Saturday, October 21, 2006

Around And About

7 comments:

radek said...

"Michal Yates has some interesting comments on teaching economics in MRZine, the online complement to Monthly Review."

Please, that's just a badly written Guide to Stupid Critiques of Economics.


"The theory has repeatedly proven itself a poor model of reality, failing test after test: there is no evidence that a higher minimum wage causes a loss of employment; there is no correlation between a worker's wages and a worker's productivity; there is no evidence that providing people with needed resources, as in a welfare system, causes them to use their time unproductively; there is no evidence that jobs requiring skills are increasing despite the fact that hundreds of millions of people would like them; there is no evidence that people inevitably destroy property held in common or that making property private guarantees that it will be used in a socially beneficial manner."

Huh? Either he's really stupid, ignorant or just plain lying through his teeth. This guy somehow got a PhD?

Robert Vienneau said...

Radek,

Thanks for the reaction. Perhaps someday, I will blog on Kalecki, Hall and Hitch, or Kahneman and Tversky. I've previously mentioned some empirical work by James Galbraith which I respect.

h economicus said...

This discussion does suggest a question to me: in a non-experimental setting, what constitutes evidence of causal relationships? Econometric studies have long been plagued with a number of problems, including (i) inappropriate use of test statistics (and thus inappropriate reference to statistical significance), (ii) limited reporting on the robustness of the results, and (iii) a general awareness by economists that results are typically obtained only after considerable massaging of the data and/or extensive specification mining. Testing of theories is further complicated by the need to control in an ad hoc manner for influential variables outside the model. In addition, models may be so abstract that variables may not have a well-defined real world analogue, requiring the researcher to posit an appropriate "proxy" variable. I think that for these reasons, it's very hard for empirical evidence to sway people from their "pet" theories.

Although there are attempts by some researchers to pay careful attention to experimental design, this does not appear to me to be the norm, raising the question of how much empirical work in economics has the status of "evidence", rather than elaborate rhetoric incorporating certain pseudo-science elements. (I'm borrowing from McCloskey here.)

Even stylized facts often involve statistic filtering of the data, making it difficult to even agree on what is even happening....

Robert Vienneau said...

H.

You've probably read:
Leamer, Edward E. (1983). "Let's Take the Con Out of Econometrics", AER, V. 73, N 1. (Mar.): 31-43.
Summers, Lawrence H. (1991). "The Scientific Illusion in Empirical Macroeconomics", Scand. J. of Economics, V. 93, N. 2: 129-148

h. economicus said...

Thanks for the references. I downloaded the papers this afternoon.

If you haven't read it yet, I recommend the following:

Ariel Rubinstein "Dilemmas of An Economic Theorist", Econometrica, 74 (2006), 865-883.

It's a curious paper. It's in a highly ranked journal and provides a critical assessment of economic methodology. Rubinstein expresses his own misgivings about the effects of economics instruction on the culture and highlights some limitations of both theory and experimental work. However, the ending takes on an almost mythical quality, with Rubinstein arguing that economic models are fables, and that there is something wonderful about these fables (though his case for the "wonderful" seems rather vague to me).

h. economicus said...

I meant mystical rather than mythical...

Robert Vienneau said...

H.,

Thanks. Yes, I downloaded the references that you and others brought up in that discussion. I generally download references people bring up in discussions, although sometime afterwards if they are a JSTOR link behind a registration wall. I suppose I ought to thank people more often for references, but I generally think I am behind in my reading.

I'm generally more impressed with experiments than surveys asking people how they think they would respond in a given situation. So I thought Rubinstein was weak there.

I think I may have read at one time or another somebody suggesting game theory was neither descriptive of how people do behave nor normative, but somewhere in between (prescriptive? performative?). I'm still trying to decide if I want to purchase Donald MacKenzie's book An Engine, Not a Camera. I really like his sociology of proof book.