Walt:
> I'm in kind of a confused state regarding economics right now. As is > obvious, I am having some theoretical problems with Marxian economics. > However, there is alot of empirical evidence I've seen that makes it > difficult for me to accept neoclassical economics, even the more > sophisticated "new keynesian" and "radical" variants.
Gil:
First, "a lot of empirical evidence...that makes it more difficult...to accept neoclassical economics"--compared to what theoretical alternative that does a demonstrably better job of accounting for that evidence? And if your answer is "none", then maybe the real question is what discipline offers a surer basis for addressing the empirical anomalies you're concerned with. To anticipate, since Jim Devine mentioned our mutual friend and colleague Peter Dorman, I'll paraphrase the passage in his book on wage differentials to the effect that Marx pretty much copped out on the question of how working conditions affect wages--arguably a product of Marx's decision to pursue Ricardian rather Smithian value theory.
Yes, Marx did "cop out." But at least in volume III, he chose to assume that _Smith's_ theory of compensating wage differentials applied (while, as many scholars have argued, rejecting Ricardo's labor theory of price except as a first approximation). Of course, that "copping out" was part of his method. He started out in volume I (after chapter 3) talking about the relationship between abstract capital and abstract labor (explicitly ignoring the specific use-values produced and the specific usefulness of concrete labor, which has the effect of implying that prices of production equal values). He then moves toward the real, empirical, world in volume II by bringing in differences amongst capitalists and then in volume III by developing the effects of these differences (implying that prices of production differ from values). As Mike Lebowitz pointed out in his two BEYOND CAPITAL books, Marx never got to the volume on Wage Labor, which likely would have brought in heterogeneous labor-power and labor. Only then would he have been able to deal seriously with the issue of compensating wage differentials within the context of his methodological discipline. For me, that means that we can't look to Marx for the solution to the compensating wage differentials issue. We have to figure it out for outselves, not just empirically but also within a coherent theoretical framework. One of the main ones is neoclassical, while the institutionalists (e.g., Clark Kerr) have written a lot about it. (The latter have a lot to contribute to Marxian political economy, in my view, because Marx was an institutionalist.)
Second, "neoclassical economics" is becoming a more and more nebulous concept as the mainstream analytical tent gets more and more diverse. Do you mean neoclassical in the narrow sense of Marshall,Walras, and Debreu, with utility-maximizing households, profit-maximizing firms, and equilibrating competitive markets (except in special cases of monopoly or monopsony equilibrium)? The former augmented by the conditions of classical noncooperative game theory, which maintains the assumptions of optimization and equilibrium, but allows for informational imperfections, expected utility maximization, and interactive optimization, and thus generalizes the possibility of inefficient equilibria beyond monopoly and monopsony? [I would include phenomena such as moral hazard, adverse selection, "efficiency" wages, and free-rider problems in this set]. The foregoing augmented by small-group experimental methods, evolutionary game theory, agent-based modelling, and modifications of classical game theory to allow for non-expected utility-maximizing individuals? [And this is just the micro side of the tent!]
Yes, criticizing NC economics is like sparring with a whirlwind (made especially hard by its ideological hegemony). In response to this problem, Christian Arnsperger and Yanis Varoufakis have an excellent piece in the _Post-Autistic Economics Review_ #38 defining what "NC economics" is. (See www.paecon.net.) They do it in terms of _methodology_, which is the way it should have been defined all along. They posit three meta-axioms which seem to capture what we usually mean by NC economics: N1. methodological individualism: "the socio-economic phenomenon under scrutiny is to be analysed by focusing on the individuals whose actions brought it about; understanding fully their 'workings' at the individual level; and, finally, synthesising the knowledge derived at the individual level in order to understand the complex social phenomenon at hand." (Sociology is bad!) N2. methodological instrumentalism: "all behaviour is preference-driven or, more precisely, it is to be understood as a means for maximising preference-satisfaction." N3. methodological equilibration: "neoclassical theoretical exercises begin by postulating the agents' utility functions, specifying their constraints, and stating their 'information' or 'belief'. Then, and here is the crux, they pose the standard question: 'What behaviour should we expect in equilibrium?' The question of whether an equilibrium is likely, let alone probable, or how it might materialise, is treated as an optional extra; one that is never central to the neoclassical project." I might add a couple meta-axioms (e.g., the NC hatred of methodology), but I can't think of them off-hand or defend them the way these authors do. But in sum, for them NC economics is "a religion with equations" totally in sync with the dominant liberal ideology (where "liberalism" is defined broadly, to include both neo-liberalism and New Deal liberalism). In contrast, we might posit three Marxian "meta-axioms": M1. non-reductionism: though the causation from the individual level to the "complex social phenomenon at hand" plays an important role, it's important to look how the structure of such phenomena shapes the character of invididual pieces of the totality. It's a two-way street, as in Levins & Lewontin's description of the dialectical method in biology. M2. anti-instrumentalism: it's not simply a matter of individuals using means to achieve goals. As Albert Camus (among others) have pointed out, the means and ends cannot be separated so neatly: the method one chooses (or has forced upon one) to achieve one's goals also affects one's goals. For example, people living in market society tend to develop "market personalities." (That's why I think _homo economicus_ is a good first approximation of human behavior in the market -- but not in other social settings.) M3. disequilibrium: yes, Marx had his own equilibrium conditions, as with his reproduction schemes in volume II of CAPITAL and profit-rate equalization in volume III. But he makes it pretty explicit that these do not represent the empirical world (which was what he was trying to understand). The reproduction schemes represent harmonious steady growth (perhaps at a growth rate of zero, as with simple reproduction), not actual growth. (They are akin to the Harrod-Domar conditions, seen in a non-NC growth theory.) Capitalism is best represented by a disequilibrium process, often seeing crises. For Marx, crises _restored_ equilibrium (in a painful way). But once equilibrium was restored, it was disturbed again, by endogenous processes (e.g., the tendency for profit rates to fall). There is a tendency for profit rates to equalize between sectors. Marx assumed that this tendency was fully realized in order to understand price/value deviations. However, with technology, etc., always changing, these conditions are unlikely to be sustained for long (or even met). They represent tendencies, but we have to bring in counter-tendencies.) I also think that Marxian political economics is open to multiple equilibria models (and totally non-equilibrium ones), as with Kaldor's non-NC model of the business cycle. Speaking of the MACRO side of the tent, the domination of neoclassical reductionism is one of the reasons why macro is always in trouble. For example, there's the whole "microfoundations" movement which undermined the coherence of practical Keynesian economics. (Such theorems as the Cambridge Capital Critique and Sonnenschein- Mantel-Debreu undermine the microfoundations method fo treating the whole as merely a reflection of a "representative" part.) Worse, there was so-called "rational expectations" (a.k.a. Nash equilibrium), which screwed everything up. Among the orthodox, the only solution was so-called "new Keynesian economics" (Mankiw, _et al_), which really created a new Monetarism using new microfoundations. It's not a very good solution (though some of their ideas, e.g. macro externalities, make sense). What distinguished Keynesian economics, I think, is something that the Keynesians weren't totally aware of: there's a _feed-back_ from macro-phenomena to micro-phenomena (as in axiom M1) which makes the whole microfoundations stuff even more laughable. For example, if we're in a Depression-type situation, the micro-components of the economy _work differently_ from the "normal" NC story (which assumes Say's "Law"). Walt:
> I can not take seriously the idea that the worse and more dangerous a job > gets, the more hours and intensity of work, that workers will necessarily > get paid more. Or at least that the amount of compensation will be so > small as to not necessitate anything approaching "efficiency," even with > reasonably informed participants and competitive markets.
Gil:
The funny thing is, if you're willing to grant the presence of "reasonably informed participants and competitive markets," then it's hard to see how compensating wage differentials *wouldn't* arise, so long as people cared about the safety of their jobs or how intensely they work. Suppose you were reasonably informed about wages and production conditions at alternative prospective jobs, and suppose that your labor market was sufficiently competitive that you could freely choose among these alternatives. If you cared about job safety, then other things equal, why would you take a more dangerous job if it didn't offer you some offsetting benefit in the form of higher wages and benefits or the like? This suggests that telling a coherent story about the *absence* of compensating differentials requires some departure from the above conditions--some form of imperfect competition, or imperfectly informed individuals (a special case of the former), or individuals that are either irrational or don't care about things like safety, at least in the sense that they don't make marginal tradeoffs between safety and other benefits. And even then it's harder to upset the compensating differential story than you'd think. For example, you can show that even labor market monopsonies would optimally pay compensating differentials, so long as prospective employees were aware of and cared about safety differentials and had *some* alternative to working for the monopsony; and even uninformed workers might nevertheless be able to elicit firms' privately held information about their safety levels, with compensating differentials resulting.
I talked about this issue in a previous missive. One thing that prevents compensating wage differentials from being realized is the normal existence of involuntary unemployment. A secondary axiom of NC economics, i.e., Say's "Law" rules that out. It's interesting that Gil ignores this issue.
> Are there any neoclassical theories which explain this, or leave it as a > possibility?
In the broader sense of "neoclassical" alluded to above, yes. More specifically, there are theories that explain why conditions that might otherwise lead to the payment of a compensating differential are swamped by offseting conditions, as when workers in "primary" labor markets enjoy both safer jobs and greater wage-setting power than their secondary-market counterparts.
I am not up on the current research, but back when I read the stuff traditional NC economics defines the dual labor market structures as arising from totally technological/natural factors (e.g., specific human capital). On the other hand, the Edwards/Reich/Gordon tradition of radical economics explained them in terms that NC economists would see as mere "sociology": there's a structurally-based conflict in capitalist production relations that must be solved by management in order to get normal profits. Sometimes more than one management strategy can be used (there are "multiple equilibria"). Together with technical/natural factors, this helps produce labor market segmentation. BTW, in terms of multiple equilibria, the article "Game Theory, Freedom and Indeterminacy" by Kevin Quinn in PAER #38 is very interesting and excellent. He argues, among other things, that "individual freedom" and "deterministic science" can be reconciled if science drops the goal of there being one specific result of theory. Game theory exercises, for example, often indicate more than one Nash equilibrium (and gets similar results with other equilibrium concepts). Rather than seeking ways to reduce the number of equilibria to one, he says that the existence of more than one represents the realm of freedom. We can choose between them. However, typically the power to do such choice is not individual but must be done collectively. Hey, sounds a bit Marxian! -- Jim Devine / "Socialist democracy is not a luxury but an absolute, essential necessity for overthrowing capitalism and building socialism." -- Ernest Mandel
