Is it harder for high-caste men to coordinate?

A new paper (pdf) by Benjamin A. Brooks, Karla Hoff, Priyanka Pandey runs at least one set of tests suggesting the answer is yes:

In an experiment in India, high-caste and low-caste men repeatedly played the Stag Hunt coordination game. This game has two equilibria, only one of which is efficient. Compared to low-caste men, high-caste men were significantly less likely to coordinate on the efficient equilibrium, and they were also 29 percentage points less likely to keep trying for efficient coordination after getting the “loser’s payoff”—the payoff to a player who attempts efficient coordination when his partner does not. We explain both findings in a model of learning where high-caste, but not low-caste men, see the loser’s payoff as an insult rather than an accident. These findings provide evidence that cultural construals can impede efficient coordination, which is a key component of economic development.

I find the distinction here between “low payoffs as insult” and “low payoffs as accident” to be especially interesting and in the broader literature underexplored.


> I find the distinction here between “low payoffs as insult” and “low payoffs as accident” to be especially interesting and in the broader literature underexplored.

Yup. It's also refreshing to see that the authors didn't try to attribute this behavior to "increased feelings of fairness" among the high caste men. I wonder if the "low payoff as insult" idea can be applied to the Ultimatum game as well.

>>Yup. It’s also refreshing to see that the authors didn’t try to attribute this behavior to “increased feelings of fairness” among the high caste men

But that would be politically incorrect (to draw any conclusions that favor the high-castes)

This same phenomena of non-cooperation has been found between pilot and co-pilot in airplane crashes in various countries.

Current study is a good argument for only hiring low caste flight crew?

Interesting. Are you intimating that a LINK (a causality) has been found between flight incidents and a state of non-cooperation in the flight crew? Your wording is not quite precise on this.

Now, to a person who has no experience as a flight crew member, I imagine such a conclusion (non-cooperation as a causative factor) would sound like a big "Duh!". However, as one who has been a flight crew member, I know that cooperation is so rigorously inculcated into the flight crew culture, that NON-cooperation would be so subtle as to perhaps be completely missed by a non-flight-crew observer.

If you have sources, I would be interested in them.

I don't have hard sources but poor flight deck communication has come up in some of the aviation incident reports I have read. Also on many PPRUNE / AVHERALD discussions. I don't think it is a common problem but it still happens.

Sometimes the senior member of the crew can be dominating and the junior member hesitates to speak up even when he finds an action questionable. Hierarchy can be very strong in the Asian nations and these problems are particularly acute there.

In particularly egregious examples I remember one member actually waiting for the situation to get pretty bad before pointing out mistakes just because they didn't like the attitude of the other member or so that they could file a deviation report. Some tapes show the flight deck atmosphere to be so degraded that routine conversation was almost dead (& I'm not talking about the sterile phase)

Recent SWA accident: lack of cooperation so bad that the flaps weren't set up for landing ... ?? no landing checklist? ... just before crashing, the captain moved the throttles while copilot was flying the aircraft. Captain was fired; copilot sent to additional training. Nine passengers injured.

Why cannot we interpret this as the "High Caste Men were more risk averse?"

Opting for a guaranteed 16% reward instead of the uncertainty of a reward-penalty (66%-50%) sounded like risk aversion to me.

It's hard to say in this scenario because the benefit has no real value attached to it.

Why is better coordination among high caste men more efficient?

Can easily make the opposite claim.

Please do.

Higher castes have denied lower castes their due.

Lack of coordination among them will help social mobility

Frankly I find it hard to believe higher caste can't coordinate (to put it politely).

The results apply to High caste men in North India or more specifically Uttar Pradesh. If they think "cultural construals" are important, then they should be careful in describing their results (surely no one needs to be reminded that we are talking about 1.2 billion people). Their abstract doesn't mention North India any where.

Least of the problems.

This paper is like a case study. Questionable external validity. No pre-registration of protocol. High probability that they went fishing. A quaint just so story. Independent third-party replication is very likely to fail. Et cetra.

That's a pretty neat experiment and result. I'm not sure about the insult and revenge seeking. Could you also frame it as strong aversion to feeling foolish?

What an interesting inquiry! Cultural quirks and differences have long been an interest of mine. I shall have to first examine the "stag hunt" game to understand the odds, actions, and consequences correctly, and then I will take a look at the rigor of the inquiry methodology.

I am not at all surprised that a caste-level difference in preference might exist. It will be interesting to me to see if I agree with the researchers that they actually found an arguably valid preference.

And one must keep in mind that this IS likely a cultural phenomenon - but then - so much is cultural!

Page 6: "The striking finding is that HH pairs are less than half as likely to coordinate on the efficient outcome as LL. Only 32 percent of the HH pairs coordinate on the efficient outcome in the last period of the pairings. In contrast, the comparable figure for LL pairs is 73 percent. LH pairs achieved efficiency at an intermediate level of 50 percent (players had two partners in turn; these percentages are an average of the final outcomes over the two pairings—see Figure 2)." How does this translate into their conclusion? The participants know the caste of their counterparty, and yet high caste men seem to be more willing to cooperate with low caste men than other high caste men? This sounds a lot more like individualism and competitiveness among HH pairs versus community camaraderie among LL pairs. Interesting experiment, poor paper. One of many howlers "High castes were historically the dominant landowners and the priests, whereas low castes were the untouchables. This makes Uttar Pradesh a nice setting...". Seriously?

Great point. There is massive scope to explore this in the Muslim world.

Comments for this post are closed