From bryancap@ocf.Berkeley.EDU Fri Sep 10 23:05:36 1993
Received: from monsoon.berkeley.edu (monsoon-ether.Berkeley.EDU) by
phoenix.Princeton.EDU (5.65c/1.113/newPE)
	id AA25170; Fri, 10 Sep 1993 23:05:34 -0400
Received: by monsoon.berkeley.edu (5.65c/CHAOS)
	id AA07942; Fri, 10 Sep 1993 20:05:30 -0700
Date: Fri, 10 Sep 1993 20:05:30 -0700
From: Bryan Caplan <bryancap@ocf.Berkeley.EDU>
Message-Id: <199309110305.AA07942@monsoon.berkeley.edu>
To: bdcaplan@phoenix
Status: R


Hello Steve,
I've been thinking about our discussion after the Tuesday section.
Here are some of my ideas and questions.
1.  You agreed that at least part of discrimination IS due to the 
Posner-signalling/information costs view.  Well, how much? If
blacks, for example, earn 27 cents less on the dollar, how much
of this is due to the fact that there are negative work characteristics
that objectively correlate with race, and how much due to purely
malicious discrimination?
2. If perfect competition is such an important assumption, then
we should expect to find that discrimination correlates with market
structure.  Does it?  One obvious counter-example that comes to mind
is rental housing.  Here is a VERY competitive market (many, many
sellers, easy substitution, etc.) if not entirely perfect. (As you
point out, nothing is.  I might make an exception in the case of
the stock market.)  But as far as I can tell, people think that
discrimination is rampant in the housing market.  It probably is,
but not the way that is usually assumed.  Namely, it is likely
that race correlates to a high degree with failure to pay rent,
among other things.  Most landlords that I know of would rent to
anyone who would pay the rent on time and not damage the property.
But if they know that there is a correlation between race and
lower landlord earnings, they will indeed "discriminate."  And
THIS kind of discrimination does not get competed away.  But is
it at all plausible to you that landlords would discriminate
in the malicious sense to any important degree if this correlation
were illusory?  Moreover, do you think it at all plausible that
landlords would SYSTEMATICALLY overrate the magnitude of the
correlation?
3. You mentioned that you have had jobs where black workers were
paid less.  Haven't you also had jobs where indistinguishable
workers of the SAME RACE were paid different amounts?  (Even more
common is paying different quality workers the same amount.)
How much of this do you think is due just to randomness, and how
much to discrimination?
4. If discrimination really ISN'T due to the signalling theory, then
firm profitability should POSITIVELY correlate with the strictness
of anti-discriminaton laws.  That is, if discrimination is malicious
rather than based on actual correlations between race and work
performance, then if the government forbids discrimination,
profits should go up!  Has this pattern ever been discovered?
If not, then the signalling theory has more support.  Note that
this test would also make sense if the signalling theory were,
as you said, correct but exaggerated. (I.e., you said that hirers
probably overestimate the differences in racial job performance).
In that case, the anti-discriminaton law would just push firms
closer to th cost-justified differential.
5. I found it interesting that you agreed with me that
anti-discrimination
laws and affirmatie action "short-circuit" market checks by making
it illegal for firms to take advantage of cost differentials (in
the former case) and, (in the latter case) making a diploma from
Harvard differ in expected value according to the race of the holder.
(I.e., since admission requirements are lower for one group than
for another, the value of their respective diplomas differs).  Doesn't
this show that the laws are at least partially counter-productive?
6. I've been thinking about whether the assumptions of perfect
competition, mobility, and information really matter.  You seem
very confident that they do.  I'm not so sure.  The really
crucial assumption is profit-maximization.  Even a monopolist
would have lower profits if it hired less qualified whites over
more qualified minorities.  The assumption of market strcute matters,
though, because the more firms there are, the wider the spread of
"tastes for discrimination" there will be.  But do we really
need an infinite number of firms before it is reasonable to believe
that at least ONE firm will be managed by non-discriminators?
Imagine that there are only 10 firms.  One of these will be
managed by relatively non-discriminatory people.  Shouldn't that
firm therefore gain market share relative to its more racist
competitors?  It doesn't seem like you need more than a handul
of firms before there will be a sufficiently wide range of
tastes for discrimination that less racist firms will
out-compete more racist ones. So I don't think that we need
anything like perfect competition before the Becker competition
process will kick in to an appreciable extent.
Second, I don't think we need PERFECT information.  Your statistics
are hardly perfect, but if they were valid, there is a HUGE
profit gain that firms could win by shifting to black workers.
They don't need to know its magnitude PERFECTLY to know that it is
large.  (Assuming you're right.)
Third, perfect mobility, while helpful, isn't necessary either.
So long as there is SOME mobility, racist firms lose out to
non-racist ones.  Gee, if we really enforced this perfect
mobility assumption to ALL of labor economics, there wouldn't
be much left to the field.  We would just say there were
a millions of monopsonies in the labor market and have nothing
more to say.
I have more thoughts, but I have to go to class now.  My e-mail
address is bryancap@ocf if you'd like to reply.
                                   Sincerely,
                                         Bryan Caplan


Hello, Steve.
If I haven't already beat the subject to death, here are some more of
my thoughts about our previous discussion.
1. You said that you thought that people who "caught-up" given a worse
initial endowment were likely to be at least average, maybe above
average.  That's possible, but not too plausible.  Other things being
equal, wouldn't we always have been better off if we had more
opportunities, a better early education, more interested parents, etc.?
Even among immigrants, probably the most motivated and ambitious
segment of the population, it usually takes a generation to catch
up to the national average. One anecdotal case dear to my heart are
my girlfriend's parents. They are Romanian immigrants, and they are
the hardest working people I have ever met. But despite their
education and ambition, they just haven't caught up to what they 
would have had if they didn't have language problems, years of
wasted time, and so on.  It would be quite amazing if, for example,
people let in under affirmative action with, on the average, lower
preparation, would, if they survived at all, be perfectly average.
As for your group that you tutored, don't you think that there is
massive self-selection? That is, aren't the very best students more
likely to pursue extra help? If so, it should come as no surprise
that THEY were all in the top half of their class.
2. I mentioned that restrictive laws in the pre-1960's south explain
why the market's checks didn't stop discrimination. If I understand
it, that is precisely the point of scholars like Roback and Walter
Williams -- namely, racists typically make discrimination MANDATORY
to prevent the market's natural tendency toward non-discrimination
from emerging.  Don't you at least think that pro-discrimination laws
explain SOME of the problem? A common example given is the Davis-Bacon
Act, which, I believe, requires that prevailing wage rates be paid
to all federal construction workers. Wasn't the motive for that law
to prevent blacks from competing away construction jobs by offering
lower prices?
One might also make the argument that discrimination is a sort of
public good for all racists, so that if the government doesn't
enforce racism, each racist might "free ride" on the discrimination
of others. (Assuming that what racists want is a racist society
rather than for THEIR particular firm to be racist.)  This is
another  reason why really racist societies often mandate rather
than merely permit discrimination.
3. One irony is that I think that most discrimination in our society
is social rather than economic. I know of no of theory that shows
that races will all learn to love each other
due to market forces. But I do know of an
economic theory that says that racist employers get outcompeted
by non-racist ones.  To the extent that market performance is related
to certain social connections, and so on, you have an argument.
4. You mentioned that stockholders might want discrimination and
thereby accept a lower return if their firm were racist. I find
this very implausible, because (a) Most stockholders know little
more about the firms in which they hold stock besides their 
profitability and (b) If they know anything at all, they will know
if their firm is "socially responsible" or not. Since firms try
to promote "socially responsible" images to investors as well
as customers, it is quite likely that things cut the other way;
being racist makes investors less rather than more likely to hold
your stock.
Incidentally, I was discussing the "social sanctions" theory of
discrimination with Prof. Dickens, and he conceded that in our
society, the social sanctions for being racist are probably much
greater than the social sanctions for hiring minorities. Hence
social sanctions encourage de-segregation rather than cementing
racism, at least in our society.
5. Another theory of discrimination that I entertain (besides
the signalling theory and the worse initial endowments theory)
is that it is caused by a principal-agent problem. Hirers may
substitute their own preferences for those of the stockholders
and managers.  In short, just as they might steal office supplies,
they might steal firm profits by hiring less qualified whites
over more qualified minorities. I'm sure this happens, but as
I said, if management believed that this were common, they
could INCREASE firm profits by adopting intra-firm quotas.
They might tell their racist hirers: "Hire 20% blacks." This
would prevent the hirer from substituing his own preferences
for the good of the firm.  Well, if this kind of discrimination
is common, wouldn't you do that if you were the CEO? Are you
confident that it would increase profits?
6. If you talk to any businessperson, they will tell you that
people differ radically in all sorts of unmeasurable attributes:
effort, interest, creativity, responsibility, etc.  Since
current wage equations can only account for (if my memory does
not fail me) 30% of wage differences, isn't it plausible to
think that non-measurable differences between people are
ubiquitous and large in magnitude?  Isn't it just perverse
to do statistics on wages, and then attribute EVERYTHING unexplained
to one factor: discrimination?  Imagine, as I suggested, that
we surveyed people's taste for discrimination and tried to explain
wage differences on the basis of it.  Wouldn't that like show that
discrimination doesn't matter much, but rather "EVERYTHING ELSE"
explains most of the differences?
7. If the signalling theory is right, then the crucial thing is to
permit free flow of information.  For example, my theory predicts that
women get discriminated against because, among other reasons, employers
fear that they will get pregnant and leave work after the firm incurs
training costs. But if women could sign a contract saying that, for
example, they forfeit their medical benefits if they get pregnant,
then discrimination against women who have no intention of getting
pregnant would disappear (though the wages of women who DID intend
to get pregnant would fall still further). This system makes
most people react with horror; but isn't it fairer to let women who
want to get ahead and don't want children prove it to potential
employers?
8. You mentioned imperfections in the capital market as an
explanation for why you, for exAmple, couldn't start up a firm
hiring minority workers who have been discriminated against.
Well, capital markets sure aren't PERFECT. But do you seriously believe
that if you could show a bank or investors that you were certain
to make a profit that you STILL couldn't get a loan/investors?
I think that the main imperfection in capital markets is in
the area of human capital, since it is hard to offer collateral.
But it is hard to believe that business ventures that would
demonstrably make money (and with 27% lower labor costs, it
sounds like a sure thing) could not get a loan. UNLESS...
the discrimination is due to signalling.
9. After the riots, all sorts of shows came on comparing the
living conditions of whites and blacks.  Blacks, for example,
are much more likely to come from a broken home, to have
illegitimate children, to be on welfare, etc.  Many people,
including many blacks, discussed the problem of self-destructive
cultural values in relation to this. That is, there are many
poor groups that neverthless don't have these problems. (If
you take the whole history of immigration, the case is clearer still.)
Well, given this, why is it so unspeakable to suggest that cultural
values influence economic success? Does anyone seriously believe
that the economic success of Jews and Asians has nothing to do with
their cultural emphasis on work and education, their family structure,
etc.? What is so amazing about this idea? I agree that it is hard to
measure, but it is evident that many groups with severe disadvantages,
such as immigrants, get ahead because they have good values -- work
ethic, ambition, etc. Why must we never discuss this in relation
to racial discrimination?  None of this is to blame people who
had a bad starting point. But whatever your starting point, you
are much better off if you adopt these values.  ONLY talking about
racial inequality as stemming from irrational hatred, and never
from oneself is, I submit, implausible.
10. I must admit that I think that you really overestimate how
racist people are.  I am half-Jewish, but I have never once in
my life heard an anti-Semitic comment said to my face, even though
most people think I'm Anglo-Saxon.  Are businesspeople really that
demonic? I suspect that most businesspeople would PREFER to hire
equally qualified people who have been treated unjustly in the past.
Why don't they? Because statistically, the differences are real,
and your profits go down if you follow your conscience. 
Sad, but, I think, true.
                                        Sincerely,
                                         Bryan Caplan