cen02-2016
Constitutional
Economics
Network
Working Paper
Series
ISSN No. 2193-7214
CEN Paper
No. 02-2016
On the Origin and Consequences of Racism
Matthew Bonick* and Antonio Farfán-Vallespín**
*Department of Economic Policy and Constitutional Economic Theory,
University of Freiburg, Germany.
E-Mail: matthew.bonick@vwl.uni-freiburg.de
** Department of International Economic Policy,
University of Freiburg, Germany.
E-Mail: antonio.farfan@vwl.uni-freiburg.de
December 29, 2016
University of Freiburg
Institute for Economic Sciences
Department of Economic Policy and Constitutional Economic Theory
Platz der Alten Synagoge / KG II D-79085 Freiburg
www.wipo.uni-freiburg.de
On the Origin and Consequences of Racism
Matthew Bonick∗
Antonio Farf´n-Vallesp´ †
a ın
December 29, 2016
Abstract
Using a novel method to measure racism at the inpidual and country level, we show, our
measure of racism has a strong negative and significant impact on economic development,
quality of institutions, education and social capital. We test different hypotheses concerning
the origin of racism and its channels of impact to establish causality. We find racism is not
correlated with measures for the coexistence of different racial or ethnic groups or ethnically-
motivated conflicts. Importantly, we show, for former colonies, racism is strongly correlated
with the presence of extractive institutions during colonial times, even after controlling for
current institutions, GDP per capita and education. We argue, extractive colonial institu-
tions not only had a negative impact on the political and economic institutions but also
shaped the cultural values of the population. We claim colonial powers instilled racism
among the population of their colonies in order to weaken their ability for collective action.
∗ University of Freiburg : email: matthew.bonick@vwl.uni-freiburg.de
† University of Freiburg : email: antonio.farfan@vwl.uni-freiburg.de
1
1 Introduction
In the literature, trying to identify the long term determinants of economic development, there
is a lively discussion on the role of culture and of institutions. We contribute to this discussion
by focusing on an important cultural value, racism, measuring it in a novel way, and estimating
its impact on GDP per capita, education, institutional quality, social capital, and conflicts. We
also examine if racism is strongly correlated with cultural values and preferences we consider
to contribute to the persistence of inefficient political and economic institutions. These cultural
values and preferences include trust, importance of obedience, feeling of control of ones life and
importance of the family, intolerance towards women, immigrants, inpiduals of other religions
and those who speak a different language. Beyond the impacts of racism, we estimate its possible
origins and find out that, for former colonies, extractive institutions are the ultimate cause of
racism.
Recent political developments in US and Europe have brought racism to the mainstream
public discourse. Examples are the victory of Brexit, the election of Donald Trump and the
success of certain parties in Europe. In all these cases their campaigns heavily relied on anti-
immigrant and anti-racial rhetoric and policy proposals are an indication racism is important in
both economic and political discourse.
In the literature, the existence of racial and ethnic discrimination has already been shown
through several empirical analyses in laboratory and field experiments. Racial and ethnic discrim-
ination are pervasive in a number of different contexts, like the labour market, consumer market
(Ayres and Siegelman 1995; List 2004), credit market (Blanchard et al. 2008; Blanchflower et al.
2003; Ladd 1998; Munnell et al. 1996) and housing markets (Ahmed and Hammarstedt 2008;
Beatty and Sommervoll 2012; Bosch et al. 2010; Hanson and Hawley 2011).
However, there is so far no measurement of racism at the country level that allows for com-
parisons and analyses of the determinants and consequences of racism at a macroeconomic level.
We use a novel way for measuring racism extracted from the World Value Survey. The variable
is derived from the question, ”On this list are various groups of people. Could you please mention
any that you would not like to have as neighbors?”. The groups in the list include criminals,
people from other religions, homosexuals, terrorists, and people from another race, among many
2
others. We are as far as we know the first ones to identify a way of measuring racism and to
attempt to address these questions at the macroeconomic level.
First, we aggregate the inpidual values to form a national value of racism for each country.
We then estimate the effect of racism on GDP per capita, institutional quality, education, social
capital and conflict. We find that racism is negatively correlated with institutional quality under
all specifications. Under the majority of specifications, racism is also negatively correlated with
all our outcome variables, except for conflict, but this significance disappears when we control
for institutional quality, log of GDP per capita or education.
Next, we attempt to identify causality by estimating the origins of racism. For this we
formulate all conceivable hypotheses about the possible origins of racism. We then test them, at
the inpidual or country level, to see which hypotheses are accepted or rejected.
First, at the inpidual level, we show that cultural and socio-economic factors are predictors
for an inpidual possessing racist attitudes. We find that having a racist attitude is negatively
correlated with an inpidual’s income level. We also find that racists tend to have a lower
education level, are more likely to be male, are slightly older, have lower life satisfaction and
tend to come from smaller population centers. These results at the inpidual level are consistent
with the results of several studies based on more limited samples. Additionally, we examine
the correlations between amoralism (importance of the family) and racism. Counter to our
hypothesis, racist attitudes are significantly associated with less amoralism.
At the country level, we find that migration or different measures of coexistence of perse
races are not correlated with racism and, religion is only a weak predictor of racist attitudes. For
former colonies, countries with higher levels of extractive institutions during the colonial period,
have higher levels of racism in the present. We claim this relationship must be causal because,
before colonization, countries that receive extractive institutions were wealthier, better educated
and had more developed institutions and, according to our findings at the inpidual level, they
should, thus, have had less and not more racism. However, given these are not the results we
find, the presence of higher levels of racism can be traced to the exogenous shift in institutions
during the colonial period.
We also attempt to answer the question whether racism was deliberately instilled in extractive
colonies in order to facilitate the persistence of extractive institutions, or whether racism is a
3
consequence of poor institutions and low education. For this, we identify cultural values that
might also be beneficial to deliberately instill by an extractive colonial power, such as, norms that
make collective action more difficult, a higher preference for extractive political and economic
institutions and, other values of intolerance. Additionally, we examine the correlations between
these values and racism.
For different measures of intolerance, the results indicate a strong correlation between extrac-
tive institutions and racism. On the other hand, there was no consistent relationship between
extractive colonial institutions and our variables for; cooperative values, or preferences for eco-
nomic and political institutions. However, racism displays a significant connection to; a lower
value placed on cooperative norms and, open economic and political institutions. Therefore,
we cannot claim that all extractive values were deliberately instilled by extractive institutions,
however, there is a clear relationship between racism and extractive values and preferences. We
suggest, there might be a difference between social structural values on which the persistence of
institutional systems rest, like racism and other values of intolerance, and other preferences that
could vary more widely as a result of short term influences, such as, preferences for democracy.
The correlations between extractive institutions, racism and other measures of intolerance sup-
ports the notion that extractive colonialists, created a hierarchical society as a tactic to protect
their economic, political and social dominance of society.
We contribute to the literature by trying to determine the long-term determinants of economic
development and whether institutions, culture or geography are the main factors. We also
contribute to the understanding of how inefficient institutions can persist over Centuries. We
show that institutions are the main determinant, but that institutions might be able to influence
culture in order to facilitate the persistence of inefficient institutions.
After the introduction, we discuss the different definitions of racism and provide our own
definition. In section three, we discuss, from a theoretical perspective, the different effects
racism could have on the economy, education, institutional quality, social capital and ethnically
motivated conflicts. Section four presents a number of hypothesis on the potential origins of
racism. Section five describes our data and how the variables interest are constructed, with the
following section displaying our empirical results. Section seven concludes.
4
2 Definitions of racism
The first problem we encounter before we define racism is how to define race. The Cambridge
Dictionaries Online1 defines race as ”A group, especially of people, with particular similar physical
characteristics, who are considered as belonging to the same type, or the fact of belonging to such
a group.” and ”A group of people who share the same language, history, characteristics, etc.”
For the Oxford Dictionaries2 race is ”Each of the major pisions of humankind, having distinct
physical characteristics: people of all races, colors, and creeds”, or ”A group of people sharing
3
the same culture, history, language, etc.; an ethnic group”. We can see that race seems to be a
rather subjective concept, since it is difficult to determine the degree to which two given groups
share a common history, or their languages are similar enough, or how close their cultures are.
The boundaries of the identification with race or ethnic group may even vary over time according
to external factors like electoral competition (Eifert et al. 2010).
We accept this relatively blurred definition of race because for the persons in the survey who,
to the question of whether they do not want somebody from a specific group as a neighbor, they
answer that they do not want somebody from a specific race, the concept of race seems to be
clear enough and an inpidual must have some expectations about the utility they would lose in
case a member of that group would live in their neighborhood. Further, these respondents must
also consider race as an important enough factor in human behavior to choose this dimension
over many other characteristics.
The next step is how to define racism. On the one hand, racism can be understood as a
distaste for certain races, a negative preference for the interaction with people from certain races
in the sense of Becker (1957). Becker defined that ”discrimination in the marketplace consists of
voluntarily relinquishing profits, wages, or income in order to cater to prejudice”.
This negative preference for certain races means that, there is a negative premium on the
utility born by the racist each time he interacts with someone from the disfavored race. According
to Becker , this prejudice should not survive under perfect competition since discriminating is
1 (http://dictionary.cambridge.org/es/diccionario/ingles/race)
2 (http://www.oxforddictionaries.com/definition/english/race)
3 It further clarifies that ”In recent years, the associations of race with the ideologies and theories that grew out
of the work of 19th-century anthropologists and physiologists has led to the use of the word race itself becoming
problematic. Although still used in general contexts, it is now often replaced by other words which are less
emotionally charged, such as people(s) or community.”
5
costly and in the long run discriminators would be driven out of the market by non-discriminating
agents not bearing such a cost.
On the other hand, most of the definitions of the word racism in dictionaries suggest that
racism is a belief, rather than a preference. For the Merriam Webster online dictionary4 racism
is ”poor treatment of or violence against people because of their race” or ”the belief that some
races of people are better than others”, so it suggests both distaste and beliefs. They also provide
a full definition of racism as the ”belief that race is the primary determinant of human traits
and capacities and that racial differences produce an inherent superiority of a particular race”
5
and also ”racial prejudice or discrimination”. According to Oxford dictionaries, ”Prejudice,
discrimination, or antagonism directed against someone of a different race based on the belief
that one’s own race is superior” and ”The belief that all members of each race possess character-
istics, abilities, or qualities specific to that race, especially so as to distinguish it as inferior or
superior to another race or races: theories of racism” and according to Cambridge Dictionaries
online6 racism would be ”the belief that people’s qualities are influenced by their race and that
the members of other races are not as good as the members of your own, or the resulting unfair
treatment of members of other races”
Some common elements emerge from these definitions. First, that racism is a belief that
inpiduals belong to a race, that this membership determines qualities that are shared by all
members of that race. Second, that these racial characteristics are the fundamental determinant
of human behavior. This implies, racism downplays the importance of inpidual differences in
behavior and fails to evaluate inpiduals based on their own merit or performance, but rather
evaluates them based on their subjective belonging to an exogenously determined group. In the
absence of perfect information, race is used as an informational trigger to help make an estima-
tion of characteristics an inpidual will display. The third element in these definitions is the
justification of discrimination, meaning that the racist considers that people of other races should
receive a different treatment than people from his own race, based on a hierarchical consideration
according to the different levels of racial worth. Therefore, races with lower valuations do not
deserve to be treated fairly and discrimination against them is not only allowed but justified.
4 (http://www.merriam-webster.com/dictionary/racism)
5 (http://www.oxforddictionaries.com/definition/english/racism)
6( http://dictionary.cambridge.org/dictionary/english/racism)
6
Finally, another important aspect that sometimes appears associated with racism is the notion
of racial animosity. The racist has a mindset that considers the relationship between racial
groups as one of conflict and hostility. For the racist, ones own racial group must be in constant
state of alert because members of other groups, (or the entire groups) might attempt to attack,
harm or take advantage of their own group if one is too weak, too confident or too relaxed see
Blumer(1958) for further discussions on the concept of group threat and its implications.
We summarize the different elements from all the definitions we have found and suggest
our own definition according to the notion an inpidual is considered a racist by holding one
or more of the following beliefs, this standard excludes the first belief specified, as almost all
inpiduals identify with a race, to varying degrees:1) that inpiduals belong to a race , 2) that
this membership to a race determines qualities that are shared by all members of that race , 3)
that these racially-determined qualities are the fundamental determinant of human behavior ,
4) that races can be classified according to their qualities in different hierarchical levels of racial
worth , 5) that when dealing with people from of certain races of lower worth it is not necessary
to follow the same standards of respect, honesty, justice, fairness or benevolence that one should
apply with members of its own race. , 6) that the relationship between races can only be of
conflict and that, as a consequence, a constant state of alert is required since the racist beliefs
that only evil can be expected from other racial groups. This is the belief behind racial hate.
The components of our definition will likely not manifest in the same way for all inpiduals.
Some people may hate other races but others may simply think other races are untrustworthy
or less productive. Thus, how racism effects behavior of a racist will not be homogeneous across
the racist population. We will explain how these different dimensions will effect outcomes in the
following section.
From this definition, we can see that these beliefs include many strong assumptions on the
part of the racist. First, as we have mentioned it is already broadly discussed on the literature
whether races can be clearly identified since inpiduals rather form a continuum of different
characteristics and it is difficult to find an exact point at which certain attributes start defining
a race in a consistent and clear way (Eifert et al. 2010). Second, it is a heroic assumption to
accept that all people in a given race have an average value of a certain characteristic that is
significantly different from the average value of other racial groups and that the within-group
7
variation of these values is so small that the probability that the lowest instance of a given racial
group has a lower value than the highest inpidual of the other group is insignificant. The
underlying ethical system of the racist according to this definition considers that it is justifiable
to give worst treatment and even to commit aggression against another group just because they
are believed to have a lower average value in certain characteristics7
Additionally, even if an inpidual does not display explicit hatred towards other races, the
racist ethical system provides a justification for inaction in fighting the persistence of an unfair
political, economic and social system based on racial pision. Thus, people may not be active
racists, but they might support, or at least not combat, political decisions that might harm
people from other racial groups.
Our definition of racism does not exclude the possibility that these beliefs might be correct, at
least for a certain restricted social environment. Some racial groups could be clearly identifiable
and actually present a statistical distribution of certain values so unfavourable that could make
racial discrimination a rational strategy. If this would be the case, we would observe that racism
would have a positive impact, since it would be a cheap way of solving problems of imperfect
information in some market transactions. There is also the possibility racism drives internal
group social cohesion, increasing efficiency and cooperation at the group level.
Further, even if beliefs about other races’ characteristics are not correct ex-ante, the literature
has shown that these beliefs could turn out to be correct ex-post. These are cases of self-fulfilling
prophecies in which even when starting from an equal distribution of values across races, racial
discrimination of one group by the other can make the actual values coincide ex-post with the
initially wrong beliefs. In this case, this rational statistical discrimination would still be the
effect of prejudice.
While it can be reasoned that statistical discrimination may be rational, we wish to highlight
some important consequences which have been emphasized in the literature. First, as Fryer et
al. (2005) show, in a class room hiring game, once a discriminatory equilibrium is established,
meaning a general view that one group is more productive, it remains persistent for a period
7 We can also connect racism with the definition of tribalism by Popper in the “Open Society and its Enemies”
according to which tribalism is the allocation of a supreme importance to the tribe, without which the inpidual
means nothing. We can also associate racism as a form of collectivism: emphasis on collective rather than
inpidual action or identity (Merrian Webster); racism would therefore be a form of collectivism since it considers
the collective as more important in determining human actions than the inpidual free will.
8
of time even after the underlying inequalities ,which determine productivity, are removed. In
the game, once the inequities were removed, it took a number of rounds for this change to be
reflected in the hiring rates. Additionally, a number of employers within the game continued to
utilize a color based strategy when engaging in hiring decisions, indicating, a scenario can emerge
were stereotypes, which at one point are rational, persist even after they are no longer justified
(Fryer et al. 2006).
We believe that our expanded definition better captures the animosity and aggressively that
we observe in many racist actions. A simple dislike for other racial groups should be similar to
a dislike for a particular ice cream flavor. Even if someone does not like chocolate ice cream,
this does mean that this person is going to burn ice cream shops selling chocolate ice cream, or
promote regulation forbidding the sale of chocolate ice cream. As it is, we observe that racist
inpiduals are not only less inclined to interact with people from other races but, we also observe
some of them commit acts of violence against people from other races, like the Ku Klux Klan
actions of intimidation or the attacks on refugee homes in Germany.
3 Theoretical consequences of racism
In this section, we derive our hypotheses about how the different dimensions of our definition
of racism could affect the economy, the quality of institutions, education and, the level of social
cohesion of a society. First, we discuss the behavior that we expect from racist agents according
to our definition. We then present the different ways in which these behavioral patterns could
affect our outcome variables and support it with insights from the theoretical literature and,
with evidence from experimental and microeconometric studies.
3.1 Dimensions of racism
There are several ways in which racism could have a long-lasting effect on our outcome variables.
For discussing this, we begin by assuming that racism is a binary characteristic, as such, there
are racist and non-racist agents. Based on our definition of racism, we expect a racist agent
to exhibit certain differential behavioral patterns in those transactions in which the race of
the participants is salient. This saliency can come from different sources, maybe there is direct
9
information about the race of the agents taking part in the transaction, like with reviewing
personal services or direct labor relationships or, the racist agent can form an expectation about
the prior probabilities of the racial group to which the transaction partner belongs, even if the
race itself is not revealed. We will refer to transactions where the agents have some information or
expectation about the racial type of the partner as racially-informed decisions. Importantly,
racism can also affect racially-informed transactions where all transaction partners are of the
same race, since potential partners from other races might have been excluded deliberately and,
thus, outcomes of that transaction are going to be different than if those excluded partners would
have participated in the transaction.
3.1.1 Racism as distaste for certain races
The most common way to understand racism in the economic literature, since the seminal work
of Becker (1957), is to treat it as a distaste for other races. According to this view, the utility of
a racist agent is negatively affected in racially-informed interactions with a partner that is from
a race in which there is a distaste. The utility will be neutrally or even positively affected if
there is a partner of the same or a different race. This utility premium, negative or positive, will
alter economic decisions of racist agents. Transaction costs for searching for racially-acceptable
exchange partners will increase and, in some circumstances the racist agent will be willing to
accept transactions in worst conditions than those offered by a partner from a disliked group, if
that is the best racially-acceptable alternative. Some transactions will not even take place if the
there is a high negative premium for a racist agent exchanging with a transaction partner of an
undesired race and, at the same time the transaction costs of finding an alternative transaction
partner of an acceptable race are too high. These transaction costs will, thus, generate an
efficiency loss to the economy. The larger the intensity and the extension of racism across the
population, the larger the efficiency loss will be.
3.1.2 Racism as belief that some racial groups have a lower level of abilities than
others
Becker(1973)claims that ”discrimination in the marketplace consists of voluntarily relinquishing
profits, wages or, income in order to cater to prejudice”. Our definition of racism suggests
10
that racism might have many more aspects rather than just a distaste for certain racial groups.
Racists may hold the belief that other racial groups, on average, have a lower average competence.
Therefore, racism might imply a racial bias in the estimation of the expected returns of racially-
identifiable factors, particularly labor or, the quality of goods acquired from racially-visible
producers of certain races. This racial bias can also lead a racist agent to hold biased expectations
on the risk premium needed for the extension of credit to inpiduals of different racial groups.
If these expectations are right, we will be talking of statistical discrimination (Arrow, 1972
and 1973) which we discuss later. If these expectations are wrong, the agent will suffer losses (or
forgo some gains) each time he engages in a racially-informed transaction. If the agent does not
change these wrong beliefs, wilfully or not and, acts according to them, then we are speaking of
racial prejudice in the sense of Becker(1973), which is another aspect of racism.
According to some psychological evidence, based on trust games (Burns, 2012), these biased
beliefs might even be accepted by the members of the disfavored racial group, leading them to
actually show less confidence, particularly when interacting with members of the racial group
considered ”superior”. This would be a form of self-accepted prejudice by which the biased
beliefs end up becoming a self-fulfilling prophecy. If this effect is sizable, it would add to the
inefficiency caused by racism.
Another way in which racial prejudice could be costly is by inducing a sub-optimal screen-
ing of the characteristics of a transaction partners depending on their race, which is particularly
relevant for; candidates in the labor market, prospective credit takers or real-state renters. In
the situation where obtaining and using truthful and relevant information about the transaction
partner is costly, in resources or time, racial bias might induce the racist agent to make decisions
based mostly on the racial identity of the transaction partner while ignoring other relevant in-
formation. The bias could also reduce the incentives to search for additional information about
important characteristics. The racist agent may thus, forgo potential gains from exchange as a
result of not utilizing or searching for all relevant information, or impose sub-optimal contract-
ing conditions on himself or the corresponding exchange partner. An example of this scenario is
higher mortgage interest rates derived from sub-optimal racial screening.
Making decisions based on racial profiling might be an efficient heuristic under certain circum-
stances. If the expectations are based on actual Bayesian updating and the cost-benefit analysis
11
says that the losses from the potential mistakes made by ignoring some relevant information by
making decisions based just on the racial profile, are compensated with the gains derived from
the expediency and energy-savings in making decisions. In this case, racism would be an efficient
heuristic and we would expect at the macroeconomic level that the more this heuristic would be
extended, the more efficient the economic system would be, all other things being equal.
3.1.3 Racism as distrust of certain racial groups
According to our definition, the racist agent considers that members of certain racial groups have
lower moral standards and are less reliable, leading to the racist having a lower level of trust
on members of other racial groups. The economic and institutional impact of trust are already
well-documented. Trust between inpiduals is fundamental to the ability to cooperate, a vital
factor in facilitating efficient interaction. Trust has a causal effect on a number of important
variables, which includes: economic development, education and the functioning of political
˜ ˜ ˜
institutions (BjA¸rnskov 2011, 2012b; BjA¸rnskov and MA c on 2013, 2015; Dearmon and Grier
2009; Dearmon and Grier 2011; Knack 2002, 2003; Knack and Keefer 1997; LaPorta et al.
1997). Further, Guiso, Sapienza and Zingales (2009) show systematic differences in the level of
trust of European managers from certain nations towards others, the results of their study find,
lower levels of bilateral trust lead to lower amounts of trade, portfolio investment and direct
investment between the two countries. The outcomes hold even after controlling for country
specific characteristics with the results becoming stronger for goods that are more trust intensive.
The lack of trust between racial groups might have similar consequences, especially if races
coincide with the borders of countries, however, this dynamic can also act inside the borders of
nations. The role of trust across racial and ethnic lines has been exemplified in a number of
experiments, indicating it is a relevant phenomenon. Fershtman and Gneezy (2001), through
an experimental approach, identify the presence of ethnic stereotypes which hinder cooperation.
This experiment detects the presence of ethnic discrimination in Israeli Jewish society by utilizing
the classic trust game, with the results showing a systematic distrust for men of Eastern origin,
even after accounting for the possibility of taste based discrimination, statistical discrimination
and in-group biases . The authors argue the outcome proves the existence of discriminatory beliefs
about trustworthy characteristics of certain ethnic groups that negatively effects the potential for
12
mutually beneficial cooperation. Burns (2012), ran a similar experiment in South Africa, showing
systematic distrust of black players, by white participants, within the trust game. From these
experiments, we can see how lower levels of trust towards different ethnic and racial groups could
prevent mutually beneficial cooperation, or increase transactions costs through, the amplified
need for costly enforcement mechanisms to overcome distrust during exchange.
3.1.4 Racism as hatred towards certain races
According to our definition of racism, a racist agent could feel animosity towards certain races,
and have a conflict mindset towards them. This animosity might lead the racist to possess hostile
attitudes towards certain racial groups, which would be translated into malevolent preferences
towards them (Hirshleifer, 1991). In occasions in which the racist with malevolent preferences
has the opportunity to hurt a member of the hated race, at a relatively low cost for himself,
given the possibly utility gains from this action, the agent would act, causing a net welfare loss
for the entire system.
Additionally, any advantage received by the hated racial group, for example, targeted policy,
would be considered unfair and immoral, leading to active opposition. This likely will lead to a
higher probability of political or even violent conflicts. The possession of malevolent preferences
will have profound political consequences, as the racist would oppose any policy that might
produce positive outcomes for the hated group, be it subsidies, asylum, education, etc. Thus,
racism can influence politics by making voters and politicians consider only the welfare of their
racial group or, even deliberately attempt to harm the hated group, with the goal of obtaining
an advantage for their own group relative to the opposed race, leading to racism-inspired
political preferences.
As a result, this could lead voters to supporting policies that might be suboptimal from an
aggregate welfare perspective, as they may only benefit one racial group or, even be designed
to harm certain racial groups. Further, racism could also reduce the willingness to contribute
to public goods if the other racial groups benefit. There is survey evidence (Bobo, 2012) of the
existence of preferences against affirmative action, school busing of blacks, and a belief there is
too much government spending on blacks.
13
3.1.5 Racism as acceptance of a lower status for other races
Even if a racist does not possess malevolent preferences, he might not actively oppose the unequal
application of rights or different forms of mistreatment of different races. One of the consequences
of this is, if another racial group suffers from lower economic success, lower educational attain-
ment or worst living conditions in general, worse treatment by the police, or unequal access to
the court system, these conditions could be accepted as natural or, not worth the cost of time
and resources to change. Racist citizens might not consider overcoming these gaps as a political
goal worth pursuing the way they might consider it if citizens of their own racial group would be
living in these circumstances. We can see this with a portion of white inpiduals opposition to
racial targeted policies in the United States, even if there is a basic understanding these groups
have been historically mistreated (Bobo 2012).
As a result, re-distributive policies benefiting these groups will receive lukewarm support from
racist voters or might even be received with opposition as a waste of resources.We can see this
is with different levels of support among US citizens of different ethnicity to policy goals like
government guaranteeing equal opportunity, government should taking steps for fair employment,
targeted government spending, government effort to improve social economic position of blacks
or, preferential hiring (Bobo 2012).
3.1.6 Racism in social norms
The effects of racism can be amplified by the fact that such behavior can influence the social
norms and force non-racists to behave in racist ways or, at least be silent when observing racist
behavior. One aspect neglected so far in the economics literature, is that racism, when it reaches
a certain critical mass, can affect the values of a society. The racist agent feels himself as holding
the moral high ground and considers non-racists as weaker, from a moral perspective. The
desirable social order for a racist is one where races are segregated and, where the disfavored
races have a lower status legally, politically and economically. Therefore, a racist agent will see
those who interact with other races as defectors, as inpiduals who do not contribute to the
higher public good of building a racially segregated social order in order to egoistically reap-off
the short-term gains from interacting with inpiduals from hated racial groups.
14
From what we know from the experimental economics literature on enforcement and punish-
ment (Fehr and Fischbacher, 2004), many inpiduals will be willing to act as voluntary enforcers
to secure contributions, punishing those who fail to contribute to the public good at their own
cost, although in this case, paradoxically, for the racist those who fail to cooperate are those
who cooperate with the disfavored racial group. Once racist agents succeed in establishing racist
social norms in a society, non-racists might be forced to behave in or, accept racist ways to
comply with social norms and avoid costly social sanctions. We believe these social sanctions
can play a fundamental role in preserving racism in societies, even if it is economically inefficient.
In some cases these social norms can even become legal norms, like the Jim Crow laws or the
South-African Apartheid.
As an illustration, Bobo(2012) identifies the following racist views which we consider social
norms: a) Separate schools, b) Not voting for officials of another race; c) Laws against intermar-
riage ; d) Right to segregate neighborhoods, as home sellers can discriminate.
3.1.7 Racism as statistical discrimination
The second explanation for the presence and persistence of racial and ethnic discrimination is
known as statistical discrimination. Arrow (1973) and Phelps (1972) models point to discrimi-
nation which is not based on personal preference. The models propose that, on average, different
racial groups may develop different levels of productivity. Arrow (1998) claims such a differential
is due to a number of factors including but not limited to: education and cultural differences. As
a result of the differential, employers will develop the expectation that some races are, on average,
less productive than others. When faced with limited information about each applicant’s level
of productivity, race based expectations can provide additional valuable information. The ob-
servable characteristic of race serves as a proxy for unobservable characteristics which may affect
productivity. In the example of an employer choosing between two applicants, under incomplete
information about productivity, the employer will rationally favour one race over another if the
observable characteristics are similar or identical.
The selection process is driven by expected productivity, which can be argued is rational and,
on average, efficient. Statistical discrimination can also be extended to other contexts beyond
markets. If the assumption develops that, on average, all or certain races, cannot be trusted or
15
possess an incompatible set of beliefs, statistical discrimination of this nature can be a driving
factor preventing inter-racial cooperation in social, economic and political exchange. In the
political sphere, if there is an assumption that one racial group is untrustworthy it will reduce
the probability of cooperation with that race, resulting in, a lower ability to solve collective
action problems.
3.2 Expected consequences of racism on outcome variables
The previous behavioral patters, motivated by the beliefs and preferences defining racism, could
affect our outcome variables, GDP per capita, Education, Social Capital, Institutional Quality
and Conflict, in different ways. Given that to best of our knowledge this is the first paper
anaylzing racism from a macroeconomic perspective, we consider necessary to spell out in detail
the channels through which racism can have an impact.
3.2.1 Impact of racism on the economy
One of the main impacts of racism on the economy is the segregation of markets with rich racial-
information into different racial sub-markets. In those markets where race is more visible, such
as labor, credit, real state or personal services, racism can cause distortions when, because of
racism, a transaction cost arises. Some racially-informed transactions do not take place or less
efficient but racially-compliant alternatives are chosen in those transactions. This means that,
for instance, one unit of labor coming from one racial group, will not be treated as a perfect
substitute of one equivalent unit of labor from another racial group even if they are equivalent
in reality. At a macroeconomic level this transaction cost will act like a tax or a tariff on the
discriminated producers, customers or workers. This will cause these markets not to be fully
integrated, with racism operating as an internal tariff inside the economy, keeping prices of
factors and goods at different levels, reducing the market size of each racial sub-market and
causing the economy to forgo important gains from trade and specialization. Racial sub-markets
will also have distorted levels of profitability and the allocation of production factors will be
accordingly inefficient. The profitability of some racial sub-markets might be below the survival
threshold and might not exist. Markets where race is not observable will not be segregated, will
16
be larger and more profitable and will receive more investment than they would receive if racism
would be absent.
There is a large amount of experimental evidence on the existence of racial discrimination
in different areas of the economy. Bertrand and Mullainathan (2004), Pager et al.(2009) and
Gaddis (2015) among many others find evidence of racial discrimination in labor markets, List
(2004) finds evidence of discrimination in the sports card market, where members of minorities
consistently received inferior initial and final offers. Acolin, Boastic and Painter (2016), Beatty
and Sommervoll (2012) and Hanson and Hawley (2011) showed the existence of racial discrimi-
nation in the housing rental market and Pager and Shepherd (2008) and Williams et al. (2005)
argue that racial discrimination also exists in the mortgage credit market. All these findings
show that prices do not equalize in the same market for equivalent factors and goods, supporting
therefore our hypothesis of internal trade barriers based on racial lines.
At the macroeconomic level, Guiso Sapienza and Zingales(2009) show that the level of trust
between different European countries lead to a difference in the volume of economic activity
between these countries. We expect racism to work in similar ways. If racism is directed towards
a racial group dominant in another country, we also expect similar outcomes at the country level.
If racism is directed towards racial groups inside the same country, the internal transactions
between these two groups will be affected, leading to a relative impoverishment of both groups.
This hypothesis is supported, at the micro level, by the results of Burns (2012), in which lower
levels of trust towards black inpiduals lead to sub-optimal outcomes at the group level.
Finally, since it has been documented in the literature that all the other outcome variables
have an impact on GDP, it is clear that any effect on education, social capital, institutional
quality or conflicts will have an effect on the economy.
3.2.2 Impact of racism on social capital
We consider racism to be a force destroying social capital in all of its aspects almost by its own
definition. According to our definition of racism, racist agents present a distaste for interacting
with members of certain races, consider them as less capable and more untrustworthy, and hate
those inpiduals belonging to these races.
The literature has identified different aspects of social capital and all are affected by racism.
17
One of these aspects is its role as a promoter of cooperation, like in the definition of social
capital by Fukuyama by which, social capital is ”an instantiated informal norm that promotes
cooperation between two or more inpiduals, the network of social connections that exist between
people, and their shared values and norms of behavior, which enable and encourage mutually
advantageous social cooperation”. It is obvious that racism deteriorates social capital, since
cooperation between a racist agent and a member of a disfavored racial group will become
extremely difficult.
Social capital has many other aspects apart from the ability to cooperate that are also affected
by racism. Fukuyama(2001) defines the radius of trust of social capital as the circle of people
among whom cooperative norms are operative. Fershtman and Gneezy (2001) and Burns (2012)
show this using the trust game. According to our definition, the racist agent considers that
members of the hated races do not share and should not be subject to the same cooperative
norms as those inpiduals of their own racial group. Therefore, we can claim that racism
reduces the radius of trust to ones own races or, to those races deemed acceptable, thus, limiting
cooperation only to the racial groups inside the radius. We see another example of this behavior
from Pecenka and Kundhlande (2013) and their experiment with the dictator game, which shows
that racial identity influenced theft decisions. The standards of moral behavior were different
depending on the racial group players were interacting. The results showed that even among
black players, participants were more likely to engage in theft when paired with black inpiduals,
highlighting the existence of inconsistent moral norms which vary along racial lines.
Another aspect of social capital affected by racism is bridging social capital, that is, ”social
networks between socially heterogeneous groups”. Bridging social capital is important because
it ”allows different groups to share and exchange information, ideas and innovation and builds
consensus among the groups representing perse interests.8 ”. Clearly, the distaste and the
hostility of the racist agent towards disliked racial groups will prevent the formation of these
networks between racist agents and members of the disliked racial groups. This problem might
be compounded if racists are able to influence social norms and impose social sanctions on inter-
racial personal relationships (weddings, friendships,...) In this case, these social sanctions will
extend the destruction of social networks to non-racist agents. This inter-racial pide would
8 (http://blogs.worldbank.org/publicsphere/bonding-and-bridging)
18
cause information not to flow between racial groups and, as a consequence, prejudice and wrong
beliefs would be more likely to persist since there is not sufficient interaction to overcome them.
Finally, using the definition of social capital of the OECD by which, social capital is defined
as ”networks together with shared norms, values and understandings that facilitate co-operation
within or among groups”, racism would then not only reduce the interaction between members
of different racial groups, but the separation of these networks across racial lines, would facilitate
the emergence of separate social norms shared only by the members of one racial group but not
the other. Different norms would make inter-racial understanding even more difficult, further
inhibiting the probability of cooperation and the exchange of ideas. For instance, Burns and
Keswell (2015) implement a public goods experiment highlighting that racial homogeneity does
not uniformly determine higher contributions to public goods but the racial makeup of each
group affects patterns of communication.
However, racism could also strengthen internal group cohesion, increasing bonding social
capital, and this in turn could positively affect generalized trust if the perception of general is
associated with ones own group. For instance, Heap and Zizzo (2009) show in their experiment
that the creation of artificial groups can induce in-group cooperation biases. The authors did
however, note that such in-group biases, comes at the cost of social capital at the aggregate.
In turn, social capital can also have an impact on the economy and institutions. In particular,
the variable we use as one of our proxies for social capital, generalized trust, has been found to
be correlated with and even having a causal effect on economic development, education and func-
˜ ˜ ˜
tioning of political institutions (BjA¸rnskov 2012; BjA¸rnskov, MA c on 2015; Dearmon, Grier
˜ ˜
2009; Dearmon, Grier 2011; Knack 2003; Knack, Keefer 1997; BjA¸rnskov 2011; BjA¸rnskov,
˜
MA c on 2013; Knack 2002; La Porta et al. 1997). Therefore, racism might have an indirect
effect on the economy and on institutional quality via its impact on social capital.
3.2.3 Impact of racism on institutions
Apart from the impact via social capital, racism can affect the quality of institutions in other
ways. One way is that when animosity among races is strong, it can give rise to racism-inspired
political preferences. The willingness of tax payers to contribute to public goods and support
for redistribution policies will be much lower if a disliked racial group is the main beneficiary
19
of these policies. Racism can even lead racist voters to support policies that harm disfavored
racial groups. Politicians can pander to their racist voters and pursue and even encourage racial
politics, ruling only for the benefit of a racial group and excluding the others, or even enacting
policies that are deliberately harmful towards other groups if they are in confrontational situation.
If politicians do not have the objective of maximizing the welfare of their entire constituency but
only of their racial or ethnic group, we will see sub-optimal policies being implemented (Easterly,
Levine 1997). Racism might also affect institutions indirectly by affecting the level of defacto
power of the different groups in society. This changes in the balance of power would affect the
institutions, the sign of this change cannot be a priori predicted.
Further, if certain groups are excluded from the public service due to the existence of a racial
bias, we can expect a lower quality of civil servants and politicians since they are selected from
a reduced pool of candidates and potential candidates with higher talent but wrong race will be
rejected.
Another potential channel of racism indirectly influencing institutions comes from the ob-
servation that racism is associated with certain political preferences. As we show in our paper
and according to the suggestions from the field of right-wing authoritarian theory (David and
Wilson (2011) and Bonilla-Silva (2000) among others), inpiduals who display racist views or
believe other races are a group threat, are more likely to have authoritarian preferences and
support authoritarian policies. Thus, racist voters are also more likely to support candidates
that exhibit racist and authoritarian views at the same time. As a consequence, if racist views
become sufficiently influential, we might also expect a change in institutions in a more authori-
tarian direction. This change might be detrimental to the quality of institutions understood in
the sense of institutions conducive to economic development, protection of economic rights or
protection of the citizens against discretion from state officers.
A pided electorate is also a weak electorate. Ferejohn(1986) shows that electoral control can
fail if the electorate does not have a certain level of social cohesion (or sociotropic voting as the
author refers to it). If voters vote motivated uniquely by their inpidual interest without taking
the entire community into consideration, then a rational politician will offer different levels of
benefits to each voter in exchange for his political support. As a result, the politician can then
play one voter against the other offering recurrently lower levels of public goods until in the final
20
equilibrium level of public goods that the citizen receives tends to zero and the incumbent gets
re-elected with certainty. A similar situation could appear if the politician in power can play the
different racial groups against each other and be reelected while offering lower levels of public
goods than the level he would have to offer in the case of a coordinated cooperative electorate.
On the other direction, good institutions might reduce the impact of racism. Adequate
legislation might protect the rights of members of inpiduals victim of racism and hence mitigate
the impact of racism on the economy and the rest of the outcome variables.
3.2.4 Impact of racism on education
There are different possible impacts of racism on education. First, discrimination in the labor
market will lead to a negative premium wage of workers of the disfavored race. Therefore, the
incentive to invest in human capital will be lower.
Pager et al. (2009) investigates the presence of discrimination by race in the low wage markets
through a field experiment in New York City. In the experiment they sent out equivalent resumes
to hundreds of entry-level jobs. The results show African American applicants were half as likely
as equally qualified whites to receive a call back or job offer. Importantly, African American and
Latino applicants with no criminal background face similar call back rates as white applicants
just released from prison. Discrimination additionally extends to the most highly educated
portion of the work force. Gaddis (2015) highlights, using an audit design, which matches
candidate pairs and applicants for 1,008 jobs on a national job-search website. The results show
African American candidates from elite universities only do as well as white candidates from less
selective universities. They show when employers respond to African Americans applicants, it is
only for jobs that come with lower starting salaries and status than white job seekers. Giddies
(2015) argues that a bachelor’s degree, even ones from elite institutions, cannot fully offset the
importance of race in the labor market.
The willingness to invest in racially-identifiable factors, specifically labor, will also be differ-
ent across racial groups, all other inpidual characteristics being equal. The expectation of a
negative premium on wages for qualified workers of certain disfavored races and of a lower pool
of available jobs will reduce the expected returns from investing in human capital for inpiduals
of disfavored races. Some of these effects have already been empirically documented in the sta-
21
tistical discrimination literature. Statistical discrimination may produce a vicious cycle in which
minority groups realize there disadvantage and under-invest in productive factors, resulting in
employer’s expectations of lower productivity of one group to be proven correct. In this case
statistical discrimination creates a self-fulfilling prophecy (Fryer et al. 2006). Another important
factor is, what determinants are producing underlying inequalities in productive characteristics
Fryer (2011) highlights the importance of policies focusing on reducing the underlying factors
producing differences in group level outcomes. For example, they highlight racial inequalities
in social and economic outcomes are substantially reduced when educational attainment is ac-
counted. Dobbie and Fryer (2011) provide support for this argument through a field experiment
and the use of charter schools, for which, the authors conclude high-quality schools significantly
increase academic achievement and, among the poor, almost fully eliminating any gap in out-
comes. As we can see the underlying factors driving the rational for statistical discrimination
play an important role.
Second, if racist-agents have racism-inspired political preferences they might have a lower
willingness to contribute to education if the other groups benefit from it. They might even
support a lower provision of education to other groups for malevolent reasons or, even forbid
the access of members of certain racial groups to higher education. There is direct evidence
for such preferences shown in a number of surveys, with views on laws protecting the right to
discriminate, views on interracial schools, government spending on other races, affirmative action
and school busing (Bobo, 2012)
Knack and Keefer (1997) present another mechanism, the credit market. If inpiduals have
difficulty obtaining credit due to low levels of trust, it will become harder to invest in human
capital accumulation. Thus, trust helps to moderate credit-market imperfections and lessen
credit constraints. Guiso et al. (2000) show trust allows inpiduals to better finance their
investments in education. We believe this line of argumentation can also be extended to racism.
If a racial bias exists in the credit market, as already discussed, inpiduals suffering this bias
would have more difficulties in funding their investment in education.
It is also important to highlight here that these impacts can be long-lasting. Since one of
the main factors predicting education and wage is the education level of the mothers, a histori-
cal exclusion of several generations of mothers from receiving education can have an impact on
22
current education levels, even if access to education is equal today. Historical policies, such as
Jim Crow in the United States, driven by discrimination and racism, were designed to provide
a systematic advantage to one group over the other. These institutional disadvantages, such
as segregated neighborhoods with worse educational systems, were designed to insured an envi-
ronment of poverty and lower educational opportunities for African Americans. The structural
disadvantages created in the past had an inter-generational effect, as African Americans are likely
to reside in the same locations as their parents and grandparents. Many of these neighborhoods
are still plagued by poverty and low quality schools, reducing the ability of these inpiduals
to attain a quality education and the skills required for the current job market. Thus, even if
the level of racism is lower today, policies of the past have consequences for the accumulation of
human capital of certain groups today. So, the main factors driving the justification of statistical
discrimination today, may have been shaped by institutions and racist policies no longer present
(Light et al 2011).
3.2.5 Impact of racism on conflicts
Racism, in its dimension of hate towards other races, might lead racist agents to accept and even
promote violent actions against members of certain other races. The victims of racially-motivated
aggressions might respond with violence and these aggressions could escalate into larger-scale
conflicts. Therefore, racial animosity could also facilitate the outbreak of inter-racial conflicts
and civil wars. These effects have also been analyzed by the ethnic fragmentation literature.
4 Origins of racism
The correlation between racism, institutional quality and education suggests a high potential for
reverse causality and the need for a deeper and more accurate analysis of the causes of racism.
If the spread of racism is related to one of the other variables which impacts the economy,
education or institutions, then our results might be showing mere correlations or statistical
artifacts. Therefore, it is necessary to test what are the origins of racism in order to validate our
results on its consequences. Our approach is to consider all the possible sources of racism and
test them. We also consider all potential confounding factors, variables that might be correlated
23
with racism which also could have an impact on our outcome variables.
4.1 Hypotheses about the origins of racism
We identify the following hypotheses about the origins of racism:
• Hypothesis 1: Racism comes from mixing racial groups and racially driven grievances
• Hypothesis 2: Religions with a vertical hierarchal structure will have lower social capital,
which may be represented by racist beliefs (Fukuyama 2001). This empirically supported
by Guiso et al. 2003, as certain inpiduals from different religions, in some contexts, are
more likely to be intolerant.
• Hypothesis 3: Racism is part of larger set of values, confounding factors
– Hypothesis 4.1 Racism as a part from a profile of bigotry
– Hypothesis 4.2 Racism is caused by or correlated with other cultural values
• Hypothesis 5: Racism comes from a cognitive bias that is mitigated with education. Kup-
pens and Spears (2014) show that inpiduals with a higher education are less likely to
display explicit racism. However, education levels did not eliminate implicit measures of
prejudice. Thus, while educations does have a positive effect on explicit racism, it did
not eliminate all forms of racial of bias. Maykovich (1975) also suggests that inpidual
respondents are less likely to exhibit racial intolerance as their education increases.
• Hypothesis 6: Racism is due to the struggle of living with low income. Bobo and Hutchings
1996 argue inpiduals with less skill and lower income are more likely to fear other races
or immigrants as competition for jobs and economic resources. Thus, it is rational they
display negative racial attitudes. There is also evidence as the economy gets worse anti-
immigrant views and policy preferences increase, even in developed countries. ((Dustmann
and Preston 2007: Meuleman et al. 2009: O’Rourke and Sinnott 2006)). From a Macro
perspective, if a country is less developed, it will have more inpiduals who fear for their
economic security, driving fear of other races perceived as competition.
• Hypothesis 7: Racism is caused by institutions
24
– Hypothesis 7.1: Racism appears when institutions fail to enforce rule of law
– Hypothesis 7.2: Racism was instilled by influential elites for political purposes
4.2 Conflict hypothesis
This hypothesis suggests that racism appears whenever different racial or ethnic groups come
into contact and coexist in the same territory. This coexistence might, according to this view,
lead to conflicts and to lower levels of cooperation.
Group conflict theorists argue, negative attitudes toward other groups stems from the view
that certain groups are perceived as competition for scarce goods. These scarce goods include,
housing, jobs, resources in the welfare state, power and status (Blalock, 1967, Blumer, 1958,
Campbell, 1965, Coser, 1956, Olzak, 1992 and Quillian, 1995). Thus, the presence of or, a
sudden influx of an outside group, animosity is a natural response to the fear of losing vital
resources.
Concerning our identification strategy, the presence of immigrants or of other races might
be an important confounding factor. It is a possibility, racial views are driven by the level of
persity within a country, thus, racism may well be a proxy for fractionalization or migration.
There is broad evidence that these variable have an important impact on our outcome vari-
ables. Alesina et al. (2003), Easterly and Levine (1997) or Hodler (2006) among many others
have empirically shown that types of societal fractionalization are negatively correlated with eco-
nomic development and the functioning of institutions. Further, Putnam (2007) highlights that,
in the short run, immigration and ethnic persity reduces social solidarity and social capital.
Koopmans and Veit (2014a and 2014b), through the use of field experiments, show that ethnic
persity reduces trust and cooperation on the neighborhood level.
Further, co-existence of different racial groups under the same political constituency might
lead to race-based politics, by which government officers maximize the utility of the racial group
rather than that of the entire citizenship. Cederman et al. (2011) empirically show the relation-
ship between ethnically-based politics and economic inequality with ethnic conflict. Cederman
et al. (2011) theorizes that concentrations of ethnic political power and ethnic income inequality
are a large driver of conflict and ethnic grievances. We, therefore, would expect ethno-linguistic
25
fractionalization, migration, ethnic power distribution, concentrations of ethnic power, and inter-
ethnic conflict to be either the source of racism or to be a confounding factor.
In order to take into account any effect coming from the conflict hypotheses, first, we estimate
whether countries with a higher contact with other races also present more racism. Second, we
systematically control for different proxies of contact between races in the regressions on the
impact of racism on our outcome variables.
In particular, we use the following proxies: for contact we use migration ratio and ethno-
linguistic fractionalization. Concerning the potential for racially-motivated conflicts, we use
the following proxies: first, our proxy for concentrations of ethnic power is the percentage of
ethnically relevant groups excluded from the executive branch as a percentage of the population
(Cederman et al. 2009). Finally, we control for new onsets of ethnic conflict to ensure ongoing
conflicts are not driving racism.
4.3 The ignorance hypothesis
As we will show, the negative correlation at the inpidual level between racism and education,
indicates, it is conceivable that racism might be a prejudice or some form of cognitive bias which
severity could be mitigated by education. It could also be considered that racism is a prod-
uct of ”ignorance”, in that education curves the innate tendency toward racism of inpiduals,
and therefore countries or inpiduals with a lower level of education might be more likely to
experience more racism.
In order to control for this, we will add education to our regression variables. If racism is
truly a consequence of lack of education, its effect should vanish when both variables are included
in the regression.
Further, it could also be that racism, or some of its associated symptoms, make inpiduals
less suitable for academic activities and therefore the relationship goes the other way. Therefore,
if we identify a correlation, we will not be able to claim the direction of causality
26
4.4 The misery hypothesis
We believe the most likely direction of causality between racism and income according to the
standard theoretical framework of economics is that racial prejudice and racial discrimination
are costly according to Becker’s (1958) definition. This bias could lead the racist inpidual to
forgo significant gains from transactions he would not take part in because prejudice.
However, it is also conceivable that inpiduals from a lower income level are more prone to
become racist for different reasons. Maybe in their working life they are less exposed to expe-
riences challenging their beliefs, maybe the struggle of long working hours, economic difficulties
and suffering from their lower social status create a psychological state more prone to becoming
racist.
Further, low qualified workers are likely to view immigrants from other racial groups as com-
petition, thus, it is in their self-interest to have and perpetuate negative views and to discriminate
against their competition(Castles 1984).
4.5 The cultural profile hypothesis
Cultural values could be clustered, forming certain profiles or ethos that define the specific
characteristics of different cultures. Certain values could appear often together for reasons not
completely clear. Racism could appear in combination with other cultural values and, these
other values would be the ones responsible, at least in part, for the outcomes we observed. If
this would be the case, the coefficients for racism would be showing the impact of other omitted
cultural variables. We handle this by controlling pairwise for each of the cultural values for which
there is evidence in the literature that have an impact on our outcome variables.
4.5.1 Racism as a particular instance of a profile of bigotry
A potential confounding factor could be the possibility that racism is an instance of a larger
psychological pattern on how inpiduals process information about the world. Forming certain
opinions, behaving according to this moral system that excludes members of other races, and
ignoring experiences when these beliefs are proven wrong, might be part of a larger heuristic
pattern of processing information and making decisions. Maybe racist inpiduals are also more
27
likely to focus on other highly visible cues, like gender, religion, political orientation, nationality,
or others, and attribute them to a larger informative content than they might have in reality.
By basing their judgement of the world on this coarse heuristic, they might save a lot of
energy in screening potential transaction partners, but incur an abundance of costly mistakes.
In certain environments this strategy might be evolutionary stable and persist. If this would be
the case, we would expect a correlation of racism with other attitudes that could be defined as
”bigotry”, like religious intolerance, sexism, political fanaticism, chauvinism, etc..
We test for this hypothesis by regressing the probability of being racist based on those atti-
tudes from the World Value Survey that we associate with bigotry.
4.5.2 Racism as part of familiar amoralism
Banfield (1958) described the set of values characteristic of the inhabitants of one village in
Southern Italy and labeled them as familiar amoralism. These values included lack of trust
towards other people, lack of interest in contributing to the common good, envy, suspicion and
others. Banfield considered that ”the extreme poverty and backwardness of (this village was)
to be explained largely (but not entirely) by the inability of the villagers to act together for their
common good or for any end transcending the immediate material interest of the nuclear family”.
In order to test whether racism is part of familiar amoralism, we examine whether racism is
associated with the variable trust in the family. We use three different measures, which capture
lower levels of amoralism: distrust in the family, unimportance of the family, and whether parents
should earn respect.
4.5.3 Other cultural confounding factors
As we have already said in our discussion on social capital, generalized trust is identified as having
a strong impact on economic and institutional outcomes. To account for this, in our regressions,
we consistently control for generalized trust. Additionally, we examine the correlations between
generalized trust and our measure for racism.
Further, Tabellini (2010) identifies and empirically shows other cultural traits favorable to
economic development which could be defined as social capital: trust and respect (appreciating
the virtue of having tolerance and respect for others in children).
28
He also identifies two values that he interprets as confidence in the inpidual, with higher
confidence having a positive economic impact. These measures are defined as: control (feeling
in control of one’s life) and obedience (appreciating obedience in one’s child), the later affecting
confidence negatively.
We estimate whether racism is also part of this set of values by regressing racism on respect,
control and obedience. Each being regressed separately under a different specification. We then
estimate the impact of racism on our outcome variables while controlling for these three values
to insure they are not confounding factors.
4.6 Institutional collapse hypothesis
Racism could be correlated with institutions through two specific channels. First, racism could
be caused by the absence of rule of law and the existence of failed institutions. Whenever the
state fails to protect inpidual rights, inpiduals may turn to racially-based groups in order to
secure clusters of cooperation and insurance, increasing the probability that other racial groups
are viewed as competitors or even threats.
It could also be that crimes committed by a particular group against another group are more
likely to escalate inter-racial tensions and animosity, which would not have occurred in a state
which protected the potential victims. However, this view does not address the question of why
are institutions weak to begin with?
4.7 Extractive institutions hypothesis
4.7.1 Hypothesis as the consequence of extractive institutions
One of the main problems with the use of cultural variables is the origin of the differences in
cultural values itself is often unclear. Who decides about which cultural values a society should
have? Some papers assume inpiduals decide which values their offspring should have, which
is based on their judgment determined by what values they believe will be beneficial. Dohmen
et al. (2011) discusses the inter-generational transmission of risk and trust attitudes. Common
sense, on one hand, would dictate, those who hold power in a society, might have also the power
to shape the values of the societies they rule. It would thus be rational to shape the values and
29
beliefs of this society to their advantage. Acemoglu et al. (2014) already suggest that elites
could take control of a civil societies organizations and use them to shape social capital in their
advantage. It is clear that if a group might control civil organizations, education, art, culture,
and religion, this group might be able to exert a large influence on shaping common values. We
believe that colonial powers, in the colonization era, enjoyed this power to a large extent and
exerted it purposefully.
The next question would then be, which values would the elite ruling a country for extractive
purposes want to promote? Maybe they would like to encourage obedience as an important
value of their forced labor? The benefits seem obvious. Would they like to instill respect for
the hierarchy? Probably yes. What about policy preferences? Would an extractive elite want
to promote values of democracy, engagement in the political process, or free entrepreneurship?
Probably not. Further, we would expect extractive elites to encourage values that make collective
action more difficult, like cheating or not respecting other inpiduals. But why would an elite
ruling an extractive institution wish to instill racism?
Acemoglu et al. (2004) show that kleptocrats can be successful in stealing the resources of
a society if they manage to prevent the coordination of the exploited by, imposing punishments
on those who attempt to organize collective action and redistributing the wealth of the punished
among other citizens in order to gain their support and break collective actions. Posner et al
(2010) provides a taxonomy of different game theoretical settings in which the logic of ”pide et
impera”, that is, piding rivals, might be in the benefit of the pider. Posner et al. (2010) cites
different historical examples like imperial Rome, who systematically pided the Germanic tribes
threatening the border of the Roman Empire by making them fight amongst themselves, instead
of fighting against the Romans. There is evidence that colonial powers also made deliberate use
of this tactic in their interaction with the different political entities in the colonized countries.
They encouraged rivalries and grievances among them with the goal of making cooperation of
local political entities against the colonizing powers more difficult. Is it conceivable that colonial
powers also applied these tactics in the education of the population of these countries? Is it
possible that people of these colonies were raised to learn to hate people from other ethnic
groups in their communities to make coordination among these different communities against
the ruling elites more difficult?
30
Some authors claim that in most cases, given that the colonizers belonged to a different ethnic
group, they promoted this differentiation with two purposes: on the one hand, make the ruled
accept that the ruling ethnic group was superior and due to this innate superiority, they deserved
to rule, facilitating the acceptance of the status quo by the colonized. Second, differentiation
among racial lines should facilitate the internal coordination among elite members. Given that
the elites tended to be a rather small group, defection among its members siding with local
entities might have had fatal consequences. Additionally, even if the ruling race was pided
into two classes, with the higher class in control, if fear and hatred was successfully promoted
among the lower class of the same race towards other races, it would hinder the probability of
coordination of the lower class with other races, which would reduce the probability of the ruling
class being overthrown. In certain instances, this could even lead to the lower classes active
participation in the perpetuation of the ruling classes control driven by fear of other races. By
making the cooperation with other racial groups in the country more difficult, the loyalty of
the inpidual elite members and possibly non elite members of the same racial group would be
almost guaranteed.
In order to test this hypothesis we use different proxies for extractive institutions and es-
timate whether they have an impact on the level of racism today, we additionally, control for
other variable offering an alternative hypothesises about the origin of racism. For this we use
population density in the 15th Century as a proxy for the likelihood of receiving extractive insti-
tutions once colonized. As explained by (Acemoglu et al. 2001; Acemoglu et al. 2002; Sokoloff,
Engerman 2000; Engerman, Sokoloff 2002) colonial powers set different types of colonial institu-
tions depending on the availability of resources and population. In those colonies were resources
and population was scarce, colonial powers created colonies with European settlements and al-
lowed them to enjoy institutions with a high level of inclusiveness. In contrast, those areas with
abundant labor force or other resources that could be exploited, received extractive institutions
aiming at deliberately eroding the ability of the colonized to resist the domination of the col-
onizers. Acemoglu et al. (2002) then makes the argument of the reversal of fortune, claiming,
countries with higher density of population in the 15th Century should have been more likely
to receive extractive institutions and, given the persistence of institutions, should have lower
quality of institutions today and as a consequence, lower level of economic development. The
31
results are likely to be causal given that economic development tends to be stable over countries
and countries more developed in the 15th Century should in principle be more developed today.
We use the same argument as above and claim, given racism is associated with lower income,
lower education and people living in smaller population centers, which we will show in our results
section, countries with a higher population density in the 15th Century should, in theory, have a
lower level of racism today. However, due to the exogenous shift in historical institutions, similar
to Acemoglu’s prediction of economic outcomes, we expect population density in the 15th century
to be associated with more racism today as a consequence of extractive institutions.
4.7.2 Deliberate instillation of racism?
Our next question is whether extractive institutions deliberately changed the cultural values
of the conquered populations or is it a consequence of the bad quality of institutions in these
countries?9 We formulate the hypothesis that if the rulers of extractive institutions deliberately
instilled racism in the population, then they might as well have instilled other values facilitating
the persistence of extractive institutions. We identify different values in the World Value Survey
that we consider to be desirable for the persistence of extractive institutions. These values can
be grouped in three categories: first, cultural values that make collective action more difficult,
beyond the cultural factors already tested, second, political preferences that facilitate the persis-
tence of non-democratic political institutions, and third, preferences for economic policies that
facilitate the persistence of inefficient and extractive economic institutions. We then estimate
the impact of extractive institutions on these extractive values and preferences. We then test
whether racism is correlated with these values and preferences. For these regression we use
9 From the structural standpoint, racism is seen as the connection of prejudice and power in which the superior
race institutionalizes its dominance at all levels of society (Alvarez et al. 1979, Carmichael 1971, Carmichael
and Hamilton 1967, Chesler 1976). Carmichael and Hamilton (1967), are one of the first to introduce the con-
cept of institutional racism. They describe this as racial inequality which results from social institutions such
as the: justice, education, and economic system which put blacks and other people of color at a systematically
disadvantageous position while, providing whites an undeserved benefit. Additionally, Blauner (1969,1972), in-
troduces a colonial perspective in which racism is the, mainly European descended, white majority utilization
of institutions to increase their social status by exploiting, controlling, and subjugation of other racial or ethnic
groups. Bonilla-Silva (1996), argues racism should be analyzed from structural standpoint. Thus, racial notions
and stereotypes are a result of the established social and institutional structures. He states, racism establishes
the rules for perceiving and treating other racial groups in society. Additionally, these racial ideals provide the
rationalization for the maintenance of current political, social, and economic status of the different races. So,
while the placement of groups of people into racial category may have originated as a consequence of powerful
actors in the social system, such as a colonizer, once the system was in place, members of the dominant race
participate in the defense and reproduction of the racial structure (Bonilla-Silva 1996).
32
inpidual data from the WVS and control for inpidual characteristics such as income level,
education level, age, sex, social class, trust and size of the town. We include country and time
fixed effects, clustered standard errors by country and run linear probability regressions.
5 Data
Our measure for racism is based on one question in the World Value Survey: ”On this list are
various groups of people. Could you please mention any that you would not like to have as
neighbors?” The answer is coded 1 if the inpiduals mention people of a different race in his or
her response. Inpiduals are restricted in the number of groups they can select.
For national regressions, we average this variable from inpidual level responses, by country,
with the provided country weights, over the last 6 waves available, which include: 1981-1984,
1990-1993, 1995-1998, 1999-2004, 2005-2009 and 2010-2014. Since many countries only have one
data point across all waves, we choose to utilize averages. This is a common strategy used in the
˜ ˜
literature on generalized trust and thus is appropriate (BjA¸rnskov, MA c on 2013). Additionally,
we standardize our measure of racism for analysis at the country level. The variables for social
capital: respect, generalized trust, obedience and control are constructed in the same fashion,
however, they are not standardized. We also test for the stability of the answers across different
waves of the survey, using pairwise testing and find that they are consistent over time. Given
the already long length of the paper, these results are not shown.
The question of the WVS can be considered a proxy for racism because:First, respondents
could have chosen any characteristic of the un-welcomed neighbor over race, like being a criminal,
etc...If they choose race over clear criminal profiles, it must be because they consider a person of
that race a more dislikeable or more dangerous than a confirmed criminal. Therefore, it judges
the entire moral character of the potential neighbor based on his race, it concludes that the
expected utility it will bring is even lower than the expected utility of a known criminal and
finally, it denies the right of the other person to live where he wants to.10
One important consideration is that our variable measures not exactly racial prejudice but a
willingness to express such an attitude. The literature has identified that sometimes expressed
10 Story of Gauland and Boateng
33
preferences might have more to do with the social identity the inpidual wants to project than
with their true preferences (Hillman 2010). Further, we are fully aware that cultural differences
exist concerning the degree of social acceptance of the expression of racism. However, we believe
that the degree of social acceptance of the expression of racism and the degree to which it might
be (un-)attractive to identify oneself as (non-)racist are obviously correlated with the level of
acceptance of racism in the society. Further, the more open racism can be expressed, the more
likely it is that the different mechanisms through which racism can affect our outcome variable,
which will be discussed later, will be in play. Therefore, the degree of acceptance of the public
expression of racism and the degree of social desirability of expressing racism publicly already
constitutes an integral part of the degree of racism of a country.
One weakness of our measure is the limited scope of the question our variable of racism is
derived from. We are not directly capturing views on racial stereotypes or racially driven policy
preferences, like whether respondent believes that other races are less intelligent or less hard-
working, or whether he approves of other racial groups receiving government benefits. However,
Case et al (1989), shows a strong correlation between allowing blacks into your neighborhood
with views on allowing blacks to dinner, laws on interracial marriage, opinion on selling your
house to blacks, segregated schools and voting for a black president if qualified. Additionally,
there is other literature supporting the connection between preferences on the racial composition
of ones neighborhood with racial stereotypes and policy preferences (Bobo and Kluegal, 1993;
Tuch and Hughes, 1996). Thus, while our measure is not perfect, there is evidence it, at least in
part, captures a broad set of racial attitudes.
Concerning our country level outcome variables, the data on GDP is taken from the World
Bank Development Indicators database. Our main measure of development is the log of GDP
per capita in constant 2005 US dollars averaged over the period 1984-2013. We take logarithms
in order, first, to permit effects to be larger in countries further away from the global production
possibility frontier and, second, to make sure that identification does not depend on the small
number of wealthy countries (Bjornskov and Meon, 2013.) In order to measure education, we
use the dataset constructed by Barro and Lee (2013). Specifically, we use the average years of
education for the population over the age of 25 which is averaged over the years 1985-2010.
Our three measures of institutional quality come from the World Governance Indicators
34
constructed by (Kaufmann et al. 2009). All measures are averaged over all available years 1996-
2012. All variables ranges from approximately -2.5 to 2.5, with a higher score indicating better
institutional quality. Our proxy for legal institutions is the rule of law. This measure captures
the level of confidence agents have in and abide by the rules of society. Specifically, the quality of
contract enforcement, property rights, the police, the courts and finally, the probability of crime
and violence. In order to capture corruption levels present in a country, we utilize the control
of corruption score. It captures the extent to which public power is implemented for the use of
private gain. This includes both petty and grand forms of corruption and capture of the state
by elites and private interests. To capture the functioning of a countries democracy, we use the
measure voice and accountability. Voice and accountability is the perception of the extent in
which a country’s citizens can participate in selecting their government, engage in freedom of
expression, freedom of association, and the level of free media.
We utilize a number of different control variables commonly used in the literature. Some
are extracted from the Ethnic Power Relations data set (Wimmer, Cederman and Min 2009).
These variables include: mountainous terrain, new onset of ethnic conflict, log of population
and excluded population. In general all variables are taken from and averaged over the period
1984-2012. Finally, to capture the immigration ratio we use two variables, population and net
migration. We create our own variable which is the net migration over total population averaged
over the time period 1984-2012 and we call it, migration ratio. In order to capture levels of
ethnic and linguistic fractionalization we apply the data set created by Alesina et al. (2003).
Additionally, we utilize regional dummies.
For our measure of historical institutions we utilize the data set from Acemoglu et al (2002)
and use three different variables including: Log of population density in the 1500s and 1000s and
the percentage of Europeans in the population in the 1900s. All other variables are described in
Table 1.
HERE TABLE 1
35
6 Results
This section presents the results of our regressions testing our hypothesises about the causes and
consequences of racism using multivariate regressions on both the inpidual and country level.
6.1 Consequences of racism
6.1.1 Racism and economic outcomes
HERE TABLE 2
In table 2, column 1, we show that racism is consistently associated with lower GDP per
capita in the negative direction, even when controlling for geographical factors, regional effects,
log of population, trust and ethnic fractionalization. We observe that trust displays a consistently
positive, though rarely significant, relationship with GDP per capita, confirming that racism is
not merely a proxy for generalized trust, but a specific and different channel. Additionally, we
observe that ethnic fractionalization is not significant in our specifications, although, negative in
sign, indicating, racism has a stronger explanatory power than ethnic fragmentation at the macro
level. In columns 2 to 4 we control for other variables related to the contact hypothesis, which
include: migration ratio, excluded population and ethnic conflict. The value of the coefficients
for racism and its level of significance, 5 percent level, remains generally unchanged across all
these specifications, with racism showing its largest reduction in magnitude when controlling for
migration ratio. In columns 1 to 4, an increase of racism by one standard deviation is correlated
with a reduction in GDP per capita by between 3.47 percent to 2.83 percent. In columns 5 and 6,
we include ”rule of law” and ”total schooling” as control variables, resulting in racism no longer
being significant, but still maintaining its negative sign. As it is known from the literature, these
variables are significantly associated with GDP per capita, thus, it is not surprising our measure
for racism loses its significance. At this point, one could tentatively interpret these results as
supporting the view of racism as a consequence of ignorance and or failure of the state to enforce
the rule of law. However, we have not yet tested other alternative hypotheses.
Given one of the main contributions of this paper is to examine the causal effect of extractive
historical colonial institutions on the current levels of racism, we also run an identical regressions
including only former colonies(table 1A), which produces similar results as table 2, except for the
36
fact, when we control for migration ratio, racism loses its significance. We subsequently preform
this procedure for all other relevant outcome variables, the results of which can be found in the
appendix. Overall, the outcomes remain consistent, across the full and the colonial samples.
6.1.2 Racism and education
HERE TABLE 3
Table 3 displays the results of the OLS regression of education, measured as the average
total years of schooling for inpiduals over the age of 25, and racism. Columns 1-4 confirm the
negative relationship between racism and education, even when controlling for different measures
of the contact hypotheses. The coefficients of racism ranges between 1 percent and 10 percent in
significance in the first 4 specifications, indicating a one point increase in the standard deviation
of racism is associated with a reduction in the average years of education by between .814 and
.603 years. When we control for rule of law and log of GDP per capita, racism is no longer
significant but still remains negative in sign, showing that there must be a relationship between
rule of law and racism, as we will see in the following table. It also seems that prosperity and
rule of law mitigate the negative influence of racism on education.
6.1.3 Racism and intuitions
HERE TABLE 4
Table 4 shows, racism is consistently associated with lower rule of law across all specifications,
with racism being significant at the 1 percent level in all columns, except for column 5 (controlling
for total schooling), where it is significant at the 5 percent level. Overall, the coefficients indicate
a 1 point increase in the standard deviation of racism is associated with the reduction of the rule
of law score, with its highest being 0.345 and at its lowest .164. When controlling for education
and current economic development we see, the coefficients are reduced in size by almost half,
providing more evidence of the connection between racism, education, and economic outcomes.
We see that, the migration ratio is positive and significantly correlated with rule of law, while,
excluded population is negative and significant at the 10 percent level. Importantly, we see
the connection between legal institutions and racism is strong and cannot be fully explained by
other factors. When these regressions are reproduced for only the colonial sample, we find nearly
37
identical results with, racism being negative in sign and statistically significant at least at the 5
percent level, providing further robustness for our results for the association between racism and
the rule of law.
HERE TABLE 5
HERE TABLE 6
Using alternative measures for the quality of institutions, we run the same regressions with
control of corruption, voice and accountability and racism in tables 5 and 6. Table 5 shows,
an even stronger result than table 4, with the coefficient for racism being significant across
all specifications at the 1 percent level, which is the case even when we control for education
and GDP per capita. Overall, we see an increase in racism by one standard deviation being
correlated with the reduction in control of the corruption score by between 0.442 and 0.261
points. Table 6 shows similar results, with racism being negatively correlated with democratic
accountability with all specifications being statistically significant ranging from between -.374
and -.180. Consistent with table 4, the coefficients for racism show its greatest reduction in
magnitude when controlling for education and log of GDP per capita. Overall, we see that
racism has a strong and consistently negative relationship with a society’s ability to control
corruption and their voice and accountability score in both the full and colonial samples, a result
that cannot be explained by other factors.
6.1.4 Racism and social capital
Table 7 shows the negative impact of racism on respect for others and generalized trust. In this
case, racism is significant and negative in sign across all specifications, even when controlling for
ethnic fractionalization, excluded population, migration ratio, and new onset of ethnic conflict.
Columns 1-4 indicate a 1 standard deviation increase in racism is associated with a reduction in
the average level of respect by 2.9 percent to 2.6 percent, with racism being significant at the 5
percent level for half the regressions and 10 percent for the rest. For generalized trust, we see
a similar magnitude as respect, with all coefficients being significant at the 5 percent level. In
table 8, racism has a strong and positive relationship with obedience, as it is significant at the 5
percent level across all regressions. On the other hand, control, at no point displays a significant
relationship with racism, Overall, we show racism is connected to lower levels of social capital,
38
shown by the significant correlations with 3 of our 4 measures.
HERE TABLE 7
HERE TABLE 8
To ensure our results in the previous tables are not due to omitting certain cultural variables,
we re-examine the relationship between racism and economic outcomes, total schooling and rule
of law while controlling for respect, obedience and control. We do not control for trust as it has
already been accounted for in previous tables. In table 9, racism continues to have a consistently
negative and significant correlation with log of GDP per capita, rule of law and education, with
all coefficients being significant at the 1 percent or 5 percent level. This table further confirms
that racism is indeed a phenomenon of its own and not a mere product of other cultural variables.
HERE TABLE 9
As a robustness test for tables 7 and 8, table 10 shows the regression of racism on the same
previous four cultural variables using inpidual level data. The overall results are consistent with
the national-level regressions for obedience and respect. However, this time the relationship be-
tween trust and racism disappears and the correlation between racism and control is statistically
significant and negative. From this section we can conclude that racism, on the inpidual level,
does not have a strong link with generalized trust, reinforcing the idea that generalized trust and
racism are two different phenomena. Further, racism is associated with the cultural variables
identified by Tabellini (2010) in the direction which could hinder economic and institutional
outcomes.
HERE TABLE 10
6.1.5 Racism and conflict
HERE TABLE 11
Table 11 shows that racism has no impact on the onset of new ethnic conflicts, even when
controlling for the same variables for which racism was previously significant.
39
6.2 Origins of Racism
6.2.1 Racism and contact
Table 12 presents the assessment of the conflict hypothesis and racism, examining the connection
between racism and ethnic fractionalization, linguistic fractionalization, excluded population and
migration ratio. Racism is never significant and as a result, we can definitively reject the conflict
hypothesis as a driver a racism on the macro level.
HERE TABLE 12
6.2.2 Racism and religion
In table 13, we test the correlation between different religions, measured by the portion of
Protestants, Catholics and Muslims within the population in 1980 and, racism. Columns 1-3 show
a higher portion of Protestants is associated with lower levels of racism, which is statistically
significant at the 5 percent level in two of the three specifications, it is not significant when
controlling for Rule of Law. The connection between the proportions of Catholics in society
displays no relationship, as the coefficients are not consistent in sign nor statistically significant
in any specification. The proportion of Muslims is positively correlated with racism, shown by
columns 7-9, but is only statistically significant in the weakest specification. From table 13,
we see the effect of religion on racism is mixed, with the proportion of Protestants having a
strong negative effect on racism, Muslims a positive, but generally weak effect, and Catholics no
measurable impact. Overall, there is some weak evidence, religions who are less hierarchical, i.e.
Protestants, tend to display less racism.
HERE TABLE 13
6.2.3 Racism,ignorance,misery and amoralism
In table 14, we examine the ignorance, misery and amoralism hypothesis together, at the inpid-
ual level. The results for income and education support our prediction that lower levels of both
education and income levels are strong predictors for higher levels of racism. Neither of these
results are surprising given the previous literature already discussed. However, the correlation
between our measures for amoralism and racism refutes our hypothesis, as racism is correlated
40
with lower trust of the family, a view that family is less importance and a higher expectation
that parents must earn respect, all of which indicate lower amoralism.
HERE TABLE 14
6.2.4 Racism and profile of bigotry
Table 15 tests the profile of bigotry hypothesis. The table clearly shows the validity of this theory,
as inpiduals who display racist views are more likely to: not want immigrants, inpiduals of
another religion, and inpiduals who speak a different language as neighbours, place less value
on insuring women have the same rights as men, and generally think ethnic persity does not
enrich their lives. Almost all measures are statistically significant at the 1 percent level. Overall,
we see a clear pattern indicating racist inpiduals also have other preferences or beliefs that can
be viewed as bigoted.
HERE TABLE 15
6.2.5 Racism and extractive colonial institutions
In table 16, columns 1 and 2, we show that population density in 1500, a proxy for extractive
institutions, is significantly correlated with racism even when controlling for total schooling,
rule of law and log GDP per capita, which are almost never significant in explaining racism.
Rule law is significant in column 2, however, it does not change the result for our variable of
interest. Mountainous terrain is also negatively correlated with racism in some specifications.
This variable cannot have any other conceivable impact on racism apart from the established
theories by (Acemoglu et al. 2001; Sokolof, Engerman 2000), where the terrain influence what
kind of economies could be established and thus, what kind of institutions were ideal. For
example, large scale plantations could be more difficult to implement in such terrain. Also, it
could be mountainous terrain constituted a protection against the conquest of colonizers and
therefore, insulated these countries from the establishment of extractive institutions (Nunn and
Puga 2010). For robustness, we test other potential proxies for extractive institutions such as log
of population density in 1000, and proportion of settlers of European decent in 1900, a proxy for
inclusive institutions. In all cases, our proxies for extractive institutions are significant increase
the levels of racism. Importantly, these results are robust. As a placebo test we run the same
41
regression on the sample of non-colonies. This time, our proxies for extractive institutions are
not significant, confirming the goodness of our results. We interpret these results as extractive
institutions being the main determinant of racism today, since education and rule of law, the
two main competing hypotheses, are rarely significant, we can reject the racism-as-ignorance
hypothesis and show the hypothesis of racism arising as a consequence of failing rule of law not
being the most important factor.
HERE TABLE 16
6.3 Racism and extractive values
6.3.1 Racism and democratic preferences
In order to further confirm that extractive institutions actually changed the cultural values of
the conquered populations, we run regressions at the inpidual level, estimating the impact of
racism on values that we consider would be ideal for the ruling class to promote through extractive
institutions in order to preserve the persistence of these institutions and facilitate their public
support. We control for inpidual characteristics such as income level, education level, age, sex,
social class, trust, size of the town and life satisfaction. We include country and time fixed effects,
clustered standard errors by country and run linear probability regressions. First, we can now
interpret the results presented in the section on the cultural hypotheses concerning obedience and
respect. Whereas obedience facilitates the submission of the inpidual to extractive institutions,
low respect to others makes collective action more difficult. We run these regressions for both
the full and colonial samples, with the full sample being shown below and the colonial samples
being placed in the appendix.
In table 17, we find that racism is associated with a lower political preference for democ-
racy. Overall we see, those who display racist beliefs are more likely to have a negative view
of democracy in a number of respects. First, they feel that democracy is less important. They
also have a stronger feeling that democracy is not decisive and does a poor job in running the
economy. Additionally, those who have anti racial preferences also have an increased belief that
alternatives to democracy are better, if the government is incompetent the army should take
over, and think it is more important religious figures have a large role in interpreting the law.
42
We show the correlation of racism and a number of preferences on the importance of civil
rights and the ability of inpiduals to shape their own political environment. Those who display
racist inclinations also believe the protection of civil rights is less important in a democracy.
They tend to consider that it is not essential to be able to choose their leaders in free and fair
elections and to change laws through referenda. Moreover, those who do not want neighbours
of another race believe it is more important to obey their leaders in a democracy. These results
are ideal for the maintenance of extractive institutions as inpiduals do not believe democracy
or legal protections are as important, favour obedience to authority and feel it is less essential
to directly shape their political environment. The results also provides evidence that inpiduals
who explicitly display racism are more likely to support authoritarian policy and place a lower
value on civil rights and democracy, a view consistent with right wing authoritarian theories.
The results presented in table 17 are in general consistent when run for the colonial sample, with
the exception of importance of obeying rulers in democracy and religious figure interpreting the
law no longer being significant.
HERE TABLE 17
6.3.2 Racism,policy preferences and civic norms
Table 18 highlights that people who express racism are more in favour of government ownership,
against competition, against unemployment aid, against taxing the rich (although not significant)
and in favour of tradition versus economic growth. All these values reveal a preference for
state control and acceptance of inefficient non-inclusive economic institutions. The results also,
partially support the misery hypothesis, as racism may be driven by economic fear, thus, if
inpiduals will rationally favour restrictive economic policy that could, in theory, shield them
from competition from other races. While the results are strong in the full sample, for the colonial
sample, these results are less consistent, in terms of, statistically significant relationships.
Additionally, the table shows, those who express racism are also more likely to find different
types of dishonest behaviour justifiable, a relationship which holds for both the full and colonial
samples. Dishonest behaviour clearly makes collective action more difficult. The tested be-
haviours include: claiming untitled government benefits, justifying violence, justifying avoiding
paying public transportation fares, cheating on taxes, and justifying officials receiving a bribe.
43
These measures, also referred to as civic norms, a common measure for social capital, highlight
the strong relationship between racism and lower civic norms, providing more evidence of the
detrimental effects of racism on social capital.
HERE TABLE 18
Table 19 shows the correlations between extractive institutions and our measures capturing
profiles of bigotry at the macro level. These variables were created in the identical fashion
as racism, meaning averaged by country. Negative views towards immigrants, inpiduals of
another religion and women having equal rights were all statistically significant in their relation
to extractive institutions and, their signs would indicate a broad level of increased bigotry due
to such institutions. Views on inpiduals who speak another language and opinion on ethnic
persity are not significantly related. However, given the sample size of only 18 for ethnic
persity, it is difficult to draw any conclusion from the results. Overall, table 19, provides
further evidence that extractive historical institutions can, for certain kinds of cultural values,
shape inpidual preferences and attitudes in the present. We refer to certain kinds of cultural
values because, almost all other extractive values have no associations with extractive colonial
institutions at the macro level and thus, these results are not shown. Even so, racism and these
other extractive values have a strong connection, shown by the previous tables.
HERE TABLE 19
7 Conclusion
We have used a novel and promising way for measuring racism at the inpidual and national
level for 82 countries. We have identified racism as an independent phenomenon which could have
consequences for a number of factors relevant to development economists at both the macro and
micro level. We have identified a number of channels through which racism may have negative
economic consequences. At the national level, we find that racism is associated with lower
GDP per capita, lower average years of schooling, lower institutional quality and lower levels
of social capital. Hence, we conclude that either racism is harmful or it arises under negative
circumstances. Either conclusion is a valuable contribution to the overall literature. We find no
significant correlation between racism and conflict. The impact of racism on institutional quality
44
remains significant across all specifications, although its coefficient is reduced when we control for
GDP per capita and education levels. The impact of racism on the rest of our outcome variables
disappears when we control for GDP per capita, institutional quality or education but the sign
of the coefficients remain negative. This implies that there is an important relationship between
racism, institutional quality and education. One possible interpretation is that the negative
effects of racism are mitigated whenever institutions are strong enough to protect the rights of
minorities and education is high enough to curb the tendency towards racist behavior.
We then test the potential origins of racism and find contact with other ethnic groups is not
correlated with current levels of racism. We find that extractive institutions cause racism and
bigotry in former colonies. We also find that racism is correlated with a large number of cultural
values and preferences which will likely contribute to the persistence of extractive institutions.
Additionally, such correlation between racism and these values support the conclusions identified
for the consequences of racism at the macro level. For example, the strong correlation between
racism and institutional quality at the macro level is supported by a strong connection between
racism and lower preferences for democratic institutions at the inpidual level.
While we cannot find evidence that extractive institutions effect cultural values beyond racism
and bigotry, we still consider a meaningful interpretation of the results is that certain values have
been deliberately instilled in order to facilitate the persistence of poor institutions.
We conclude that institutions are still the ultimate cause of long-term economic development,
but this does not mean that cultural values should be ignored. Institutions could instill cultural
values and preferences on society against the interests of the population, in order to further the
interest of those cultural influencers. In our case, influencers seem to be attempting to facilitate
the persistence of these inefficient institutions by instilling one cultural value, racism, that is
harmful for the economy, for the quality of institutions and for social cohesion. We believe this
interaction between institutions and cultural values constitutes an important contribution to the
literature.
45
Table 1 : Additional Variable Descriptions
Racism The question in the survey is : "On this list are various groups of people. Could you
please mention any that you would not like to have as neighbors?" Answer is coded
(1) if people of a different race is mentioned and 0 if not . For the country level, the
variable is averaged by country, over all the avaliable waves and then standardized.
Source: World Value Surveys.
Inpidual level
Civic Norms The question in the survey is : “Please tell me for each of the following actions
whether you think it can always be justified, never be justified, or something in
between" The answer is coded with (1) being never justifiable and (10) being always
justified”. The specific topics are in reference to the following actions: claiming
government benefits to which you are not entitled, cheating on taxes if you have a
chance, avoiding a fare on public transport, someone accepting a bribe in the course
of their duties and violence against others. Source: World Value Surveys.
Respect The question in the survey is: Here is a list of qualities that children can be
encouraged to learn at home. Which, if any, do you consider to be especially
important? If tolerance and respect for other people is mentioned, it is coded as (1,) if
not mentioned (0). Source: World Value Surveys.
Obedience The question in the survey is: Here is a list of qualities that children can be
encouraged to learn at home. Which, if any, do you consider to be especially
important? If obedience is mentioned, it is coded as (1), if not mentioned (0). Source:
World Value Surveys.
Control The question in the survey is: Some people feel they have completely free choice over
their lives, while other people feel that what they do has no real effect on what
happens to them. Please use this scale where (1) means "no choice at all" and (10)
means "a great deal of choice" to indicate how much freedom of choice and control
you feel you have over the way your life turns out. Source: World Value Surveys.
Trust Generally speaking, would you say that most people can be trusted or that you need to
be very careful in dealing with people? Answer is coded (1) if people can be trusted
and (0) if you cannot be to careful. Source: World Value Surveys.
Education The question in the survey is: What is the highest educational level that you have
attained? (1) Inadequately completed elementary education, (2) Completed
(compulsory) elementary education, (3) Incomplete secondary school:
technical/vocational type, (4) Complete secondary school: technical/vocational type,
(5) Incomplete secondary: university-preparatory, (6) Complete secondary: university-
preparatory, (7) Some university without degree/higher education, (8) University with
degree/higher education. This variable was changed to one with pseudo years of
education, according to each level: To (1) we assigned 3 years of schooling; to (2), 6;
to (3), 8.5; to (4), 11; to (5), 12.5; to (6), 14; to (7), 14.5; and to (8), 16. Source:
World Value Surveys.
Age Respondent's age. Source: World Value Surveys.
Gender Gender of the respondent. (0) Female and (1) Male. Source: World Value Surveys.
Table 1 :Continued
Variable Description
Life Satisfaction The question in the survey is : All things considered, how satisfied are you with your
life as a whole these days? Using this card on which (1) means you are “completely
dissatisfied” and (10) means you are “completely satisfied”where would you put your
satisfaction with your life as a whole? Source : World Value Surveys.
Scale of income A scale of incomes in which the household falls into, before taxes and other
deductions. This variable takes values from 1 to 10, (1 ) being the lowest decile and
(10) the highest. The data is recollected in local currency, scaled and then aggregated
so the deciles represent a country level income ranking. Source: World Value
Surveys.
Size of Town Categorical variable: (1) Under 2,000; (2) 2–5,000; (3) 5–10,000; (4) 10–20,000; (5)
20–50,000; (6) 50–100,000; (7) 100–500,000; and (8) 500,000 and more. Source:
World Value Surveys.
Social Class The question in the survey is: People sometimes describe themselves as belonging to
the working class, the middle class, or the upper or lower class. Would you describe
yourself as belonging to the (1) Upper class, (2) Upper middle class , (3) Lower
middle class, (4) Working class and (5) Lower class. Source: World Value Surveys.
Importance Democracy Created from the question: “How important is it for you to live in a country that is
governed democratically? On this scale where 1 means it is “not at all important” and
10 means “absolutely important” what position would you choose. Source: World
Value Surveys.
Democracy : Decisive Variable based on the response to the statement Democracies are indecisive and have
too much squabbling. The variable is on the scale 1-4 with 1 being strongly agree and
4 strongly disagree. Source: World Value Surveys.
Democracy : Economy Response to the question, in democracy, the economic system runs badly which is on a
scale from 1-4 with 1 strong agree and 4 strongly disagree Source: World Value
Surveys.
Democracy : Alternative Variable is based on the response to the statement Democracy may have problems but
is better than the alternative. It is on a scale from 1-4 with 1 being strongly agree and
4 strongly disagree. Source: World Value Surveys.
Civil Rights, Choosing Leaders, These variables measure attitudes about specific aspects of democracy and is derived
Democracy : Army take over, from the question, “Many things are desirable, but not all of them are essential
Change Laws , Obey Rulers, characteristics of democracy. Please tell me for each of the following things how
and Democracy : Womens essential you think it is as a characteristic of democracy. Use this scale where 1 means
Rigths “not at all an essential characteristic of democracy” and 10 means it definitely is “an
essential characteristic of democracy.” The variables are focus on the opinions on:
civil rights protect people from state oppression, the ability to choose leaders in free
elections, the army takes over when government is incompetent, people can change
the laws in referendums, people obey their rulers , religious authorities interpret the
law and women have the same rights as men. : World Value Surveys.
Immigrants, Other Religions The question in the survey is : "On this list are various groups of people. Could you
and Different Language please mention any that you would not like to have as neighbors?" Answer is coded
(1) if immigrants, people of another religion or people who speak another language
is mentioned and 0 if not . Each variable is coded independently of the others. Source:
World Value Surveys.
Diversity : Enriches Life This variable is based on people’s belief about ethnic persity. Variable is on a scale
from 1-10 with 1 being the belief that ethnic persity erodes a country and 10 being
ethnic persity enriches ones life
Table 1 :Continued
Variable Description
Government Ownership and Two variables based on the question: “I'd like you to tell me your views on various
Competition Harmful issues. How would you place your views on this scale? 1 means you agree completely
with the statement on the left; 10 means you agree completely with the statement on
the right; and if your views fall somewhere in between, you can choose any number in
between”. The first question is in regard to government ownership of businesses. A
score of 1 indicates a preference for an increase in private ownership and 10 a
preference for an increase in government ownership. The second question focuses on
the preference for inpidual responsibility. A score of 1 shows inpiduals agree with
the following statement, “government should take more responsibility to ensure to that
everyone is provided for” and a score of 10 indicates a preference for people taking
responsibility for their own actions. The next variable sheds light on people’s view of
competition. The variable is coded as 1 when inpiduals believe competition is good,
it stimulates people to work hard and develops new ideas. The measure is coded 10
when they believe it is harmful and brings out the worst in people.
Unemployment Benefits and The next variables are based on the question “Many things are desirable, but not all of
Taxing the Rich, them are essential characteristics of democracy. Please tell me for each of the
following things how essential you think it is as a characteristic of democracy. Use
this scale where 1 means “not at all an essential characteristic of democracy” and 10
means it definitely is “an essential characteristic of democracy.” We focus on the
answers for importance of following characteristics: people receive state aid for
unemployment, taxing the rich, State making incomes equal, Woman having equal
rights. World Value Survey
Tradition Vs Economic Growth Question based on someone’s view on what’s more important, Tradition vs. high
economic growth. On a scale from 1-2 with 1 being tradition and 2 being economic
growth. World Value Survey
Unimportance Family Question based on someone’s view on importance of family, On a scale from 1-4 with
1 being very important and 4 not at all important. World Value Survey
Earned Respect from Parents Question based on someone’s view on respect for their parents, On a scale from 1-2
with 1 always respect amd 2 respect must be earned. World Value Survey
Distrust Family Question based on someone’s view on how much they trust their family, On a scale
from 1-4 with 1 being trust completely and 4 do not trust at all. World Value Survey
Country Level
Historical Insitutions
Log of Population Density : Extracted from Acemoglu et al (2002)
1500s and 1000s
Eupean Settlement 1900s Percentage of settlers of European decent 1900s , extracted from Acemoglu et al
(2002)
Other Variables
Regional dummies Latin America, Europe and Central Asia, South Asia, SubSaharan Africa, East Asia
and the Pacific and Western Europe
Religion The respective percentage of Protestants (protmg80), Catholics (catho80), Muslims
(muslim80) living within a country in 1980, Source : La Porta (1997)
Table 2 : Racism and Economic Outcomes
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Log GDP per cap
Racism -0.329** -0.283** -0.347** -0.322** -0.096 -0.029
(0.142) (0.137) (0.134) (0.143) (0.094) (0.112)
Mountainous
-0.121* -0.110* -0.080 -0.108 -0.096 -0.028
Terrain
(0.063) (0.060) (0.062) (0.072) (0.071) (0.059)
logpop -0.115 -0.112 -0.110 -0.112 -0.046 0.008
(0.085) (0.079) (0.080) (0.086) (0.066) (0.064)
Latin America -0.509 -0.054 -0.620 -0.505 -0.537* 0.150
(0.437) (0.374) (0.432) (0.439) (0.283) (0.352)
Eastern Europe
-0.953** -0.250 -1.063*** -0.897** -1.822*** -0.328
and Central Asia
(0.393) (0.381) (0.391) (0.421) (0.351) (0.288)
South Asia -2.009*** -1.603*** -2.187*** -1.859*** -1.322*** -1.912***
(0.468) (0.426) (0.435) (0.538) (0.352) (0.371)
Africa -2.443*** -2.148*** -2.541*** -2.395*** -1.759*** -1.862***
(0.482) (0.450) (0.481) (0.494) (0.399) (0.415)
East Asia -0.474 -0.159 -0.666 -0.468 -0.856** -0.724**
(0.551) (0.494) (0.535) (0.553) (0.354) (0.352)
Western Europe 0.865* 1.040** 0.622 0.876* 0.272 0.204
(0.474) (0.405) (0.472) (0.471) (0.313) (0.332)
Trust 1.496 1.047 1.527* 1.513 0.311 0.623
(0.913) (0.835) (0.866) (0.921) (0.841) (0.611)
Ethnic Frac -0.405 -0.237 -0.200 -0.353 -0.063 -0.006
(0.523) (0.498) (0.500) (0.534) (0.423) (0.431)
Migration Ratio 19.651***
(5.465)
Excluded
Population -1.552***
(0.582)
New Onset Ethnic
Conflict -2.312
(3.400)
Total Schooling +
25 0.350***
(0.042)
Rule of Law 0.904***
(0.136)
Constant 10.889*** 10.533*** 10.961*** 10.800*** 7.467*** 8.418***
(1.471) (1.366) (1.411) (1.508) (1.247) (1.179)
Observations 82 82 82 82 72 82
R-squared 0.686 0.742 0.705 0.688 0.821 0.816
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 3 : Racism and Education
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Total Schooling + 25
Racism -0.770** -0.603* -0.814** -0.754** -0.206 -0.290
(0.344) (0.327) (0.334) (0.332) (0.254) (0.205)
Mountainous Terrain -0.037 0.024 0.060 -0.015 0.157 0.106
(0.149) (0.148) (0.128) (0.138) (0.142) (0.131)
logpop -0.144 -0.191 -0.161 -0.081 0.061 -0.018
(0.212) (0.214) (0.203) (0.226) (0.171) (0.157)
Latin America -0.014 0.814 -0.316 0.001 1.262 0.698
(1.149) (1.027) (1.175) (1.128) (0.816) (0.747)
Eastern Europe and Central Asia 2.943*** 4.231*** 2.703*** 3.127*** 3.911*** 3.982***
(0.886) (0.862) (0.895) (0.862) (0.637) (0.606)
South Asia -1.947 -1.180 -2.269* -0.731 -1.679** 0.682
(1.198) (1.112) (1.245) (1.235) (0.816) (0.806)
SubSaharan Africa -1.547 -1.013 -1.829 -1.235 -0.524 1.472*
(1.214) (1.135) (1.177) (1.018) (1.017) (0.841)
East Asia 1.116 1.606 0.740 1.067 0.679 1.726**
(1.156) (1.075) (1.138) (1.107) (0.818) (0.705)
Western Europe 1.703 1.836* 1.190 1.624 0.498 0.564
(1.158) (1.086) (1.156) (1.105) (0.832) (0.860)
Trust 3.160* 1.643 2.993* 3.514** 1.761 1.299
(1.662) (1.480) (1.586) (1.732) (1.369) (1.305)
Ethnic Frac -0.857 -1.142 -0.489 -0.442 -0.122 -0.381
(1.118) (1.091) (1.064) (1.035) (0.968) (0.810)
Migration Ratio 46.735***
(14.104)
Excluded Population -3.210**
(1.538)
New Onset Ethnic Conflict -20.528**
(10.084)
Rule of Law 1.672***
(0.333)
Log GDP per cap 1.312***
(0.205)
Constant 8.764** 9.428** 9.472** 7.535* 4.446 -5.061
(3.646) (3.634) (3.575) (3.959) (3.119) (3.769)
Observations 72 72 72 72 72 72
R-squared 0.655 0.716 0.680 0.680 0.785 0.814
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 4 : Racism and Rule of Law
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Rule of Law
Racism -0.332*** -0.307*** -0.345*** -0.328*** -0.164** -0.182***
(0.093) (0.090) (0.093) (0.092) (0.064) (0.068)
Mountainous Terrain -0.103** -0.097** -0.075 -0.097** -0.108** -0.048
(0.043) (0.046) (0.047) (0.047) (0.046) (0.040)
logpop -0.136** -0.135** -0.133** -0.135** -0.090* -0.084*
(0.058) (0.062) (0.056) (0.058) (0.046) (0.044)
Latin America -0.728** -0.476 -0.805** -0.727** -0.760*** -0.496*
(0.346) (0.348) (0.337) (0.348) (0.242) (0.269)
Est Europe and Central
Asia -0.690** -0.301 -0.766*** -0.662** -1.240*** -0.256
(0.278) (0.326) (0.264) (0.297) (0.216) (0.225)
South Asia -0.108 0.118 -0.230 -0.032 0.278 0.809*
(0.550) (0.539) (0.509) (0.566) (0.437) (0.467)
SubSaharan Africa -0.643** -0.479 -0.711** -0.619* -0.264 0.472
(0.318) (0.297) (0.308) (0.331) (0.332) (0.317)
East Asia 0.277 0.451 0.144 0.279 0.010 0.493**
(0.356) (0.338) (0.318) (0.357) (0.233) (0.222)
Western Europe 0.731* 0.828** 0.563 0.736* 0.338 0.336
(0.377) (0.353) (0.351) (0.377) (0.234) (0.243)
Trust 0.965 0.716 0.987 0.974 0.126 0.283
(0.647) (0.630) (0.592) (0.646) (0.621) (0.461)
Ethnic Frac -0.442 -0.348 -0.300 -0.416 -0.247 -0.257
(0.370) (0.354) (0.367) (0.371) (0.338) (0.311)
Migration Ratio 10.903**
(4.793)
Excluded Population -1.070*
(0.613)
New Onset Ethnic
Conflict -1.153
(2.576)
Total Schooling + 25 0.225***
(0.034)
Log GDP per cap 0.456***
(0.063)
Constant 2.732*** 2.535** 2.782*** 2.688*** 0.613 -2.236**
(0.985) (1.058) (0.955) (0.980) (0.889) (1.105)
Observations 82 82 82 82 72 82
R-squared 0.624 0.665 0.646 0.625 0.762 0.779
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 5 : Racism and Corruption
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Control of Corruption
Racism -0.431*** -0.409*** -0.442*** -0.428*** -0.261*** -0.292***
(0.096) (0.092) (0.093) (0.094) (0.058) (0.066)
Mountainous Terrain -0.070* -0.064 -0.043 -0.065 -0.069 -0.019
(0.041) (0.044) (0.046) (0.046) (0.051) (0.040)
logpop -0.173*** -0.172*** -0.170*** -0.173*** -0.135*** -0.125***
(0.059) (0.062) (0.057) (0.059) (0.050) (0.045)
Latin America -0.574* -0.355 -0.645* -0.573* -0.602*** -0.360
(0.337) (0.328) (0.333) (0.339) (0.225) (0.252)
Est Europe and Central
Asia -0.848*** -0.511* -0.918*** -0.829*** -1.479*** -0.447**
(0.261) (0.285) (0.251) (0.276) (0.210) (0.196)
South Asia -0.193 0.002 -0.305 -0.141 0.187 0.652
(0.521) (0.515) (0.499) (0.526) (0.377) (0.437)
SubSaharan Africa -0.526* -0.385 -0.589* -0.510 -0.166 0.501*
(0.300) (0.281) (0.306) (0.312) (0.300) (0.258)
East Asia 0.202 0.354 0.081 0.204 -0.082 0.402*
(0.332) (0.312) (0.303) (0.333) (0.215) (0.205)
Western Europe 0.760** 0.844** 0.606* 0.764** 0.344 0.396
(0.372) (0.352) (0.358) (0.373) (0.240) (0.265)
Trust 1.279** 1.063* 1.299** 1.285** 0.517 0.650
(0.633) (0.631) (0.586) (0.636) (0.603) (0.504)
Ethnic Frac -0.464 -0.383 -0.334 -0.446 -0.345 -0.294
(0.351) (0.341) (0.357) (0.355) (0.328) (0.294)
Migration Ratio 9.447**
(4.617)
Excluded Population -0.984*
(0.582)
New Onset Ethnic
Conflict -0.799
(2.355)
Total Schooling + 25 0.221***
(0.032)
Log GDP per cap 0.421***
(0.052)
Constant 3.236*** 3.065*** 3.281*** 3.205*** 1.282 -1.344
(1.007) (1.055) (0.966) (1.015) (0.925) (1.016)
Observations 82 82 82 82 72 82
R-squared 0.705 0.733 0.721 0.705 0.817 0.826
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 6 : Racism and Voice and Accoutability
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Voice and Accountability
Racism -0.359*** -0.335*** -0.374*** -0.356*** -0.195** -0.180*
(0.121) (0.115) (0.121) (0.119) (0.088) (0.105)
Mountainous Terrain -0.131** -0.125** -0.096* -0.127** -0.146*** -0.065
(0.053) (0.053) (0.051) (0.057) (0.050) (0.045)
logpop -0.112 -0.111 -0.108 -0.112 -0.031 -0.050
(0.074) (0.073) (0.067) (0.075) (0.057) (0.051)
Latin America 0.253 0.488 0.160 0.254 0.269 0.530**
(0.379) (0.369) (0.381) (0.380) (0.245) (0.257)
Est Europe and
Central Asia 0.054 0.417 -0.038 0.071 -0.385 0.572**
(0.330) (0.363) (0.333) (0.349) (0.277) (0.222)
South Asia 0.650 0.860 0.502 0.695 1.047** 1.742***
(0.691) (0.681) (0.678) (0.669) (0.509) (0.578)
SubSaharan Africa -0.190 -0.038 -0.272 -0.176 0.280 1.138***
(0.378) (0.357) (0.369) (0.384) (0.338) (0.285)
East Asia 0.730* 0.893** 0.570 0.732* 0.458* 0.987***
(0.377) (0.354) (0.372) (0.379) (0.235) (0.239)
Western Europe 1.322*** 1.413*** 1.120*** 1.325*** 0.930*** 0.852***
(0.387) (0.364) (0.390) (0.386) (0.224) (0.277)
Trust -0.006 -0.238 0.020 -0.001 -0.832 -0.819
(0.699) (0.665) (0.650) (0.706) (0.634) (0.596)
Ethnic Frac -0.147 -0.060 0.024 -0.131 0.112 0.074
(0.394) (0.369) (0.388) (0.405) (0.346) (0.356)
Migration Ratio 10.168***
(3.784)
Excluded Population -1.293**
(0.531)
New Onset Ethnic
Conflict -0.693
(3.069)
Total Schooling + 25 0.241***
(0.042)
Log GDP per cap 0.544***
(0.071)
Constant 1.967 1.783 2.026* 1.940 -0.896 -3.954***
(1.262) (1.232) (1.151) (1.286) (1.064) (1.097)
Observations 82 82 82 82 72 82
R-squared 0.524 0.559 0.556 0.525 0.738 0.741
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 7: Racism and Social Capital
1 2 3 4 5 6 7 8
VARIABLES Respect Respect Respect Respect Trust Trust Trust Trust
Racism -0.026* -0.027** -0.026* -0.026* -0.029** -0.027** -0.029** -0.030**
(0.014) (0.014) (0.014) (0.014) (0.013) (0.013) (0.013) (0.013)
Mountainous -0.018** -0.019** -0.018** -0.018** -0.005 -0.004 -0.005 -0.005
Terrain
(0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.009) (0.009)
logpop -0.011 -0.011 -0.011 -0.011 0.003 0.003 0.003 0.003
(0.008) (0.008) (0.008) (0.008) (0.010) (0.010) (0.010) (0.010)
Latin America -0.008 -0.019 -0.010 -0.008 -0.158*** -0.138*** -0.157*** -0.158***
(0.035) (0.040) (0.037) (0.036) (0.042) (0.044) (0.043) (0.043)
Est Europe and -0.054 -0.068 -0.055 -0.051 -0.050 -0.023 -0.049 -0.053
Central Asia
(0.038) (0.044) (0.039) (0.041) (0.040) (0.044) (0.041) (0.043)
South Asia -0.029 -0.037 -0.031 -0.020 0.016 0.031 0.017 0.008
(0.068) (0.070) (0.070) (0.076) (0.060) (0.061) (0.059) (0.060)
SubSaharan Africa -0.086* -0.093* -0.087* -0.083* -0.148*** -0.134*** -0.147*** -0.150***
(0.046) (0.049) (0.046) (0.047) (0.037) (0.039) (0.037) (0.039)
East Asia -0.023 -0.028 -0.025 -0.022 0.125** 0.134** 0.126* 0.124**
(0.038) (0.040) (0.039) (0.038) (0.061) (0.059) (0.063) (0.062)
Western Europe 0.115*** 0.112*** 0.112** 0.115*** 0.118* 0.122* 0.120* 0.117*
(0.041) (0.042) (0.044) (0.042) (0.068) (0.067) (0.069) (0.068)
Ethnic Frac 0.061 0.057 0.063 0.063 -0.007 -0.001 -0.009 -0.010
(0.052) (0.051) (0.054) (0.053) (0.054) (0.054) (0.054) (0.055)
Migration Ratio -0.397 0.730
(0.630) (0.696)
Excluded -0.020 0.012
Population
(0.078) (0.081)
New Onset Ethnic -0.129 0.115
Conflict
(0.361) (0.349)
Observations 82 82 82 82 82 82 82 82
R-squared 0.414 0.418 0.415 0.415 0.508 0.516 0.508 0.508
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table 8 : Racism and Social Capital 2
1 2 3 4 5 6 7 8
VARIABLES Obedience Obedience Obedience Obedience Control Control Control Control
Racism 0.037** 0.033** 0.038** 0.036** -0.033 0.001 -0.036 -0.026
(0.015) (0.015) (0.015) (0.015) (0.071) (0.067) (0.071) (0.070)
-0.021** -0.022** -0.025** -0.023** 0.000 0.008 0.007 0.013
Mountainous Terrain
(0.010) (0.010) (0.011) (0.010) (0.044) (0.042) (0.046) (0.048)
logpop 0.005 0.005 0.004 0.004 -0.067 -0.066 -0.066 -0.065
(0.011) (0.011) (0.011) (0.011) (0.050) (0.047) (0.051) (0.051)
Latin America 0.094** 0.064 0.103** 0.094** 0.721*** 1.026*** 0.702*** 0.722***
(0.043) (0.045) (0.040) (0.043) (0.251) (0.231) (0.242) (0.250)
Est Europe and Central
Asia -0.128*** -0.169*** -0.119*** -0.134*** -0.581** -0.161 -0.599** -0.526**
(0.045) (0.048) (0.042) (0.048) (0.240) (0.245) (0.234) (0.258)
South Asia -0.091 -0.113 -0.077 -0.108 -0.596* -0.365 -0.625* -0.448
(0.110) (0.114) (0.119) (0.119) (0.338) (0.324) (0.340) (0.368)
SubSaharan Africa 0.130** 0.109* 0.138** 0.124* -0.476* -0.267 -0.493* -0.431
(0.062) (0.063) (0.058) (0.063) (0.275) (0.248) (0.269) (0.283)
East Asia -0.119** -0.134** -0.104* -0.120** 0.482* 0.632*** 0.451* 0.490**
(0.056) (0.055) (0.056) (0.056) (0.245) (0.219) (0.238) (0.242)
Western Europe -0.076 -0.083 -0.057 -0.078 0.415 0.486* 0.376 0.429
(0.054) (0.052) (0.051) (0.054) (0.267) (0.248) (0.265) (0.268)
Ethnic Frac 0.193** 0.184** 0.177** 0.188** 0.824*** 0.923*** 0.858*** 0.875***
(0.075) (0.074) (0.076) (0.073) (0.266) (0.253) (0.271) (0.280)
Migration Ratio -1.112 11.379***
(0.838) (2.905)
Excluded Population 0.126 -0.257
(0.111) (0.476)
New Onset Ethnic
Conflict 0.261 -2.268
(0.600) (2.115)
Observations 82 82 82 82 82 82 82 82
R-squared 0.584 0.597 0.593 0.585 0.494 0.574 0.496 0.500
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 10 : Social Capital Inpidual Level
(1) (2) (3) (4)
VARIABLES Trust Respect Obedience Control
Neighbours: People of a different race 0.002 -0.035*** 0.017* -0.099***
(0.007) (0.006) (0.009) (0.027)
Scale of incomes 0.005*** -0.003** -0.007*** 0.037***
(0.002) (0.001) (0.001) (0.006)
Highest educational level attained 0.007*** 0.007*** -0.017*** 0.027***
(0.002) (0.001) (0.002) (0.004)
Age 0.001*** 0.001*** -0.000 0.001
(0.000) (0.000) (0.000) (0.001)
Sex 0.005 -0.032*** 0.000 0.132***
(0.003) (0.004) (0.004) (0.028)
State of health (subjective) -0.032*** 0.002 0.005* -0.142***
(0.003) (0.003) (0.003) (0.014)
Class2 0.003 0.002 -0.007** 0.070***
(0.003) (0.002) (0.003) (0.013)
Size of town -0.003* 0.002* -0.007*** 0.009
(0.001) (0.001) (0.001) (0.006)
Most people can be trusted -0.003 -0.042*** 0.007
(0.005) (0.006) (0.030)
Satisfaction with your life 0.005*** 0.003*** 0.005*** 0.311***
(0.002) (0.001) (0.001) (0.012)
Observations 161,435 161,429 161,421 157,623
R-squared 0.127 0.056 0.133 0.216
Country FE Yes Yes Yes Yes
Time FE Yes Yes Yes Yes
Method OLS OLS OLS OLS
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 11 : Racism and Conflict
(1) (2) (3) (4) (5)
VARIABLES Dependent Variable : New Onset Ethnic Conflict
Racism 0.003 0.003 0.003 -0.002 0.002
(0.003) (0.004) (0.003) (0.003) (0.004)
Mountainous Terrain 0.006** 0.006** 0.005* 0.001 0.005*
(0.003) (0.003) (0.003) (0.002) (0.003)
logpop 0.001 0.001 0.001 0.003 0.001
(0.003) (0.003) (0.003) (0.003) (0.003)
Latin America 0.002 -0.003 0.002 0.001 -0.000
(0.009) (0.010) (0.009) (0.008) (0.010)
Eastern Europe and Central Asia 0.024** 0.018 0.025** 0.019* 0.023**
(0.011) (0.013) (0.011) (0.010) (0.011)
South Asia 0.065** 0.062* 0.066** 0.052 0.065**
(0.030) (0.031) (0.031) (0.033) (0.031)
SubSaharan Africa 0.021 0.018 0.021 0.010 0.019
(0.013) (0.014) (0.013) (0.014) (0.014)
East Asia 0.002 -0.001 0.003 0.002 0.003
(0.008) (0.008) (0.009) (0.008) (0.008)
Western Europe 0.005 0.003 0.006 0.002 0.007
(0.008) (0.008) (0.008) (0.005) (0.008)
Trust 0.007 0.011 0.007 0.028 0.009
(0.022) (0.022) (0.022) (0.022) (0.022)
Ethnic Frac 0.023 0.021 0.022 0.017 0.022
(0.016) (0.016) (0.017) (0.015) (0.016)
Migration Ratio -0.180
(0.177)
Excluded Population 0.007
(0.025)
Total Schooling + 25 -0.004**
(0.002)
Rule of Law -0.002
(0.005)
Constant -0.039 -0.035 -0.039 -0.029 -0.033
(0.046) (0.048) (0.045) (0.057) (0.050)
Observations 82 82 82 72 82
R-squared 0.325 0.336 0.326 0.392 0.327
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 12 : Conflict Hypothesis
1 2 3 4
VARIABLES Dependent Variable : Racism
Ethnic Frac -0.069
(0.625)
Lingustic Frac -0.608
(0.478)
Excluded Population -0.538
(0.638)
Migration Ratio -6.043
(5.717)
Mountainous Terrain 0.019 0.042 0.036 0.016
(0.058) (0.064) (0.059) (0.060)
logpop -0.035 -0.036 -0.035 -0.037
(0.093) (0.099) (0.087) (0.087)
Latin America -1.504*** -1.622*** -1.569*** -1.684***
(0.420) (0.383) (0.404) (0.421)
Est Europe and Central Asia -0.976*** -0.941*** -1.042*** -1.213***
(0.339) (0.315) (0.344) (0.382)
South Asia 0.488 0.634 0.395 0.328
(1.008) (0.989) (1.022) (1.009)
SubSaharan Africa -0.672 -0.381 -0.734** -0.840**
(0.449) (0.458) (0.348) (0.362)
East Asia -0.436 -0.411 -0.530 -0.526
(0.545) (0.562) (0.555) (0.532)
Western Europe -1.650*** -1.672*** -1.754*** -1.666***
(0.344) (0.340) (0.397) (0.347)
Constant 1.389 1.502 1.462 1.523
(1.514) (1.633) (1.456) (1.451)
Observations 82 82 84 84
R-squared 0.344 0.370 0.361 0.368
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 14: Income, Education and Amoralism
(1) (2) (3) (4) (5)
VARIABLES Dependent Variable : Racism
Scale of incomes -0.002* -0.002* -0.001 -0.003**
(0.001) (0.001) (0.001) (0.001)
Highest educational level attained -0.008*** -0.007*** -0.007*** -0.008*** -0.005***
(0.001) (0.001) (0.001) (0.001) (0.001)
Social class (subjective) -0.003 -0.004* -0.004* -0.005 -0.002
(0.002) (0.002) (0.002) (0.003) (0.003)
Age 0.000*** 0.000*** 0.000*** 0.000* 0.000**
(0.000) (0.000) (0.000) (0.000) (0.000)
Sex 0.005** 0.005** 0.004 0.005* 0.004
(0.002) (0.003) (0.003) (0.003) (0.003)
State of health (subjective) 0.004* 0.003 0.003 0.004 0.003
(0.002) (0.002) (0.002) (0.003) (0.003)
Size of town -0.005*** -0.005*** -0.005*** -0.004** -0.006**
(0.001) (0.001) (0.001) (0.002) (0.002)
Satisfaction with your life -0.003*** -0.003*** -0.002** -0.004*** -0.002
(0.001) (0.001) (0.001) (0.001) (0.001)
Unimportant in life: Family 0.028***
(0.006)
How much you distrust: Your family 0.019***
(0.006)
Earned respect from parents 0.015**
(0.006)
Observations 178,651 166,248 165,881 95,243 66,763
R-squared 0.105 0.106 0.107 0.111 0.116
Country FE Yes Yes Yes Yes Yes
Time FE Yes Yes Yes Yes Yes
Method OLS OLS OLS OLS OLS
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 15 : Profile of Bigotry
(1) (2) (3) (4) (5)
Democracy: Diversity:
Other Different Womens Enriches
VARIABLES Immigrants Religions Language Rights Life
Neighbours: People of a different race 0.449*** 0.430*** 0.394*** -0.319*** -0.365**
(0.016) (0.014) (0.018) (0.055) (0.157)
Scale of incomes -0.002** -0.001 -0.001 -0.009 0.053***
(0.001) (0.001) (0.001) (0.010) (0.015)
Highest educational level attained -0.003*** -0.006*** -0.006*** 0.057*** 0.104***
(0.001) (0.001) (0.001) (0.008) (0.018)
Age 0.000 -0.000* -0.000** 0.005*** -0.002
(0.000) (0.000) (0.000) (0.001) (0.002)
Sex 0.006* -0.002 0.002 -0.241*** -0.083*
(0.003) (0.003) (0.003) (0.045) (0.047)
State of health (subjective) 0.006** 0.002 0.005** -0.025 -0.002
(0.002) (0.002) (0.002) (0.020) (0.029)
Social class (subjective) -0.001 -0.004** 0.000 0.012 0.023
(0.002) (0.002) (0.002) (0.018) (0.031)
Size of town -0.001 -0.004*** -0.002** 0.020** 0.017
(0.001) (0.001) (0.001) (0.010) (0.016)
Most people can be trusted -0.012*** -0.010*** -0.008* 0.006 0.483***
(0.004) (0.004) (0.005) (0.041) (0.073)
Satisfaction with your life -0.001 -0.000 -0.001 0.061*** 0.054***
(0.001) (0.001) (0.001) (0.009) (0.012)
Observations 158,406 114,050 93,259 93,228 31,442
R-squared 0.273 0.282 0.239 0.137 0.178
Country FE Yes Yes Yes Yes Yes
Time FE Yes Yes Yes Yes Yes
Method OLS OLS OLS OLS OLS
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 19 : Extractive Colonial Insitutions and Profile of Bigotry
(1) (2) (3) (4) (5)
Democracy: Diversity:
Other Different Womens Enriches
VARIABLES Immigrants Religions Language Rights Life
log population density 1500 (baseline) 0.051*** 0.043** 0.022 -0.152** -0.175
(0.015) (0.017) (0.014) (0.068) (0.134)
Mountainous Terrain -0.036*** -0.030* -0.009 0.022 0.060
(0.012) (0.016) (0.014) (0.066) (0.145)
logpop 0.013 0.012 0.014 -0.076 -0.238
(0.014) (0.018) (0.020) (0.079) (0.246)
Trust 0.385*** 0.268** -0.035 2.759** 10.378*
(0.129) (0.095) (0.108) (1.043) (4.058)
Ethnic Frac 0.008 0.031 0.093 0.148 2.949*
(0.090) (0.089) (0.073) (0.377) (1.426)
Latin America -0.102 -0.123* -0.070 0.588** 1.956**
(0.075) (0.059) (0.056) (0.258) (0.596)
South Asia -0.044 -0.158 -0.009 -0.599 -1.534
(0.067) (0.127) (0.193) (0.497) (0.863)
SubSaharan Africa 0.003 -0.116** -0.032 -0.434 0.263
(0.063) (0.048) (0.060) (0.315) (0.925)
East Asia -0.064 -0.085* 0.070 -0.198 -0.350
(0.043) (0.046) (0.043) (0.293) (1.072)
Total Schooling + 25 0.016 -0.002 0.007 0.120* -0.196
(0.012) (0.011) (0.010) (0.061) (0.136)
Rule of Law -0.102** -0.061 -0.031 0.138 0.012
(0.048) (0.039) (0.044) (0.175) (0.423)
Log GDP per cap 0.022 -0.001 -0.018 -0.476** -0.015
(0.035) (0.028) (0.036) (0.169) (0.470)
Constant -0.284 0.086 0.023 11.735*** 7.909
(0.332) (0.344) (0.366) (2.032) (4.313)
Observations 34 30 28 28 18
R-squared 0.747 0.778 0.582 0.798 0.848
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 1A : Racism and Economic Outcomes Colonial Sample
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Log GDP per cap
Racism -0.375** -0.118 -0.381** -0.364* -0.117 -0.050
(0.175) (0.130) (0.174) (0.178) (0.114) (0.131)
Mountainous Terrain -0.133 -0.132* -0.104 -0.106 -0.034 -0.060
(0.099) (0.069) (0.112) (0.104) (0.072) (0.086)
logpop -0.126 -0.231* -0.127 -0.121 -0.111 -0.053
(0.127) (0.123) (0.129) (0.128) (0.092) (0.093)
Latin America -0.369 0.121 -0.390 -0.312 -0.362 0.282
(0.453) (0.348) (0.454) (0.461) (0.247) (0.381)
South Asia -1.913*** -1.348** -1.921** -1.527* -1.157*** -1.532***
(0.675) (0.552) (0.696) (0.750) (0.345) (0.476)
SubSaharan Africa -2.593*** -2.706*** -2.582*** -2.413*** -1.653*** -2.168***
(0.576) (0.466) (0.575) (0.633) (0.408) (0.498)
East Asia -0.986 -0.741 -1.009 -0.986 -1.166*** -0.652
(0.652) (0.522) (0.654) (0.654) (0.323) (0.446)
Trust 3.683** 0.923 3.494** 3.949** 1.885** 1.070
(1.373) (1.365) (1.544) (1.486) (0.875) (1.172)
Ethnic Frac 1.055 1.268** 1.065 1.140 0.698* 1.404**
(0.715) (0.553) (0.721) (0.727) (0.403) (0.577)
Migration Ratio 57.495***
(11.731)
Excluded Population -0.543
(1.018)
New Onset Ethnic Conflict -5.497
(5.691)
Total Schooling + 25 0.311***
(0.045)
Rule of Law 0.878***
(0.231)
Constant 9.908*** 12.138*** 9.982*** 9.661*** 7.870*** 8.621***
(2.139) (2.069) (2.231) (2.169) (1.478) (1.610)
Observations 37 37 37 37 34 37
R-squared 0.715 0.844 0.717 0.723 0.889 0.809
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 2A : Racism and Education Colonial Sample
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Total Schooling + 25
Racism -0.925** -0.229 -0.953** -0.897** 0.077 -0.121
(0.394) (0.296) (0.400) (0.400) (0.313) (0.248)
Mountainous Terrain -0.267 -0.195 -0.181 -0.193 0.003 -0.033
(0.285) (0.205) (0.306) (0.299) (0.247) (0.204)
logpop -0.048 -0.331 -0.067 -0.038 0.090 0.202
(0.372) (0.362) (0.398) (0.378) (0.277) (0.260)
Latin America -0.049 1.192 -0.138 0.167 1.820 0.702
(1.372) (1.006) (1.372) (1.355) (1.070) (0.764)
South Asia -2.396 -0.973 -2.376 -0.641 -1.056 1.388
(1.883) (1.532) (2.018) (2.002) (1.213) (0.942)
SubSaharan Africa -2.072 -2.613*** -2.089 -1.386 -1.389 2.497**
(1.447) (0.900) (1.444) (1.343) (1.108) (1.119)
East Asia 0.356 0.985 0.304 0.414 1.482 2.456**
(1.788) (1.620) (1.781) (1.775) (1.009) (0.923)
Trust 6.916 -0.150 6.235 7.840* -1.708 -1.108
(4.317) (4.274) (4.749) (4.520) (3.279) (2.714)
Ethnic Frac 1.152 1.513 1.163 1.588 1.973* -0.949
(1.753) (1.221) (1.730) (1.625) (1.110) (1.004)
Migration Ratio 143.688***
(25.208)
Excluded Population -1.654
(3.092)
New Onset Ethnic Conflict -24.367
(15.403)
Rule of Law 2.641***
(0.637)
Log GDP per cap 1.990***
(0.343)
Constant 6.215 12.036* 6.753 5.454 3.915 -13.287**
(6.136) (5.830) (6.795) (6.297) (4.487) (5.761)
Observations 34 34 34 34 34 34
R-squared 0.509 0.739 0.514 0.556 0.741 0.812
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 3A : Racism and Rule of Law Colonial Sample
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Varriable : Rule of Law
Racism -0.370*** -0.228** -0.380*** -0.370*** -0.214*** -0.228**
(0.097) (0.085) (0.101) (0.097) (0.076) (0.083)
Mountainous Terrain -0.083 -0.083 -0.035 -0.083 -0.054 -0.033
(0.070) (0.055) (0.090) (0.076) (0.067) (0.063)
logpop -0.084 -0.142** -0.085 -0.084 -0.044 -0.036
(0.075) (0.066) (0.073) (0.075) (0.056) (0.057)
Latin America -0.741** -0.470 -0.777** -0.741** -0.699** -0.602*
(0.351) (0.352) (0.350) (0.360) (0.262) (0.307)
South Asia -0.434 -0.122 -0.448 -0.431 -0.079 0.289
(0.549) (0.474) (0.569) (0.538) (0.368) (0.421)
SubSaharan Africa -0.484 -0.546** -0.465 -0.483 0.112 0.496
(0.317) (0.253) (0.308) (0.350) (0.295) (0.424)
East Asia -0.381 -0.245 -0.419 -0.381 -0.490*** -0.008
(0.348) (0.293) (0.350) (0.355) (0.168) (0.208)
Trust 2.977*** 1.450* 2.655*** 2.980*** 2.028*** 1.586***
(0.877) (0.799) (0.834) (0.897) (0.477) (0.549)
Ethnic Frac -0.398 -0.280 -0.382 -0.397 -0.517 -0.797
(0.568) (0.418) (0.546) (0.578) (0.448) (0.501)
Migration Ratio 31.818***
(8.615)
Excluded Population -0.923
(0.794)
New Onset Ethnic Conflict -0.046
(3.878)
Total Schooling + 25 0.179***
(0.039)
Log GDP per cap 0.378***
(0.088)
Constant 1.466 2.700** 1.592 1.464 -0.241 -2.278
(1.254) (1.142) (1.264) (1.217) (0.953) (1.411)
Observations 37 37 37 37 34 37
R-squared 0.622 0.743 0.642 0.622 0.814 0.747
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 4A : Racism and Corruption Colonial Sample
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Control of Corruption
Racism -0.443*** -0.291*** -0.449*** -0.440*** -0.277*** -0.280***
(0.108) (0.084) (0.110) (0.109) (0.083) (0.076)
Mountainous Terrain -0.044 -0.043 -0.015 -0.036 0.003 0.014
(0.069) (0.055) (0.087) (0.074) (0.073) (0.057)
logpop -0.138 -0.201*** -0.139 -0.137 -0.108 -0.083
(0.083) (0.070) (0.082) (0.083) (0.079) (0.057)
Latin America -0.571 -0.280 -0.592 -0.555 -0.535** -0.410
(0.358) (0.318) (0.360) (0.367) (0.257) (0.285)
South Asia -0.524 -0.189 -0.532 -0.414 -0.135 0.310
(0.551) (0.449) (0.567) (0.561) (0.354) (0.352)
SubSaharan Africa -0.369 -0.436* -0.358 -0.318 0.197 0.761**
(0.334) (0.242) (0.333) (0.358) (0.274) (0.358)
East Asia -0.395 -0.249 -0.418 -0.395 -0.504** 0.035
(0.368) (0.286) (0.371) (0.370) (0.205) (0.206)
Trust 3.396*** 1.759** 3.204*** 3.472*** 2.375*** 1.792***
(0.953) (0.825) (0.935) (0.996) (0.639) (0.607)
Ethnic Frac -0.513 -0.387 -0.503 -0.489 -0.683 -0.973**
(0.576) (0.429) (0.573) (0.583) (0.436) (0.449)
Migration Ratio 34.097***
(7.724)
Excluded Population -0.551
(0.781)
New Onset Ethnic Conflict -1.566
(3.739)
Total Schooling + 25 0.184***
(0.036)
Log GDP per cap 0.436***
(0.074)
Constant 2.271 3.594*** 2.346* 2.201 0.686 -2.045
(1.347) (1.185) (1.346) (1.315) (1.235) (1.300)
Observations 37 37 37 37 34 37
R-squared 0.685 0.802 0.691 0.687 0.840 0.825
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 5A : Racism and Voice and Accountability Colonial Sample
(1) (2) (3) (4) (5) (6)
VARIABLES Dependent Variable : Voice and Accountability
Racism -0.405*** -0.262** -0.414*** -0.402*** -0.234** -0.218*
(0.133) (0.108) (0.140) (0.135) (0.104) (0.115)
Mountainous Terrain -0.180** -0.179*** -0.137 -0.172** -0.136** -0.113*
(0.073) (0.062) (0.082) (0.077) (0.061) (0.057)
logpop 0.017 -0.042 0.015 0.018 0.059 0.080
(0.087) (0.084) (0.087) (0.087) (0.059) (0.069)
Latin America 0.351 0.625* 0.319 0.368 0.398 0.536*
(0.374) (0.360) (0.390) (0.379) (0.274) (0.304)
South Asia 0.182 0.498 0.170 0.295 0.582 1.140**
(0.743) (0.654) (0.773) (0.755) (0.511) (0.503)
SubSaharan Africa -0.075 -0.138 -0.057 -0.022 0.639** 1.223***
(0.373) (0.316) (0.364) (0.397) (0.296) (0.388)
East Asia 0.206 0.343 0.171 0.206 0.068 0.699*
(0.520) (0.477) (0.511) (0.529) (0.325) (0.341)
Trust 2.334 0.791 2.046 2.412 1.355 0.491
(1.397) (1.464) (1.414) (1.454) (1.086) (1.023)
Ethnic Frac 0.153 0.272 0.168 0.178 0.003 -0.375
(0.578) (0.449) (0.573) (0.591) (0.400) (0.472)
Migration Ratio 32.127***
(10.169)
Excluded Population -0.825
(0.710)
New Onset Ethnic Conflict -1.606
(4.826)
Total Schooling + 25 0.199***
(0.049)
Log GDP per cap 0.500***
(0.095)
Constant -0.676 0.570 -0.564 -0.748 -2.573** -5.635***
(1.383) (1.357) (1.428) (1.345) (1.061) (1.474)
Observations 37 37 37 37 34 37
R-squared 0.537 0.659 0.553 0.539 0.783 0.754
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 8A: Profile of Bigotry Colonial Sample
(1) (2) (3) (4) (5)
Democracy: Diversity:
Other Different Womens Enriches
VARIABLES Immigrants Religions Language Rights Life
Neighbours: People of a different race 0.444*** 0.457*** 0.452*** -0.364*** -0.031
(0.028) (0.027) (0.032) (0.083) (0.201)
Scale of incomes -0.002 -0.000 -0.001 0.003 0.069**
(0.001) (0.001) (0.002) (0.014) (0.027)
Highest educational level attained -0.004*** -0.005** -0.007*** 0.059*** 0.098***
(0.001) (0.002) (0.002) (0.015) (0.032)
Age -0.000 -0.000 -0.000*** 0.004** -0.001
(0.000) (0.000) (0.000) (0.002) (0.002)
Sex 0.007 0.000 -0.005 -0.265*** -0.079
(0.004) (0.004) (0.004) (0.079) (0.066)
State of health (subjective) 0.009** 0.004 0.003 -0.024 -0.026
(0.004) (0.003) (0.003) (0.036) (0.053)
Class2 0.004 0.002 0.002 -0.012 -0.091**
(0.003) (0.002) (0.003) (0.026) (0.041)
Size of town -0.002* -0.002 -0.002 0.030* 0.022
(0.001) (0.002) (0.001) (0.015) (0.018)
Most people can be trusted -0.015** -0.004 -0.004 0.058 0.406***
(0.007) (0.006) (0.009) (0.077) (0.096)
Satisfaction with your life 0.000 -0.000 -0.000 0.082*** 0.045***
(0.002) (0.001) (0.001) (0.017) (0.015)
Observations 74,663 52,811 44,629 43,950 14,953
R-squared 0.308 0.330 0.279 0.111 0.181
Country FE Yes Yes Yes Yes Yes
Time FE Yes Yes Yes Yes Yes
Method OLS OLS OLS OLS OLS
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
References
Acemoglu, D., S. Johnson, and J. A. Robinson. 2002. “Reversal of Fortune. Geography and Institutions in the Making of the Modern
World Income Distribution.” The Quarterly Journal of Economics, 117(4): 1231–94.
Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2001. “The Colonial Origins of Comparative Development. An Empirical
Investigation.” American Economic Review, 91(5): 1369–401.
Acemoglu, Daron, Tristan Reed, and James A. Robinson. 2014. “Chiefs: Economic Development and Elite Control of Civil Society in
Sierra Leone.” Journal of Political Economy, 122(2): 319–68.
Acemoglu, Daron, James A. Robinson, and Thierry Verdier. 2004. “Alfred Marshall Lecture. Kleptocracy and Divide-and-Rule: A
Model of Personal Rule.” Journal of the European Economic Association, 2(2-3): 162–92.
Acolin, Arthur, Raphael Bostic, and Gary Painter. 2016. “A field study of rental market discrimination across origins in France.” Journal
of Urban Economics, 95: 49–63.
Ahmed, Ali M. and Mats Hammarstedt. 2008. “Discrimination in the rental housing market: A field experiment on the Internet.” Journal
of Urban Economics, 64(2): 362–72.
Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and Romain Wacziarg. 2003. “Fractionalization. Journal
of Economic Growth.” Journal of Economic Growth, 8(2): 155–94.
Alvarez, Rodolfo and Kenneth G. Lutterman. 1979. Discrimination in organizations: Jossey-Bass Inc Pub.
Arrow, Kenneth J. 1972. “Gifts and Exchanges.” Philosophy & Public Affairs, 1(4): 343–62.
Arrow, Kenneth J. 1973. “The Theory of Discrimination.” Discrimination in labor markets: 3–33.
Arrow, Kenneth J. 1998. “What Has Economics to Say about Racial Discrimination?” The Journal of Economic Perspectives,
12(2): 91–100.
Ayres, Ian and Peter Siegelman. 1995. “Race and Gender Discrimination in Bargaining for a New Car.” The American Economic Review,
85(3): 304–21.
Banfield, Edward C. 1958. The Moral Basis of A Backward Society: New York: Free Press.
Barro, Robert and Jong-Wha Lee. 2013. “A New Data Set of Educational Attainment in the World, 1950-2010.” Journal of Development
Economics, 104: 184–98.
Beatty, Timothy K. and Dag E. Sommervoll. 2012. “Discrimination in rental markets: Evidence from Norway.” Journal of Housing
Economics, 21(2): 121–30.
Becker, Gary S. 1973. The Economics of Discrimination. Chicago & London: University of Chicago Press.
Bertrand, Marianne and Esther Duflo. 2016. Field Experiments on Discrimination. Cambridge, MA: National Bureau of Economic
Research.
Bertrand, Marianne and Sendhil Mullainathan. 2004. “Are Emily and Greg More Employable Than Lakisha and Jamal? A Field
Experiment on Labor Market Discrimination.” American Economic Review, 94(4): 991–1013.
Bjørnskov, Christian. 2011. “Combating Corruption: On the Interplay between Institutional Quality and Social Trust.” The Journal of Law
& Economics, 54(1): 135–59.
Bjørnskov, Christian. 2012. “How Does Social Trust Affect Economic Growth?” Southern Economic Journal, 78(4): 1346–68.
Bjørnskov, Christian and Pierre-Guillaume Méon. 2013. “Is trust the missing root of institutions, education, and development? Public
Choice.” Public Choice, 157(3-4): 641–69.
Bjørnskov, Christian and Pierre-Guillaume Méon. 2015. “The Productivity of Trust.” World Development, 70: 317–31.
Black, Dan A. 1995. “Discrimination in an Equilibrium Search Model.” Journal of Labor Economics, 13(2): 309–34.
Blanchard, Lloyd, Bo Zhao, and John Yinger. 2008. “Do lenders discriminate against minority and woman entrepreneurs?” Journal of
Urban Economics, 63(2): 467–97.
Blanchflower, David G., Phillip B. Levine, and David J. Zimmerman. 2003. “Discrimination in the Small-Business Credit Market.”
Review of Economics and Statistics, 85(4): 930–43.
Blauner, Robert. 1969. “Internal Colonialism and Ghetto Revolt.” Social Problems, 16(4): 393–408.
Blauner, Robert. 1972. Racial oppression in America. New York, London: Harper and Row.
Blumer, Herbert. 1958. “Race relations as a sense of group position.” Pacific Sociological Review, 1: 3–7.
Bobo, Lawrence. “The Real Record on Racial Attitudes.” In Social Trends in American Life : Findings from the General Social Survey
since 1972, 38–83.
Bobo, Lawrence and Vincent L. Hutchings. 1996. “Perceptions of Racial Group Competition: Extending Blumer's Theory of Group
Position to a Multiracial Social Context.” American Sociological Review, 61(6): 951–72.
Bobo, Lawrence and James R. Kluegel. 1993. “Opposition to Race-Targeting: Self-Interest, Stratification Ideology, or Racial Attitudes?”
American Sociological Review, 58(4): 443–64.
Bonilla-Silva, Eduardo. 1997. “Rethinking Racism: Toward a Structural Interpretation.” American Sociological Review, 62(3): 465–80.
Bosch, Mariano, M. A. Carnero, and Lídia Farré. 2010. “Information and discrimination in the rental housing market: Evidence from a
field experiment.” Regional Science and Urban Economics, 40(1): 11–19.
Burns, Justine. 2006. “Racial stereotypes, stigma and trust in post-apartheid South Africa.” Ninth Annual Conference on Econometric
Modelling for Africa, School of Economics, University of Cape Town, 2004, 23(5): 805–21.
Burns, Justine. 2012. “Race, persity and pro-social behavior in a segmented society.” Journal of Economic Behavior & Organization,
81(2): 366–78.
Burns, Justine and Malcolm Keswell. 2015. “Diversity and the provision of public goods: Experimental evidence from South Africa.”
Economic Experiments in Developing Countries, 118: 110–22.
Carmichael, Stokely and Charles V. Hamilton. 1992. Black power. The politics of liberation in America / Kwame Ture & Charles V.
Hamilton ; with new afterwords by the authors. New York: Vintage Books.
Carmichael, Stokely and Stokely Speaks. 1971. “Black Power Back to Pan-Africanism.” New York: RandomHouse, Vintage Books, 92.
Case, Charles E., Andrew M. Greeley, and Stephan Fuchs. 1989. “Social Determinants of Racial Prejudice.” Sociological Perspectives,
32(4): 469–83.
CEDERMAN, LARS-ERIK, NILS B. WEIDMANN, and KRISTIAN S. GLEDITSCH. 2011. “Horizontal Inequalities and
Ethnonationalist Civil War: A Global Comparison.” The American Political Science Review, 105(3): 478–95.
David Card and Orley Ashenfelter, ed. 2011. Handbook of Labor Economics: Elsevier.
Dearmon, Jacob and Kevin Grier. 2009. “Trust and development.” Journal of Economic Behavior & Organization, 71(2): 210–20.
Dearmon, Jacob and Robin Grier. 2011. “Trust and the accumulation of physical and human capital.” European Journal of Political
Economy, 27(3): 507–19.
Dobbie, Will and Roland G. Fryer. 2011. “Are High-Quality Schools Enough to Increase Achievement Among the Poor? Evidence from
the Harlem Children's Zone.” American Economic Journal: Applied Economics, 3(3): 158–87.
Dohmen, Thomas, Armin Falk, David Huffman, Uwe Sunde, Jürgen Schupp, and Gert G. Wagner. 2011. “INDIVIDUAL RISK
ATTITUDES. MEASUREMENT, DETERMINANTS, AND BEHAVIORAL CONSEQUENCES.” Journal of the European Economic
Association, 9(3): 522–50.
Dustmann, Christian and Ian P. Preston. 2007. “Racial and Economic Factors in Attitudes to Immigration.” The B.E. Journal of Economic
Analysis & Policy, 7(1).
Easterly, William and Ross Levine. 1997. “Africa's Growth Tragedy: Policies and Ethnic Divisions.” The Quarterly Journal of Economics,
112(4): 1203–50.
Eatwell, John, Murray Milgate, and Peter Newman, ed. 1991. The World of Economics. London: Palgrave Macmillan UK.
Eifert, Benn, Edward Miguel, and Posner Daniel N. 2010. “Political Competition and Ethnic Identification in Africa.” American Journal
of Political Science, 54(2): 494–510.
Engerman, Stanley and Kenneth Sokoloff. 2002. Factor Endowments, Inequality, and Paths of Development Among New World
Economics. Cambridge, MA: National Bureau of Economic Research.
Fehr, Ernst and Urs Fischbacher. 2004. “Third-party punishment and social norms.” Evolution and Human Behavior, 25(2): 63–87.
Ferejohn, John. 1986. “Incumbent Performance and Electoral Control.” Public Choice, 50(1/3): 5–25.
Fershtman, Chaim and Uri Gneezy. 2001. “Discrimination in a Segmented Society: An Experimental Approach.” The Quarterly Journal
of Economics, 116(1): 351–77.
Fryer, Roland G., Jacob K. Goeree, and Charles A. Holt. 2005. “Experience-Based Discrimination: Classroom Games.” The Journal of
Economic Education, 36(2): 160–70.
Fryer Roland G. Jr. 2011. “Chapter 10 - Racial inequality in the 21st century: the declining significance of discrimination.” In Handbook of
Labor Economics, ed. David Card and Orley Ashenfelter, 855–971: Elsevier.
Fryer. R, L. Anderson, and C. Holt. 2006. “Discrimination: Experimental Evidence from Psychology and Economics.” In Handbook on
Economics of Discrimination, ed. William Rogers: Edward Elgar Publishing.
Fukuyama, Francis. 1995. Trust. The social virtues and the creation of prosperity. New York, London: Free.
Fukuyama, Francis. 2001. “Social capital, civil society and development. Third World Quarterly.” Third World Quarterly, 22(1): 7–20.
Gaddis, S. M. 2015. “Discrimination in the Credential Society. An Audit Study of Race and College Selectivity in the Labor Market.” Social
Forces, 93(4): 1451–79.
Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2000. The Role of Social Capital in Financial Development. Cambridge, MA: National
Bureau of Economic Research.
Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2003. “People's opium? Religion and economic attitudes.” Journal of Monetary
Economics, 50(1): 225–82.
Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2009. “Cultural Biases in Economic Exchange? *.” Quarterly Journal of Economics,
124(3): 1095–131.
Hanson, Andrew and Zackary Hawley. 2011. “Do landlords discriminate in the rental housing market? Evidence from an internet field
experiment in US cities.” Journal of Urban Economics, 70(2–3): 99–114.
Hargreaves Heap, Shaun P. and Daniel J. Zizzo. 2009. “The Value of Groups.” American Economic Review, 99(1): 295–323.
Hillman, Arye L. 2010. “Expressive behavior in economics and politics.” European Journal of Political Economy, 26(4): 403–18.
Hirshleifer, Jack. 1991. “Conflict and Settlement.” In The World of Economics, ed. John Eatwell, Murray Milgate, and Peter Newman,
117–25. London: Palgrave Macmillan UK.
Hodler, Roland. 2006. “The curse of natural resources in fractionalized countries.” European Economic Review, 50(6): 1367–86.
Knack, Stephen. 2001. Trust, assocational life, and economic performance.
Knack, Stephen. 2002. “Social Capital and the Quality of Government: Evidence from the States.” American Journal of Political Science,
46(4): 772–85.
Knack, Stephen. 2003. “Groups, Growth and Trust: Cross-Country Evidence on the Olson and Putnam Hypotheses. Public Choice.” Public
Choice, 117(3-4): 341–55.
Knack, Stephen and Philip Keefer. 1997. “Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.” The Quarterly
Journal of Economics, 112(4): 1251–88.
Koopmans, Ruud and Susanne Veit. 2014. “Cooperation in Ethnically Diverse Neighborhoods: A Lost-Letter Experiment.” Political
Psychology, 35(3): 379–400.
Koopmans, Ruud and Susanne Veit. 2014. “Ethnic persity, trust, and the mediating role of positive and negative interethnic contact: A
priming experiment.” Social Science Research, 47: 91–107.
Kuppens, Toon and Russell Spears. 2014. “You don’t have to be well-educated to be an aversive racist, but it helps.” Social Science
Research, 45: 211–23.
Ladd, Helen F. 1998. “Evidence on Discrimination in Mortgage Lending.” The Journal of Economic Perspectives, 12(2): 41–62.
Lang, Kevin, Michael Manove, and William T. Dickens. 2005. “Racial Discrimination in Labor Markets with Posted Wage Offers.”
American Economic Review, 95(4): 1327–40.
LaPorta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny. 1997. “Trust in Large Organizations.” American
Economic Review Papers and Proceedings, 87: 333–38.
Light, R., V. J. Roscigno, and A. Kalev. 2011. “Racial Discrimination, Interpretation, and Legitimation at Work.” The ANNALS of the
American Academy of Political and Social Science, 634(1): 39–59.
Marsden, Peter V. 2012. Social trends in American life. Findings from the General Social Survey since 1972 / edited by Peter V. Marsden.
Princeton, N.J., Woodstock: Princeton University Press.
Maykovich, Minako K. 1975. “Correlates of racial prejudice.” Journal of Personality and Social Psychology, 32(6): 1014–20.
Meuleman, Bart, Eldad Davidov, and Jaak Billiet. 2009. “Changing attitudes toward immigration in Europe, 2002–2007: A dynamic
group conflict theory approach.” Social Science Research, 38(2): 352–65.
Munnell, Alicia H., Geoffrey M. B. Tootell, Lynn E. Browne, and James McEneaney. 1996. “Mortgage Lending in Boston: Interpreting
HMDA Data.” The American Economic Review, 86(1): 25–53.
Nunn, Nathan and Diego Puga. 2010. “Ruggedness: The Blessing of Bad Geography in Africa.” Review of Economics and Statistics,
94(1): 20–36.
Nunn, Nathan and Leonard Wantchekon. 2011. “The Slave Trade and the Origins of Mistrust in Africa.” American Economic Review,
101(7): 3221–52.
O'Rourke, Kevin H. and Richard Sinnott. 2006. “The determinants of inpidual attitudes towards immigration.” European Journal of
Political Economy, 22(4): 838–61.
Ostby, G. 2008. “Polarization, Horizontal Inequalities and Violent Civil Conflict.” Journal of Peace Research, 45(2): 143–62.
Pager, Devah and Hana Shepherd. 2008. “The Sociology of Discrimination: Racial Discrimination in Employment, Housing, Credit, and
Consumer Markets.” Annual review of sociology, 34: 181–209.
Pager, Devah, Bruce Western, and Bart Bonikowski. 2009. “Discrimination in a Low-Wage Labor Market: A Field Experiment.”
American Sociological Review, 74(5): 777–99.
Pecenka, Clinton J. and Godfrey Kundhlande. 2013. “Theft in South Africa: An Experiment to Examine the Influence of Racial Identity
and Inequality.” The Journal of Development Studies, 49(5): 737–53.
Phelps, Edmund S. 1972. “The Statistical Theory of Racism and Sexism.” The American Economic Review, 62(4): 659–61.
Posner, E. A., K. E. Spier, and A. Vermeule. 2010. “Divide and Conquer.” Journal of Legal Analysis, 2(2): 417–71.
Putnam, Robert D. 2007. “E Pluribus Unum. Diversity and Community in the Twenty-first Century The 2006 Johan Skytte Prize Lecture.”
Scandinavian Political Studies, 30(2): 137–74.
Quillian, Lincoln. 1995. “Prejudice as a Response to Perceived Group Threat: Population Composition and Anti-Immigrant and Racial
Prejudice in Europe.” American Sociological Review, 60(4): 586–611.
Tuch, Steven A. and Michael Hughes. 1996. “Whites' Opposition to Race-Targeted Policies: One Cause or Many?” Social Science
Quarterly, 77(4): 778–88.
William Rogers, ed. 2006. Handbook on Economics of Discrimination: Edward Elgar Publishing.
WILLIAMS, RICHARD, REYNOLD NESIBA, and EILEEN D. MCCONNELL. 2005. “The Changing Face of Inequality in Home
Mortgage Lending.” Social Problems, 52(2): 181–208.