Before yesterday, I didn’t really think so. However, Emmanuel Saez changed my outlook a bit. I attended a seminar he gave on a paper that he’s working on concerning optimal US tax and transfer policy. I liked the general thrust of his presentation; he was concerned with both the effects that income elasticities for taxation have on revenue collection and the implications of some behavioral insights for administering a tax system.
Explicit in all of this is the idea that redistribution is a good thing. In other words, if we can soak the rich to maximize revenue that can be transferred to the poor, without the rich “going Galt” or the poor deciding not to work, then we should. The question of taxation then becomes the issue of finding the marginal tax rate where the “rich,” however we choose to define them (Saez likes top 1%, but also drills down intop 0.1% and 0.01%), pay as much revenue to the government as possible.
The questions he received throughout did not challenge his prior that redistribution is good (I think he got one about this topic really being welfare economics, but there was a murmuring consensus in the room that this wasn’t true). Instead, they focused on details such as, “how do you make the tax base neutral,” or “where does the estate tax play into this?” In other words, their analysis and critique of his arguments seemed to be positive, but the entire premise for his argument was normative. And they want along with it (“they” being a group of neoclassical economists from Brookings and Urban). What other priors ostensibly derived from “welfare economics” would neoclassical economists accept/reject?
Update (12:14 AM): I should mention the couple comments on this post are quite good- I’ve given one of them its own post.
It is important to distinguish economic as a social science from political economy as a branch of social and political philosophy. Alfred Marshal is the first scientific economist. Those prior to him, like Smith, Ricardo, Mathus, and Marx, were political economists, or in Heilbroner’s terminology, “worldly philosophers.”
To the degree that economics is a science it is entirely descriptive, and not at all normative, if norms are taken to be values. Science does have norms, of course, but they are supposed to be entirely methodological instead of ethical.
Political economy is a branch of ethics as the study of human action in terms of principles and values. Economics as a science finds practical application in political economy, which is essentially normative in that it is concerned with achieving “the public purpose.” Public purpose is defined normatively. Different political persuasions posit different norms, but the facts and causal implications are the same for all.
While one view is that political economy is essentially the same as descriptive economics, taking the ideal system for meeting public purpose to be the most efficient one in terms of the science of economics, regardless of specific outcomes. Positions based on say, “the general welfare,” take a normative stance and allow certain inefficiencies, e.g., in the cause of serving distributive justice.
I think this distinction is an important one. Is the lesson to be learned that in the policy world, even neoclassical economics throw aside pure positivity because they have these policy goals in mind? Am I just naive to be surprised by this instance? I have a lot to learn…
Neoclassical economics (aka “Chicago School”) has a very definite political agenda that it doesn’t admit largely because it identifies “economic efficiency” with ideal policy, regardless of outcomes. The fact of the matter, however, is that social sciences are always normative to a degree, because universes of discourse are bounded by norms, and method is also. and these universes are far more complex than the universe of physics.
Even physicists have run up against a methodological boundary with quantum mechanics, which has no overarching interpretation. The prevalent interpretation is the Copenhagen interpretation, which more or less prescinds from the difficulties by holding that descriptions in QM are theoretical (nominal) rather than real, in the sense that we really don’t know exactly what corresponds to them. (I hope that’s not an oversimplification.)
The basic idea of scientific enterprise in general is to describe what is the case in terms of variables, which establish the parameters of the discipline, and measurement, which links specific data to theoretical models. Because models are necessarily simplifications of what they model, the parameters determine what is in, what is out, and what the priorities are. In the social sciences, these parameters are based on assumptions, and it is often at the level of the assumptions that norms migrate from methodological to value-laden.
Progressive economists generally admit that their assumptions have a value-laden bias. However, conservative economists often do not, but act as if they are describing what is. As a result, they represent what they are doing as positive science even though it often is not, in that assumptions especially, and often data too, are not evidence-based.
The philosophy of economics is concerned with the logic of the universe of discourse that characterizes a subject. The logic of science, for example, looks at models as generalized descriptions and seeks to determine to what degree data processing, norms, and assumptions may add bias. The logic of justification looks at the norms that determine what is admitted and rejected, and how data is assessed and evaluated in terms of specific criteria. Then it examines criteria for assumptions and bias.
Most economists don’t bother with this critical procedure. They just assume what they learned is the way that economics is properly done, especially when a school of thought is dominant. This leads to a lot of unexamined assumptions and sloppy thinking posing as science. It also leads to dismissing opponents out of hand, or marginalizing them as out of touch with mainstream thought.
One of the biggest criticisms of neoliberalism, for example, is unrealistic assumptions, e.g., basing models on a single representative agent as a rational actor that always acts on self-interest, taking the future into full account. This is just not the way way the world is, as behavioral economists, psychologist, cognitive scientists, and sociologists point out. However, assumptions like this make the building of complex mathematical models possible, whereas assuming different types of agent would make modeling too complex to be tied up neatly. Another example, is the principle of aggregation leading to the fallacy of composition. Australian econ prof. Steve Keen has addressed a lot of these criticisms in Debunking Economics (2001).
Do neoliberal economists in the US tend to have a policy agenda? I would say so. One of their key conclusions is that full employment is incompatible with price stability; therefore, the government and central bank should accept a certain level of non-frictional unemployment as necessary for managing inflation. Modern Monetary Theory aka Neo-Chartalism has shown that this based on a lack of understanding of national accounting in a post-1971 world, in which the US as a sovereign is the monopoly provider of a nonconvertible flexible rate currency of issue. Another erroneous ass assumption is the desirability of balancing the budget or running surpluses as a matter of principle, even though facts show that this results in underutilization of capacity and unnecessary unemployment when there is an output gap. These ideas are simply hold-overs from pre-1971, when currencies were convertible, and even then it was dubious as to whether they resulted in maximum economic efficiency given that the real economy is not only influenced by business cycles but also financial crises leading to debt deflation, as in the Great Depression and the present crisis. Neoliberal economists totally missed the onset of the present crisis and still can’t explain, other than to say, “No one could have foreseen it.”
Great stuff, particularly what you said about the lack of critical procedure.
You use the word ‘prior’ in a very odd way. You use the word to mean ‘utility function’ or ‘policy preferences’, but I think the word ‘prior’ is generally used to refer ‘beliefs about the state of the world before updating on the evidence at hand’ (see http://en.wikipedia.org/wiki/Prior_probability).
I never thought that neoclassical economics and a policy preference for redistribution were incompatible.
@jsalvatier: I guess I’m not seeing the distinction. A utility function is another representation of a ‘belief about the state of the world.’
@Darren: I suppose much of my exposure to NCL has indicated that they are, because arguments not geared directly towards efficiency are either diminished or dismissed out of hand. I had an economics of poverty class in which my professor invited Cecilia Rouse to speal, who does work on education at Princeton. Rouse made a number of backhanded jabs at ‘equity’ (I guess my prof and she had a debate before class) and carefully geared all of her arguments to efficiency. This is just one instance, but it characterizes much of my exposure to mainstream economists to date.
In philosophy ( I can’t speak for economics since I’m not familiar with all the jargon), “prior” generally means “before” the facts and their relationships, as in “a priori” in contrast to “a posteriori.”
Naive realists hold that mind is a mirror for the world, and if it is disciplined to act as such by eliminating subjective bias, as scientific method is designed to do, then descriptions mirror facts, general descriptions aggregate facts, and logic relates facts through implication. A positive science is a purely descriptive one, based on observation and entailment. In economics, the idea is that economics as a positive science can describe maximum efficient for any set of data.
There are lots of things that are wrong with this view. Cognitive scientists, for example, have shown that perception is actually quite a bit more complex than this, and human beings simply do not have the capacity for isolating “priors” that manifest as hidden assumptions. Moreover, Wittgenstein and others showed that the world is perceived through the lens of language, and a universe of discourse, along with its norms, shapes a worldview. Observation constructs a view of reality, rather than reflecting reality as it is. In the 18th century, Kant speculated that the mind contributes to structuring the data based on a priori categories, and cognitive science bears this out out and adds to it. See, for instance, Demasio, Descartes Error.
The naive realists’ claim that the commonsense view of the world reflects reality is just not supported by the evidence. Many economists, especially one’s that think that their models reflect maximum efficiency and therefore are the grounds for effective policy -making, are of this simplistic view.
In economics, Tony Lawson is one of those becoming well known for attempting to correct this erroneous stance. See Tony Lawson, Economics and Reality (1997), for instance. He was marginalized for some time, but now more people are taking him seriously. Several papers of his are available online.
BTW, the concept of utility in economics is a can of worms, not only philosophically (meaning of term) but also economically (constructing workable models that are evidence-based). See Steve Keen, Debunking Economics (2001), chapter 2, The calculus of hedonism. He shows how neoclassical economists construct utility models that work brilliantly mathematically but at the expense of being evidence-based, owing to the assumptions necessary to get the models to work so brilliantly mathematically.
>@jsalvatier: I guess I’m not seeing the distinction. A utility >function is another representation of a ‘belief about the >state of the world.’
There is some mistake here…the commenter was right. A “prior” is nothing like a utility function, at least in common parlance of probability theory. In fact priors belong to probability theory as such, while utility functions arise only in decision theory.
A “prior” is the probability equivalent of what in logic would be the assignement of a truth value to a hypothesis. For instance, say you are interested in knowing the average weight of some critter. Call that “theta”. At first you have some probability distribution (say, from a previous study, or just a wild guess). That is your prior. Then you sample the population and get (by bayes rule) a new distribution for theta, called the posterior distribution.
So I guess your description (“a belief about the world”) is an appropriate description of what a prior is.
Now, a utility function is a different thing altogether. It is a statement of your preferences, or of the value you assign to some possible state of the world. At most you could say it reveals “a description of what you’d prefer the world to be”.
Say you wanted to hunt the critter for food. Then the utility of hunting one of the critters is a function of the value of theta. If theta is 1 gram then its pretty low, if it is 10kg its probably pretty high (a good meal), but if it is too high it again goes down because the critter will probably eat you instead.
So, from the prior you could calculate for instance the prior expected utility by averaging over the possible states of theta multiplied by the prior of theta and the utility of theta. And, of course, after the sampling you could calculate expected utility of that action from the posterior instead of the prior, and with this data you could decide to hunt or not according to the expected utility of each action.
So, to sum up:
A prior is a belief about the state of the world, irrespective of personal preference.
Utility is a statement of personal preference about possible states of the world (or outcomes of possible actions)
These are standard concepts in probability theory. Of course, maybe they have other meanings in other disciplines?