"An increase in the minimum wage raises the costs of fast foods and other goods produced with large inputs of unskilled labor. Producers adjust both by substituting capital inputs and/or high-skilled labor for minimum-wage workers and, because the substitutes are more costly (otherwise the substitutions would have been made already), by raising prices. " -- Gary Becker and Richard PosnerI know about this editorial from posts from Frederic Sautet and Don Boudreaux. Both Sautet and Boudreaux endorse Becker and Posner's "reasoning". When will economists accept arithmetic?
Monday, January 29, 2007
Mumpsimus
Apparently, Gary Becker and Richard Posner have an editorial in the 26 January 2007 issue of the Wall Street Journal. They oppose an increase in the minimum wage:
When will economists accept arithmetic?
ReplyDeleteYou seem either to believe that a particular counterexample constitutes a general proof, or that considering a particular counterexample to be empirically irrelevant - and those posts did refer to empirical data - amounts to not accepting arithmetic. Huh.
Welp, here I went and broke my self imposed ban on commenting on anything to do with the minimum wage.
The Becker and Posner quote that Robert cites seems to be a reference to theory.
ReplyDeleteAs Robert has stated repeatly, the textbook case is not a general theory itself; it is not clear what restrictions in a Arrow-Debreu world support the conclusion that higher minimum wages lead to lower employment *when* capital goods are produced.
Radek, I'm interest in knowing whether you view yourself as an instrumentalist. The way that you refer to empirical results suggests it, though I have no recollection of you stating so explicitly. In the context of the Cohen quote I posted previously, I would hope that you would understand why certain kinds of empirical findings are not viewed a compelling evidence for theories that are viewed to be internally inconsistent...
I come across a claim. That claim states that certain assumptions (e.g., competitive profit-maximizing) imply certain conclusions (e.g., that firms will want to employ less labor at a higher wage). I know of many examples with those assumption and the opposite of the conclusion. Yes, I think, that such counterexamples prove the claim, not logic, incorrect.
ReplyDeleteAnd no, Becker and Posner are not drawing on a phenomelogical model, supported by the data.
Thanks for playing.
Well here is the rest of the quote which is about empirics. If you add that on it doesn't seem like Becker and Posner (neither of whom I particularly like) are referring to pure (or especially, absolute) theory but rather to "how things work in the real world" combined with a plausible story to explain what is going on.
ReplyDeleteAs far as instrumentalism. Let's just say that I occasionally have warm fuzzy feelings about it. But I hate to say it. The model is not "inconsistent". It makes sense on its own assumptions. You might not like the assumptions and are welcome to substitute in corn and iron instead but that only shows that a model with different assumptions has different conclusions, not that the original model is "inconsistent".
ome economists deny that a minimum wage reduces employment, though most disagree. And because most increases in the minimum wage have been slight, their effects are difficult to disentangle from other factors that affect employment. But a 40% increase would be too large to have no employment effect; about a tenth of the work force makes less than $7.25 an hour. Even defenders of minimum-wage laws must believe that beyond some point a higher minimum would cause unemployment. Otherwise why don't they propose $10, or $15, or an even higher figure?
A number of countries, including France, have conducted such experiments; the ratio of the minimum wage to the average wage is much higher in these countries than in the U.S. Economists Guy Laroque and Bernard Salanie find that the high minimum wage in France explains a significant part of the low employment rate of married women. Mr. Salanie has argued that the minimum wage also contributes to the dismal employment prospects of young persons in France, including Muslim youths, an estimated 40% of whom are unemployed.
That'd be me typing in wrong thing in the wrong place. God I'm tired today.
ReplyDeleteI understand *why* you think that the model is consistent (the results follow from the assumptions). I would argue that the model is nonsense because it involves assigning a property to capital (an economically meaningful measure of supply that is invariant to factor prices) that it clearly does not have. Your willingness to accept a model based on its predictions rather than the "realism" of its assumptions is consistent with my understanding of instrumentalism. The issue is whether such a criteria results in a "deep" understanding of causal relationships.
ReplyDeleteBut that's an argument about definitions and assumptions. So "inconsistent" is not the correct word here. Of course a model can be consistent and be nonsense (in the every day sense of the term) at the same time though, obviously-squared, I, personally, don't think this is the case on this issue.
ReplyDeleteAt this moment in time "correct predictions" gets about a 70% weight and "realism of assumptions" gets a 30%. I might change my mind tomorrow. The obvious counter to your argument is that no model is going to have "realistic assumptions", whatever those are. I think that realism of assumptions is 'weakly desirable', ceteris paribus - it's sort of like getting a higher N in your statistics; works fine if the extra observations are not too noisy.
This is even completely ignoring the fact that one person's "realistic assumption" is another's crazy talk. Capital is a good example. Is it a factory? Is it iron? Is it a social relation? Is it embodied labor? I'm not sure that there is any hard reason why things in real world need to map into concepts in our heads 1-1.
Ay, too much metaphysical cream in my economic coffee, I admit.
To be a bit more on topic. Yes, it is possible to write down a model in which an increase in wages will actually lead to higher employment. It is also possible to write down a model where an increase in wages will lead to less employment. (There's a "Summers meta-theorem" out there which says something like "any policy recommedation can be derived from SOME model of maximizing behavior").
ReplyDeleteSo if I'm gonna be generous (not prefer one model to another) I'd say that the theory is ambigous. Which is why you look at the data. It's hard to see how the fact that data tend to favor one particular relationship (not necessarily a model) involves "instrumentalism". It's just what the data say. I know it's deeper than that but at some point you gotta pick a point at which you're willing to settle up.
I do not see two internally consistent models here. The assumptions of the one consistent model (of a multi-commodity economy) do not imply that firms will hire less labor at a higher wage. One can prove that by an example in which all the model assumptions are met.
ReplyDeleteIf one appends the quantity of capital (in numeraire units) as an exogenous parameter, the model is overdetermined and, in general, logically inconsistent.
I don’t know what “empirically irrelevant” means in this context. First, empirical evidence exists (http://robertvienneau.blogspot.com/2006/05/empirical-evidence-exists-on-sraffa_16.html) on the presence of Sraffa effects. Second, one might object that a comparison of equilibria cannot tell us anything about real economies. An empirically applicable model, to be realistic, must be set in historical time. But that cannot be Radek’s view here.
The assumptions of the one consistent model (of a multi-commodity economy) do not imply that firms will hire less labor at a higher wage. One can prove that by an example in which all the model assumptions are met.
ReplyDeleteUmmm, look, even if I accept that this is the only consistent model then I can just as easily argue that the assumptions of this model "do not imply that firms will NOT hire less labor at higher wage. One can prove by an example in which all the model assumptions are met."
So to repeat myself "if I'm gonna be generous (not prefer one model to another) I'd say that the theory is ambigous.". Even in this case the argument begins to turn on 1) how likely is one case compared to another and relatedly 2) empirical data. There's no "not accepting arthimetic" here.
And incidentally, this is from the abstract of the Han and Schefold paper you quoted earlier and which you allude to again below;
As a result, one envelope is found which involves reswitching. Reverse substitution of labour or reverse capital deepening are observed in about 3.65% of tested cases: they involve at least two switchpoints.
I don't think 3.65% of tested cases inspires much confidence in the pervasiness of these effects. (I can't get the full article so I don't know for sure if I'm not misinterpreting the importance of the result). This combined with the literature which finds significant effects on employment due to substantial increases in minimum wages (none of which are positive) would suggest that even when playing on your turf your conclusions/allusions/accusations are way overstated.
For the record, I don't dispute the possibility of a negative relationship between the minimum wage and employment. From what I've read of the literature, I'm not convinced that such a relationship has been definitively displayed. However, I will concede that I'm not familiar with studies published in the last two or three years.
ReplyDeleteWhat I do question is whether, such a relationship, presuming it exists, is explained by neoclassical theory. I've been influenced as much by Kirman as by post-Keynesians to view the neoclassical practice of imposing ad hoc restrictions in their models to generate "predictions" to be anything but scientific. Labour-capital substitutability seems to be a convention rather than either a "first principle" or a theory, given the inability of economists to indicate what they mean by capital.
I'd agree that often times the labor-substitutability story is made in polemics without little justification. And often times when you scratch it falls apart. But the labor-substitutability story in not a necessary condition for a negative relationship between wages and employment.
ReplyDeleteIn general, the relationship between labor intensity (in terms of person-years per unit output) and wages is not even monotonic.
ReplyDeleteBecker and Posner, in the post to which I object, tell the incorrect labor-substitution story. To me, "substitution" is a term of art, denoting an exploded neoclassical theory.
And it's a matter of mathematics what the assumptions of a theory imply. The linked bloggers in my original post have their mathematics wrong.
I think H. E. would find Zambelli (2004) of more direct interest to his aggregation concerns than Han and Schefold (2003).
Also, capital-reversing is just one Sraffa effect. Those Steedman references I gave in the previous post illustrate many more.