16 Jan 2018

Published! Group cooperation and public goods

Back in grad school, I was doing experimental economics in the lab to test whether water managers who worked together were more cooperative than random students. The experiments were based on a "public goods game" in which each player in a random group of four decided to help themselves or help others in the group. Players were anonymous to each other because they were hidden behind computer screens, but there was a clear benefit to all from cooperation, just as there was a clear benefit to the individual "freerider" who took advantage of others' cooperation.

My results were surprising to anyone who believes that "public servants" focus on the group rather than themselves, as the water managers were no more cooperative than the students. Economists would not be surprised, as they would assume public servants behave just like other people. Skeptics would laugh at the idea of cooperation, of course, as they tend to think as follows:
SOCIAL scientists tell a story about a peasant called Vladimir. One day God comes down to him and says: “I will give you one wish. You can name anything you want and I will grant it to you.”
Vladimir starts to celebrate, but then God lays down a condition. “Whatever you choose,” He says, “I will give to your neighbour twice over.”
Vladimir frowns and thinks. And then he clicks his fingers. “I have it,” he says. “Lord, please take out one of my eyes!”
Taking such a diversity of opinions into account, I was glad to add some data to the discussion.

These experimental results were central to my dissertation ("Conflict and cooperation inside an organization"), and the chapter on the experiment ("Water managers are selfish like us") was improved and published as a chapter [pdf] in a handbook on experimental environmental economics.

After doing those experiments, I got excited about testing other questions regarding cooperation within and across groups. I was mostly interested to try to find the line where people acted strategically regarding their cooperation decisions in a group, i.e., when did it make sense to freeride and when did it make sense to cooperate. In a small group, for example, it may be harder to freeride because others could observe your actions more easily than they could in a larger group. (These ideas date back to Olson's 1965 Logic of Collective Action, so I wasn't innovating on hypothesis as much as testing one.)

In the end, I decided to test a few ideas:
  1. Would students in groups of 20 cooperate less if their grades were "curved" such that grades depended on relative performance, as opposed to depending on absolute performance? My hypothesis is that group size would be too large to induce strategic (more defection) behavior.
  2. Would students cooperate more if their earnings depended on how well they did compared to everyone else not in their group? My hypothesis was that such "in group" cooperation incentives would have a positive impact.
  3. How would my four treatments compare on impacts? My hypothesis is that the four could be separately ranked, thus providing some empirical advice.
I conducted most of the experiments in 2007 and got useful responses but never got 'round to sending the paper into a journal. Fast forward to 2016, and I got a call for papers on strategic behavior and cooperation. I sent in the paper, made lots of very significant and useful changes in response to referee comments (yay peer review!) and got published.

You can download the paper here until 19 Jan, along with all the other papers in the special issue. After that point, you should email me if you want a copy or download it from here.

Oh, and what about the hypotheses? On #1, there was no real difference in cooperation, so "curve grading" will probably not reduce cooperation, efficiency or output among students or workers. If you have experience for or against, then please comment!

On #2, there was a big difference when students were rewarded to relative performance compared to everyone outside their group/team. This "in-group" result is no shocker, but it's important in reference to any firms or bureaucracies that might want to set up teams. The other part goes with #3.

For question #3, there was a clear ranking of "in group" cooperating the most, followed by curve grading/absolute grading, with the fourth treatment (how well you do compared to members in your group) having the lowest cooperation. This ranking is useful -- as the paper discusses -- in thinking of ways to structure regulatory, pollution or collective action incentives.

Bottom line: Read my paper if you want to think more about how people do (not) cooperate in small groups, organizations or bureaucracies.


Anonymous said...

Would it be OK if I cross-posted this article to Writer Beat? There is no fee, I'm simply trying to add more content diversity for our community and found this article insightful. I'll be sure to give you complete credit as the author. You can learn more about Writer Beat by checking out my profile or just say "sure" in an email to me and I'll handle the rest.


David Zetland said...


Post a Comment

Note: only a member of this blog may post a comment.