Why are Goodhart, Dukenfield and Campbell important for development?

Charles Goodhart, William Claude Dukenfield and Donald Thomas Campbell are three quite different people. Goodhart, born in 1963, is an economist professor emeritus at the London School of Economics. Dukenfield, 1880-1949, better known as W.C. Fields, was an American comedian, actor, juggler and writer. Campbell, 1916-1996, was an American social scientist. What do they have in common? And why do I think that they have some special relevance for development and particularly for what is called ‘managing for results’? It all comes down to the three ‘laws’ named after the three gentlemen.

Goodhart’s law

Let’s start with Goodhart’s law. It’s original formulation (taken from Wikipedia) was

Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.

Today, it’s most popular formulation (coined by Marilyn Strathern, a British anthropologist) is:

When a measure becomes a target, it ceases to be a good measure.

Goodhart’s law tells us for example that when a project’s target is to increase income for a number of poor people, then measure of success for a project should not – or not alone – be the target people’s income.

The reason for this can be explained through Dukenfield and Campbell’s laws.

Dukenfield’s law of incentive management

Dukenfield’s Law of Incentive Management was proposed as such by Mark Kleiman in the Atlantic Magazine. In it’s original wording (written in You Can’t Cheat an Honest Man), it states:

If a thing is worth winning, it’s worth cheating for.

Kleinman proposes the following general formulation for the law:

Any incentive to create a result also creates an incentive to simulate the same result.

Kleinman proposes two corollaries:

One: the greater the incentive, the greater the temptation.

Two: holding the level of audit effort constant and other things equal, the reliability of a measure will decline as the importance attached to it grows.

Translated into the world of development, this means that if we define for example increased income as the successful result of a project and incentivize the achievement of this result, then there is an temptation to take short-cuts to achieve the results, even if only at the time when it is being measured.

Going back to Goodhart, it becomes clear that measure and result need to be separated. While it is indeed an important result of a development project to increase the income of poor people, the measure of success should actually be on a deeper level in terms of achievement, e.g. the sustainability of this change or deeper systemic change that lead to higher income.

Campbell’s law

Campbell takes the same logic into the realms of decision-making (again from Wikipedia):

The more any quantitative social indicator (or even some qualitative indicator) is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor.

This basically says that if the achievement of a result is used for decision-making, e.g. whether a project contract is extended, then there is a tendency to specifically work to achieve this particular result, even if a shift in strategy would be better to achieve the overall goal of a project.

In conclusion

So in conclusion, if a measure becomes a target, then there is an incentive for people to either cheat to achieve it or to adapt their strategy to achieving this particular result even if the achievement of the overall development goal would require a change in strategy.

This closely links to the following quote from the New Scientist (12th April 2011, pp 40-43):

The facts are absolutely clear.  
There is no question that in virtually all circumstances in which people are doing things in order to get rewards, extrinsic tangible rewards undermine intrinsic motivation.

Hence, we have to rethink how we treat and measure results, how we motivate project staff and how we finance projects. The currently fancy idea to pay organisations by the results they achieve seems in the light of the above cited laws a pretty bad idea.


2 thoughts on “Why are Goodhart, Dukenfield and Campbell important for development?

  1. Michael Eddy (@MichaelEddy)

    Thanks for blogging this. While the discussion it invites is welcome, the conclusions drawn here appear overly broad.

    First, while you’re right that incentives can in certain circumstances crowd out intrinsic motivation, it’s a very incomplete reading of the literature. More broader reviews identify a number of conditions where well-designed incentives can complement rather than crowd out intrinsic motivation (see, for example, Geezy, Uri 2011, link at bottom).

    Second, compared to the status quo, where organizations are typically held accountable for receipts rather than results, well-designed results-based finance at least manages to things that have intrinsic worth (e.g. learning outcomes) rather than only instrumental worth (% budget executed, # of textbooks purchased, etc).

    Third and related, while sometimes results-based finance is framed merely as a way to place carrots in front of individuals/organizations, it goes well beyond that, enabling greater autonomy for service providers and front-line workers, allowing them to use data to make better decisions, course correct and adapt to complex and changing environments.

    Of course, this doesn’t mean that incentives are a panacea. Poorly implemented results-based finance (just as poorly implementing any program) can lead to unintended consequences, gaming and corruption. Organizations wishing to achieve greater value for money through results-based finance should be thoughtful and considerate in how they implement it, ensuring that new instruments build on both the successes and failures of previously implemented results-based finance instruments. That said, early experiments with RBF show that when well designed, RBF can achieve better outcomes at a fraction of the cost.


    1. Marcus Jenal

      Hi Michael. Thanks for your comment. Very interesting. I’ll read the referred paper with interest. I’m still skeptical about working too strongly with results as driving force of development, especially as our ability to predict great solutions for complex problems (on which in the end the fixing of contractually agreed results depends) isn’t all too great. I see your point that the current situation isn’t ideal either. I’m curious in any case to observe the on-going discussion.


Leave a Reply