Friday, March 13, 2015

Keeping Employees from Leaving: How an Art is becoming Scientific

There is now a report in Wall Street Journal on how firms are increasingly using statistical analysis to find out which employees are more likely to leave, and using this information to improve personnel management and target employees for interventions to make staying more worthwhile. Firms do this because replacing employees who leave can be very expensive, making the analysis and the responses cheap in comparison. It says a lot that Wal-Mart, a firm known to be careful about its expenses, is investing in such analysis. It is probably less surprising that Credit Suisse does so, given the importance of keeping staff in banking, or that some human-resource analytics firms do, given that they can use these results to keep their own staff and sell the methods to client firms.

What have they discovered by analyzing their employees? Well, they are more likely to leave if they have problematic managers or little contact with co-workers, less likely to leave when they are given opportunities to change jobs internally (especially promotions), plus a variety of other smaller discoveries.

Is this surprising? Actually we have known it for a long time. Job mobility is an established field of sociology and management research, and as far as I can see the statistical analysis done by the firms re-discovers what is already known. So, I would probably not go to a firm statistician expecting to learn much new about job mobility, though I would still find it interesting to see what they are doing.

Does that mean it is worthless? It does not. There are two really significant pieces of progress in the news that firms are doing this analysis. The first is that the whole point of studying job mobility is to understand what happens to the lives of people and the fates of organizations, and it is wasteful to have this understanding without also using it to improve the lives of people and the fates of organizations. Job mobility is often valuable, but there are also many cases of job mobility that is wasteful for the employee and the firm. It is better to reduce them. The second piece of progress is that firms are now gaining knowledge that lets them address the situation of each employee, and they can often intervene in positive ways such as improving job content or opening for promotions when they see a risk of that employee leaving.

There is of course some potential that this gets intrusive or used in troublesome ways, so it is worth watching. Firms are after all able to track health coverage decisions and health care use with enough detail that they can start linking them to job mobility, something that would be new to academic researchers and potentially troublesome. They could also track emails, which academic researchers have already done but always anonymously.  There are good reasons to limits such data collection and analysis.

Even with these potential problems, it is nice to see business catching up to the value of research. Of course, it has only done so to a limited extent. The number of statistical analysts involved in this work is far fewer than the number of human resource managers (and other managers) who thinks that such management is an art that calls for their unique experience and cannot be understood by others. Maybe some of these managers are right, but on average I would place my bets on the statistician.