This book is recommended by Joel (mentioned in his post on “Econ 101 Management”) so we read it recently in our book club at work.
The premise is that measuring employee performance is guaranteed to distort an organization’s desired results. This assertion contradicts management mantras everywhere, such as “You can’t improve what you can’t measure”. How can a manager know her employees are doing the right thing for the company unless she measures their performance?
Austin attacks this conventional wisdom by creating an economic model of how managers and employees will interact, and then exploring the consequences of that model. The model assumes a certain amount of intrinsic motivation – that the employee wants to do the right thing, but is only motivated to work so hard on their own. So the manager has to offer incentives to get the employee to work harder.
In the model, the employee has two dimensions along which their performance could be measured. If the manager offers incentives based on only one of the dimensions, that will distort the employee’s efforts. Austin uses the example of a job placement service which offered incentives based on the number of job applicants each employee interviewed. The inevitable result was that the employees spent all of their time interviewing applicants and none of their time calling companies to find jobs for those applicants. The goal of the organization was to connect applicants with jobs, but because of the distorted incentive system, it failed completely. Another example is from 21 Dog Years, where customer service representatives at Amazon were measured by how many phone calls they answered per hour. Mike Daisey hung up on every third caller and won a customer service award because he answered so many calls.
At this point, many managers are saying “That’s because the measurements were stupid – if I measure enough aspects of performance, then I can construct an appropriate incentive system.” The problem is that not only do managers have to measure all aspects of performance that contribute to company goals, they have to measure them all equally well. Otherwise, Austin’s model shows that the poorly measured aspects will get ignored in favor of the well measured aspects, and the same sorts of performance distortion occur.
Furthermore, measurement has a cost. An organization has to spend time and resources constructing measurement systems and reviewing the results. If an organization has to measure many aspects of performance to monitor employee performance, the cost of such measurement may outweigh the short-term benefit in increased employee performance.
Austin acknowledges the possibility that in a system with perfect knowledge of what an employee should be doing, such as an assembly line, it might be possible to perfectly measure performance along all relevant axes. But in a knowledge and service based economy, the measurement of performance is always going to be inexact. Because measuring performance directly is impossible, organizations use easier-to-measure proxies for performance. And because there is not a direct correlation between measurement and performance, distortions are introduced into the system.
Austin recommends managing by increasing the intrinsic motivation of the employee to do a good job. He makes the assumption that employees know the most about how to do their own job, so they are the only ones who can optimize their effort among the different aspects of their job. So a non-measuring manager needs to convince the employees to do their best “by example and through persuasion, and in clearly communicating direction to employees”.
The weakest part of the book is that Austin’s thesis rests on the model he constructed for how employees might behave in a measured environment. While the model is appealing, he provides very little data justifying the assumptions he builds into the model. For instance, he assumes that the employee knows better than the manager on how to optimize effort, but that assumption would not hold in a situation where a recent college graduate is being managed by a twenty year veteran. He also assumes employees want to do the right thing for the company/customer, which is not universally true in my experience. Austin is playing the same game as the managers he criticizes – he makes convenient assumptions and constructs a system that will work for those assumptions.
Despite having issues with his methodology, I agree with Austin’s recommendations for management. I hate being measured, as my rant about timesheets illustrated. Don’t revert to Taylorist management methods that treat the employee as an automaton that has to be bribed into doing the right thing. Hire skilled people who want to do a good job, point them in the right direction, and then clear obstacles from their path so they can get there.
I recommend the book as a thought-provoking read despite its weaknesses. Austin’s model is a good counterweight to the pithy aphorisms spouted by management consultants. His thought experiment provides ammunition for a more human-oriented style of management, where employees are treated as more than numbers in a spreadsheet.
The book I’ve been meaning to recommend in response to this for a week is Trust in Numbers, which talks about this issue in the sciences and management and the like. It’s very readable, and I think you’d like it.