Development agencies are increasingly interested in making aid more transparent, stakeholder-led, and effective by expanding the use of payment by results (PbR) — rewarding those implementing projects on the basis of results delivered instead of paying for inputs. For payment by results to work, you have to get a lot of things right. It has to be for the right kind of programme targeting the right results, properly measured and rewarded in the right way. These issues, and more, are laid out in Stefan Dercon and Paul Clist’s 12 principles for payment by results (PDF).
We agree with much of what they have to say, but their suggested principles would be most appropriate if development cooperation were characterized by stable and predictable relationships between inputs such as teachers hired and outcomes such as basic literacy, albeit subject to unexpected shocks and principal-agent constraints. (Their analysis is set out in more depth in a previous paper co-authored by Clist). Some of their principles would make sense if the comparison were between payment by results and the hypothetical administration of inputs by aid agency staff who have perfect information about the ‘production function’ that mediates the relationship between inputs and outcomes.
We don’t think development cooperation works this way. Below, we set out an alternative 12 principles for payment by results in the real world alongside the corresponding principles proposed by Clist & Dercon.
In our view, development and service delivery are usually the outcomes of complex relationships between inputs, outputs, and outcomes that are not completely known in advance, but discovered through a process of experimentation, learning, and adaptation. We take the view that the advantages and disadvantages of PbR should be judged in comparison to existing input-based aid programmes, which can have serious shortcomings.
Where We Agree
There are significant points of agreement between our principles and the corresponding ones proposed by Clist & Dercon, and we have also benefited from a useful exchange with them about these points. We agree that:
PbR is not a panacea for all the problems of development cooperation, and applied badly or in the wrong circumstances it may be worse than the alternatives;
PbR is unlikely to succeed if it is conceived primarily as a way to use financial incentives to oblige governments or contractors to do something that they would not otherwise do;
the definition and measurement of a performance measure is essential to the success of PbR;
risk transfer from donors to delivery organisations is not an objective of PbR;
monetary incentives might weaken intrinsic motivation of individuals, and alternative mechanisms to create effective incentives should also be considered; and,
evidence of the success of any aid programme should be in terms of impact rather than inputs or intermediate effects.
Where We Disagree
In other important respects our principles differ from those of Clist & Dercon, as a consequence of a different view about development cooperation in practice, particularly the challenge of managing complexity:
For Clist & Dercon, the main purpose of PbR is to solve principal-agent problems in service delivery by improving incentives. Their discussion appears to focus mainly on output-based contracts with service delivery organisations. For us, the main purpose of PbR is to enable autonomy for local implementers, which is essential to find solutions to complex problems — including results-based aid agreements with government agencies.
For Clist & Dercon, PbR is less attractive in uncertain environments such as fragile states; in our view PbR may be more attractive relative to other aid instruments in these contexts, because they are precisely the situations in which local autonomy and experimentation are most useful.
Clist & Dercon characterise measuring results as a cost; we believe that measuring results is more useful — and may be no more expensive — than measuring inputs and activities because it generates useful information, feedback, learning, and improvement.
Finally, we agree with a point made by Paul Clist in the longer paper from which their principles are drawn: donors need to consider the contracts they use for PbR and, where appropriate, make sure that they create sufficient opportunity for implementing agencies to experiment, fail, learn, and adapt. On average, payment by results requires that the payments for successful outcomes must also cover the costs of failures along the way, but this logic is not meaningful if project horizons are so short or working capital so scarce that implementers only have one chance at getting it right. PbR may create circumstances in which experimentation and adaptation are feasible; but this does not mean that all problems can be tackled this way, nor can every implementing partner take advantage of the freedoms that PbR should offer.
We contend that our 12 principles for payment by results are more relevant in the real world, which is characterised by high levels of ambiguity, uncertainty, and unexpected change.
12 Principles for Payment by Results