. 22
( 32 .)


cumventing the ban on the immigration of Jews into Russia, rather
than take the politically unpopular step of lifting the ban.
In short, whether judgments or actions toward particular groups
are favorable or unfavorable, these actions cannot be automatically
equated with prejudgments. Indeed, it is a sweeping prejudgment
to do so, especially when those who attribute prejudice to others
often have less direct knowledge of the groups in question, at the
times in question, than those who made the favorable or unfavor-
able judgments.
Even when favorable or unfavorable judgments about groups are
based on knowledge, judgments of particular individuals from
within these groups may be made on the basis of prejudice. As
W.E.B. DuBois pointed out about the hiring of blacks in the nine-
teenth century, "the individual black workman is rated not by his
own efficiency, but by the efficiency of a whole group of black fel-
low workmen which may often be low. "Therefore, if white people
were to lose their racial prejudices overnight, he said, this would
make very little immediate difference in the economic condition of
most blacks. According to DuBois, writing in 1898, "some few
would be promoted, some few would get new places" but "the mass
would remain as they are," until the younger generation began to
improve in response to greater job opportunities due to reduced
racial barriers.
In short, DuBois saw white employers' adverse judgments of
black workers at that time as being generally accurate, though not
accurate as regards all individual black workers currently or the po-
tential of black workers in the future. Whatever the empirical va-
lidity of his assessments, what is relevant here is that he made the
crucial analytical distinction between adverse judgments based on
experience and unsubstantiated prejudgments. Moreover, while
the lumping together of disparate members of a group can explain
why particular individuals from that group earn less than compara-
The Economics of Discrimination 165

ble individuals from other groups, such individual disparities can-
not explain group disparities. For that, the generalization about the
group's performance as a whole must be erroneous. Seldom is any
serious effort made to demonstrate such group-wide errors. More
often, the simple fact that an adverse generalization has been
made about a group is taken as itself evidence of prejudice.

Biases differ from prejudices and there are at least two kinds of
bias which also differ from each other.
The first kind of bias is what might be called cognitive bias. The
person making judgments or taking actions may not begin with
any specific adverse beliefs about a particular group, nor any hos-
tility toward them, but his mode of judgment may cause individu-
als of equal ability to be ranked differently when they come from
one group rather than another. For example, someone who places
great weight on how an applicant for a job or college admissions
dresses may pass up many highly qualified members of groups that
do not put much emphasis on dress or who have different patterns
of dressing from those of the evaluator or of other groups being
Even in the absence of any adverse prejudgments or negative
feelings toward the group in question, such an evaluator may thus
systematically undervalue the achievements or potential of indi-
viduals from some groups. In addition, circumstances may also
generate bias, in addition to human bias. As a scholarly study of
India noted:

Aspiring members of previously victimized groups encounter biased
expectations, misperceptions of their performance, and cultural bias
in selection devices; they suffer from the absence of informal net-

works to guide them to opportunities, entrenched systems of senior-
ity crystallize and perpetuate the results of earlier discriminatory se-
lections. Thus norms of non-discrimination in present distributions
are insufficient to erase or dislodge the cumulative effects of past

This kind of situation is sometimes characterized in the United
States as "institutional racism," as distinguished from the inten-
tional and overt racial bias of individuals. This kind of cognitive
bias can also be present in impersonal criteria drawn up in good
faith, and many claims have been made that various mental tests
are "culturally biased" in this sense.
Before evaluating such claims, it is necessary to be clear as to
what is and is not being alleged. Nor can we simply dismiss criteria
which produce adverse results for particular groups as "irrelevant"
or substitute plausibility for an empirical test of whatever correla-
tion may or may not exist between particular criteria and subse-
quent performances. While careful empirical analysis is necessary,
that analysis must be preceded by clear definitions of what we do
and do not mean by the words being used.
Another and very different kind of bias is based on favoritism
for one's own group, which may exist independently of any be-
lief, presumption, or bias about inferior abilities in other groups.
Indeed, this kind of bias can co-exist with a belief that some
other group is not merely equal but superior. This is in fact a
phenomenon found in various countries around the world. For
example, an advocate of affirmative action programs for Malays
in Malaysia declared: "Whatever the Malays could do, the Chi-
nese could do better and more cheaply." Nor was he the only
leader to see his own group as less capable than other groups in
the same society. In Nigeria, preferential hiring of people from
the northern part of the country was advocated on grounds that
otherwise "the less well educated people of the North will be
The Economics of Discrimination i6j

swamped by the thrusting people of the South." The same has
been true of leaders of the Sinhalese in Sri Lanka, Turks in
Cyprus, Fijians in Fiji, Lulu in Zaire, and Assamese in India's
state of Assam.
A political movement organized to ban Japanese immigration to
the United States, early in the twentieth century, sought to do so
precisely on the ground that these immigrants were of high ability
and formidable competitors:

We have been accustomed to regard the Japanese as an inferior race,
but are now suddenly aroused to our danger. They are not window
cleaners and house servants. The Japanese can think, can learn, can
invent. We have suddenly awakened to the fact that they are gaining
a foothold in every skilled industry in our country. They are our
equal in intellect; their ability to labor is equal to ours. They are
proud, valiant, and courageous, but they can underlive us. ... We
are here today to prevent that very competition.

The bias of those seeking favoritism for themselves is wholly in-
dependent of any claim that others are inferior and is often insisted
upon most fervently when others are considered to have superior
performance. There may of course also be favoritism for groups who
consider themselves superior”whites in the American South dur-
ing the Jim Crow era, whites in South Africa under apartheid, or
Germans under the Nazis”but such a belief is not inherent in pro-
grams of favoritism, which are perfectly compatible with the oppo-
site belief that others are too formidable as competitors to be dealt
with without some offsetting official advantage.

Both bias and prejudice are attitudes. The practical question is
how and to what extent such attitudes are translated into acts of

discrimination. But before addressing that question, we must first
be clear as to what we mean by the word "discrimination."
Policies of treating members of particular groups less favorably
than similar members of other groups are usually called "discrimi-
nation" when practiced by the group with dominant political
power and "reverse discrimination" or "affirmative action" when
practiced against members of the group with dominant political
power. Discrimination may also be practiced by private employers,
landlords, or institutions. However, if words are to have any fixed
meanings”without which discussions are fruitless”the term can-
not be extended to all situations in which some groups turn out to
have less favorable outcomes than others.
While biases and prejudices are conditions in people's minds,
discrimination is an overt act taking place outside their minds in
the real world. Nor is there necessarily a one-to-one correlation
between the two, as so often assumed by those who make the fight
against "racism" their number one priority or by those who claim
that opponents of affirmative action must be assuming that preju-
dice and bias have been eradicated.
It is not only theoretically possible to have more discrimination
where there is less bias or prejudice, and less discrimination where
there is more bias and prejudice, this has in fact happened in more
than one country. The degree to which subjective attitudes are
translated into overt acts of discrimination depends on the costs of
doing so. Where those costs are very high, even very prejudiced or
biased people may engage in little or no discrimination. One need
only imagine the owner of a professional basketball team who has
deeply racist beliefs but even deeper commitments to making
money. The first black football player signed by the Washington
Redskins in the 1960s was hired by a man reputed among sports
writers who knew him to be deeply racist. Yet he broke a long tradi-
tion of all-white football teams in Washington by hiring a top run-
The Economics of Discrimination 169

ning back who was black, at a time when most of the leading run-
ning backs in the league were black”and when the Redskins' run-
ning game was very ineffective.
There is no inherent contradiction in a racist breaking the
color line to hire blacks”or in someone who is not a racist fail-
ing to do so. The costs they face when making their decisions
must be taken into account, along with their predispositions, in
any causal analysis. During the last years of the Jim Crow era in
the South, there were stories of Southerners like Arkansas' Sen-
ator J. William Fulbright who voted against their consciences to
continue racial segregation, because to do otherwise would be to
jeopardize their political careers. Personal costs can lead to ac-
tions either more adverse or less adverse than the individual's
own beliefs and feelings.


Where the costs of discrimination are low or non-existent for
those making hiring and promotions decisions, then discrimina-
tion can become extensive, not only in terms of decisions not to
hire or promote members of particular groups, but also in terms of
extending such discrimination to a wider range of groups.
At one time, there were said to be American railroads in which
Catholics could rise only so high and others in which this was true
of Protestants. Within living memory, there was a time when not a
single Ivy League college or university had a Jew among its
tenured faculty, despite a large number of Jewish intellectuals and
scholars available. Nor was a black allowed in the Marine Corps,
at even the lowliest rank, when World War II began. In other
parts of the world as well, there has been similarly severe discrimi-
nation, sometimes even broader in its scope.

While many discussions of discriminations ignore the cost of dis-
crimination to those doing the discriminating, on the most ele-
mentary economic principle that more is demanded at a lower
price than at a higher price, we should expect to see the severity of
discrimination vary with the cost to the discriminator. That is in
fact what we find in country after country and in era after era”but
only if we look at evidence.
In Poland between the two world wars, for example, an absolute
majority of all the physicians in the country were from the Jewish
minority, which was only 6 percent of the population. Yet the Pol-
ish government did not hire Jewish physicians, though many other
Poles obviously became patients of Jewish physicians in the private
sector, or else so many Jewish doctors could not have made a liv-
ing. What was the difference between the public sector and the
private sector in this regard?
In both sectors, there were both financial and medical costs to re-
fusing to use Jewish physicians. To staff a government hospital with
all-Gentile physicians, in a country where such physicians were in
the minority in that profession, meant either having to pay more to
attract a disproportionate share of the available Gentile physicians
or accepting lesser-qualified physicians than some of those available
from the Jewish community. In either case, financial or medical
costs were entailed, if not both. However, in the case of those who
made decisions within the Polish government, there was no cost at
all to be paid by them. Financial costs were paid by the taxpayers
and the human costs were paid by patients in government hospitals,
subject to lower quality medical treatment than was available in the
society at the time. Neither of these costs was a deterrent to dis-
crimination by government officials.
In the private sector, both kinds of costs were paid by sick peo-
ple. Concern for ones own personal health, especially in an emer-
gency situation or when confronted with a potentially crippling or
The Economics of Discrimination 171

fatal disease, could easily overcome whatever anti-Jewish attitudes
one might have. Given the respective incentives in the govern-
ment and private sectors, the different levels of discrimination
against Jews is very much what one might expect, on the basis of
the most elementary economic principles.
Poland provides examples of another phenomenon”more dis-
crimination where there was less hostility and less discrimination
where there was more hostility. Anti-Jewish feelings tended to be
stronger in eastern Poland than in western Poland. Yet Jewish arti-
sans were more prevalent in eastern Poland, just as black artisans
once had better job opportunities in the American South, where
racism was most rampant. In both cases, organized labor affected
the cost of discrimination.
Guilds were stronger in western Poland than in eastern Poland
and American labor unions were stronger in the North than in the
South during the eras under discussion. To the extent that orga-
nized labor succeeds in raising pay levels above where they would
be under supply and demand in a free market, they provide incen-
tives for employers to hire fewer workers because labor is now
more costly, both absolutely and relative to the cost of capital that
may be substituted for it. At the same time, wage rates raised
above the level that would prevail under supply and demand at-
tract more workers who apply for jobs that have higher pay. The


. 22
( 32 .)