Concepts / Measurement
Goodhart's law.
When a measure becomes a target, it can stop being a good measure. An indicator chosen to track something important starts to be managed for its own sake, and the number improves while the thing it was meant to show may not.
Named after the economist Charles Goodhart, who observed in 1975, writing about monetary policy, that a statistical regularity tends to collapse once it is relied upon for control. The familiar one-line version above is a later paraphrase, popularised by the anthropologist Marilyn Strathern in 1997.
How it works
From measure to target, in four steps.
Nobody needs to act in bad faith. The shift follows from what is rewarded.
- 01
An indicator is chosen
A number is picked because it moves with something that matters, and it is easy to collect and compare.
- 02
It is tied to rewards
Promotion, funding, recognition or political credit start to depend on the number.
- 03
Behaviour shifts
Effort moves towards improving the number: easier cases, reclassification, activity counted in place of results.
- 04
The number stops informing
The indicator improves while the outcome may not. It no longer tells decision-makers what they need to know.
Where it can show up
Wherever numbers carry rewards.
These are general patterns, not findings about any particular programme, department or government. Whether they occur in a given case is a question for evidence.
Welfare and services
Counting enrolments or beneficiaries, rather than whether people are better served.
Activity reporting
Counting meetings, events and reports submitted, rather than assessing their effect.
Recorded figures
Some statistics change with how cases are recorded and reported, not only with what actually happens.
Infrastructure
Counting what has been built, rather than whether it is used, maintained and useful.
How to guard against it
Keep the number honest.
Indicators remain useful. The aim is to stop any single number from becoming the goal.
- Use several measures, not one, so that improving a single number is not enough.
- Keep some measures out of rewards, and use them to learn rather than to judge.
- Read numbers alongside field evidence: visits, interviews and complaints show what a dashboard cannot.
- Review indicators regularly, and retire those that have stopped tracking the outcome.
- Ask the awkward question: what would improve this number without improving the result?
Origins and further reading
Where the idea comes from.
A closely related idea, Campbell's law, holds that the more a quantitative social indicator is used for decisions, the more it is exposed to pressures that distort both the indicator and the process it is meant to monitor.
- Goodhart, C. A. E. (1975). Problems of monetary management: the U.K. experience. Papers in Monetary Economics, Reserve Bank of Australia.
- Campbell, D. T. (1979). Assessing the impact of planned social change. Evaluation and Program Planning, 2(1), 67–90.
- Strathern, M. (1997). ‘Improving ratings’: audit in the British university system. European Review, 5(3), 305–321.
- Muller, J. Z. (2018). The Tyranny of Metrics. Princeton University Press.
About this concept
A general idea, not a verdict.
This page explains a general concept from economics and social science. It does not describe or assess any particular party, government, programme or election. Related concept: The McNamara fallacy. Related perspectives: Consistency is not accuracy and When the ground stops reaching the top.