<p><strong>Social comparisons, perceptions of fairness and wellbeing<br />
</strong></p>
<p class="MsoNormal"><em>Romesh Vaitilingam interviews Armin Falk for Vox<o:p></o:p></em></p>
<p class="MsoNormal"><em>October 2008 </em></p>
<p><em><o:p></o:p>Transcription of an VoxEU audio interview<o:p></o:p> [http://www.voxeu.org/index.php?q=node/3002]</em></p>
<p><br />
<strong>Romesh Vaitilingam: </strong>Welcome to Vox Talks, a series of audio interviews with leading economists from around the world. My name is Romesh Vaitilingam. Today's interview is with Professor Armin Falk, who is Director of the Bonn Laboratory for Experimental Economics at IZA, the Institute for the Study of Labor at the University of Bonn.<br />
Armin and I met at the Center for Economic Performance in London in October 2008, where Armin was speaking at a workshop on happiness research. He was discussing recent papers he had done on social comparisons and how our individual perspectives of these social comparisons affect our wellbeing. He began by explaining how he uses experiments to conduct this research.</p>
<p><strong>Armin Falk: </strong>I use experiments because I am interested in causal influences. That is basically the major advantage of running experiments in comparison to, say, using survey data, which is also very important and complements our findings, but it is, at some point, very important to establish cause and effects.<br />
The two papers I presented today are, in fact, related to the issue that you mentioned which is how and whether people compare with others and how this is affecting their subjective wellbeing, and we measure subjective wellbeing in these two papers in two different ways.<br />
One is we measure heart rate variability, which is an indicator of stress, and the other indicators are activation and so-called reward circuitry in the brain. There are two workers in this experiment, two subjects, if you like, and both of them are scanned in a brain scanner.<br />
Both are doing a job, a simple task, and if both of them do the task correctly, both of them get paid by the experimenter. We basically study three conditions. One condition is both get the same wage. In another condition, one worker gets more, the other gets less. And the third condition is the other way around.<br />
So what we can study is for a given, our own absolute income, how is the brain encoding the information that a co-worker who did the same job, not better or worse than I did, receives the same versus receiving more or receiving less. So from a traditional economic perspective, you would expect that what matters is absolute income, and of course, the more the better. And this is also what we find in the scanner study.<br />
What is more interesting, though, is relative income matters as well, and it matters dramatically. We find, in particular, that reward activation is measured in terms of ventral striatum activity (activity in the ventral striatum is an important part of the reward circuitry in human brains). So activation is strongest if I have more money than the other person, followed by if with the same amount of money, and activation is weakest, and by far weakest, if I have less money than the other person.<br />
This establishes, if you like, that social comparison matters, that the brain encodes these differences instantaneously, immediately. Social contact really matters, and we can reject the hypotheses of the traditional economic model that it is only absolute income that really matters.</p>
<p><strong>Romesh: </strong>How widely applicable do you think this is? You are testing students; presumably, you are putting them in these machines and then asking them questions. How much does that tell us about what people are operating like in their real lives as it were?</p>
<p><strong>Armin: </strong>This is a good question. It is true we are studying a very stylized situation to start with, and the subject pool is typically students, but I mentioned in the beginning that we are complementing studies that use real people, if you like, people working in the workplace, using much larger datasets, long run studies. They also find strong correlations between unfair treatment at the workplace. So for example, a mismatch of what I think I should get with what I actually do get. And later, maybe years later, adverse effects on health, stress, cardiovascular diseases, and so on and so forth.<br />
The problem with these studies is to identify cause and effects, because it could go both ways. You are sick, and therefore, you have lower wages, or you have lower, unfair wages, and therefore, you become sick. That is the advantage of our study, and I think, in this sense, these two ways of getting at the problem are complements, and I think we need both types of studies in fact.</p>
<p><strong>Romesh: </strong>Now, tell me about the second study, the one about unfairness or attitudes for unfairness in the way people are paid at work.</p>
<p><strong>Armin: </strong>The other paper is to some extent related. Again, we are interested in how you compare what you get to what you think you should get or what you would consider as fair.<br />
The set up is very simple. There is a principal, a boss if you like, in a firm, and a worker, an agent. The agent works on a tedious task and generates revenue and money. All the money he generates by working on this tedious task is going to the principal. After some time is over in the experiment, like 30 minutes, the principal receives all the money and can then decide on how to allocate between himself, the principal, and the agent.<br />
Now, given that the agent did all the work and the principal did not work at all, most agents would consider a wage of at least 50%, a share of the total pie of at least 50%, as appropriate. And in fact, most people want more, the average being something like 66%, two-thirds of the pie.<br />
Now, the principals offer, on average, only about 40%. So we create some discrepancy between what principals are willing to share and what agents think they should get, what they think is appropriate. And we take this difference, or this discrepancy between what you think you should get and what you actually do get, as a predictor for stress. And we measure stress in the experiment by measuring heart-rate variability. And if heart-rate variability slows twice as high, we find a very, very strong correlation between the amount of unfair pay, as measured by this discrepancy, between actual and appropriate pay, and heart-rate variability. And again, this is complementing field studies, but nicely showing cause and effect of unfair pay, perceptions of unfairness, and adverse effects on health, via stress reactions.</p>
<p><strong>Romesh: </strong>It seems to be these experiments, you're really digging deep into the human nature. You're trying to understand physiological reactions to experiences in the workplace.</p>
<p><strong>Armin: </strong>That's correct. And I think this is a trend in the social sciences, and economics in particular, which comes under the head of neuroeconomics, which is basically trying to look at two things: biomarkers as dependent variables, but also treatment variables, if you like - for example, if you do hormone treatments and so on - to extend our knowledge about the nature and the principles, the structures of economic decision making.<br />
So, for example, in brain research, we have now, already, a better understanding, I think, of how we take, for example, intertemporal decisions, or risky decisions. Instead of assuming one utility maximizer, for example, it is probably better, in light of the neural evidence, to think of a dual-self approach, or a multiple-self approach, which means that parts of the brain want immediate gratification. They want the money now, and they want sex now, and they want chocolate now. They want the nice things now. And then there's a controlling part, inhibiting some of these pretty important impulses, which says, &quot;Look, wait a minute.&quot; It's more long-term oriented and so on.<br />
And it's kind of a rivalry between these two systems, which explains many of our decisions. And this is a totally new way of thinking about how decisions come into being, and this competition between several systems could also lead to different policy implications, in the end.<br />
I'll take a different example. It's a long discussion. It's a no-brainer that economic preferences are extremely important for decision-making. For example, risk preferences are important for occupational choice, for how many stocks we hold as part of our wealth portfolio. It's important for what type of wages we get. If wages are higher, people are more willing to take risks, and so on and so forth. It's related to migration. It's related to a lot of economically relevant behaviors.<br />
But economists talk and know little about where preferences are coming from. And we're just starting to better understand the social-demographic determinants of, say, risk preferences, but also other preferences. And it's only a natural extension to dig deeper and to ask, maybe there are even neurocorrelates, genetic dispositions, and so on. And what we try at this point is to extract both behavioral measures of social risk and time preferences and economic decision-making, and so on, on one side, and DNA on the other side, and then to relate genetic dispositions to economic behaviors to really better understand where preferences are coming from.<br />
This is, again, also relevant for policy, if you like. For example, if you think about social mobility. One of the issues in many European countries is that the economic success of kids is much determined by economic success of the parents. For example, in Germany, the likelihood that you have higher education depends extremely strongly, more than in other countries, I think in Western Europe, on income of your parents. So there's slow social mobility. Now, if our preferences are based on genetics, and if genetics are passed from parents to kids, the government can't do much about it. But if this low social mobility is coming from the institutional environment, for example, the context of the social environment and so on, there's more scope for policy intervention.<br />
And I think, of course, even if we find that genes are important, it's not determining all of our behaviors. That is absolutely clear. But, in its own right, it's interesting to study whether they are some of these dispositions.</p>
<p><strong>Romesh:</strong>And final question, going back to the research you were talking about on attitudes to fairness at work. What policy implications are we in a position to draw policy implications from that about, for companies thinking about how they reward their staff, and also for governments making decision about equality legislation?</p>
<p><strong>Armin: </strong>I do think we can draw policy implications from this. One central message&nbsp;&nbsp; my central message, if you like - to policy-makers in firms and managers and so on is, don't think that fairness issues are just distribution issues. They are efficiency issues. If you treat people in an unfair way, this is bad for your revenues, this is bad for your performance, because people who are not motivated will not work. Since contract enforcement is a big problem, because work is not enforceable, you will rely on voluntary cooperation of workers, you better treat them in a nice way, in a fair way. And if you can't, because, say, the economy is in bad conditions or the firm is in bad conditions, then communicate why you can't. So it is very important to take people seriously, to give them social approval, and to treat them with respect, ultimately. And I think that will pay off, tremendously.</p>
<p><strong>Romesh: </strong>Armin Falk, many thanks.</p>
<p><strong>Armin:</strong>Thanks.</p>