I’m improving my Public Policy Economics course ready for the next academic year at Manchester (hello, prospective ECON20431 students, if you’re reading this blog!) One of the areas that troubled me last year was how economics treats questions of social welfare in public policy analysis. I thought it was just me, as it had been many years since I thought about it. Having just finished Daniel Hausman’s [amazon_link id=”1107695120″ target=”_blank” ]Preference, Value, Choice and Welfare[/amazon_link], I’m reassured that it is just difficult territory.
[amazon_image id=”1107695120″ link=”true” target=”_blank” size=”medium” ]Preference, Value, Choice, and Welfare[/amazon_image]
The book has two big points. One is to argue that ‘preferences’ in economics is used, and ought to be used, to mean subjective comparative evaluations that include all relevant considerations – including moral imperatives, social norms, passions, whims. This obviously does not mean the same as everyday usage of ‘preference’; in everyday language we consider moral duties can conflict with preferences, for example. It also means self-interest can’t be built in to preferences. “Preferences are the outcome of demanding and poorly understood processes of comparative evaluation.” People might often take short cuts (as per [amazon_link id=”0141015918″ target=”_blank” ]Gerd Gigerenzer[/amazon_link]) or be vulnerable to certain biases (per [amazon_link id=”B00SSKM714″ target=”_blank” ]R Thaler[/amazon_link] et al).
Preferences are then combined with beliefs to determine choices. Hausman argues that economists often ignore the role of beliefs or judgements about the choice environment because they tend to assume people’s beliefs are correct.
His second big point is that economists should not and cannot then say that what determines preferences is a matter only for psychologists or other social scientists; economics itself has to engage with the question of preference-formation.
I agree with Hausman so far. I am less clear about the part of the book that is about preferences and welfare. He argues that the satisfaction of people’s preferences is evidence that their welfare is being improved, but that welfare cannot be defined as the satisfaction of preferences, as economists assume. I don’t understand the second part of this: he says it is because preferences are not defined by self-interest, but I don’t follow why welfare has to be defined by self-interest rather than being defined by preferences (including moral imperatives, emotions etc). Hausman then goes on to assume policies should aim to improve social welfare, which is fine if you agree with the definition of welfare. But he asks why policy should be sensitive to people’s preferences beyond the promotion of their welfare, and goes on to discusss the behavioural economics findings that people might often not be good judges of their own interest. He challenges methods such as contingent valuation, because they reflect people’s sense of moral obligation, for example. That doesn’t seem to me to be a problem at all.
I think my difficulty with this is that as soon as you posit an ‘objective’ concept of welfare as what is in people’s best interests, and allow this to diverge from their preferences – especially in this wide sense which can include concern for others or for absolute moral imperatives – you are in the territory of social scientists determining what is best. Indeed, Hausman agrees that if people do include the seeking of their own benefit and are good judges of circumstances, preference satisfaction is a good indicator of welfare; and even if not, it might be better than a third party deciding.
Anyway, an interesting book (& there’s much more than I’ve summarized here including a discussion of [amazon_link id=”0444851275″ target=”_blank” ]Sen[/amazon_link]’s approach), even if it hasn’t saved me from having to continue puzzling over this territory.
Pingback: Whose choice? | Homines Economici
There are lots of cases where social scientists are forced to “determine what is best,” in particular when people’s preferences are intransitive. Take hyperbolic discounting for example. A hyperbolic discounter can be thought of as having two simultaneous but contradictory preference relations: one of them is myopic, placing no weight on future periods, and the other is long-sighted, caring what happens in the future. Both relations are, by themselves, totally rational (economically), and either could be used to make an internally consistent welfare-maximizing policy. But absolutely no one would argue that we should include the myopic preference relation in designing policy–we make the values judgement that the long-sighted version of a hyperbolic discounter’s preferences is the right one to maximize over.
We make values judgements like this all the times, and they usually aren’t justifiable with the same kind of argument about intransitivity as above. For example, when designing welfare-maximizing crime policy, absolutely no one gives a positive weight to the preferences to murderers who desire to kill. And in optimal tax theory it is sometimes customary to place zero weights on the extreme right tail of the income distribution, in part because it makes math easier, but also because it’s very hard to argue that giving Bill Gates an extra billion worth of consumption is something that society should care about.
Economics has a long venerable tradition of classical liberalism–“let the people do as they please” so to speak. But, I don’t really see the problem with, every once and a while, letting social scientists acknowledge that certain preferences and behaviors are just totally nutters.
I agree with what you say but neither I nor the book touch on aggregation issues – Hausman was arguing at the level of evaluating individual welfare.
There are no right answers and a lot of wrong ones. The trouble is that past data and old policies lead to the wrong ones.
This is a broad issue and my thinking has involved overtime to the point where I now lean towards academic economists sticking to positive economics and limiting normative conclusions. One reason is because things like moral imperatives are ambiguous (what weight on different utilities? Utilitarianism is arbitrary). I do not like researchers imposing their value judgements. That is for elected officials to do, not a bunch of academics. This is different than an economist answering a question like according to my model a proposed policy will have these effects.
The second reason is because even if a model performs well on a positive metric that does not mean it is good for policy. There are many assumptions that have second order effects in terms of say impulse response functions, but the welfare consequences may be first order.
As an aside, an extreme version of 2 (which I am guilty of) occurs making policy recommendations from DSGE models. To make policy you need to know the shocks hitting the economy. Of course under the null hypothesis that your model is correct you have these shocks. But many would say these shocks are a manifestation of where the model fails, a measure of our ignorance.