Why Measuring Economic Performance Always Ruin the Metric
Formulated by British economist Charles Goodhart in 1975, Goodhart's Law states that when a measure becomes a target, it ceases to be a good measure. When central banks or governments target a specific economic metric, people alter their behavior to optimize for that metric rather than the underlying goal. This distorts data, encouraging gaming of the system instead of genuine economic progress.
The Origin in Monetary Policy
Goodhart's Law was originally formulated in 1975 by British economist Charles Goodhart during his tenure as an adviser to the Bank of England. In a research paper addressing monetary management in the United Kingdom, Goodhart observed a persistent frustration facing policymakers: whenever a central bank identified a specific monetary aggregate that appeared reliably correlated with broader economic variables like inflation or output, attempting to control that aggregate caused the historical relationship to collapse.
In his original formulation, Goodhart wrote that 'any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.' The problem was rooted in human adaptability. Financial institutions and market participants naturally changed their practices, introduced new financial instruments, and restructured their balance sheets to circumvent regulatory controls. As a result, the measured indicator ceased to reflect the economic reality it had previously tracked.
From Economic Rule to Universal Aphorism
Although Goodhart's observation began as a technical insight about central banking and monetary aggregates, it quickly resonated across other disciplines. In 1997, British anthropologist Marilyn Strathern summarized the principle into the concise, memorable aphorism most commonly quoted today: 'When a measure becomes a target, it ceases to be a good measure.' Strathern used the concept to critique the rise of audit culture, bureaucratic evaluation, and performance management systems in higher education.
Strathern highlighted how auditing regimes intended to ensure institutional accountability ended up reshaping academic practices in unproductive ways. When university departments were evaluated on specific quantitative indicators, staff and administrators inevitably began managing the indicators themselves rather than focusing solely on educational quality and scholarly inquiry. This transformation demonstrated that the distortion Goodhart identified in financial markets applies to virtually any organizational hierarchy that relies on proxy metrics.
Parallel Concepts: Lucas and Campbell
Goodhart's Law does not exist in isolation; it shares deep intellectual roots with parallel findings in macroeconomics and the social sciences. In 1976, economist Robert Lucas published what became known as the Lucas critique. Lucas argued that it is naive to predict the effects of an economic policy purely on the basis of historical data, because economic agents alter their behavior and expectations in response to the policy rules themselves. When governments change how they intervene, the underlying structural parameters of the economy shift as well.
A few years earlier, in the 1970s, social scientist Donald T. Campbell formulated what is known as Campbell's law, focusing on quantitative social indicators and educational testing. Campbell observed that the more any quantitative social indicator is used for decision-making, the more subject it becomes to corruption pressures, and the more apt it is to distort and corrupt the very social processes it was designed to monitor. Together, Goodhart, Lucas, and Campbell established that human reflexivity fundamentally limits the stability of social and economic measurements.
The Mechanics of Metric Distortion
The breakdown described by Goodhart's Law occurs because there is almost always a gap between a true objective and the quantitative proxy chosen to represent it. True objectives—such as high-quality healthcare, genuine student learning, scientific innovation, or financial stability—are multidimensional and difficult to measure directly. Administrators and policymakers select observable proxies, such as wait times, test scores, citation counts, or specific capital ratios, assuming the proxy will reliably track the underlying goal.
Once stakes or rewards are attached to the proxy, individuals and organizations face strong incentives to optimize for the measure rather than the objective. This can occur through gaming, outright manipulation, or simply shifting effort away from unmeasured aspects of performance toward measured ones. Because optimizing a proxy is usually easier than improving the complex underlying reality, the correlation between the metric and the actual goal deteriorates rapidly under pressure.
Formal Taxonomies of the Law
In modern decision theory and artificial intelligence safety research, researchers have formalized Goodhart's Law into distinct failure modes. Notably, researchers David Manheim and Scott Garrabrant categorized the phenomenon into four primary mechanisms: regressive Goodhart, extremal Goodhart, causal Goodhart, and adversarial Goodhart. Each mechanism explains a different mathematical or behavioral path through which optimization breaks a proxy relationship.
Regressive Goodhart occurs when a proxy is an imperfect, noisy measurement of the true value; maximizing the proxy tends to select for measurement error alongside real quality. Extremal Goodhart happens when a model operates well within historical parameters but breaks down when pushed to extreme, unprecedented levels. Causal Goodhart involves mistaking a mere statistical correlation for a causal relationship, and adversarial Goodhart occurs when intelligent agents actively strategize and manipulate their behavior to achieve higher scores on the metric.
Real-World Manifestations
The effects of Goodhart's Law appear across diverse sectors. In academic publishing, reliance on citation counts and the h-index to determine hiring and grant funding has led to citation cartels, salami slicing of research into minimal publishable units, and an emphasis on trendy topics over methodical, long-term scholarship. In financial regulation, as economist Jon Danielsson has noted, when regulators require all financial institutions to use the same standardized risk models, banks adjust their portfolios to meet the models' constraints, which can paradoxically generate systemic risk by causing all institutions to act identically during market stress.
Similarly, in workplace management and public administration, tracking software bugs fixed can lead engineers to resolve superficial issues while avoiding complex structural problems, and targeting emergency room wait times has historically led hospitals to hold patients in ambulances outside before officially checking them in. In all these cases, the metric successfully hits its target while the overarching mission is compromised.
Key takeaways
•Formulated by Charles Goodhart in 1975, the law originally observed that historical statistical relationships in monetary policy collapse once used for control purposes.
•The famous aphorism 'when a measure becomes a target, it ceases to be a good measure' was coined by anthropologist Marilyn Strathern in 1997.
•Goodhart's Law aligns closely with the Lucas critique in macroeconomics and Campbell's law in social science, all describing how people alter their behavior in response to measurement systems.
•Modern decision theory categorizes metric failure into distinct mechanisms, including regressive, extremal, causal, and adversarial Goodhart effects.