IntroductionSatisfying the participation constraint: getting doctors to serve in rural areasCompetition as an incentive for performanceConclusionsUncited referenceAcknowledgementsReferencesDesigning incentives for rural health care providersin developing countriesJeffrey Hammera,*, William Jackb,1aDevelopment Economics Research Group, The World Bank, 1818 H St., NW, Washington, DC 20433, USAbDepartment of Economics, Georgetown University, Washington, DC 20057, USAReceived 1 March 2001; accepted 1 November 2001AbstractIn many developing country settings, and particularly in rural areas, the implementation ofanything more than very rudimentary contracts for medical care providers, including publicemployees, is virtually impossible. In this paper, we examine the kinds of policy levers thatgovernments might conceivably have available to induce physicians to serve in rural areas. Usingsimple models of screening and spatial competition, we investigate how the government can sortbetween physicians with low and high opportunity costs of relocation, and how the quality ofexisting providers (e.g., traditional healers) might affect the government’s training policies.D 2002 Elsevier Science B.V. All rights reserved.JEL classification: H42; I11; O18Keywords: Physician incentives; Rural health care1. IntroductionSince 1977, the international health community has put a great deal of emphasis onensuring universal access to basic primary care as a high priority for public action. Whilethe rationale for this emphasis is questionable,2we take this goal as given. In practice,attempts at universal provision have often been disappointing. Reliance on the privatesector to put trai ned professionals in such areas is not warranted as demand for the services0304-3878/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.PII: S 0304-3878(02)00063-9* Corresponding author. Tel.: +1-202-458-1410.E-mail addresses: [email protected] (J. Hammer), [email protected] (W. Jack).1Tel.: +1-202-687-0773.2See Filmer et al. (2000) for a critique of this approach to health care policy.www.elsevier.com/locate/econbaseJournal of Development Economics 69 (2002) 297 – 303at the true opportunity cost of the professionals’ time is simply too low. State interventionis obviously needed. On the other hand, public performance has met with, at best, variablesuccess. Disappointment stems from several sources but a common problem is the inabilityto staff and supply medical posts in rural areas. High vacancy rates, rates of absenteeism,simple lack of conscientious or courteous care, and frequent lack of supplies such asessential drugs are common in many public facilities.A vast literature examines how to optimally pay physicians, in the context of theprincipal-agent model.3These models are applicable to institutionally rich environments,such as exist in urban areas of industrialized countries. However, sophisticated perform-ance-based contracts are difficult to enforce in poorer countries, so we restrict ourselves tosimple mechanisms. The first is a menu of rural service options that specifies alternativerewards for different lengths of rural service. If the state provides medical training, theserewards might be implemented through adjustments to tuition fees or repayments ofstudent loans, conditional on the length of service.The second model builds on the first, and examines the endogenous behavior ofphysicians who operate in rural areas. We assume that explicit incentives cannot beprovided by the central authorities, and that performance responds only to competitionwith local alternative providers (e.g., traditional healers). Consistent with virtually allstudies of physician behavior in developing countries, we assume that both providerscharge fees. We use a differentiated goods model of competition that allows us to ask howwell a rural physician who locates in a rural village should be trained. The level of trainingaffects both the average quality of service provided in the village, but also the income ofthe physician. This feeds ba ck into the optimal rewards that must be offered to inducerelocation in the first place.In this paper, we do not address the public/private issue: that is, we make no distinctionbetween public servants and private providers contracted by the governm ent. These issues,related to the question of asset ownership, have been dealt with elsewhere (e.g., Hart et al.,1997; Glaeser and Shleifer, 1998; Besley and Ghatak, 2000). Instead, we simply ask howphysicians should be compensated and trained, when explicit incentive mechanisms areinfeasible. The next section models the design of rural service options, taking theperformance of the physician at village level as fixed. Following this, Section 3incorporates competition at the village level to endogenize physicians’ market incomes,under the assumption that explicit incentives cannot be provided from the center.4Section4 briefly concludes.2. Satisfying the participation constraint: gettin g doctors to serve in rural areasThis section examines how a government might pay physicians to serve in rural areaswhen some physicians are more willing to move than others. This heterogeneity could3See, e.g., Ma (1994), Ma and McGuire (1997), McGuire (2000), and Newhouse (1996).4The underlying models are familiar to students of basic microeconomics, so are not presented in any detailhere. The interested reader is referred to the working paper version of this paper, Hammer and Jack (2001).J. Hammer, W. Jack / Journal of Development Economics 69 (2002) 297–303298derive, for example, from differences in underlying preferences or opportunity costs. Theonly variable upon which payment can be conditioned is length of service, denoted q.Weassume for simplicity that the quali ty of servic e is independent of the time spent on thejob,5that it is independent of the physicians’ willingness to move,6and that the socialvalue of rural service per unit time is constant, b.7The total payment received by thephysician from the government is P, and the cost of staying in the rural area can be high orlow, and is increasing and convex in the time spent. Denote the cost by ci( q), wherei = L,H, cL( q)<cH( q), and cLV( q)<cHV( q), for q>0. A proportion / of physicians have lowopportunity costs, and a proportion (1  /) have high costs. A physician’s net utility issimply P  ci( q).The government’s objective is to maximize the social benefit of medical