Re: (OPE-L) Re: tendencies for equalization

From: Ian Wright (iwright@GMAIL.COM)
Date: Mon Oct 04 2004 - 12:22:27 EDT

Hi Jerry

Thanks for answering the questions. The objections you raise to
tendencies to equalisation are not germane, primarily because I would
expect pronounced income inequalities in a market economy, as I stated
in a previous post. Hence, I do not expect:

> On the contrary, we would expect -- given existing inequalities --
> that if there were such a tendency than it would manifest itself
> empirically in a long-term _decline_ in wage disparities rather than
> relative stability in wage disparities.

I have never claimed that a tendency would manifest as a long-term
decline in income dispersion, because I think that income inequality
is a property of the statistical equilibrium state of the economy, and
therefore invariant over time. I agree that a market economy produces
income inequality; indeed, I provided an explanation that accounted
for the functional form of this income inequality, an explanation very
different from the enumeration and classification of causal factors
that affect wages.

The link between the real tendency to equalise wages and its empirical
manfiestation is complex, but I mentioned two possible manifestations,
one counterfactual, one actual. First, the income dispersion may be
narrower than would otherwise be the case if workers did not try to
exit low-pay and enter high-pay jobs. Second, the unimodality of the
income distribution I think indicates that workers equal in their
productive capabilities are distributed around a single mode. Both of
these possibilities require more careful work to evaluate, but they
were my responses to your original question of how this tendency might
manifest empirically.

I mentioned in a previous post that the exponential distribution is a
reasonable fit for 90-95% for the income of all groups in
industrialised countries over a period of several decades. As with all
empirical data, there are other fits, such as lognormal or gamma, and
also slightly different measures can be employed, but that is not too
important here. There is no trend, because this is a functional form.
As I also mentioned previously, the dispersion may wax and wane (maybe
with the business cycle, an interesting question), and the mean may
change, but the functional form is invariant.

I provided a theoretical explanation for this fact, that it is a
maximum entropy distribution under a money conservation constraint.
This kind of explanation is probabilistic, and so it can take a little
getting used to. The important point, however, is that this
explanation is intended to include many if not all of the concrete
determinations that have been mentioned as causal factors that affect
wages. It is a highly parsimonious explanation because it collects all
those concrete determinations together and assumes they contribute
entropy-increasing noise.

Contrast to the methodological approach of enumerating all the
possible events that can occur in the labour market. It becomes
difficult to understand how all these factors interact. It can be very
complex. A better approach is a top-down one: assume that all events
can happen, subject to global constraints, and then start counting
probabilities. We then get the most probable statistical state of the
complex system. This is why I asked for other possible explanations of
the exponential form of the income distribution, so we can compare and
evaluate them to the maximum entropy argument.

I think that listing lots of causal factors without a theoretical
scheme to put them back together in order to explain empirical data is
incomplete. Enumeration and classification is only a step towards
understanding. Also it is too hasty and has a hint of empiricism to
reject a real tendency even if its empirical manifestation may be

Would the claim that there is a tendency to equalise profits raise
similar objections?

Best wishes,


This archive was generated by hypermail 2.1.5 : Wed Oct 06 2004 - 00:00:01 EDT