Experimental evidence: how wealth affects cooperation

What is your thesis is all about?
Three are the main topics of the papers: inequality, cooperation and identity. Inequality is pervasive, and while we might not consciously act in response to it, it does affect how we behave to a large extent. Because we were born in a specific socio-economic bracket we make behaviours, tastes and mind-sets proper of our entourage our own.

Inequality determines choices such as education, health-related behaviours, profession. Who we perceive we are, who we perceive others to be and how connected we feel to them determines who we choose to interact with and how. Since much in our societies relies on interpersonal cooperation, it is fundamental to understand how the inequality characterising our societies impacts cooperation, and if anything can be done about it with minimal intervention.

Why does this topic interest you?
Looking at the contemporary world, it is evident how identity and diversity are possibly the two main hurdles our societies will have to tackle in the 21st century. People move to and from anywhere in the globe. Globalisation and information technology force us to be in touch with very distant people, both geographically and culturally. We need to understand the consequences of this, figure out how to make things work at best, what works and what does not work to improve on social and individual wellbeing and to make economic relationships as fruitful as possible for everyone. This seems to me an important challenge, and one I’m willing to pursue.

What are the main results?
The take-aways from my work so far can be summarised as follows: social structures matter. People care about social structures, especially rich individuals will try to preserve them, even at the expense of their own and collective wellbeing. While cooperation would make everyone better off indiscriminately, if the can choose they tend to cooperate among each other and exclude the poor from benefiting of their greater capacity to contribute to collective wellbeing. This will tend to accentuate inequality in the long run. Next question is whether this generalises to status hierarchies different from wealth.

Moreover, social structures influence what individuals believe about each other. Who others are, for example rich or poor, influences what is believed about them. In particular, when it comes to cooperation, the rich (those who can contribute the most to collective wellbeing) are believed by both the rich and the poor to be the ones who will be most cooperative. This expectation is then disconfirmed by actual behaviour, showing the rich to be less cooperative than the poor in relative terms.

How can lessons learned from these experiments be transferred to real situations?
Experiments allow us to break down very complex situations into their simplest components and to look at them in isolation from anything else that might influence outcomes. When we observe cooperation breaking down between, say, rich and poor or natives and immigrants in the real world, it is very hard to conclude that it is the income differential or the different ethnic backgrounds that are causing the breakdown. This because income differences and ethnic differences also come with, for instance, fundamental differences in culture or education. Often, socio-economic and native/immigration status are tightly connected themselves. Is it then the socio-economic differences per se that cause those cooperation failures, or is it instead something else that is connected with those differences?

If our aim is that of minimising and possibly correcting for lack of cooperation, we are then forced to finding the root causes of those cooperation failures rather than limiting ourselves to observing when such failures occur. This is something we can do by breaking up the situation into its fundamental pieces and observing each of them separately. Of course, this does not mean that what we observe by doing so will translate 1-to-1 to the real world: other intervening factors will for sure affect outcomes in unpredictable ways. It is a first step in a long and complex incremental work aiming at painting as complete a picture as possible of how reality works. Understanding how different factors add onto each other comes next.

Can you give us an example to illustrate your findings?
The spatial organisation of cities tends to reflect income segregation. The poor and the rich tend to live in their respective neighbourhoods, commonly contiguous and often separated by a single street at most. Still they form part of the same township, and as such cooperation must be achieved between the two neighbourhoods to ensure its efficient management. Now, segregation into distinct and identifiable areas makes income class distinctions salient. My findings tell us that in this situation the rich tend to cooperate mostly among themselves and to be relatively uncooperative with the poor. This has consequences on the distribution of wealth and on income disparities, which tend to become more pronounced over time (which is often heard in the news).

This situation can also lead to downward spiralling feedbacks between cooperation and confidence in others’ willingness to cooperate. I show that people do cooperate insofar and inasmuch they believe others, particularly the rich, will cooperate as well. As soon as others, especially the rich, are observed cooperating less than expected, this can deteriorate people’s trust in others’ cooperativeness, leading to reduced cooperation, which will further decrease expectations, and so on.

Did anything surprise you?
Indeed. Especially the rich’s dual pattern of cooperation. I could expect, based on previous findings, that the rich would be less cooperative than the poor, especially when cooperation with the poor is involved, but I was not anticipating the full extent of it. Definitely, I was not expecting to see them increasing cooperation in their group.

Can you tell us briefly about the way you conducted the experiments?
The idea is that of eliminating any interfering factor from the situation being analysed in order to isolate the effect of interest. This means that the situation must be artificial (naturally occurring interactions are ridden with socio-psychological context), ensuring participants’ socio-demographic and individual characteristics are uncorrelated with the variables of interest.

The first step is therefore that of gathering a large enough sample of participants which are then randomly allocated the experimental characteristics, and allowing them to interact in a controlled way. For instance, a major variable of interest in my experiments is being rich or poor: the point is then to randomly allocate the participants to being rich or poor so that the two groups can only be distinguished in terms of being rich or poor. If, for instance, most of the rich are also women, it would be hard to say whether the effects observed are originating from being rich, or from being women.

Next, we allow them to interact such that specific mechanisms are activated. In public good games, individuals interact by choosing whether to allocate their own resources to a common "project" allowing them to increase everyone’s welfare, or not. Allocating an additional euro to the common project increases the group’s welfare but is individually a bad idea unless the other group members contribute an additional euro too. Their choice will be informative about their willingness to cooperate with others.

Finally, it is crucial to incentivise people’s choices. This ensures they will behave as is in their best interest, reflecting what they truly want to do having considered the effect of their choices on themselves and on others. In simple terms, their choices in the laboratory have a real impact on the amount of money they and others will walk away with. This allows us to measure individual and group responses to specific stimuli.

"Bitter divisions: inequality, identity and cooperation" by Andrea Martinangelis was defended on 17 May 2017 at the School of Business, Economics and Law, University of Gothenburg.