Saturday, April 13, 2013

The Economist: a menace?

The Economist this week posted a study from Evolv, showing certain correlations between unusual worker choices and their talent. Among other factoids, it was argued that workers who had switched from their default browser (Internet Explorer or Safari) to another (Firefox or Chrome) were more likely to stay with the company and perform well - 15% more in fact. The browser is assumed to be a pure signal of initiative, rather than actually raising employee performance.

So far so good. However, the article concludes by encouraging all wannabe workers to change their browsers. This may be a good thing for some, but the advice will also definitely make some people worse off and will probably make society as a whole worse off. Let's take a look at who and how.

Suppose there are three types of workers: high, medium, and low quality. High quality employees also have high initiative, leading them to change their browser without suggestion (as I have). Medium quality employees have enough initiative to change their browser when it is suggested there is a benefit to doing so, but otherwise do not. Low quality employees don't read The Economist ;) or don't follow the suggestion.

How does the published article change the pooling equilibrium? The table below shows the average skill level of a person depending on what internet browser they use before the article and after. Before the article, if someone switched their browser you could reliably know they were high quality and everyone else was mixed. After the article, if someone did not switch their browser, you could reliably know they were low quality and that everyone else was mixed.

n = number of people; theta = skill level; 1, 2, 3 for high, middle, low skilled respectively.
If workers are paid their average expected wages, it is trivial to show that high skilled workers will now be paid less because it is harder for firms to identify whether they are high or middle quality. It is equally trivial to show that low skilled workers will also be paid less - firms can identify them pretty clearly. Middle quality workers will be paid more because they are now being confused with a better class of worker. The bigger the difference between skill levels and the smaller any group is, the bigger the wage change.

Therefore, by publishing the article, The Economist has chosen to benefit middle skill workers at the expense of high and low skilled workers. The question then becomes by how much and how many people are middle vs. low. If there are a lot of low-initiative workers out there, The Economist just made society worse off. If there are a lot of middle skilled workers, it is possible for their good to possibly outweigh the loss to the low skilled ... but it's a) definitely not Pareto optimal, b) definitely not Rawlsian optimal, and c) likely inequality enhancing.

Thanks a lot.

It is difficult to know whether they also made econometricians' jobs easier or harder. Identifying the effect depends on the difference in the average skill level of the two groups. It can be shown that, depending on how many people are in each group and how different their skill levels are, the size of the measured effect may go up or down. To be more precise, the size of the coefficient will fall if:
Roughly, that means if middle skilled people are more like high skilled people than they are like low skilled people, the measured effect will go down. If middle skilled people more closely resemble the low skilled, the measured impact will go up.

No comments:

Post a Comment