How service providers are paid matters as much as how much they are paid

What key insights have emerged from development economics in the past decade, and how should they impact the work of the World Bank? A new working paper Toward Successful Development Policies: Insights from Research in Development Economics from the Bank’s research department captures 13 of the most significant insights in the world of development economics.

Here’s insight #3 on how health and education service providers should be paid in order to ensure the delivery of quality services. See all previous insights here: Thirteen insights for successful development policies

Service providers, such as schoolteachers or health workers, must be paid. How they are paid shapes their incentives to provide high-quality services as much – if not more than – how much they are paid. Service providers around the world typically are paid in one of three ways, each of which is problematic:

• According to the specific services they provide (fee-for-service). This approach is quite common in health. This approach incentivizes the provision of services, and if the payer can observe quality, it may also encourage the provision of quality services. But it also incentivizes the provision of unnecessary services in environments where it is hard for the payer to establish whether the marginal service is necessary. Fee-for-service also encourages providers to “upcode” – to pretend the service was more complex than it really was.

• According to the number and characteristics of the people they serve (capitation). A school could get a fixed amount per enrolled pupil. A primary health care provider could get a fixed amount per person on their list. These amounts could vary according to the demographics of the population served, with schools getting more generous capitation payments for older or poorer children and primary health care providers getting more generous payments for seniors and women. A hospital could get paid a specific amount for each patient with a specific diagnosis or for each patient in a specific diagnosis group. However, capitation encourages providers to list more people as being served, but disincentivizes delivering services to them – since providers get the capitation payment anyway. It also discourages providers from taking on hard-to-serve groups, e.g. very old people or students with learning difficulties, and to skimp on quality.

• According to their own characteristics (budget-and-salary). A doctor or a teacher could get a salary, linked perhaps to their seniority; a hospital could get a budget linked perhaps to the number of beds it has or the number of operating rooms; a school could get a budget based on the number of classrooms it has. But salaries and budgets fail to reward effort, and often, as one commentator^ref2 puts it, “condone on-the-job leisure”.

Unsurprisingly, sticking with the same payment method and simply making it more generous is not the answer. A study of public-sector health providers in India^ref3 found that the size of the provider’s salary made no difference to the quality of their performance. This is consistent with an experiment in Indonesia^ref4 where teachers’ salaries were doubled with no impact on student learning.

Some countries have decided to switch between these three approaches. During the 1990s and early 2000s, countries in Europe and Central Asia (ECA) moved from paying hospitals through budgets to either fee-for-service or some form of method based on patient characteristics. As expected, in countries in the first camp, inpatient admissions increased, while in countries in the second camp, average length-of-stay (a measure of the volume of care) fell.^ref5

Reforms along these lines fail to address the key point that all three payment methods have their strengths and weaknesses. Moving from one payment method to another simply means trading one set of strengths and weaknesses for another.  A more sophisticated approach has been to search for a blend of one or more of the payment methods in the hope of encouraging providers to deliver only necessary quality services. In health, getting the right formula in blended schemes has proved a challenge – in both developed and developing countries. So too has setting up the necessary additional administrative machinery, making some ask whether the extra benefits are worth the extra cost.^ref6

In China, where primary health providers have traditionally been paid fee-for-service, the concern has been over-provision of care, especially of drugs and tests which were priced above cost in the government-set fee schedule.^ref7 Recently the country has explored alternative payment methods. In one experiment,^ref8 one group of health centers was left being paid fee-for-service while the other group was switched to a mix of capitated global budget and performance-related payments based on the appropriateness of drug prescribing and patient satisfaction. The blended approach led to less antibiotic use and lower patient out-of-pocket expenditures.

An increasingly common hybrid approach in budget and salary systems is to add a pay-for-performance element to the payment system. In education, schools and individual teachers might receive additional payments linked to student learning outcomes. In health, facilities and individual health workers might receive additional payments linked to the delivery of specific services (e.g. antenatal visits, the delivery of babies in a health facility, childhood immunization) or to the quality of care (e.g., a specific set of interventions being delivered, and specific checks being done during the antenatal visits).

Many of these are supported by the World Bank and many have been or are being rigorously evaluated. An ongoing Development Research Group study of the experiences to date in the health sector uses meta-analysis to distill results across these evaluations. Preliminary results point to substantial heterogeneity in impacts across countries and across indicators. Emerging findings suggest relatively large and statistically significant effects of pay-for-performance on facility deliveries and on the content of antenatal care visits, but limited or negligible effects on other indicators, including the likelihood of a woman having four or more antenatal care visits, the likelihood of a baby receiving postnatal care and a child being fully immunized. In education, too, the results to date are mixed. Some studies^ref9,ref10 have found positive effects of pay-for-performance schemes on teacher behaviors and student learning, while others^ref11,ref12 have found no effects.

The health meta-analysis is exploring the causes of the heterogeneity in results, which may be due to differences in the design of the pay-for-performance scheme (e.g., the generosity of the payments relative to cost) and its implementation. Evidence from the education sector is already pointing in this direction. A recent study^ref13 found a pay-for-performance scheme in Tanzania’s schools was more successful if combined with additional spending on school inputs.

In summary, all payment schemes shape service provider incentives, and all three basic schemes have advantages and disadvantages. Blending elements from one or more holds promise but getting the design and implementation of a blended provider payment scheme right is not straightforward.