The costs of performance-based financing

Paying people based on their performance has been under discussion throughout the history of work contracts. The earliest document describing performance contracts, the Hammurabi Code, is nearly 4000 years old and was literally written in stone.1 Since that time, various institutions including Roman society, the churches, trade unions and modern governments concerned with certain bankers’ performance2 have tried to emphasize the “true value of labour”1 and have made a stand against performance contracts. In the March issue of the Bulletin of the World Health Organization, Meessen et al.3 rightly raise the question of whether performance-based financing is just a donor fad or a catalyst for wider reform. Looking at the broader evidence, one finds the following arguments against performance-based financing, based on three main issues. First, there is the issue of its effect on worker motivation in the health sector. It is argued that the introduction of financial incentives into a working environment characterized by a high degree of idealism might actually erode workers’ intrinsic motivation, i.e. their personal satisfaction gained from working in the health sector. This theory that incentives have a negative effect on motivation has been labelled “crowding out”.4 Second, performance-based financing focuses on a certain range of indicators. This can often lead to “gaming”, i.e. the neglect of non-remunerated aspects of work and the focus on remunerated ones (including potentially false reporting). Gaming tends to distort the original purpose of the indicators. It has resulted in such absurd and varied phenomena as: the large-scale shift from normal deliveries to Caesarean sections often conducted at night (observed in Brazil and other Latin American countries);5 the construction of tents for waiting outpatients to “avoid” queues inside British hospitals,6 the refusal of Rwandan hospital pharmacies to deliver the last box of any drug to the ward (thus “preventing” stock-outs);7 and the nearby collapse of the world economy in 2009 due to performance-financed bankers’ interest in short-term benefits. Third, the hidden costs of performance-based financing are not limited to emotional costs (concerning the self-esteem of health workers) and technical costs (due to misdirected focus on indicators). There are considerable costs (both financially and in working hours invested) in establishing a performance-based financing system that continuously monitors the quantity and perceived quality of health-sector performance. These costs can be even higher if the performance contracts are managed by nongovernmental organizations. Overhead costs as high as 50% are reported in areas with difficult geographical access such as eastern parts of the Democratic Republic of the Congo (personal communication, Cordaid, 2009). Governments might be reluctant to accept such costs for their own staff under the management of nongovernmental organizations. In addition, it is frequently reported that health staff who are paid according to their performance spend a considerable percentage of their working time filling out – and sometimes “modifying” – the assessment sheets. All these negative side-effects of performance-based financing are consistently depicted in broader reviews and in detailed examination of its use in Rwanda.7–9 In such a context as Rwanda, it is naturally expected that performance-based financing is likely to bring about systemic changes. Nonetheless, the effect of these systemic changes should be thoroughly scrutinized to ensure that they do not seriously affect health workers’ self-esteem and ethics, the technical quality of their work and the sharing of costs and decision-making between governments, nongovernmental organizations and the health-care providers themselves. For the moment, it seems difficult to answer the question initially raised by Meessen et al.3 Performance-based financing is definitely – as the authors write themselves – not a panacea or a magic bullet. Yet, perhaps we should reformulate the question to ask if it is even a bullet?