Without being too over dramatic, it is so curious to watch the norms of the modern academy – things like equality, freedom of speech and non-violence coming to seem as quaint as the divine right of kings. Its not that practices in the world ever corresponded to these norms but there was a time when they were at least the dominant norms.

The time of every norm for itself is here.

Likely I’m overdramatizing, but I see a major effect of September 11th, 2001 being a legitimation of unlikely norms. Its the equality of cranks and wackos. Its so odd that the effect of a physical threat is a more concrete intellectual freedom.

I thought the problems of welfare economics were pretty well-known by now and that (besides the relentless law and economics school) most economists and philosophers who are interested in normative issues in economics have moved on to other projects, e.g., Amartya Sen, who takes a “resourcist” rather than a “welfarist” approach? Why bother with welfare economics at all?

Unfortunately, I think that economists have been slow to develop the theory and practical application of preference formation. It is so much easier to take them as a given. There are some exceptions and things are starting to change but for the most part economists (along with all the rest of social scientists) have a poor track record of predicting human behavior in a general way.

I think there is an issue beyond the aggregation of preferences that needs to be clarified.

You refer to what individual preferences God “uses”. And then you also say “we” could “use” this function to “construct” Zion.

I think preferences and “actions under preferences” are being conflated.

In other words, does “use” mean to describe or imagine the state of the world most preferred. This woud be equivalent to restating your close as ‘If so, then we could, indeed, use the rational actor model to construct …a model of … Zion!

Or does “use” mean change either our incentives or preferences in such a way that Zion actually occurs in the world. If this is the correct interpretation, unless knowing God’s welfare function changes the individual welfare functions of a significant number of people, then I don’t see how knowing the social welfare function “constructs” anything other than what we currently have. For example, does knowing that it is desirable that lions to lay down with lambs actually cause lions to lay down with lambs?

Furthermore, adopting something of theologically econo-Panglossian mode, how does one argur from either the rational actor model or from a theology of a just and perfect God that the world could be any other way than it actually is? In other words, if there was a way rational actors could make the world any better right now, than it is right now, they would have done it already.

Likewise, if a just and perfect God could make the world any “better” right now, than it is right now, he would have done it already.

Under these models all one needs to do to see the ideal welfare function is to look around.

But as I stated I don’t think our social science predicts very well. For example, who among us would predict the most popular video game in 3 years, or the fertility rate in Oman in 10 years?

But here is my question. Let’s say I wish to raise my children and my children’s children to become much better at predicting (or heaven help us) controlling the behavior of people in the world in the future, what education and/or indoctrination would I give them?

What mix of religion, social science, neuro-pharmacology, humanities and time in the school of hard knocks would be optimal? Does a religious education make one a better predictor of atheists than an atheistic education makes of predicting the behavior of believers?

2 : “since this thread represents my total exposure to the notion of a welfare function, I’m confident that there are certainly some subtleties I’m missing.”

My abstraction about people not knowing what was going on was not meant to be a roundabout way of talking about you, though I can certainly see that it could be read that way. Don’t worry David, if I think you don’t understand, I’ll use your name!

On the other hand, you assert in 2 that you may be missing some subtleties. Aren’t you therefore saying that you “may not know what is going on”? Or maybe I’m missing some subtleties…

And although you’re right about the purpose of modeling, different models serve different purposes; viz., some are for clarifying and some are for operating. It seems to me that the model that Nate is working with (from what you’ve described) is a proposed clarification of the notion of optimal welfare. And while it can be used (in some sense) to show what an optimal welfare state may look like, it is not a practicable model. I had taken the topic of this post to indicate that Nate was considering this within the context of practicability.

Moreover, since this thread represents my total exposure to the notion of a welfare function, I’m confident that there are certainly some subtleties I’m missing.

Obviously it all depends on what one is trying to model. If one is looking at decisions to work or not work then a longer time frame makes more sense, whereas decisions to buy Kool-Aid might be more weekly. Empirically, the decision is based on the data available. And the data collection is often base don some guess as to what time frame is of interest to the researchers’ questions. And so one tries to use a welfare function that is accurate for the things one is interested in, even if it is a lousy one for other things.

But yes, in any model there is pretty much always some amount of tradeoff between accuracy and usefulness. The point of good modeling is to capture what is relevant to what you are trying to model, while leaving out irrelevant detail that makes the model needlessly complicated. This can be infuriating to people who don’t understand what is going on. They see the model as being unrealistic and so dismiss it, but may fail to recognize that it is sufficiently realistic to accomplish the purpose for which it was created.

This is not such a concern for people like Nate who are just using it as an asbtract philosophical notion. If you are never really going to manipulate the function, then you can make it completely accurate across all time, containing an infinite number of goods. It only becomes and issue when one is trying to draw testable implications or use it empirically to estimate something.

Moreover, it seems that the judgment of how long we make the time horizon of the welfare function pits accuracy against usefulness. Specifically, it seems to me that if the time horizon is too short, then the the functions values have very accurate meaning but are too volatile to be accurate. If the time horizon is too long, then its values are more accurate, but less meaningful. For example, over time having sustenance is surely a baseline welfare condition. Shorter time frames allow the term having sustenance to be defined more specifically, longer time frames require it to be defined more generally. But the more generally it is defined, the less information it conveys.

All you do is define the welfare function over a longer period of time. A full-blown welfare function would include all the things you made choices on from today until the end of time. Each choice could be a seperate argument. This is unwiedly, and so when operationalizing preferences over time economists make additional assumptions. But the basic theory has no problem with what you describe. Water today and water tomorrow can, if desired, simply be treated as two different goods.

I have to say that the whole thing seems to be an ill-founded endeavor. From an economic point of view, the fact that preferences are transitive also makes them relative. Thus, the realization of any given preference is the result of a trade-off. Presumably, the optimal welfare case obtains when all welfare trades have been made. But because welfare needs are fluid (now I may be willing to trade water for food, tomorrow vice versa) homeostasis is never reached, trades will never cease, and the optimal case is never obtained.

Perhaps I’m engaged in sophistry here, since I seem to have simply defined the optimal welfare function out of existence, and this doesn’t really solve anything. But my welfare function seems to include something regarding the expression of this opinion; hence this comment.