Debt, Hybrid Securities and the Effective Tax Rate” (luglio) 0805 – Michele MORETTO, Sergio VERGALLI “Managing Migration Through Quotas: an Option-Theory perspective” (luglio) 0806 – Francesco MENONCIN, Paolo PANTEGHINI “The Johansson-Samuelson Theorem in General Equilibrium: A Rebuttal” (luglio) 0807 – Raffaele MINIACI – Sergio PASTORELLO “Mean-variance econometric analysis of household portfolios” (luglio) 0808 – Alessandro BUCCIOL – Raffaele MINIACI “Household portfolios and implicit risk aversion” (luglio) 0809 – Laura PODDI, Sergio VERGALLI “Does corporate social responsability affect firms performance?” (luglio) 0810 – Stefano CAPRI, Rosella LEVAGGI “Drug pricing and risk sarin agreements” (luglio) 0811 – Ola ANDERSSON, Matteo M. GALIZZI, Tim HOPPE, Sebastian KRANZ, Karen VAN DER WIEL, Erik WENGSTROM “Persuasion in Experimental Ultimatum Games” (luglio) 0812 – Rosella LEVAGGI “Decentralisation vs fiscal federalism in the presence of impure public goods” (agosto) 0813 – Federico BIAGI, Maria Laura PARISI, Lucia VERGANO “Organizational Innovations and Labor Productivity in a Panel of Italian Manufacturing Firms” (agosto) 0814 – Gianni AMISANO, Roberto CASARIN “Particle Filters for Markov-Switching StochasticCorrelation Models” (agosto) 0815 – Monica BILLIO, Roberto CASARIN “Identifying Business Cycle Turning Points with Sequential Monte Carlo Methods” (agosto) 0816 – Roberto CASARIN, Domenico SARTORE “Matrix-State Particle Filter for Wishart Stochastic Volatility Processes” (agosto) 0817 – Roberto CASARIN, Loriana PELIZZON, Andrea PIVA “Italian Equity Funds: Efficiency and Performance Persistence” (settembre) 0818 – Chiara DALLE NOGARE, Matteo GALIZZI “The political economy of cultural spending: evidence from italian cities” (ottobre) Anno 2009 0901 – Alessandra DEL BOCA, Michele FRATIANNI, Franco SPINELLI, Carmine TRECROCI “Wage Bargaining Coordination and the Phillips curve in Italy” (gennaio)
© Copyright 2026 Paperzz