Performance-based payments (P4P) are seen by leading policymakers as a potential solution to the slow progress in achieving the main target of Millennium Development Goal 5: reducing the maternal mortality ratio by two-thirds by 2015 [1–3]. The basic principle behind P4P is that payments are contingent on performance. For example, an increase in the utilization of health services can trigger a bonus to the responsible health workers/managers . It is assumed that payments to health-care providers will induce them to offer maternal health-care services of a higher quality, and that this in turn will increase the number of women and children who receive high-quality care. In Africa, 17 countries are now implementing P4P, 14 of them as pilots, and three as nationwide programmes .
Critics argue that P4P need not result in improved health outcomes, and that it can have adverse effects [6–10]. In particular, P4P may crowd out motivation and/or attention for tasks for which health workers are not paid. Health workers may also play with the books and alter numbers and not behaviour. Lastly, P4P programmes are criticized for focusing on the quantity- rather than the quality of care.
Proponents of P4P refer to a relatively limited number of studies that demonstrate that P4P can have a substantial effect on the utilization of health-care services and to some extent on health outcomes [11–15]. A weakness of these studies, however, is that they do not tell us precisely how or why these changes take place. If we obtain a better understanding of what health providers think about P4P programmes , and what they actually do to make changes occur, it will help us shed some light on any potential adverse effects, as well as helping us understand why P4P appears to work well in some settings and not in others.
The majority of P4P studies that have found positive results in terms of utilization and/or outcomes have used data from programmes in which substantial technical and managerial resources accompany the intervention. In countries with weak health sector institutions, this is unlikely to be the case during nationwide roll-out organized by both national and local governments. It is therefore important to also study interventions that are implemented by governments in low-income countries using their own resources.
This paper offers insights into two of the aspects lacking in the current literature on P4P: Why changes may occur in response to the introduction of P4P, and how the intervention plays out when implemented by local government as part of a national programme that does not receive donor funding. The paper is structured as follows: The first section looks at the underlying conceptual framework of P4P, while the second section describes the nationally funded P4P programme in Tanzania. The third section lays out the methods that we employed, whereas the fourth section reports on the findings from the in-depth interviews with health workers and health administrators. The fifth and sixth sections offer a discussion of the results and concluding remarks.
Payment for performance (P4P), or performance-based payment (PBP), can be defined as: “the transfer of money or material goods conditional upon taking a measurable action or achieving a predetermined performance target” . Underlying P4P is the principal-agent model. The principle behind this model is that there is a lack of alignment of the preferences (interests) of the principal (employer) and the agent (employee) when it comes to the goals to be achieved by an organization. The principal therefore attempts to find ways of aligning the agent’s goals to the goals of the organization . In the context of P4P, health workers, or 'agents’ , are provided with performance bonuses by the principal in order to achieve health outputs and outcomes.
Nevertheless, even if a health worker is strongly motivated by the reward the principal offers and changes his behaviour in response, the intervention may not necessarily improve health outcomes.
First, financial incentives may crowd out attention to tasks important for high-quality care. Health workers may focus on aspects of health care for which they are rewarded, while ignoring other aspects of care for which they are not rewarded yet are nonetheless important for quality . Holmstrom and Milgrom coin the former type of behaviour multitasking, and argue that financial incentives may not be very effective in the health sector, as employers pay for input rather than for health outcomes .
Second, external rewards may crowd out health workers’ intrinsic motivation to do the job . Clearly, if health workers were intrinsically motivated to deliver high-quality health care, there would be no need for employing financial incentives as a motivator. Studies comparing what health workers in low-income countries can do with what they actually do [21–23] suggest that the intrinsic motivation for the average health worker is low. However, the use of external rewards may still crowd out the motivation for those health workers who are intrinsically motivated, and if the use of external rewards for some reason is discontinued, motivation and effort may end up being lower than before these rewards were introduced.
Third, P4P is vulnerable to corruption, i.e. health workers may be changing the numbers rather than the indicators themselves (see Oxman and Fretheim for a review of adverse effects . Fourth, a major assumption of P4P is that workers are able to offer high-quality care if they choose to. This may not be the case, as knowledge of guidelines and access to equipment and medication may be inadequate. Lastly, P4P presupposes that if health workers deliver high-quality care, women will come to the facility. This may not be the case since women may deliver at home for reasons outside the control of health workers.
In conclusion then, agency theory tells us that offering health workers an incentive for an increase in the number of deliveries need not result in improved health outcomes, primarily because the incentive is related to effort and not to outcomes.
P4P in a Tanzanian context
The productivity of health workers in Tanzania has been proven to be low. One study shows that less than 60% of working hours are used for productive activities, (:3) whereas another demonstrates that few health workers follow clinical guidelines, and that low motivation is a central factor . Lastly, a number of studies have shown that health workers in Tanzania are unhappy with their working environment and their salaries [25, 26].
Norway, one of Tanzania’s long-term development partners, took a leading role in introducing the idea of result-based financing in the country’s health sector in 2008 [2, 27]. The government of Tanzania was very receptive to the idea and wanted to launch a national P4P pilot programme in 2009 . Tanzania’s development partners in the health sector were reluctant to endorse the idea due to many contested issues. First, there was a strong feeling that the state of Tanzania’s health management information system (HMIS) was not ready for P4P, which had been documented in an appraisal study carried out in 2009 . Second, other perceived preconditions for a successful P4P, such as a satisfactory staff situation and adequate access to essential drugs, equipment and supplies, were lacking . The government of Tanzania was therefore not allowed to use the funds in the health basket earmarked for P4P. While acknowledging that the proper conditions for P4P were lacking, the government of Tanzania proceeded with the implementation of P4P in 2009, choosing to employ a 'learning by doing approach’ .
The donor community requested that the government of Tanzania halt P4P . At this point, however, the government of Tanzania had already issued a directive that P4P should be included as an activity in the districts’ Comprehensive Council Health Plan (CCHP) for 2009/10. Even so, not all the districts followed the directive, and in some places health workers were eagerly awaiting a P4P programme that was never implemented . On the other hand, the district administration in Mvomero District decided to follow the directive and P4P was consequently budgeted for in the health plan and implemented in 2009, and health workers received their first bonuses in 2010. To distinguish the P4P scheme that we study from the donor-funded scheme which was later launched as a pilot in Pwani Region in 2011 , we will refer to it as the 'locally funded P4P’.
The design of the locally funded P4P in Tanzania (2009–2011)
The main aim of the locally funded P4P in Tanzania was to “provide better motivation and explicit attention to results, by ensuring that health workers and their supervisors are motivated to strive for better results in Maternal, Newborn and Child Health Services and other health services in the districts” . The bonuses were to be paid based on achievements using the following indicators: antenatal care, institutional deliveries, post-natal care, and Health Management Information Systems (HMIS). Council Health Management Teams (CHMTs) were to monitor and ensure that health facilities in their mandated area were submitting their reports in time, and they were also to review and verify the reports. In turn, the CHMTs were to be monitored by the Regional Health Management Teams (RHMTs) .
At the facility level, a maximum annual bonus was to be achieved if the facility met all the targets for all the indicators, whereas a partial bonus was to be paid if only some of the targets were met. For deliveries, the target for dispensaries was that 60% or more of all the expected deliveries of the catchment area should take place at the dispensary. At the national level, 51% of all women with a live birth received delivery care from a skilled provider in 2010, although the percentages vary between 21% and 91% across regions . New targets were to be set at the beginning of each year and the basic rule for target setting was the requirement of improvements from the previous performance . Furthermore, the bonuses were to differ according to facility type. Dispensaries had a maximum bonus limit of T.Shs 1 million (approximately USD 676) and health centres, CHMTs and RHMTs’ maximum annual bonus was T.Shs 3 million (USD 2,000), while hospitals had the highest maximum annual bonus of T.Shs 9 million (USD 6,000). Payments at the health facility were to be shared equally among the staff regardless of grade, qualifications or position. If a health facility reached all targets, each individual was supposed to get a maximum annual bonus of approximately T.Shs 200,000 (USD 136) .