Working Paper
                                  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
On the Origin and Consequences of Racism

                     Matthew Bonick∗

                   Antonio Farf´n-Vallesp´ †
                         a     ın

                     December 29, 2016


    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   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


  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

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

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.

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
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.”

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”
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)

  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

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.

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

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


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

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

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

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.

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.

  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

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

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.

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)

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

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


  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

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-

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

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

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

    – 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


  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

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

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


  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

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).

  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

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?

  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

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).

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

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

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.


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


  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

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


  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


  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

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.



  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,

shown by the significant correlations with 3 of our 4 measures.



  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.


  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



6.1.5  Racism and conflict


  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.

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.


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.


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

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.


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.


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

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.


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.

  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.


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.

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.


  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.


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

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


           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
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
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
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)
           -0.121*    -0.110*   -0.080   -0.108   -0.096   -0.028
           (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***
Population                  -1.552***
New Onset Ethnic
Conflict                          -2.312
Total Schooling +
25                                  0.350***
Rule of Law                                    0.904***
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***
Excluded Population                     -3.210**
New Onset Ethnic Conflict                         -20.528**
Rule of Law                                      1.672***
Log GDP per cap                                         1.312***
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**
Excluded Population                  -1.070*
New Onset Ethnic
Conflict                              -1.153
Total Schooling + 25                             0.225***
Log GDP per cap                                      0.456***
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**
Excluded Population                 -0.984*
New Onset Ethnic
Conflict                              -0.799
Total Schooling + 25                             0.221***
Log GDP per cap                                     0.421***
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***
Excluded Population              -1.293**
New Onset Ethnic
Conflict                           -0.693
Total Schooling + 25                          0.241***
Log GDP per cap                                   0.544***
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
            (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
                          (0.078)                     (0.081)
New Onset Ethnic                          -0.129                     0.115
                                 (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
Excluded Population                        0.007
Total Schooling + 25                             -0.004**
Rule of Law                                         -0.002
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
Lingustic Frac                       -0.608
Excluded Population                           -0.538
Migration Ratio                                     -6.043
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***
How much you distrust: Your family                      0.019***
Earned respect from parents                               0.015**

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***
Excluded Population                  -0.543
New Onset Ethnic Conflict                     -5.497
Total Schooling + 25                            0.311***
Rule of Law                                       0.878***
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***
Excluded Population                   -1.654
New Onset Ethnic Conflict                     -24.367
Rule of Law                                  2.641***
Log GDP per cap                                      1.990***
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***
Excluded Population                   -0.923
New Onset Ethnic Conflict                      -0.046
Total Schooling + 25                              0.179***
Log GDP per cap                                       0.378***
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***
Excluded Population                   -0.551
New Onset Ethnic Conflict                      -1.566
Total Schooling + 25                              0.184***
Log GDP per cap                                      0.436***
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***
Excluded Population                   -0.825
New Onset Ethnic Conflict                       -1.606
Total Schooling + 25                               0.199***
Log GDP per cap                                        0.500***
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

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
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
  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.