Measuring Performance According to Goodhart
By now, only 20% of the people reading this post are sticking to their 2019 New Year’s resolutions. To understand why, we dig through the work of economist Charles Goodhart in this post to see if we can learn why so many performance goals are left unmet.
Enormous Nails, Goodhart, and Purpose:
You might have noticed it in your own workplace. Someone higher up the ladder introduces a new performance measure and immediately, everyone prioritizes that measure – often with unintended and at times bizarre consequences.
A classic anecdote of this in action involves a nail factory in the former Soviet Union. The government put in place a measure of factory performance based on the number of nails produced. So what did the workers produce? Millions of nails that were way too tiny and not of high enough quality for construction. To try to make sure this didn’t happen again, the government changed the measure of productivity to weight of nails produced. The workers responded by producing enormous nails, way too large to be used in the production of anything practical. Hilarious!
While the story above is likely false, it illustrates very well Goodhart’s law, named after economist Charles Goodhart. He stated that “Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.” Famously rephrased by Marilyn Strathern in an article about financial auditing of universities, it means that “when a measure becomes a target, it ceases to be a good measure.” As soon as you use something as a performance metric, it stops being an accurate measure because we will always look for ways to game it.
Goodhart’s law can be seen at play across many industries. Call centers who measure employee performance by the number of calls answered naturally have a lower satisfaction rate. Hospitals that measure custodian performance by how fast they can clean a patient’s room are often dirtier. Salespeople rewarded for sales volume will sell more product by slashing margins. Evaluating a dog’s loyalty by how much it barks at dangerous strangers will instill a pavlovian response to bark at everything to the point of driving their owner crazy.
With any performance metric, there will be a way to optimize it at the detriment of something that really matters. And if a dog can play into Goodhart’s law, you certainly can, too.
Goodhart's Law and You
Goodhart’s law is not limited to work environments; it also permeates our personal lives. We all set goals that we want to accomplish, be it losing 15 pounds, waking up earlier, reading more books, or spending more time with family. Depending on our measures of progress on these goals, we may fall victim to Goodhart’s law. People starve themselves to lose weight, causing many negative side effects. People may take an extra vacation so as to spend more time with their family, but spend the whole time checking work emails on their cell phone. Yes, they logged more clock-hours with their family, accomplishing their goal on paper, but intrinsically important outcomes of are lost due to the progress measure used.
It’s important to employ a different philosophy when it comes to goals and self-improvement. To lose 24 pounds and reach your optimal BMI, a simple algorithm can say that you need to walk 12,214 steps and consume 1654 calories a day to reach your optimal weight in 46 days. But this way, the reason why we want to live a healthier lifestyle is far removed from meeting a step goal for the day.
It is important to understand and play into the motivations for our goals—and these motivations are often aligned with our purpose. Having a clearly articulated purpose makes it easier to see where our motivations are coming from. What’s more, purpose-oriented goals are also more sustainable. Focusing just on what we want to improve about ourselves and not why we want to change these things is a surefire way to hit a dead end or play into Goodhart’s law and make an improvement in a superficial, unsustainable way. Aligning goals with a sense of purpose forces us to rely on internal motivation, rather than falling back on the external motivation of seeing what the number on the scale is.
Using Why Chains to Evade Goodhart's Law
One way to uncover these deeper motivations is to take a look at your purpose and personal values and create a “why chain” that links your goal to your purpose and values; keep asking yourself Why do I want to do this? until you hit something more profound. Take the following example:
I want to save 20% of my paycheck for an emergency fund. Why?
I want to be financially stable. Why?
I want to be financially stable because it will help me feel more relaxed. Why?
I want to be financially stable because it will help me feel more relaxed, which will help me be more present in my daily life.
After a while, you can stop asking Why? and start seeing the connections that naturally emerge between what was originally a surface-level goal and what is now a reflection of your values and purpose. Linking goals to our personal values and purpose instead of just static performance metrics is a helpful—and effective—way to encourage internal motivation, and internal motivation is powerful in maintaining goal-achieving behavior. A study on obesity and long-term weight loss found that successful maintenance of a target weight is associated with internal motivation. One 2017 master’s thesis even found that ESL kindergarteners are more successful at learning the English alphabet when they are internally motivated by setting goals.
Above all, just being aware of Goodhart’s law is a step in the right direction. Whether you are the person in charge of performance incentives at your organization or looking to improve your personal life, the implications of Goodhart’s law are clear. It is not enough to just set a goal—understanding why we are reaching for this goal is perhaps more important in motivating us to deliberately follow through. And connecting this goal back to our purpose helps ensure that the way we are measuring goals results in strong nails – not millions of useless ones.