Policy Reformers Should Randomize Reforms

Dark clouds over Norway.

Photo by Odd Andersen/AFP/Getty Images

I’m very interested in the question of gender diversity on corporate boards, so at the American Economic Association annual meeting I was eager to listen to a presentation of research by Adriana Lleras-Muney, Sissel Jensen, Sandra Black, and Marianne Bertrand on the impact of a policy in Norway that forces many large companies to have boards that are at least 40 percent women.* Their conclusion is that 10 years into the process there seem to have been very few effects. Nothing bad has happened, and a bunch of women got board gigs they otherwise wouldn’t have had, but there’s no medium-term impact on gender wage gaps or executive hiring or anything else dramatic.

They do seem to find a substantial impact on the attitudes of Norwegian M.B.A. candidates so maybe there will be a long-term impact.

At any rate, with that inclusive precedent as the backdrop a whole bunch of other European countries are at least considering a Norwegian-style quota system.

I’d like to beg them to try something else—conduct an experiment.

After all, the 40 percent threshold is totally arbitrary. If you think the gender composition of boards is a big deal, there are a few plausible candidates for tipping points. One is when a board goes from zero women to one woman. Another is when it goes from mostly men to mostly women (which essentially never happens). A third is when it goes from mostly women to all women (which, I believe, is literally never the case). I think politicians feel that a “one woman” mandate is too small a reform to be worth fighting for, and a “mostly women” mandate is too extreme to be feasible. But the interesting compromise would be to put your firms into three buckets. One bucket is the firms mandated to have at least one woman on the board, one bucket is the firms mandated to have a mostly female board, and the third bucket is firms not subject to any mandates. Then randomly assign firms to the buckets.

The random assignment would make it much easier to draw research conclusions. And because only a minority of firms are subjected to the tough mandate, you can actually push for a big change that would let you answer a big question—what if governance of major firms was in the hands of women rather than men?

Obviously the point here has applications beyond gender and corporate governance. It’d be nice to see more “let’s try an experiment” thinking work its way into policy reforms of all kinds. But sitting in the audience this struck me as a particularly plausible case for an experiment since there’d be nothing about randomization that smacks of unethical human experimentation.

*Correction, Jan. 8, 2014: A previous version of this post misspelled the last name of Adriana Lleras-Muney.