“lab in the field” experiments, behavioral games and real life outcomes.

Lab in the field setting... (Uganda, summer 2009)

Behavioral games are abstract situations in which individuals have to allocate resources between themselves and other players and allow to study strategic interdependence in decision making: participants have to take into account the objectives and strategies of the other players. For instance, in the Public Goods game each player is allocated an endowment (i.e., 10 coins, generally corresponding to a consistent proportion of a daily wage), and must decide how much of his endowment to keep in his personal pocket and how much to put in a group pot. The total amount donated to the group pot is then doubled and redistributed evenly among all the players. For the individual, the optimal strategy is to give nothing and “free-ride” on other players’ contributions to the collective fund, while the most profitable outcome for the group is that all players contribute everything.

Differently from what a rational choice model would predict, experimental results have shown that in the first round of a public goods game players contribute, on average, between 40 to 60% of their endowment. Although cooperation in real life occurs more often than a rational choice approach would suggest, most economists did not consider observational evidence sufficient to assess between egoistic and altruistic models of human behavior, because in real life (seemingly) altruistic/non-selfish behavior can always be explained as covertly oriented to increase one’s reputation, psychological well-being or being enforced by internalized social norms and fear of social sanctions. Indeed, it was experimental evidence coming from behavioral games to put in jeopardy the micro-foundation of the rational choice approach by documenting people’s predisposition to reciprocity, cooperation, and altruism in completely anonymous settings.

I have to admit that my first reaction to experimental literature was: “only an economist would need laboratory experiments to realize that selfish motives are not the only drive in human behavior.” Nonetheless, I am grateful that these games have been developed, since they represent an extremely useful tool for social science research, as I had the chance to experience first-hand in my current research on farmer organizations in rural Uganda.

In collaboration with Guy Grossman, a graduate student in political science at Columbia University, I conducted a quite innovative research that combines experimental evidence from behavioral games played in the field with social networks information and observational data. The research involved more than 3,000 farmers and local leaders from 50 farmer cooperatives through Uganda. Goal of the research was to understand how producer organizations in development countries solve classic problems of collective action. The study focuses on the role of social and spatial networks, associational capital, and leadership accountability in affecting economic and social outcomes.

To give you a sense of how behavioral games added to our study consider the following example. One of our hypotheses was that leaders’ accountability and their willingness to monitor and sanction non-cooperative behavior greatly influences group outcomes. Unfortunately, this hypothesis is hard to test relying exclusively on observational data, because of selection and measurement issues. Thus, to capture the role of legitimacy and the effects of a centralized sanctioning system we designed a novel adaptation of the public goods game in which players were randomly assigned to three different conditions. The first condition (baseline), simply replicates 6 rounds of a conventional public goods game. In the two other conditions, after two preliminary rounds of play, one of the players was selected out of the session to become a monitor endowed with sanctioning power. The monitor could spend 1 coin to take away 3 coins from players whose contribution level he disapproves. In the second condition (random monitor), the monitor was selected through a random lottery, while in the third condition (elected monitor) the monitor was elected by the players using a secret ballot. Results, reported in the figure below, suggest that in the presence of a centralized sanctioning system, players significantly increase their contribution to the public good: in both elected and random monitor conditions players contribute more, on average, than in the baseline. Moreover, the process of monitor selection is consequential: elected monitors are perceived as more legitimate and thus exert a greater authority.

Average contribution in the Public Goods game

Interestingly, we find that behavior in games is related to real life economic performance: the more productive members of the farmer organizations (namely, those who sell their crop in bulk through the organization) give higher contributions in the elected monitor condition, thus suggesting that centralized sanctioning and leader legitimacy are relevant factors in explaining organizational outcomes. This finding shows that behavioral games can be used not only to capture underlying behavioral tendencies common to all human beings (or profound cultural differences across societies) but also to measure differences between individuals (or groups) that derive from their individual and group experiences. In this respect I think lab in the field experiments that incorporate behavioral games into socially meaningful settings are an interesting addition to the social science research (out of the lab) tool-kit.


Written by deliabalda

July 23, 2010 at 3:26 pm

3 Responses

Subscribe to comments with RSS.

  1. Thanks for this post. I enjoyed hearing about your work.

    Just something I’d like to clarify. You wrote: “[I]t was experimental evidence coming from behavioral games to put in jeopardy the micro-foundation of the rational choice approach by documenting people’s predisposition to reciprocity, cooperation, and altruism in completely anonymous settings.”

    The foundations of rational choice are about consistency in preferences/choices not selfishness, and I think that you conflate the two. Conditional cooperation, altruism, etc., are all OK with rational choice as long as the behavior is consistent.


    Mike M.

    July 23, 2010 at 5:21 pm

  2. You want to dedicate your life to games? Study sociology!



    July 23, 2010 at 6:36 pm

  3. Abstracting away from field experiments such as the one described here, which are unquestionably useful, I think the general economics critique of these lab experiment results holds up better than as implied above. Consider some real world example of an anonymous, one-shot, public goods game: for instance, say the US funded national parks completely out of (anonymous) donations. Does anyone expect 40-60% of the population to contribute? Or even 1%? I certainly don’t. And when it comes to games that more closely resemble the experiments – say, if unemployment insurance were funded by anonymous voluntary donations over the next year only, which more closely resembles the kind of cash transfer in the experimental setting – I think you would see arbitrarily close to zero people doing something other than free riding.

    The point is that, yes, economists do need to see some evidence that altruistic behavior occurs, outside of situations like groups (where behavior is based on assumed repeated games) and non-anonymous settings (where its tough to disentangle desire for reputation from altruism). In particular, in a lot of market settings, which is the heart of economics, I certainly don’t see a lot of altruism “in the field”, no matter what laboratory results say about how people will behave.



    July 23, 2010 at 7:05 pm

Comments are closed.

%d bloggers like this: