I recently attended a conference session that ended with a debate on whether social preferences should be taken into account when measuring social welfare. That might not sound like a particular exciting issue but I think it's an interesting and important one. So, lets look at the issues.
We can all agree on the idea that social welfare should guide policy. A policy can be considered good if and only if it improves social welfare. The difficulty is measuring social welfare. How can we reconcile the differing desires and preferences within a population? How can we take into account the desires and preferences of future generations who will be influenced by a policy? And so on.
The presence of social preferences, such as envy and altruism, muddies the waters even more as the following example illustrates. Robinson and Friday are the only people living on a desert island. Robinson is selfish and envious. Friday is altruistic and generous. You arrive with a boat full of goodies and have to decide how to distribute your cargo. What should you do?
If your measure of social welfare takes account of social preferences then you should give all your cargo to Robinson. This keeps Robinson happy because he has no reason to be envious and it keeps Friday happy because he likes to see Robinson happy. The selfish guy gets everything! This outcome strikes many as worrying. Worrying enough that they argue social welfare should not take account of social preferences. But, I find this argument unconvincing. Here are three reasons why.
1. The Robinson, Friday example is just an example. We know that the social preferences of real people are far more subtle. For instance, overwhelming evidence suggests that generosity is always given conditionally. In short, Friday's simply don't exist. And if Friday's don't exist then we have much less of a problem incorporating social preferences into social welfare. We can rule out the extreme kinds of preferences that people don't like and not miss anything important.
2. The Robinson, Friday example gives a negative picture of social preferences, which is at odds with the positive view given elsewhere. Social preferences are about sharing things, rewarding those who work hard, punishing those who free-ride, cooperating for mutual benefit etc. Social preferences are almost always a good thing! So, to not take account of them seems strange. If for example, 'rich' people want 'poor' people to have a reasonable standard of living it seems strange to not take account of that when measuring the welfare benefits of a redistributive policy.
3. Social welfare is essentially about judging fairness. This means that social preferences naturally underpin any measure of social welfare. So, we allow the policy maker to have social preferences. And, we allow the economist who comes up with measures of social welfare to have social preferences. But, we don't take account the social preferences of others? That seems way too egocentric. For instance, you might consider it fair to split equally the cargo between Robinson and Friday. The problem is, neither Robinson nor Friday consider it fair! And why should your definition of fairness trump theirs?
I think, therefore, measures of social welfare should take account of social preferences, and should take account the diversity of social preferences. But, that challenges us to improve our understanding of what fairness means to people.