cities, income, and demand for...
TRANSCRIPT
![Page 1: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/1.jpg)
Cities, Income, and Demand for Redistribution
Kent FreezeCarleton College
Abstract
It has often been assumed that wealthier individuals should exhibit less concernwith inequality than poorer individuals. In this paper, I investigate urbanization maymediate the degree with which income predicts support for redistribution, finding thatthis assumed relationship may not always have an empirical basis. Urban countriesgenerally see greater income-based divides over issues of economic inequality and re-distribution.
1
![Page 2: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/2.jpg)
At the beginning of the American republic, James Madison in Federalist 10 noted that
“from the protection of different and unequal faculties of acquiring property, the possession
of different degrees and kinds of property immediately results; and from the influence of these
on the sentiments and views of the respective proprietors ensues a division of the society into
different interests and parties” (p. 78). Debate over the appropriate role of the government
in the economy and in rectifying inequalities has continued to be one of the most contentious
issues in politics. Indeed, Madison also referred to inequality as being “the most common
and durable source of factions” (p. 79), and conflict over inequality has been an enduring
dimension of political competition in many democracies.
Madison’s insight regarding the divisive nature of inequality certainly seems applicable in
many countries today. Political competition over inequality has become increasingly intense
over recent years, and the public is beginning to polarize around the issue of inequality as
well (McCarty, Poole and Rosenthal 2006; Bartels 2008). Madison’s assumption that one’s
material wealth influences one’s interests and attitudes has been applied (and sometimes
simply tacitly assumed) by many models in political science. Often this assumption helps
link economic inequality and some outcome, such as the size of government expenditures,
political behavior or democratization. From this approach, individuals are viewed as rational
income–maximizers who view redistribution spending in terms of how it will potentially
affect their own pocketbook. Perhaps the most well–known example of this is the median
voter model of redistribution, which has been formalized in the work of Romer (1975) and
Meltzer and Richard (1981) and is also implied in Madison’s above assertion that inequality
will change the “sentiments and views of the respective proprietors.” This model has been
adapted and applied in many recent comparative studies (Boix 2003; Acemoglu and Robinson
2006; Perotti 1996; Persson and Tabellini 1994). There has been some support found for this
relationship between income and support for redistribution, however, this finding has largely
been limited to studies examining advanced capitalist countries (Cusack, Iversen and Rehm
2
![Page 3: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/3.jpg)
2006; Fong 2001). Given that developing countries are the subject of many macro-level
studies examining the relationship between inequality and economic growth or democracy
and redistribution, this empirical oversight is particularly problematic.
This paper is an examination of the relationship between income and support for redis-
tribution in both developing and developed countries. I argue that the connection between
individual income and support for redistribution is not nearly as simple as is often assumed,
as it relies on the ability for individuals to be able to form objective judgments regarding
their own relative position on the entire income distribution. It is entirely possible for in-
dividuals who are relatively poor to feel satisfied with their income, and have little concern
for income inequality. Conversely, individuals who are actually upper middle class may erro-
neously feel themselves to be relatively poor. These misperceptions of one’s objective income
status can be at least partially overcome when individuals reside in social contexts in which
information concerning the distribution of income is more broadly available. In this paper,
I examine one contextual factor that may shape and influence the ability for individuals to
make these judgments: the level of urbanization in the country.
1 Income and Support for Redistribution
The median voter model of redistribution has not been without criticism. Some scholars
have noted that there may be circumstances when even rational income–maximizers may
not exhibit preferences for redistribution in the narrow manner predicted by Meltzer and
Richard. One approach has argued that individuals maintain their self–interested priorities,
but income maximization is over a longer time horizon, as in models emphasizing occu-
pation/skill specificity/risk exposure (Iversen 2005; Moene and Wallerstein 2003) or social
mobility (Benabou and Ok 2001; Piketty 1995). Alternatively, individuals may consider
alternative sources of welfare support, such as the support obtained through membership
3
![Page 4: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/4.jpg)
in religious organizations, when evaluating the need for a government role in redistribution
(Scheve and Stasavage 2006). Perhaps the most prominent empirical failure of the median
voter model is what Lindert (2004) coins as the “Robin Hood paradox,” where the countries
which have the greatest level of inequality (and therefore the greatest need for a “Robin
Hood” type figure) actually have the lowest amounts of redistributive social spending (Ko-
rpi and Palme 1998). Kenworthy and Pontusson (2005) further describe the failures of the
model, and show that increasing levels of inequality in advanced capitalist countries were not
matched by corresponding increases in public support for redistribution. In the American
context, for example, Bartels (2008) found that individuals often grossly underestimate the
the amount of economic inequality. Such empirical failures suggest more attention should be
given how individuals view inequality and the process by which preferences for redistribution
are formed.
While relative income status may generally correlate with support for redistribution,
there is wide variation between countries in the strength of this correlation. In line with
findings from Dion and Birchfield (2010) and Beramendi and Rehm (2016), I find a wide
variety in the relationship between income status and support for redistribution. Figure 1
shows the simple regression coefficient of relative income on support for redistribution for
each country covered in the 2012 European Social Survey (ESS), 2014 Latin American Public
Opinion Project (LAPOP), the 2009 International Social Survey Project (ISSP), and 2013
PEW data sets.1 While most countries have a statistically significant negative regression
coefficient between relative income and support for redistribution, there are many countries
(especially in Latin America) for which there is no relationship between income and support
for redistribution. This suggests that the simplifying assumption of the median voter model
1Relative income is measured by one’s household income relative to the mean for the country. Forexample, an individual earning an income of $30,000 in a country with a mean income of $60,000 would havea relative income of 0.5. The data used here are described in greater detail in the data section of the articleand in the appendix.
4
![Page 5: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/5.jpg)
SWENZLDEUSVKAUTCZECHEFIN
DNKAUSHUNPOLGBRJPNFRALVAUSANORCYPESTCHLUKRBGRTWNRUSZAFSVNHRVARGBEL
KORPHLTURCHNESPISR
order
-.8 -.6 -.4 -.2 0 .2Relative Income Coefficient
ISSP
CZENLDFRACHEDEU
ISLBEL
GBRNORSVKPOLISREST
SWEDNKBGRUKRSVNLTUFINITAIRL
RUSESPHUNALBCYPPRT
order
-.6 -.4 -.2 0 .2Relative Income Coefficient
ESS
VENURYJAMECUBRACHLCOLNIC
ARGPANSLV
DOMPRYBOLGUYMEXHNDCRIBLZPERGTM
HTI
order
-.5 0 .5 1Relative Income Coefficient
LAPOP
JORCZEDEUJPNUSAISR
SENCANFRARUSZAFVENAUSEGYARGMEXGRCLBNGBRCHLESPITA
PHLNGASLVTUNKORKENUGAIDN
TURGHAPOLBRABOLPAKCHNMYS
order
-.6 -.4 -.2 0 .2Relative Income Coefficient
PEW
Figure 1: Relative Income Coefficient and Support for Redistribution
5
![Page 6: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/6.jpg)
of redistribution is far from universal.
2 Urbanization and the Development of Preferences
for Redistribution
It is often assumed that left-leaning individuals tend to reside in cities while more parochial
and conservative residents inhabit the countryside. Marx and Engels felt cities were necessary
for the mobilization of the poor into working class movements, famously arguing that “[The
bourgeoisie] has created enormous cities, has greatly increased the urban population as
compared with the rural, and has thus rescued a considerable part of the population from
the idiocy of rural life” (Marx and Engels 2002, pg. 224). Such assumptions about the
urban-rural divide in political attitudes also have important empirical applications. For
example, Rodden (2010) argues the geographic concentration of left-leaning voters in cities
may result in lower representation in countries with first past the post electoral districts.
For the vast majority of human history, nearly all of humanity lived a rural, agrarian life.
Beginning with the industrial revolution, this nearly uninterrupted continuum of rural life
was uprooted. With the increasing economies of scale of urban environments and higher levels
of agricultural productivity came a massive movement of individuals from the countryside to
the city. The rapid movement of individuals to cities over the past three centuries represents
a dramatic shift in the sociological organization of humanity that is without parallel. I
consider several mechanisms whereby urbanization creates increased concern for and conflict
over economic inequality.
In order for an individual’s income to predict support for redistribution, individuals must
first be aware of their own relative income status. However, an individual’s subjective eval-
uation of their relative status may not always match reality. Ravallion and Lokshin (1999,
pg. 21) found that in Russia “measured household incomes cannot account well for self-
6
![Page 7: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/7.jpg)
reported assessments of whether one is ‘poor’ or ‘rich’ ”. Similarly, the economic literature
on happiness or life satisfaction finds that individual’s objective income status is often poorly
correlated with subjective well-being (Graham 2009; Clark, Frijters and Shields 2008). There
are multiple reasons why individuals may fail to even be aware of their objective income sta-
tus. One possibility is that they simply do not have contact with individual’s from other
income strata – Knight and Gunatilaka (2010) found that despite massive income differences
between urban and rural China, those living in the countryside were more likely to report
higher levels of subjective well-being. Geographic segregation by income may decrease the
capability for individuals to connect their objective income with their subjective socioeco-
nomic status, particularly when this geographic segregation is coupled with interregional
inequalities.
Cities are likely to make individuals more concerned about inequality for several reasons.
First, urban environments increase the mobilization potential of the poor around the issue
of redistribution. Differences in mobilization capabilities play a key role in urban–bias argu-
ments where urban regions hold an advantage over the countryside in resource allocations by
the central government (Lipton 1977; Bates 1981). The ability to better mobilize the poor in
cities can also aid in the spread of new (and potentially subversive) ideas and attitudes more
effectively than in the countryside. This notion was discussed by one of the earliest Chicago
urban sociologists, Robert Park, in the process that he termed “social contagion.” According
to Park (1915), cities play a key role in fostering and spreading disparate norms and values,
as the density of urban environments allow such ideas to thrive and spread. Since individuals
are subject to a “just world” bias, in which individuals are likely to believe that the existing
system they live in is just and fair (Lerner 1980), challenging existing inequalities requires
individuals to become convinced somehow that the world they live in is not fair. Urban
environments, with higher density of population, greatly facilitate the ease with which new
ideas challenging the status quo can survive and spread.
7
![Page 8: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/8.jpg)
Second, cities tend to exhibit higher levels of conspicuous consumption. This phenomenon
was noted by Thorstein Veblen over a century ago in his seminal work The Theory of the
Leisure Class. Veblen noted that in urban societies where individuals lacked the dense con-
nections found in more rural communities, wealthy individuals increasingly relied on “con-
spicuous consumption” over leisure to indicate their higher status as a method of signalling
to the masses their relatively higher status. In rural settings, however, where “everyone
knows everyone,” conspicuous consumption is less necessary (Veblen 1899, p. 60). Many
anthropologists have also argued that many of the social mores of rural peasant societies
guard against the conspicuous display of wealth as a means of preserving stability, with
many better off individuals in the countryside even forgoing important expenses such as
home repair in order to maintain an appearance of equal social status with their neighbors
(Wolf 1955). These social pressures to preserve a face of equality serve to dampen concern
for inequality in the countryside.
Third, cities have the economies of scale which facilitate the spread of public services,
such as education or health care. These often form the core of redistributive demands, and
tend to be cheaper to provide in a smaller, more densely populated urban region than a
sparsely populated rural area. As cities urbanize, the cost of providing public services that
are at the core of redistributive social spending will fall, which may make redistribution more
acceptable. Individuals may potentially be very dissatisfied with the level of inequality, but
if they reside in a rural area more removed from the activities of a central state, it may make
little sense for them to support redistribution as there is a lack of state capacity to even
implement any form of redistribution. With greater levels of urbanization, redistributive
social spending becomes an increasingly plausible policy option as the government can more
easily implement such spending.
Finally, from a psychological viewpoint, cities tend to break down stable, hierarchical so-
cial relations, which can lead to greater feelings of relative deprivation as individual reference
8
![Page 9: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/9.jpg)
groups broaden to include different classes of people with whom they may previously have
been unlikely to compare themselves with. Many traditional, rural societies are marked by
firm hierarchical systems of social stratification in which an individual’s social status is largely
determined by heredity. As noted by Runciman (1966), individuals will be most likely to
compare themselves with other individuals with whom they share common attributes. Stouf-
fer et al. (1949), who pioneered the concept of relative deprivation in his landmark study
of the American soldier, noted that married men were more likely to compare themselves
to other married men; or African Americans would be more likely to compare themselves
to local African Americans. In a rigid class-based society the poor may be very unlikely to
be concerned with inequality if the wealthy belong to a more separate class of people with
which they share little in common, such as nobility or higher status caste.
Systems of inherited status are challenged by mass urbanization. With larger populations
concentrated in less geographic space, it becomes possible for individuals to have greater
levels of social mobility than were previously possible. Sociologist Wirth (1938, p. 16) noted
this phenomenon, arguing that
“The social interaction among such a variety of personality types in the urban
milieu tends to break down the rigidity of caste lines and to complicate the
class structure, and thus induces a more ramified and differentiated framework
of social stratification than is found in more integrated societies. The heightened
mobility of the individual, which brings him with the range of stimulation by a
great number of diverse individuals and subjects him to fluctuating status in the
differentiated social groups that compose the social structure of the city, tends
toward the acceptance of instability and insecurity in the world at large as a
norm.”
The breakdown of rigid social relations allows individuals to compare themselves to a
9
![Page 10: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/10.jpg)
broader segment of society, which will prompt greater support for redistribution among the
poor who may find themselves frustrated by mobility for other similar individuals which they
perceive themselves as lacking.
The above discussion suggests two possibilities in how urbanization may influence in-
dividual support for redistribution. First, urbanization may lead to greater support for
redistribution overall. Moving from an agrarian to industrial society, with all of its inher-
ent dislocations and spatial concentration of individuals should, in aggregate, lead to higher
support for redistribution.
Second, urbanization may shape the degree to which individual attributes, especially an
individual’s relative income status is salient. Whereas in an agrarian society the poor and
rich find themselves segregated, urbanization makes these income differences much more
apparent and salient. The higher mobility between classes brought about by urbanization
also can polarize perceptions of inequality by income as well. The poor are likely to place the
blame for their lower income on systemic causes outside of their control, while the wealthy
are likely to take personal responsibility (or work ethic) as the most appropriate explanation
for their higher status. Similarly, the greater state capacity (and economies of scale in
administering many social programs made possible by greater urban density) serves to make
the possibility of redistribution simultaneously an enticing possibility to the poor – or a
frightening reality to the rich. As a result, urbanization may serve to polarize support for
redistribution by an individual’s relative income standing.
3 Data Description
3.1 Survey Data
Although most empirical examinations of preferences for redistribution have focused on de-
veloped countries, the recent expansion of cross–national polls, especially the various regional
10
![Page 11: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/11.jpg)
Survey CountriesISSP 2009 Argentina, Australia, Austria, Belgium, Bulgaria, Chile,
China, Croatia, Cyprus, Czech Republic, Denmark, Estonia,Finland, France, Germany, Hungary, Israel, Japan, Korea,South, Latvia, New Zealand, Norway, Philippines, Poland,Russian Federation, Slovakia, Slovenia, South Africa, Spain,Sweden, Switzerland, Taiwan, Turkey, Ukraine, United King-dom, United States
ESS 2012 Albania, Belgium, Bulgaria, Cyprus, Czech Republic, Den-mark, Estonia, Finland, France, Germany, Hungary, Ireland,Israel, Italy, Lithuania, Netherlands, Norway, Poland, Portu-gal, Russian Federation, Slovakia, Slovenia, Spain, Sweden,Switzerland, Ukraine, United Kingdom
LAPOP 2014 Argentina, Belize, Bolivia, Brazil, Chile, Colombia, CostaRica, Dominican Republic, Ecuador, El Salvador, Guatemala,Guyana, Haiti, Honduras, Jamaica, Mexico, Nicaragua,Panama, Paraguay, Peru, Uruguay, Venezuela
Pew 2013 Argentina, Australia, Bolivia, Brazil, Canada, Chile, China,Czech Republic, Egypt, El Salvador, France, Germany,Ghana, Greece, Indonesia, Israel, Italy, Japan, Jordan,Kenya, Lebanon, Malaysia, Mexico, Nigeria, Pakistan, Philip-pines, Poland, Russia, Senegal, South Africa, South Korea,Spain, Tunisia, Turkey, Uganda, United Kingdom, UnitedStates, Venezuela
Table 1: Country coverage by survey
barometer surveys, have begun to provide a global picture on the topic that was previously
never available. Public opinion data comes from several cross-national data sets in which
similar survey items have been rescaled or recoded to a common coding scheme to allow
cross-survey comparability of responses. The surveys which I have collected data on include
the 2013 Pew Global Survey, European Social Survey (ESS), the International Social Sur-
vey Programme (ISSP), Afrobarometer, and the Latin American Public Opinion Project
(LAPOP). Taken together, these surveys cover a total of 78 unique countries across the five
data sets.
11
![Page 12: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/12.jpg)
There are numerous difficulties involved when attempting to compare cross-national pub-
lic opinion data that has been gathered and collected from different sources. Question order-
ing effects, translations and different question wordings can potentially have a large influence
on survey responses. As a result, I analyze data from the individual surveys separately. With
results coming from both global (ISSP and Pew) and regional (ESS, Afroabarometer, and
LAPOP) public opinion sources, a more robust picture results.
The dependent variable used is support for government led redistribution to reduce eco-
nomic gaps between the rich and poor. The exact wording and scaling for this particular
question varies between the surveys and is shown in Table 2.
These questions are quite different from each other. Those from the ISSP, ESS and
LAPOP are generally similar in wording (although LAPOP employs a different scale), asking
whether or not the government should be involved in the reduction of gaps in income between
the wealthy and poor. This wording is not perfect. Because it combines concern over
inequality with the role of the government in redistribution, it is difficult to determine
whether an individual answering in the negative is not concerned about inequality, or simply
feels it is outside the purview of appropriate government action. Second, the question fails to
include any costs or drawbacks that could be associated with redistribution. With inequality
generally being seen as “bad,” this may bias responses in favor of redistribution upward
compared to a question that imposes costs or constraints, such as slowing economic growth
or incurring budget deficits.
The question from the PEW, however, is markedly different. Rather than asking a
prospective question, as in the ISSP, ESS and LAPOP surveys, respondents are asked
whether or not they view inequality as a major problem. A respondent answering that
inequality is a big problem may not necessarily feel that it should be addressed by govern-
12
![Page 13: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/13.jpg)
Survey Question ScaleISSP 2009 It is the responsibility of the gov-
ernment to reduce the differencesin income between people withhigh incomes and those with lowincomes.
4=Strongly agree3=Agree2=Neither agree nor disagree1=Disagree0=Strongly disagree
ESS 2012 The government should take mea-sures to reduce differences in in-come levels.
4=Agree Strongly3=Agree2=Neither agree nor disagree1=Disagree0=Disagree Strongly
Afrobarometer How well or badly would you saythe current government is han-dling the following matters, orhaven’t you heard enough to say?Narrowing gaps between rich andpoor
3=Very Badly2=Fairly Badly1=Fairly Well0=Very Well
LAPOP 2014 The [country] government shouldimplement strong policies to re-duce income inequality betweenthe rich and the poor. To whatextent do you agree or disagreewith this statement?
7 point scale:1=Disagree Strongly7=Agree Strongly
Pew 2013 Do you think the gap between therich and the poor is a very bigproblem, a moderately big prob-lem, a small problem or not aproblem at all in our country?
3=Very big problem2=Moderately big problem1=Small Problem0=Not a problem at all
Table 2: Questions on Redistribution and Inequality by Survey
13
![Page 14: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/14.jpg)
ment action through redistributive spending.
Although these questions all have different drawbacks, and are not directly comparable
to each other, they do all tap some opinions concerning gaps between the rich and poor
and the justice of those gaps. Because of these differences, I analyze each survey separately
rather than attempting to combine them into an overarching analysis.
At the individual-level, my primarily variable of interest is relative household income.
In each survey, income is coded into different categories (with some exceptions in the ISSP
where raw income is reported). Relative income is calculated by first taking the average
of the top and bottom values of the quantile (for example, if the decile is a range between
$15,000 and $20,000 a year, the respondent is scored as having an income of $17,500). Top
categories are given a value of 1.5 times the top category (eg. more than $50,000 would
become $75,000). Relative income is then calculated by dividing the respondents income
score by the weighted average for the country. I used a logarithmic transformation of this
relative income measure to increase the linearity of the measurement in the models and to
reduce the impact of outliers.
In addition, several additional individual-level control variables were used for each of the
five surveys. These include marital status (1 if currently married, 0 otherwise), rural vs.
urban residence (1 if urban, 0 if rural), gender (1 if female, 0 if male), employment status
(unemployed dummy variable), age (in years), and religiosity (measured either in terms of
frequency of attendance of religious services or in importance of religion.2 The description
and coding of these individual-level controls is provided in the appendix.
3.1.1 Country-Level Data
As a measure of democratic political competition, I use the Polity2 scale of democracy and
autocracy, which ranges from -10 (fully autocratic) to +10 (fully democratic).
2The religiosity question is not asked in 6 countries for the PEW survey, so I do not include it as a control.
14
![Page 15: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/15.jpg)
One important issue when using cross-national public opinion data is the anchor point
problem – an issue that has seen relatively little coverage in existing studies of comparative
support for redistribution. This issue occurs when respondents view questions on the need
for government intervention in light of the current status quo, as opposed to some objective
cross-national level. In countries where redistributive social spending is particularly high,
a respondent may decide that further reduction of inequality is unnecessary, and indicate
their disapproval toward it in a survey response. However, they may still feel that some
redistribution is necessary – perhaps if the same exact person were located in a country with
extremely low levels of redistributive social spending they may indicate a higher support for
redistribution. To illustrate this point more concretely, it is useful to imagine a hypothetical
individual who has a preference for a fixed amount of redistribution to occur - perhaps a
total amount of social spending of around 30% of GDP. In a country were social spending is
above that amount, this person would indicate opposition to redistribution, while a resident
of a country with less than 30% would indicate support, even though the preference for
redistribution is identical. When preferences are themselves a function of social policy itself,
it is necessary to correct for this problem of anchor points. The use of anchoring vignettes
such as those proposed by King et al. (2004) and King and Wand (2007) may help to alleviate
some of this problem, but appropriate vignettes are not available in the surveys examined
here.
Ideally, I would be able to use an appropriate cross-national measure of redistribution
as a means of controlling for different anchor points, but such a measure is unavailable for
the majority of countries examined here, especially developing countries3. As a very rough
alternative, I use government revenue as a percent of GDP, available from the IMF World
Economic Outlook database for October 2014, which has broad coverage of countries, as
3The most accurate measure of redistribution currently available is the gini difference measure usedby Bradley et al. (2003), often calculated from data available from the Luxembourg Income Study (LIS).Unfortunately, the LIS has only a limited number of developing countries in its data set.
15
![Page 16: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/16.jpg)
well as being related to overall government spending. While such spending is not necessarily
redistributive, it seems likely that more government spending correlates with greater redis-
tribution of income. Another potential control for anchor point problems is to use the level of
economic development, measured by log GDP per capita at PPP and taken from the World
Bank. Generally speaking, countries with a higher level of economic development should
have a more developed welfare state (Wilensky 1975). These measures likely do a better
job at tracking differences in social spending patterns between developed and developing
countries than within each group.
I also include two additional control variables. To measure economic inequality, I use the
University of Texas Inequality Project (UTIP) estimated household income inequality gini
coefficient for the year 2008, or most recent available year. This is the most recent available
year in the data set, but is also relatively close to the years for the surveys included. The
UTIP data set has fairly broad coverage in terms of countries, and uses a similar data source
for each country, making the inequality measures more comparable than other inequality
data sets which aggregate and include gini coefficients from numerous different sources.
Due to missing data issues in developing countries, I do not use this as a control for the
Afrobarometer and LAPOP surveys. Finally, I also include a measure of urbanization from
the World Bank, indicating the percentage of the overall population which resides in cities.
4 Empirical Analyses
4.1 Urbanization and Aggregate Support for Redistribution
I begin by considering how urbanization influences average support for redistribution in a
country. To begin exploring this question, I created simple scatter plots from each of the four
surveys examining the relationship between urbanization and average, country-level support
for redistribution. From much of the literature discussed, it seems likely that urbanization
16
![Page 17: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/17.jpg)
should generally be associated with an increase in support for redistribution – more urbanized
locations should on average exhibit higher aggregate support for the reduction in inequality
by redistributive social spending.
SE
NZ
DE
SKAT
CZ
CH
FI
DKAU
HU
PL
GB JP
FRLV
US
NO
CY
EE
CL
UA
BG
TW
RU
ZA
SIHR
AR
BEKR
PH
TR
CN ES
IL
1.5
22.
53
3.5
Aver
age
Supp
ort f
or R
edis
tribu
tion
40 60 80 100Urbanization
Correlation: -0.33
ISSP
CZ
NL
FR
CH
DEIS
BE
GB
NO
SKPL
IL
EE
SE
DK
BGUASI LT
FI
IT
IE
RU ES
HUAL
CY
PT
22.
53
3.5
Aver
age
Supp
ort f
or R
edis
tribu
tion
50 60 70 80 90 100Urbanization
Correlation: -0.49
ESS
VE
UY
JM
EC
BR
CL
CO
NI
AR
PA
SV
DO
PY
BO
GYMX
HNCR
BZ PE
GT
HT
4.5
55.
56
Aver
age
Supp
ort f
or R
edis
tribu
tion
20 40 60 80 100Urbanization
Correlation: 0.23
LAPOP
JO
CZDE
JP
US
IL
SN
CA
FRRU
ZA
VE
AU
EG
ARMX
GR LB
GB
CL
ESIT
PH
NG
SV
TN
KR
KEUG
ID
TR
GH
PL
BR
BO
PK
CN
MY
22.
22.
42.
62.
8Av
erag
e Su
ppor
t for
Red
istri
butio
n
20 40 60 80 100Urbanization
Correlation: -0.29
PEW
Figure 2: Scatter of Average Support for Redistribution and Urbanization
The results in Figure 2, however, suggest that this simple relationship between urbaniza-
tion and support for redistribution does not exist, at least not uniformly across the surveys
examined in this paper. In three of the four surveys, urbanization actually appears to have
a negative relationship with support for redistribution. In only one case (LAPOP), does
urbanization have a positive relationship with support for redistribution, with a modest
positive correlation of 0.23.
17
![Page 18: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/18.jpg)
While these scatter plots provide a rough overview of the data, they do not form an appro-
priate statistical test of the hypotheses because they do not adequately control for variance
at both the individual and country levels of analysis, nor do they control simultaneously
for variables at both levels. In order to overcome these shortcomings, I examine the data
using a hierarchical linear model (HLM) which can control for both individual and country
level variables, and account for variance at each level of analysis.4 At the individual-level,
there are no consistent patterns between the different surveys, although females are gener-
ally more supportive of redistribution than males, this effect is only statistically significant
and positive in three of the four surveys (with the exception of LAPOP). Relative income is
negatively associated with support for redistribution in three of the surveys as well, but also
not in the LAPOP surveys. Interestingly, being unemployed is only positively predictive of
support for redistribution in the ESS and PEW surveys. Religiosity, which is often claimed
to have a negative relationship with support for redistribution, is only negatively associated
in the ESS. In the LAPOP survey, religiosity is actually positively associated with support
for redistribution.
There are even fewer significant results at the country-level. Of interest for this paper,
urbanization does not seem to have any country-level influence on support for redistribution,
with each model failing to find any statistical significance. One possible reason for this
finding lies in the anchoring effect – it could be that urbanized countries tend to also have
greater amounts of social spending, leading to a moving baseline in public opinion support
for redistribution. As a result, looking just at public opinion support in aggregate may mask
important ways in which urbanization potentially shapes individual-level preferences.
4All HLM models reported here were calculated using within country survey weights in Stata 14 usingthe command xtmixed.
18
![Page 19: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/19.jpg)
Table 3: Urbanization and Redistribution Support HLM
(1) (2) (3) (4)ISSP ESS LAPOP PEW
Individual LevelMarried -0.0246 -0.00498 0.0225 -0.00569
(0.0200) (0.0144) (0.0357) (0.0103)Urban -0.00678 -0.00839 -0.116 0.0555∗∗
(0.00985) (0.0196) (0.0894) (0.0194)Female 0.0674∗∗ 0.0996∗∗∗ -0.0209 0.0459∗∗
(0.0224) (0.0187) (0.0178) (0.0158)Unemployed 0.0374 0.0770∗∗∗ -0.0266 0.0319∗
(0.0364) (0.0186) (0.0662) (0.0144)Age 0.00156∗ 0.00429∗∗∗ 0.000755 0.00108∗
(0.000628) (0.000637) (0.00106) (0.000484)Yrs. Educ. -0.0133∗∗∗ -0.0110∗∗∗ -0.00363
(0.00340) (0.00244) (0.00379)Religiosity -0.0150 0.0132 0.0638∗
(0.00808) (0.0215) (0.0294)Ln. Rel. Inc. -0.142∗∗∗ -0.178∗∗∗ 0.0453 -0.0675∗∗∗
(0.0321) (0.0209) (0.0348) (0.0195)Country-LevelGov. Rev. 0.0151∗ -0.000316 0.00642 0.00749
(0.00678) (0.00694) (0.0108) (0.00478)Ln. GDPPCPPP -0.393 -0.194 -0.171 -0.111
(0.245) (0.118) (0.201) (0.0782)Gini 0.00206 0.0225 0.0115
(0.0164) (0.0153) (0.00636)Polity2 0.00218 -0.00875 0.0819∗∗ 0.00728
(0.0279) (0.0242) (0.0251) (0.00499)% Urban -0.00312 -0.00969 0.0101 -0.00117
(0.00645) (0.00554) (0.00748) (0.00281)Constant 6.458∗ 4.799∗∗∗ 5.469∗∗∗ 2.727∗∗∗
(2.558) (1.293) (1.273) (0.713)Individual Level SD 0.353∗∗∗ 0.223∗∗∗ 0.318∗∗∗ 0.166∗∗∗
(0.0425) (0.0392) (0.0417) (0.0151)Country Level SD 1.014 0.942∗∗ 1.699∗∗∗ 0.743∗∗∗
(0.0294) (0.0187) (0.0359) (0.0172)N 38307 40766 26165 30138Countries 36 27 21 37
Standard errors in parentheses∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
19
![Page 20: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/20.jpg)
4.2 Urbanization, Relative Income Status and Support for Redis-
tribution
In this section, I examine how urbanization interacts with relative income to shape support
for redistribution. The effect of urbanization on support for redistribution may not be
uniformly felt across the population – rather, it could potentially lead to greater polarization
of opinions on redistribution across income lines. In countries which are more urbanized, the
wealthy will be at a greater risk of expropriation of their income, while the poor stand as more
likely to benefit. The expectation would then be that urbanization does not raise support
for redistribution, but rather makes personal income status more salient when evaluating
the issue of inequality and need for redistribution of income. This predicts an interactive
relationship between urbanization, income and support for redistribution.
The relative income coefficients shown in Figure 1 early in the article were calculated by
running a country-level regression while controlling for individual attributes such as urban
residence, years of education, gender, interest in politics, and employment status. In order
to examine the general relationship between urbanization and these relative income coeffi-
cients, I generated simple scatter plots for each survey examined in this paper, including the
95% confidence interval for each estimate as well. In Figure ??, I examine the relationship
between relative income coefficients on support for redistribution against the aggregate level
of urban development in a country. In each survey, the correlation between urbanization
and the relative income coefficient is negative, in the expected direction, signifying that as
countries become more urban, income has a stronger negative relationship with support for
the reduction of inequality. This relationship between urbanization and the relative income
coefficient is strongest in the LAPOP and surveys, with a correlation coefficient of -0.48 and
-0.46 respectively, which is statistically significant at the 0.01 level. In the other surveys, the
correlation is lower, at -0.37 and -0.17 for the PEW and LAPOP surveys respectively.
20
![Page 21: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/21.jpg)
SE
NZ
DESK AT
CZCH FI DKAUHUPL GB JPFRLV USNOCYEE
CLUA BG TWRUZASI HR AR BEKRPH TRCN ES IL
-.8-.6
-.4-.2
0.2
Rel
ativ
e In
com
e C
oeffi
cien
t
40 60 80 100Urbanization
ISSP
CZ
NLFR
CHDE IS BEGBNOSK PL ILEE SE DK
BGUASI LT FIIT
IE RU ESHUAL
CYPT
-.6-.4
-.20
.2R
elat
ive
Inco
me
Coe
ffici
ent
50 60 70 80 90 100Urbanization
ESS
VE
UYJM EC BR CL
CONI ARPASV DOPY BO
GY MXHN CRBZ PEGT
HT
-.50
.51
Rel
ativ
e In
com
e C
oeffi
cien
t
20 40 60 80 100Urbanization
LAPOP
JO
CZDE JP
US ILSN CAFRRUZA VEAUEG ARMXGR LBGB CLESITPHNG SVTN KRKEUG ID TRGH PL BRBOPK CN
MY
-.6-.4
-.20
.2R
elat
ive
Inco
me
Coe
ffici
ent
20 40 60 80 100Urbanization
PEW
Figure 3: Scatter of Relative Income Coefficient and Urbanization
I again run a series of HLM for each survey examined in this paper, controlling for both
individual and country-level effects. To investigate how urbanization mediates the influ-
ence of relative income on support for redistribution, I include an interaction term between
country-level urbanization and relative income, with the expectation that this interaction
will have a negative relationship (indicating that in more urbanized countries, income will
have a stronger negative relationship with support for redistribution or concern with in-
equality). In all four models, this interaction term is statistically significant in the expected
direction – with a notably strong relationship being evident in the ISSP and PEW surveys.
Marginal effects plots calculated from these models, shown in Figure 4, allow for a bet-
21
![Page 22: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/22.jpg)
Table 4: Urbanization, Income and Redistribution Support HLM
(1) (2) (3) (4)ISSP ESS LAPOP PEW
Individual LevelMarried -0.0188 -0.00430 0.0210 -0.00211
(0.0190) (0.0141) (0.0359) (0.00992)Urban -0.0113 -0.0120 -0.120 0.0493∗∗
(0.00853) (0.0187) (0.0888) (0.0184)Female 0.0665∗∗ 0.0998∗∗∗ -0.0226 0.0440∗∗
(0.0229) (0.0188) (0.0181) (0.0155)Unemployed 0.0366 0.0822∗∗∗ -0.0300 0.0276
(0.0292) (0.0185) (0.0650) (0.0146)Age 0.00155∗ 0.00436∗∗∗ 0.00104 0.00106∗
(0.000613) (0.000636) (0.00104) (0.000480)Yrs. Educ -0.0126∗∗∗ -0.0109∗∗∗ -0.00217
(0.00356) (0.00244) (0.00350)Religiosity -0.0158 0.0115 0.0606∗
(0.00815) (0.0219) (0.0300)Ln. Rel. Inc. 0.282∗ 0.0946 0.402∗∗ 0.0496
(0.110) (0.129) (0.147) (0.0465)% Urb X Rel. Inc. -0.00602∗∗∗ -0.00367∗ -0.00532∗∗ -0.00187∗∗
(0.00162) (0.00172) (0.00199) (0.000690)Country LevelGov. Rev. 0.0150∗ -0.000238 0.00639 0.00754
(0.00673) (0.00693) (0.0108) (0.00483)Ln. GDPPCPPP -0.398 -0.195 -0.171 -0.118
(0.240) (0.118) (0.200) (0.0802)Gini 0.00193 0.0227 0.0110
(0.0162) (0.0151) (0.00634)Polity2 -0.00245 -0.00893 0.0816∗∗∗ 0.00716
(0.0269) (0.0241) (0.0247) (0.00500)%Urban -0.00467 -0.0103 0.00861 -0.00173
(0.00637) (0.00561) (0.00734) (0.00293)Constant 6.668∗∗ 4.848∗∗∗ 5.560∗∗∗ 2.857∗∗∗
(2.524) (1.293) (1.280) (0.724)Individual SD 0.347∗∗∗ 0.221∗∗∗ 0.318∗∗∗ 0.167∗∗∗
(0.0418) (0.0390) (0.0419) (0.0153)Country Level SD 1.012 0.941∗∗ 1.698∗∗∗ 0.743∗∗∗
(0.0295) (0.0187) (0.0357) (0.0171)N 38307 40766 26165 30138Countries 36 27 21 37
Standard errors in parentheses∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
22
![Page 23: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/23.jpg)
-.4-.2
0.2
.4M
argi
nal E
ffect
20 90Urbanization
ISSP
-.3-.2
-.10
.1.2
Mar
gina
l Effe
ct
20 90Urbanization
ESS-.2
0.2
.4.6
Mar
gina
l Effe
ct
20 90Urbanization
LAPOP
-.2-.1
0.1
Mar
gina
l Effe
ct
20 90Urbanization
PEW
Figure 4: Marginal Effects of Relative Income on Support for Redistribution Across Urban-ization
ter understanding of the substantive and statistical interpretation of the interaction effect.
While the same general negative relationship is shown across all surveys, there are important
differences between them as well. In the LAPOP countries, it is only in the most urbanized
countries where income seems to have the predicted negative relationship. For most coun-
tries in the region, income is actually positively associated with support for redistribution.
The ESS and PEW surveys generally show that in most countries, both urbanized and rel-
atively rural, income has a negative relationship with support for redistribution, although
this relationship becomes more strongly negative in the most urbanized countries.
23
![Page 24: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/24.jpg)
5 Conclusions
Urban contexts play an important role in shaping demand for redistribution and partic-
ularly the relationship between income and support for redistribution. This paper shows
income does not have a universal relationship with support for redistribution, in contrast to
a key assumption behind seminal models of political economy. Rather than relative income
differences naturally leading to a variety of preferences for or against the redistribution of
income, my findings suggest this relationship is shaped by the context in which an individual
currently resides.
These findings are extremely relevant to rapidly urbanizing countries today. Whyte
(2010), found that individuals in contemporary China were relatively acceptant of inequality
compared to other post-transition economies, and that the “social volcano” of protests of
inequality was not born out in China today. However, over the long run, greater splits
income-based splits over issues of economic inequality and support for redistribution can be
expected to expand into the future.
24
![Page 25: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/25.jpg)
References
Acemoglu, D. and J. Robinson. 2006. Economic Origins of Dictatorship and Democracy.
Cambridge University Press.
Bartels, L.M. 2008. Unequal Democracy: The Political Economy of the New Gilded Age.
Princeton University Press.
Bates, R.H. 1981. Markets and states in tropical Africa. University of California Press
Berkeley.
Benabou, R. and E.A. Ok. 2001. “Social Mobility and the Demand for Redistribution: The
POUM Hypothesis.” Quarterly Journal of Economics 116(2):447–487.
Beramendi, Pablo and Philipp Rehm. 2016. “Who gives, who gains?: Progressivity and
Preferences.” Comparative Political Studies 49(4).
Boix, Carles. 2003. Democracy and Redistribution. Cambridge Univ Press.
Bradley, D., E. Huber, S. Moller, F. Nielsen and J.D. Stephens. 2003. “Distribution and
redistribution in postindustrial democracies.” World Politics 55(2):193–228.
Clark, A.E., P. Frijters and M.A. Shields. 2008. “Relative income, happiness, and utility: An
explanation for the Easterlin paradox and other puzzles.” Journal of Economic Literature
46(1):95–144.
Cusack, T., T. Iversen and P. Rehm. 2006. “Risks at work: The demand and supply sides
of government redistribution.” Oxford Review of Economic Policy 22(3):365.
Dion, M.L. and V. Birchfield. 2010. “Economic Development, Income Inequality, and Pref-
erences for Redistribution1.” International Studies Quarterly 54(2):315–334.
25
![Page 26: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/26.jpg)
Fearon, J.D. 2003. “Ethnic and Cultural Diversity by Country.” Journal of Economic Growth
8(2):195–222.
Fong, C. 2001. “Social preferences, self-interest, and the demand for redistribution.” Journal
of Public Economics 82(2):225–246.
Graham, Carol. 2009. Happiness around the world: The paradox of happy peasants and
miserable millionaires. Oxford University Press.
Iversen, T. 2005. Capitalism, democracy, and welfare. Cambridge Univ Pr.
Kenworthy, L. and J. Pontusson. 2005. “Rising inequality and the politics of redistribution
in affluent countries.” Perspectives on Politics 3(03):449–471.
King, G., C.J.L. Murray, J.A. Salomon and A. Tandon. 2004. “Enhancing the validity
and cross-cultural comparability of measurement in survey research.” American Political
Science Review 98(01):191–207.
King, Gary and Jonathan Wand. 2007. “Comparing incomparable survey responses: Evalu-
ating and selecting anchoring vignettes.” Political Analysis 15(1):46–66.
Knight, J. and R. Gunatilaka. 2010. “The Rural–Urban Divide in China: Income but Not
Happiness?” The Journal of Development Studies 46(3):506–534.
Korpi, Walter and Joachim Palme. 1998. “The paradox of redistribution and strategies of
equality: Welfare state institutions, inequality, and poverty in the Western countries.”
American sociological review 63(5):661–687.
Lerner, M.J. 1980. The belief in a just world: A fundamental delusion. Plenum Pub Corp.
Lindert, Peter. 2004. Growing public: Social spending and economic growth since the eigh-
teenth century. Cambridge University Press.
26
![Page 27: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/27.jpg)
Lipton, M. 1977. Why poor people stay poor: urban bias in world development. Harvard
Univ Pr.
Madison, James. 1966. Paper 10. In The Federalist Papers. New Rochelle, New York:
Arlington House pp. 77–83.
Marx, Karl and Friedrich Engels. 2002. The Communist Manifesto. London: Penguin
Classics.
McCarty, Nolan M., Keith T. Poole and Howard Rosenthal. 2006. Polarized America: The
dance of ideology and unequal riches. MIT Press.
Meltzer, A.H. and S.F. Richard. 1981. “A Rational Theory of the Size of Government.”
Journal of Political Economy 89(5):914.
Moene, K.O. and M. Wallerstein. 2003. “Earnings inequality and welfare spending: A
disaggregated analysis.” World Politics 55(4):485–516.
Park, Robert E. 1915. “The City: Suggestions for the Investigation of Human Behavior in
the City Environment.” American Journal of Sociology 20(5):577–612.
Perotti, Roberto. 1996. “Growth, income distribution, and democracy: What the data say.”
Journal of Economic growth 1(2):149–187.
Persson, Torsten and Guido Tabellini. 1994. “Is Inequality Harmful for Growth?” The
American Economic Review 84(3):600–621.
Piketty, T. 1995. “Social mobility and redistributive politics.” The Quarterly journal of
economics pp. 551–584.
Ravallion, Martin and Michael Lokshin. 1999. Subjective economic welfare. Vol. 2106 World
Bank Publications.
27
![Page 28: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/28.jpg)
Rodden, Jonathan. 2010. “The geographic distribution of political preferences.” Annual
Review of Political Science 13:321–340.
Romer, T. 1975. “Individual welfare, majority voting, and the properties of a linear income
tax.” Journal of Public Economics 4(2):163–185.
Runciman, W.G. 1966. Relative deprivation and social justice: A study of attitudes to social
inequality in twentieth-century England. University of California Press.
Scheve, K. and D. Stasavage. 2006. “Religion and preferences for social insurance.” Quarterly
Journal of Political Science 1(3):255–286.
Stouffer, S. A., E. A. Suchman, L. C. DeVinney, S. A. Star and R. M. Williams Jr. 1949.
The American soldier: Adjustment During Army Life. Princeton University Press.
Veblen, Thorstein. 1899. The theory of the leisure class: An economic study in the evolution
of institutions. The Macmillan Company.
Whyte, Martin K. 2010. Myth of the social volcano: Perceptions of inequality and distributive
injustice in contemporary China. Stanford Univ Press.
Wilensky, Harold L. 1975. The welfare state and equality: structural and ideological roots of
public expenditures. Univ of California Press.
Wirth, L. 1938. “Urbanism as a Way of Life.” American journal of sociology pp. 1–24.
Wolf, Eric R. 1955. “Types of Latin American Peasantry: A Preliminary Discussion.”
American Anthropologist 57(3):452–471.
28
![Page 29: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/29.jpg)
Appendix A: Variables and Coding
European Social Survey
ESS Round 4: European Social Survey Round 4 Data (2008). Data file edition 4.0. Norwe-
gian Social Science Data Services, Norway ? Data Archive and distributor of ESS data.
Government Redistribution. Five point Likert scale from the question: “The government
should take measures to reduce differences in income levels.” (4) Agree Strongly, (3) Agree,
(2) Neither Agree nor Disagree, (1) Disagree, (0) Disagree Strongly. Don’t Knows recoded
as missing. To make this variable of government redistribution comparable to the four point
scale used in other surveys, this measure is multiplied by 0.75.
Urban. Dummy variable recoded from a five–point response to the question “Which phrase
on this card best describes the area where you live”: (1) Urban: “A Big City; The suburbs
or outskirts of a big city; A town or a small city.” (0) Rural: “A country village; A farm or
home in the countryside.” Don’t know is recoded as missing.
Years of Education. Continuous: Education in years.
Gender. Dummy variable: (1) Female, (0) Male.
Age. In years.
Married. Dummy variable taken from the question: “Could I ask about your current legal
marital status? Which of the descriptions on this card applies to you?”. (1) Legally Married,
or in a civil partnership (0) Separated, Divorced, Widowed, Never married or entered into a
civil partnership.
Religiously Active. Dummy variable taken from the question: “Apart from special occasions
such as weddings and funerals, about how often do you attend religious services nowadays?”
(1) At least once a month; (0) Less than once a month.
Interest in Politics : Four point Likert scale from the question: “How interested would you
say you are in politics?” (4) Very interested; (3) Quite interested, (1) Hardly interested, (0)
29
![Page 30: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/30.jpg)
Not at all interested. Don’t Know is recoded as missing.
Unemployed. Dummy variable taken from the following question: “Which of these descrip-
tions applies to what you have been doing for the last 7 days?” (1) Unemployed and actively
looking for a job. (0) Other response (In paid work, student, doing housework, looking after
children, sick or disabled, unemployed but not actively looking for a job.).
Voted. Dummy variable taken from the question: “Some people don?t vote nowadays for one
reason or another. Did you vote in the last [country] national8 election in [month/year]?”
(1) Yes; (0) No.
Party ID. Taken from the question “Is there a particular political party you feel closer to
than all the other parties? Which one?” This nominal variable preserves the original coding.
Retrospective Party Vote. “Which party did you vote for in that [most recent] election?”
Original coding is preserved.
Ethnic Identity. Coded according to the Fearon (2003) classification scheme. This is pri-
marily coded from the question “What language or languages do you speak most often at
home?” Respondents could specify up to two languages. This coding was sometimes sup-
plemented with the variables “What country were you born in?” and “What country were
your parents born in?” for certain immigrant groups – for example, an individual living in
Germany was coded as Turkish if either they were born in Turkey, or both of their parents
were born in Turkey (but not if they were born in Germany, and one of their parents was
born in Turkey). Sometimes religious affiliation was also used to supplement – for example,
Pomaks in Bulgaria were coded using language and religion (Pomaks are Bulgarian speaking
Muslims). Finally, the region of the respondent was occasionally used – for example, Scots
and Welsh in Great Britain were coded according to region rather than language. All coding
decisions of ethnicity for the ESS are available on request.
Relative Household Income. Taken from a household income variable asking respondents
to place themselves into predetermined income deciles from the following question: “Using
30
![Page 31: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/31.jpg)
this card, please tell me which letter describes your household’s total income, after tax and
compulsory deductions, from all sources?” Deciles are converted to relative income by first
taking the average of the top and bottom values of the decile (for example, if the decile is
a range between $15,000 and $20,000 a year, the respondent is scored as having an income
of $17,500). Top categories are given a value of 1.5 times the top category (eg. more
than $50,000 would become $75,000). Relative income is then calculated by dividing the
respondents income score by the weighted average for the country.
ISSP
ISSP Research Group, International Social Survey Programme (ISSP): Role of Government
IV. Distributor: GESIS Cologne Germany ZA4850, Data Version 2.0.0.
Government Redistribution. Four point Likert scale from the question: “On the whole,
do you think it should or should not be the government’s responsibility to reduce income
differences between the rich and the poor?” (3) Definitely Should be, (2) Probably should
be, (1) Probably should not be, (0) Definitely should not be. “No Answer” and “Can’t
Choose” are recoded as missing.
Urban. Question phrasing is idiosyncratic to each country, or sometimes taken from an
alternative question on community size. For this variable I use the harmonized “urbrural”
variable, which gives five possible categories: (1) Urban: “A Big City; The suburbs or
outskirts of a big city; A town or a small city.” (0) Rural: “A country village; A farm or
home in the countryside.” Don’t know is recoded as missing.
Years of Education. Continuous: Education in years.
Gender. Dummy variable: (1) Female, (0) Male.
Interest in Politics : Five point Likert scale from the question: “How interested would you
say you personally are in politics?” (4) Very interested, (3) Fairly interested, (2) Somewhat
interested, (1) Not very interested, (0) Not at all interested. Don’t Know is recoded as
31
![Page 32: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/32.jpg)
missing.
Unemployed. Question phrasing is idiosyncratic to each country, but usually asks current
employment status (1) Currently looking for a job. (0) Other response (working, student,
doing housework, sick or disabled, doesn’t work and is not looking for a job.).
Partisanship. Questions of partisanship in the ISSP range from prospective and retrospective
vote choice, as well as partisan affiliation, and are country-specific. Original codings for this
ordinal measure are preserved.
Relative Household Income. Question phrasing and measurement of household income is
idiosyncratic to each country. For some countries, the source variable of household income is
continuous, while for others it is categorical. When categorical, relative income is calculate
by first taking the average of the top and bottom values of the quantile (for example, if the
decile is a range between $15,000 and $20,000 a year, the respondent is scored as having
an income of $17,500). Top categories are given a value of 1.5 times the top category
(eg. more than $50,000 would become $75,000). Relative income is then calculated by
dividing the respondents income score by the weighted average for the country. When
continuous, relative income is calculated by dividing respondents stated household income
by the weighted average for the country.
Latin American Public Opinion Project (LAPOP)
I thank the Latin American Public Opinion Project (LAPOP) and its major supporters
(the United States Agency for International Development, the United Nations Development
Program, the Inter-American Development Bank, and Vanderbilt University) for making the
data available.
Government Redistribution. Seven point Likert scale from the question: “The government
should take strong policies to reduce inequality between the rich and the poor. How much do
you agree or disagree with this statement?” (6) Strongly Agree, (0) Strongly Disagree. Don’t
32
![Page 33: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/33.jpg)
Knows recoded as missing. To make this variable of government redistribution comparable
to the four point scale used in other surveys, this measure is multiplied by 0.5.
Urban. Dummy variable completed by the interviewer: “Urban or Rural Primary Cluster.”:
(1) Urban, (0) Rural.
Age. In years.
Years of Education. Continuous: Education in years.
Gender. Dummy variable: (1) Female, (0) Male.
Interest in Politics : Four point Likert scale from the question: “How much interest do you
have in politics?” (4) A lot; (2) Some, (1) Little, (0) None. Don’t Know is recoded as
missing.
Unemployed. Dummy variable taken from the question: “How do you mainly spend your
time?” (1) Actively looking for a job. (0) Other response (working, student, doing house-
work, sick or disabled, doesn’t work and is not looking for a job.).
Married. Dummy variable taken from the question: “What is your marital status?”. (1)
Legally Married, or in a civil partnership (0) Separated, Divorced, Widowed, Never married
or entered into a civil partnership.
Religiously Active. Dummy variable adapted from the question: “Do you attend religious
services.....?” (1) At least once a month; (0) Less than once a month.
Voted. Did you vote in the last general elections of [year]? (1) Yes; (2) No.
Party ID. Response to the question “Which political party do you identify with?” Original
codings are preserved. In the United States and Canada, the question on partisanship is
retrospective vote choice: “What party did you vote for in the last election?”.
Ethnic Identity. Coded according to the Fearon (2003) ethnicity categories. In most coun-
tries, (other than the United States and Canada), a country specific question was asked with
preset ethnic categories. For example, in Bolivia, respondents were asked “Do you consider
yourself white, mestizo, indigenous, Black or Afro-Bolivian, Mulatto or something else?”
33
![Page 34: Cities, Income, and Demand for Redistributionsites.duke.edu/preferencesoverredistribution/files/... · studies examining the relationship between inequality and economic growth or](https://reader034.vdocument.in/reader034/viewer/2022042120/5e9a2baffb11b90ddd7c1fbe/html5/thumbnails/34.jpg)
Occasionally religion was used (eg. Jews in Argentina).
Relative Household Income. Taken from a household income variable asking respondents to
place themselves into predetermined income categories from the following question: “Into
which of the following income ranges does the total monthly income of this household fit,
including remittances from abroad and the income of all the working adults and children? ”
Quantiles are converted to relative income by first taking the average of the top and bottom
values of the quantile (for example, if the decile is a range between $15,000 and $20,000 a
year, the respondent is scored as having an income of $17,500). Top categories are given a
value of 1.5 times the top category (eg. more than $50,000 would become $75,000). Relative
income is then calculated by dividing the respondents income score by the weighted average
for the country.
Cross–National Indicators
GDP per capita at PPP. From the World Bank.
Gini Coefficient. From the World Bank. Missing years are imputed by using the average of
the first available year prior and after the missing year, weighted by the length of the gap.
(For example, if the Gini Coefficient was listed as 30 in 1990, and 40 in 2000 with missing
values for years 1991 to 1999, the imputed gini coefficient would be 31 in 1991, 32 in 1992
and so forth).
Urban Percentage of the Population. The proportion of the population in the country residing
in urban areas. From the World Bank.
Polity2. Level of democracy/autocracy in a country. Ranges from -10 (full autocracy) to
+10 (full democracy).
34