1

International Development Institute Working Paper 2014-03

What are the Political and Social Implications of the New Middle Classes in Developing Countries?

Frank-Borge Wietzke

Andy Sumner

King's International Development Institute Room 8C, Chesham Building

Strand, London

WC2R 2LS

+44 (0)20 7848 1514

kings-idi@kcl.ac.uk

2

What are the Political and Social Implications of the New Middle Classes in Developing Countries?

Frank-Borge Wietzke and Andy Sumner 1

27 November 2014

Abstract: This is a survey paper that considers the emergence of new middle classes in

largely middle-income developing countries and proposes a set of research questions arising

for future exploration. Rapidly falling global poverty rates are leading to changes in the

income distribution of many developing countries. Yet, the political consequences of these

developments have received relatively little attention to date. We review emerging evidence

on the living conditions and attitudes of upwardly mobile lower middle-income groups and

discuss how they compare to theories of middle or working class-led development in the

political regime transition literature. We conclude that middle income groups in emerging

economies still face important economic and organizational challenges that prevent them

from playing the transformative role often attributed to working and middle classes in the

historical development of todays advanced welfare state democracies. Pressures for political

reform are more likely to arise around individual property rights and government corruption,

as formerly poor households aim to protect gains in economic wellbeing. We also identify

important divides in social policy and redistribution preferences between upwardly mobile

lower middle-income groups and more affluent middle classes.

1 Correspondence to: f.b.wietzke@lse.ac.uk and andrew.sumner@kcl.ac.uk Many thanks for comments to Nancy Birdsall and Paul Segal.

3

1. Introduction There is a long tradition of linking poverty reduction and middle class growth to

political change and democratisation. Modernisation theorists, and more recently analysts of

political transition, often suggest that falling poverty numbers and middle class growth are

conducive to the emergence of stable democracies. This argument is based on historical

observations about the role of bourgeois middle classes in the development process of

advanced European democracies, as well as on observed positive cross-national associations

between the size of middle-income groups and political and social cohesion outcomes (Barro,

1999; Easterly 2001; Easterly et al. 2006; Przeworski, 1985).

By most of these accounts we should expect to see important changes in the political

landscape of many of todays emerging economies. The global population of the developing

world living above the absolute (moderate) global poverty line of PPP$2/day (in 2005

Purchasing Power Parity terms) but below $10/day (which is associated with decreased

vulnerability to poverty) has doubled since the early 1990s.2 The group on incomes between

PPP$2$10/day now accounts for almost half of the developing worlds population and could

measure 34bn by 2030 (Edward and Sumner, 2014a; 2014b). There is also dramatic growth

of affluent income groups above $10/day whose circumstances resemble more closely those

of conventional middle classes in advanced economies. Even though these groups still

account for relatively small population shares in their countries, their actual numbers are

beginning to exceed those in advanced economies by some estimates (Wilson 2013: 2,

Kharas 2010).3

The expansion of both these middle classes (the lower insecure middle and the more

prosperous upper middle noted above) is by now a hot topic in the media and among the

major aid organizations and global consultancy agencies.

4

2 See Sumner (2012). All dollar values in the remaining parts of the article are reported in 2005 PPP terms.

However, political scientists have

only paid limited attention to recent lower middle class growth. Most of the political science

literature on the politics of economic transition from the late 1990s and early 2000s does not

yet take into account more recent estimates of lower income growth. Available literature also

does not provide significant theorising of potential new conflict lines and political coalitions

that might emerge in the process of rapid expansion of the middle-income groups.

3 Projections differ depending on assumptions about growth and inequality trends and use of national accounts or survey means (see Wilson 2013; Edward and Sumner, 2014a; Goldman Sachs 2008; Kharas and Gertz 2010). 4 See, for example, The Economist, 1 June and 29 June 2013, the New York Times, 2 June 2013, reports by private sector consultancies (for instance, Goldman Sachs 2008; Wilson 2013), as well as by the growing grey literature of donor agencies (ADB 2010; AfDB 2011; Chun 2010; Ferreira et al. 2013 for the World Bank and Kharas 2010 for the OECD).

4

The objective of this paper is to fill this gap by discussing the political and social

consequences of rapid lower middle income growth. Our theoretical starting point is the

standard three-class model that distinguishes between the poor or working classes,

middle classes and elites in the analysis of political regime transitions (Acemoglu and

Robinson, 2006; Boix, 2003; Levantoglu 2013; Rueschemeyer et al. 1992). We review

emerging evidence on the economic and social circumstances of the new middle income

groups in emerging economies and discuss how their living conditions compare to attributes

typically associated with middle and working classes in accounts of the historical

development of todays advanced welfare state democracies. We conclude that in particularly

the large population majorities between extreme poverty and affluent middle class incomes

still lack the level of economic security and political (or work-related) organization that

would enable them to play, at least in the near future, the transformative functions attributed

to working and middle classes in the literature. However, we also identify increasing amounts

of disposable income, human capital investments and durable assets among the new middle

income groups. Building on earlier work by Ansell and Samuels (2010) we speculate that this

may create new demands for the protection of individual property rights and government

accountability that may, in themselves, constitute an important impetus for future political

reforms.

Another major influence on our discussion is the literature on social mobility, relative

deprivation, and redistribution preferences. Important theories of political preference

formation suggest that people who believe that they or their children will have strong

prospects for upward mobility in the future are less likely to support redistribution in the

present, because they want to lower the tax burden for themselves and their children in future

time periods (Hirschman 1973; Benabou and Ok 2001; Alesina and La Ferrara 2005; see also

Piketty 1995). But observers typically agree that these attitudes weaken as more permanent

inequalities emerge, or as groups begin to face real or perceived threats of falling behind

(Hirschman 1973). We discuss evidence on social and political attitudes in emerging

economies and identify emerging divides between lower and middle income groups. Even at

the comparatively low levels of economic development that still characterise most emerging

nations today, experiences and preferences of affluent groups are often already shaped by

fears of downward mobility and feelings of relative deprivation that resemble those of the

squeezed middle classes in advanced industrialised economies (Frank 2007). This is in

contrast to lower-income and vulnerable groups for whom concerns about absolute economic

deprivation and uncertainty still dominate daily reality. We suggest that these differences in

5

social perceptions may have contributed to recent violent events in countries like Egypt,

Venezuela or Thailand, where social inequalities appear to interact dangerously with pre-

existing divides along religious, ethnic and regional lines.

The paper proceeds as follows. After an overview of the theoretical literature that

informs our discussion (Section 2) we turn to prevailing definitions of the middle classes.

We focus on relative as well as absolute definitions of middle class status, which we feel are

central to explaining different responses to social mobility between lower and affluent middle

income groups (Section 3). Section 4 discusses lower and middle class divides over

government redistribution and Section 5 reviews emerging evidence on the consequences of

lower middle class growth for actual social policy choices. The last section concludes and

outlines avenues for future research.

Before we enter the discussion it is important to be clear about the strengths and

weaknesses of the evidence reviewed here. We believe that the primary strength of the

studies we discuss comes from the greatly enhanced quality and frequency of micro data on

poverty trends and lower middle-income growth. Most of the economic studies and grey

literature we cite draw on repeated household income and expenditure surveys and databases5

By contrast, evidence is much weaker when it comes to establishing causal evidence

on the consequences of poverty change and upward mobility for individual behaviours and

attitudes. In the absence of panel surveys that track values and behaviours over time most

studies of middle class preferences and behaviours reviewed here only provide cross-

sectional snapshots of the distribution of preferences across relevant income groups. These

studies do not permit claims that personal experiences of upward mobility cause new

political preferences and demands for reforms. For the time being we limit ourselves to

largely descriptive accounts of the distribution of political preferences across relevant income

groups in emerging economies. It will be reserved for future analysis to document how

relevant attitudes and behaviour change over time as formerly poor individuals move along

the income ladder.

that are relatively well tested and provide much more detailed information on within-country

income distributions t than simpler indicators of economic and distributional change typically

used in the academic literature on political transitions (such as per capita GDP or the Gini).

2. Theoretical points of departure

5 See for instance the World Banks PovcalNet: http://iresearch.worldbank.org/PovcalNet/index.htm

6

A natural starting point for our discussion is the literature on the political economy of

middle class growth. Classical models of middle class development, associated with

modernisation thinkers like Barrington Moore (1966), Seymor Martin Lipset (1959, 1960) or

David Landes (1998) argue that growth in middle-income groups is associated with higher

entrepreneurial activity, investments in human capital and democratic development. This is

because middle classes economic prospects traditionally depend more on professional skills

and income than on inherited capital (Acemoglu and Zilibotti 1997; Doepke and Zilibotti

2007), progressive political and market-friendly value systems of middle class citizens

(Easterly 2001; Lopez-Calva et al. 2011; Lipset 1960; Przeworski (1985), and because of

middle classes ability to mitigate conflicts between elites and extremist movements (Lipset

1959: 58; Easterly 2001; Easterly et al. 2006; Barro, 1999).

Several recent extensions of these theories have linked arguments about middle class

growth to processes of political transition. For instance, drawing on the well-worn median

voter model, Boix (2003) and Acemoglu and Robinson (2006) respectively argue that falling

or middling inequality levels increase the chances of political reform/democratisation,

because elites are more likely to agree to power sharing when they face fewer threats of

expropriation by the poor masses. In this framework middle classes primarily function as

deal makers or buffers between poor groups and elites, implying again that higher middle

class shares are associated with a greater likelihood of democratic transition (Acemoglu and

Robinson 2006: 253ff).

Other theories have highlighted the role of lower income groups and working classes.

For example Power-Resource Theory and historical accounts of the development of welfare

state institutions in todays advanced economies identify conflicts between urban working

classes and capital holders as the primary drivers of democratisation and social reforms

(Korpi 1983; Therborn 1977; 1979; Rueschemeyer et al. 1992; Lindert 2004). These theories,

albeit to varying degrees, emphasize rising real wages and advances in the political

organization and unionization of the working classes as the most important historical

developments facilitating political change. By contrast, they are more sceptical about the

transformative role of middle classes. For example Rueschemeyer et al. (1992) suggest that

middle classes first and foremost sought their own inclusion and formed the alliances

necessary to achieve this end (p.168). But middle classes often oppose suffrage extension to

working classes as such a move posed a potential threat to their interests (ibid.: 6). Building

on these ideas Levantoglu (2013) presents a model where middle classes in developing

country autocracies only demand political reforms when they feel insecure about their future

7

status, but these groups will collaborate with elites when they feel secure about their

economic prospects.

Our primary reason for questioning these theories in the context of recent global

lower middle income growth is that there is neither strong evidence for the development of a

broad-based middle class culture in emerging economies, nor that fast growing lower

middle income groups work in conditions that would enable them to organize as effectively

around shared class interest as working classes in the historical development process of

Europes and North Americas advanced welfare states. Our review suggests that political

outcomes in emerging countries are more accurately described by growing divides between

the social protection needs of the new upwardly mobile, but economically vulnerable

population majorities and affluent middle classes that attempt to protect their status

advantages. Low rates of formal sector employment and weaker institutions for work-related

collective action among lower middle income groups also lead us to question whether

political relations in emerging economies will be dominated by the type of organized class

struggle between labor and capital often described in historical accounts of European and

North American welfare state development.

Our own theoretical preference tends to gravitate towards more recent contributions

that emphasise political negotiations over property rights and social protection over

conventional models of working-, middle class, and elite bargaining. The primary change we

observe among populations who have escaped extreme poverty is the increased accumulation

of consumer assets, housing, health and education. This makes it more likely that these

groups will oppose unjustified claims on their property by corrupt or unaccountable

governments and bureaucrats. This trend is consistent with historical scenarios described by

Ansell and Samuels (2010), who argue that the accumulation of assets by upwardly mobile

groups can create strong demands for the extension of economic and political freedoms, as

newly rich households are particularly likely to defend their hard-earned economic gains

against infringement of their property rights by elites. In a different vein, Reenock et al.

(2007) have suggested that radical demands for redistribution and government poverty

alleviation policies are more likely to occur if income growth occurs in the presence of wide-

spread incidences of absolute basic needs deprivation.

Building on the work of these authors we posit that the implication of recent lower

middle income growth is not the linear progression towards more peaceful and democratic

development described by standard political economy models. Most of the evidence reviewed

here suggests that conflicts over political outcomes are likely to increase as growing numbers

8

of people join the lower middle-income ranks. More peaceful outcomes typically associated

with middle income growth will only emerge after more substantive and broad-based

improvements in emerging economies living standards.

3. Who are the new middle classes?

Scholarship on middle class development has traditionally not agreed on a consistent

definition of middle class status. The recent literature on emerging economies middle classes

is no exception. Economists, whose work dominates research on lower middle income

growth, typically rely on statistical definitions of middle class membership.6 For example, an

early study by Easterly (2001) focuses on the 20th and 80th percentiles on the consumption

distribution to document the existence or absence of a middle class consensus. More

recently Birdsall et al. (2000) use the range from 75% to 125% of national median income,

allowing the population size of the middle class to vary across countries. On the other hand,

sociologists and political scientists working in the Marxist or Weberian tradition define

middle class status by the type of education, skills, or assets individuals bring to the market.

In recent empirical studies this is typically operationalized by peoples position on an asset

index, that summarizes information on household wealth and economic attributes, see for

instance Cramer and Kaufman 2011, Lopez-Calva and Torche, 2011; Lopez-Calva et al.

2011).7 Finally several studies have used peoples subjective assessment of their class status,

drawing on subjective income and class categories regularly included in international value

and perception surveys (Amoranto et al. 2010; Ferreira et al. 2013).8

Despite their focus on different aspects of wellbeing, the above approaches all have in

common that they define middle class status in relative terms. Membership in the middle

class is determined in relation to some average measure of material or subjective wellbeing

that is allowed to vary across countries and time. These relative definitions are popular

among analysts because they capture social comparisons and psychological processes that are

at the centre of many theories of middle class behaviour (see below and Birdsall et al. 2000).

6 The preference for relative measures in earlier studies is likely related to the comparatively better availability of relative income scales, such as through the UN-WIDER or Deininger/Squire inequality databases. 7 Education is sometimes excluded from these models, because it serves as an independent predictor in the explanation of political behaviour and mobility experiences (Lopez-Calva and Torche, 2011; Lopez-Calva and Ortiz-Juarez 2011). 8 These studies typically have to deal with the problem that self-reported middle class shares are typically over reported, leading to larger middle class shares than would be practically feasible.

9

However, their primary drawback is a lack of information on peoples objective living

conditions that can be compared across countries. Notwithstanding some recent economic

convergence, differences in average incomes between advanced and emerging economies are

still so large that a person living in the middle of his or her countrys distribution may find

herself in starkly different economic circumstances, depending on whether he or she is was

fortunate enough to be born into a developed nation or not. This makes it harder to compare

changes in actual living standards across countries of the world and it omits important

information about economic risks and deprivations still faced by many people in the middle

income segments of emerging countries. For instance, Ravallion (2010: 4) notes in a telling

example that the Xiaokang in China, an official income category that resembles the idea of

a middle class in local terms, required an income of PPP$2.24 per day in rural areas and

PPP$3.47 a day in urban areas in 1991,9 far below the official poverty lines of most advanced

economies (for example, US$13/day in the US). Earlier work by Milanovic and Yitzhaki

(2002) finds that only an estimated 11% of the worlds population can be considered as

middle class, using Italys and Brazils mean incomes as the respective upper and lower

cut-offs.10

The bottom line that emerges from these studies is that the bulk of recent income

growth in the emerging world is concentrated among populations who are still far from the

incomes associated with middle class status under the relative definitions typically used in

advanced economies. The aforementioned study by Ravallion (2010) shows that income

growth was bundled just above the moderate international poverty line of PPP$2 per day.

In comparison, 78% of the worlds population are still poor by this definition and

only 11% would be counted among the rich. More recent studies that have used absolute

income thresholds that approximate difference between non-poor individuals and middle

classes in advanced economies (typically using a range of PPP$10-100), estimate that only

about a quarter of the worlds population would be categorized as middle class under these

definitions. Of these 60% live in the developed world and only 20% in the fast growing

BRICs economies (Wilson 2013; Kharas and Gertz 2010).

11

9 Similar income ranges are often used in the debate about Indian middle classes (Ravallion 2010).

He consequently proposes a definition of a developing world middle class that explicitly

distinguishes between people who are middle class by emerging country standards and

poor by advanced nations official poverty benchmarks (using the US poverty line of

10 Milanovics and Yitzhak note that at the time of their study Brazils mean income was close to the average level of Western countries relative poverty lines (PPP$10/day, p.172). 11 Such trends are mostly driven by Asia (with China alone accounting for half of the developing worlds share in lower middle-incomes) and to a lesser extent by sub-Saharan Africa (AfDB 2011).

10

PPP$13/day as the upper cut-off). Banerjee and Duflo (2008) similarly observe that groups

between the 20th and the 80th percentile of the income distribution, which was used to define

developing countries middle classes under the relative definition of Easterly (2001), live on

precarious incomes between the global PPP$2 poverty line and PPP$4/day (see also Birdsall

et al. 2013).12 In a slight departure from Ravallion (2010), these authors use the more

conservative upper threshold of PPP$10/day for their definition of the developing worlds

middle class, based on observed differences in economic vulnerability around this income

benchmark (see below).13

3b. Non-monetary outcomes

The relatively precarious income situation of the fast growing lower-middle income

groups in emerging economies is also reflected in several non-monetary dimensions. On most

of these dimensions lower middle income groups differ quite considerably from the

circumstances typically attributed to working and middle classes in advanced economies.

Occupation

As noted before, work-related behaviours and collective action play an important role

in the literature on working classes and welfare state development. Economic theories of

middle class development similarly highlight occupational attitudes and behaviours, such as

increased entrepreneurial activity, professionalism, and greater human capital investments

among the bourgeoisie (Acemoglu and Zilibotti 1997; Doepke and Zilibotti 2007).

Most of the available evidence suggests that the working conditions of todays lower

middle income groups are still a long away from the scenarios described in these theories.

Birdsall et al. (2013) document that Latin American households on incomes between PPP$4

and PPP$10/day are more likely to be in informal, self-employed relationships in the primary

sector than the more affluent middle classes. This result is supported by Banerjee and Duflo

(2008), who find that work conditions of vulnerable groups resemble those of the poor more

than the affluent middle classes. Households on incomes just above the absolute poverty line

are still often engaged in informal, self-employed activities that do not provide much security

or chances for professional improvement. The only difference relative to poorer groups is that

12The exceptions are rural India and Pakistan where the majority of the population still live on less than PPP$2/day and Panama where middle classes are wealthier. 13 Most studies concur that the extreme poverty line of around PPP$ 2 should be used as the lower middle class threshold. However, a higher poverty cut-off of PPP$ 4 for more advanced emerging regions like Latin America (Birdsall et al. 2014; Ferreira et al. 2013).).

11

households in the vulnerable income bracket tend to have steadier and better-paid jobs.

Lopez-Calva and Ortiz-Juarez (2014) similarly document that occupational categories,

especially in interaction with rural-urban differences, are important predictors of movements

into upper middle class status, but with few differences between the poor and lower middle

income groups on incomes below PPP$10. This study is based on panel data and is thus able

to account for initial household attributes. These results are also consistent with other

emerging evidence that in many developing countries self-employment in relatively small

informal non-form enterprises account for a large share of incomes for rural populations close

to extreme poverty (Davis et al. 2010).

The important short-term implication of these results is that reform processes in

todays emerging economies are less likely to be driven by conflicts around organized

occupational interests that were important in the political development of todays advanced

welfare states. The continued high incidence of informality among people just out of poverty

creates obvious barriers for collective action. This is compounded by relative weak

institutions for organized labour in many developing regions, caused by the uneven coverage

and demise of unions during the adjustment period of the 1980s and 1990s in Latin America

and Eastern Europe, and historically more restrictive labour regulations in Asia (McGuire

1994, 1999; Haggard and Kaufmann 2008).

There is similarly little evidence that lower middle class growth will have the positive

macro-economic consequences associated with middle class-centred theories of capitalist

development. Most of the self-employed activities pursued by people close to the poverty line

still tend to represent economic coping strategies, rather than the type of enterprising

behaviours described by these theories. In the words of Banerjee and Duflo (2008:26):

While there are many petty entrepreneurs among the middle class, most of them do not seem to be capitalists in waiting. They run businesses, but for the most part only because they are still relatively poor and every little bit helps. If they could only find the right salaried job, they might be quite content to shut their business down. If the middle class matters for growth, it is probably not because of its entrepreneurial spirit.

Risk and uncertainty

Another widely accepted attribute of middle class status is economic security.

Sociologists in the Weberian tradition typically consider peoples economic opportunities

(life chances) and the security afforded by ones employment contract as an important

marker of middle class status (Goldthorpe and McKnight 2004; Grusky and Kanbur 2006).

12

This idea has also been taken on board in recent attempts to define middle class status in

developing economies. What emerges from this literature is, again, the large difference in

economic security between lower middle income groups and more affluent middle classes.

Lopez-Calva and Ortiz-Juarez (2014) and Birdsall et al. (2013) note for Latin America that

households below PPP$10/day have much higher probabilities of sliding back into poverty

(Lopez-Calva and Ortiz-Juarez estimate a probabilities of about 10% at the PPP$10 line, with

much higher values for lower-incomes). They conclude that this justifies distinguishing

explicitly between non-poor vulnerable groups and more affluent middle classes who have

attained a higher degree of economic security. Banerjee and Duflo (2008) and Ravallion

(2010) also acknowledge the high degree of economic insecurity of households in this

income range. All of this suggests that concerns about absolute deprivations (or the risk

thereof) are still a very tangible aspect of the daily realities of the fast growing population

majorities in emerging economies. We argue below that this is likely to lead to very different

reactions to economic growth and mobility experiences than would be predicted by standard

political economy models.

Education

Education is another important marker of middle class. It matters both directly, as a

determinant of individual economic and professional opportunities (Grusky and Kanbur

2006; Ferreira et al. 2013), and indirectly, through its well-documented association with

progressive social and political attitudes (Lopez Calva et al. 2011).

However, there are again few indications that upwardly mobile population majorities

in developing countries resemble affluent middle classes on this dimension. In Lopez-Calvas

and Ortiz-Juarezs (2014) vulnerability-based approach to middle class measurement,

education, especially tertiary, emerges as a strong predictor of differences between lower

income groups between PPP$ 2-10 and more affluent middle classes. Similar evidence is

produced by Castellani and Parent (2011), although the authors do not specify how middle

and affluent classes are defined.. Birdsall et al. (2013) also find that education distinguishes

vulnerable from affluent middle classes. Affluent middle class households are much more

likely to be headed by an individual with completed secondary education than the strugglers

who survive on less than PPP$10 a day.

There are also signs that education still represents a barrier to upward mobility that

affects poor and vulnerable households more than more affluent classes. Ferreira et al. (2013)

and Torche and Lopez-Calva (2011) find for Latin America that upward mobility was less

13

likely for individuals from households with lower educational background. In particular,

movements into the middle class were still much likelier for people who had some tertiary

education (Ferreira et al. 2013: 7).

There are, however, few signs that these educational differences go along with the

cultural changes and progressive values commonly attributed to middle classes. Lopez-Calva

and Ortiz-Juarez (2014), who compare Latin American survey respondents attitudes on

composite indices relating to trust in institutions, perceptions of opportunity, legitimisation of

political violence and left-right partisanship, fail to find strong evidence of a discrete middle

class identity. Typically attitudes change monotonically with income, but affluent middle

classes do not stand out as a group with particularly distinct preference structures (if there are

groups that differ systematically, it is generally the poorest or wealthiest).14 Moreover,

responses vary little across income classes with generally much larger differences between

countries (see also Cramer and Kaufman 2011). Amoranto et al. (2010: 10f) find on a global

sample from the World Values Survey that people who see themselves as middle class are

more likely to report preferences for liberal economic policies, gender equality, and higher

levels of perceived trust than lower status groups. However, this relationship is again

monotonic with more progressive values and attitudes among respondents who place

themselves in the upper class.15

All of these results thus undermine the idea that middle

classes would act as socially coherent and politically distinct groups in ways that would

establish them as agents of change within their societies.

Assets and human capital investments

The primary area where circumstances of lower middle classes begin to differ from

those of the extreme poor involves household spending on consumer assets and human

capital. Birdsall et al. (2013: 7f) find that people below PPP$10/day spend just 2436% of

their disposable income on food and other necessities, the rest going to middle class goods

like appliances or vehicles (by comparison poor households spend an estimated 7090% of

expenditure on food). Banerjee and Duflo (2008) also report strong differences between

households below the absolute PPP$2 poverty line and vulnerable groups with respect to

dwelling characteristics. In particular in the upper regions of the PPP$210/day income

14 Lopez-Calva et al. (2011) find that middle classes appear less willing to support individual rights even though they show higher values of social tolerance. 15 Both studies deal with missing survey information on household income by estimating regression models that predict permanent household income (Lopez Calva et al. 2011) and subjective class status (Ameranto et al. 2010) on the basis of observed household attributes like assets or educational and occupational outcomes.

14

bracket, households are much more likely to have better access to basic sanitation, electricity

and bigger houses than the extreme poor.16

Households in the vulnerable income bracket further begin to care more about their

physical wellbeing and the health of their household members. Banerjee and Duflo (2008)

show that the vulnerable and middle-income groups spend more in absolute and even in

relative terms on health than the poor. These greater health investments are reflected in

higher life expectancy,

Better access to electricity also goes along with a

higher likelihood of television ownership, which could influence behaviours and aspirations

by improving access to news and information.

17 and increased insurance coverage (most of which for health).18

There is also evidence of increased education spending that could eventually offset

school-related differences between vulnerable and affluent middle-income groups. Banerjee

and Duflo (2008) find that the share of household expenditures devoted to education remains

roughly similar with rising incomes. This not only implies a net increase in absolute amounts

spent on schooling. Investments in human capital also tend to be concentrated among fewer

children. The same study finds that the number of children per adult woman in the household

drops by at least 0.5 when households move from extreme poverty to the PPP$24/day

category and again by about the same amount when households move from the $24/day

category to the $610/day category (the number of children actually drops by more than one

in Nicaragua, Pakistan and Peru). At the same time, the share of children aged 1318 in

school and the number of children in higher quality private schools increases dramatically as

income rises. This suggests that income growth, in combination with accompanying

demographic transformations, leads to changes in human capital investments among

vulnerable groups that increasingly resemble the patterns commonly associated with middle

class development. It appears that even households who are still in relatively precarious

circumstances are making very conscious decisions to improve the prospects for upward

mobility for their children.

Taken together all of these trends are consistent with our theory that future political

organization among upwardly mobile lower income groups is likely to revolve more around

the protection of economic and property rights than around conventional (work-related)

class interests. The increased ownership of assets and human capital creates strong incentives

16 This may have to do with the fact that many lower/middle class households live in urban areas, but the trends broadly hold for rural areas as well. 17 This result also holds in a small number of panel surveys analysed by the authors. 18 However, coverage barely extends to one third in most rural samples and two thirds in urban areas (Banerjee and Duflo 2008, Table A15).

15

for this group to obtain credible guarantees that their hard-earned gains in private wealth will

be protected in the future (see also Ansell and Samuels 2010). At the same time, the high

level of economic risk and increased investments in insurance suggest that these groups are

averse to the uncertainties associated with rapid economic change. The following sections

will review the consequence of this development for political preference formation and

aggregate political outcomes.

4. Consequences of lower middle class growth for redistribution preferences Changing living standards of upwardly mobile populations are not only of interest in

their own right. They also have potentially far-reaching implications for political behaviours.

According to most political economy models discussed above changes in incomes

primarily influence political outcomes through their effects on redistribution preferences.

Widely-used theories of political decision-making, like Meltzer and Richards (1981) version

of the median voter model (originally outlined by Hotelling), suggest that people on rising

incomes should gradually lose their taste for radical redistribution. Other models, such as

Benabou and Oks (2001) prospect of upward mobility (POUM) hypothesis, posit that

people who believe they or their children will have strong opportunities for upward mobility

are less likely to support redistribution in the present, because they want to lower the tax

burden for themselves and their children in the future (see also Piketty 1995). According to

the political calculus of the political economy models proposed by Acemoglu and Robinson

(2006), Boix (2003) and Easterly (2001) this should result in more peaceful and democratic

outcomes, by reducing conflict over redistribution and increasing elites willingness to share

power with the less well-off.

Against these predictions stands the possibility of new social divides, related to

uneven experiences of relative inequality and the comparatively precarious living conditions

of the large populations on lower middle incomes. Relative deprivation theories, like

Hirschmans (1973) Tunnel Paradox, suggest that rapid economic transformation and

lower-income growth would lead to heightened concerns about relative downward mobility.

This applies in particular to more affluent middle classes, for whom fears of falling behind

begin to outweigh positive expectations about the future. As a result these groups would be

less able, or willing, to play the mediating function attributed to them in the political

economy literature on middle class.

Drawing on the discussion above there are also reasons to assume that, among lower

income groups, attitudinal effects associated with positive upward mobility prospects will be

16

crowded out by persistent experiences of absolute deprivation. In particular for households

just above the poverty line, who face a realistic risk of sliding back into absolute poverty,

high levels of economic insecurity and limited occupational prospects make it more likely

that these groups will demand more redistribution and/or protection, even if its members

perceive higher prospects for upward mobility. The outcome would be deeper divides

between affluent middle classes and lower-income groups and higher risk of political

conflict.

Extant scholarship has found it notoriously challenging to provide empirical support

for the predictions of standard political economy models like the median voter theorem.

These difficulties generally already start with the explanation of individual redistribution

preferences (Kenworthy and McCall 2007; Lbker 2007). In particular in emerging regions

with traditionally high levels of inequality, like Latin America, people often do not report

strong demands for redistribution (Blofield and Luna 2011; Cramer and Kaufman 2011;

Kaufman 2009, 2009b). Evidence is also weak for more specific predictions about the effect

of individual incomes on redistribution preferences. In most cases relative household income

predicts demand for redistribution less well in countries with lower levels of economic

development and deeper inequalities (Ansell and Samuels 2011; Cramer and Kaufman 2011;

Dion and Birchfield 2010; Haggard et al. 2013). .

There are some indications that social experiences associated with recent lower

middle income growth may have something to with the poor predictive performance of

standard political economic models in emerging economies. The first finding that emerges

from this literature is that, even in countries where vulnerable groups account for the majority

of populations in the middle of the local income distribution, self-assessments are still

influenced by the largely precarious nature of living conditions of populations closer to the

extreme poverty line. For instance Ferreira et al. (2013), who compare economic and

subjective class categories in Latin America, fail to find strong evidence of an emerging

middle class identity among vulnerable lower income groups. In perception surveys from

Colombia, Mexico and Peru there is a remarkable correspondence between the widely used

vulnerability-threshold of PPP$10 and peoples self-placement as belonging to the lower-

income groups or lower middle classes.19

19 More specifically the authors find that approximately PPP$10/day emerge as the threshold below which people consider themselves as belonging to lower or lower middle classes than other (wealthier) income groups. However, this subjective threshold rises in wealthier countries like Brazil and Chile (the cut-offs are PPP$16 and PPP$20/day respectively), suggesting that relative living standards have some role in these assessments.

Other studies also indicate that psychological and

social perceptions in the vulnerable group are strongly influenced by the experience of

17

economic insecurity. Yuan et al. (2010) find for China that, while lower middle class status is

generally associated with small increases in life satisfaction, the group on incomes from

PPP$210/day still reports above average concerns about the stability of life in the future.20

For wealthier groups subjective status assessments overlap more closely with material

attributes typically associated with middle class status. Amoranto et al. (2010: 8f) find in a

large cross-national sample of responses from the World Value Surveys that higher

educational attainment and more skilled occupations are associated with higher probabilities

of self-reporting a higher class status. This is consistent with the results above, which suggest

that more secure formal types of employment and better schooling are still more common

among affluent middle classes than among lower income groups.

There are also signs that different social experiences contribute to growing divides

between lower and affluent middle income groups. Consistent with the relative deprivation

argument, attitudes of affluent middle classes increasingly appear to shift from a

preoccupation with material outcomes to relative status concerns and other qualitative aspects

of wellbeing. For example, the aforementioned study by Yuan et al. (2010) finds that Chinese

respondents whose incomes approach upper middle class status (PPP$610) also reported

considerably lower values of self-perceived health. While it is not clear if this reflects actual

health problems or changes in sensitivity to physical wellbeing, the results are remarkably

consistent with the well-documented finding for developed economies that money

increasingly loses ground to qualitative aspects in the wellbeing evaluations of individuals on

higher incomes (Easterlin, 1974). In a similar vein Graham and Pettinato (2002) find for

Latin America that, while absolute income levels matter more for the subjective wellbeing of

lower-income groups, relative differences become much more important for those in more

comfortable economic circumstances. They also identify a large group of frustrated

achievers in Peru, who, despite experiencing significant improvements in their incomes,

remain unsatisfied with their relative status in society. They speculate that this has to do with

the fact that these individuals compare themselves to groups higher up on the income ladder.

Ravallion and Lokshin (2000) report for Russia that non-poor households who feared a fall in

their incomes supported stronger redistributive policies. Both of these studies use panel data

and are therefore able to make very precise statements about individual mobility

trajectories.21

20 This holds for the urban sample only.

21 However, Ravallion and Lokshin cannot account for initial redistribution preferences.

18

By contrast, preferences of poorer and vulnerable households still appear to be

dominated by the largely precarious nature of these groups living conditions. While many

vulnerable households probably experienced periods of upward mobility, and even though

their investment in insurance and human capital of their children reflect positive expectations

about the future, it does not appear that this group is willing to trade off potentially lower

future taxes for its current preferences for redistribution. Cramer and Kaufman (2011) find

that, even though redistribution preferences of poorer and middle-income groups defined by

an asset index- did not respond in the expected way to average inequality, their satisfaction

with distributional outcomes was significantly reduced in times of high growth.22

Recent work by Wietzke (2014) underlines that these differences between lower and

middle-income groups can contribute to very strong disagreements over social policy

priorities. This study overcomes the problem of lacking income data that has affected the two

aforementioned studies by tracking changes in average support for redistribution along with

transformations in poverty headcounts and growth in various lower/middle-income groups. It

finds that public support for redistribution increased primarily in societies where poverty

reduction was accompanied by growth in the share of economically vulnerable groups up to a

poverty threshold of around PPP$4 a day. By contrast, the estimated association with

distribution preferences gradually disappears in societies that experienced growth among

more affluent middle-income households.

Their

findings are also broadly consistent with the work of Reenock et al. (2007), who show that

democratic breakdowns are more likely to occur in rapidly growing societies that still have

large shares of households with unmet basic needs.

There is also support for our hypothesis that tensions arise increasingly over the

political representation of the growing economic aspirations of upwardly lower middle

income groups. Lopez-Calva et al. (2011) find for Latin America that, while respondents on

incomes of around PPP$ 4-10 report more positive perceptions of economic opportunity than

the poor, lower middle classes generally do not appear to have much trust in the political

institutions of their countries. Of all classes considered only more affluent groups report

significantly weaker feelings of political alienation and higher levels of trust in the political

system. Somewhat worryingly lower middle classes are also more likely than the poor and

the wealthy to express a preference for order and stability at the expense of individual rights.

22 The authors hypothesize that feelings of relative deprivation explain this result, but economic risk and volatility may also play a role.

19

At the same time there are indications that conflicts along economic class line still

interact quite strongly with other salient social identities in emerging economies. Haggard

et al. (2013) document that unskilled and semi-skilled workers and those residing in large

cities are more inclined to support redistribution. This result is broadly consistent with recent

events in Egypt, Thailand, or Turkey, where protests by urban middle classes seem to be as

much concerned with economic and social policy as with questions of cultural identity and

the growing influence of upwardly mobile but less liberal rural populations (New York

Times, 2 June 2013). Interestingly Haggard et al. also find lower support for redistribution

among unionised workers pointing to possible insider-outsider conflicts between those within

and outside of the formal sector. The literature on ethno-linguistic fractionalization similarly

has documented often considerable differences in political preferences across groups

(Habyarimana et al. 2007; Lieberman and McClendon 2012).

5. Consequences of lower middle class growth for political and social reforms Turning to aggregate outcomes, the literature has produced similarly little evidence on the

postulated link between distributional change and democratic development. Studies of the

relationship between democracy and inequality generally suggest that large differences in

incomes are more likely to be accompanied by instability and reversions to autocratic rule

(Dutt and Mitra 2008; Houle 2009). However, there is only mixed evidence that falling

inequality leads to democratic transition, with positive results (Boix 2003) outweighed by

contradictory or ambiguous ones (Teorell 2010; Haggard and Kaufman 2012). Ansell, and

Samuels (2010), in the study that resembles our framework more closely, find that land

concentration reduces the probability of democratization, while rising income increases it.

They hypothesize that rising income inequality reflects the emergence of new upwardly

mobile classes that demand stronger protection of their economic property rights from landed

elites, a theory that is broadly consistent both with the three class model of Boix and

Acemoglu and Robinson (Haggard and Kaufman 2012: 500) and the theoretical arguments of

this paper.

In the context of recent lower middle income growth the weakness of empirical

evidence is compounded by a general lack of detailed information on the growth rates of

specific income groups. Beyond income or land inequality, cross-national studies of political

transition and democratisation processes have typically focused on relatively general

measures of economic development, such as per capita GDP, or average indicators of basic

20

needs satisfaction like mean caloric consumption (Reenock et al. 2007) or child mortality

(Gerring et al. 2012; Ross 2006). However, these measures are not sufficiently disaggregated

to test hypotheses about the effects of growth in specific lower or middle income segments

(Wietzke 2014).

In spite of this limitation a small number of studies have tried to link growth of

middle incomes to social and political outcomes. Easterly (2001), in his study on the middle

class consensus, finds that a higher share of income going to groups between the 20th and 80th

percentiles of income distribution is empirically related to better economic policies, greater

political stability, as well as higher income and higher growth. Larger middle class income

shares are also associated with better educational and health outcomes, better infrastructure,

and more democratic governance systems. Based on the historical distribution of incomes in

most low-income countries it is fair to assume that this captures a majority of poor and

vulnerable groups in the developing world (Banerjee and Duflo 2008). However, Easterlys

study generalises across a large group of countries at very different stages of development.

As a consequence it does not provide a direct test of the effect of lower middle-income

growth in emerging economies.23

Cobham and Sumner (2013a; 2013b), with a slight shift in focus, concentrate on the

ratio of incomes held by elites relative to the poorest groups. They draw on Palmas (2006;

2011) proposition that changes in distribution are determined by the top decile and poorest

four deciles, as the middle of decile 5 to decile 9 capture half of GNI in any given country.

They corroborate Palmas finding on the surprisingly stable capture of decile 5 to decile 9

across countries and note (in Cobham and Sumner, 2013a) that rising relative incomes of the

poorest groups are also associated with improvements in non-monetary outcomes like clean

water supply and lower child mortality, hinting at greater public investments in these areas.

However, this is a statistical correlation rather than a policy explanation.

Probably the most detailed evidence about the consequences of lower and middle

class growth comes from Loayza et al. (2012). They predict changes in government health

and education spending and governance quality over a sample of 670 annual observations

across approximately 130 countries. The analysis distinguishes as main explanatory variables

between the proportion of individuals in extreme poverty (below PPP$2.5 a day), populations

on PPP$ 2.5 to 10 per day, and the proportion of individuals who have reached middle class

23 Easterly also includes ethno-linguistic fractionalisation in his operationalization of the middle class consensus.

21

status above PPP$10 a day.24

The results suggest that growth in the vulnerable income

range (PPP$ 2.5-10) does not lead to significantly different results than changes in the share

of poorer households below PPP$2.5. By contrast there are much stronger signs for

improvements when growth occurs at higher income levels. Expansions of incomes above

PPP$10 have a robust positive influence on social policy and governance outcomes that

reduces or entirely crowds out estimated effects of GDP obtained in earlier specifications.

These changes in social policy also do not imply a move towards a more state-driven

economy, as increased upper incomes are associated with reduced tariffs on international

trade and a higher degree of financial liberalisation. It is important to note, however, that it is

not possible to conclude from this study that countries that implemented successful reforms

benefited from a middle class dividend. Because of the difficulty of comparing middle class

incomes across emerging and advanced economies, the authors do not use an upper income

ceiling. This makes it hard to determine if there is a genuine middle class effect or whether

the estimates just pick up positive institutional correlates of growth occurring in higher

incomes.

Evidence from contextualized and case study research

In practice, stronger insights about the link between recent lower middle income

growth and regime change are often obtained from more contextualized analysis. In most

emerging economies economic and political transformations have to be interpreted in the

context of often extreme swings in national political and macro-economic environments. In

particular the third wave of democratisation, which swept over Latin America, Eastern

Europe and parts of Asia and Africa in the 1980s and 1990s often coincided with radical

market reforms and cuts to social expenditures during the period of macro-economic

adjustment and the Asian crisis of the late 1990s (Haggard and Kaufman 2008; Huber et al.

2008; Huber and Stephens 2012).25

On the other hand the historical background of economic adjustment probably offset

many of the postulated mechanisms between inequality and political outcomes. Large-N

This, on the one hand, increased inequalities and shifted

the burden of social assistance for lower-income households to informal and family networks,

which were ill-equipped for providing protection against systematic macro-economic shocks

(Cook and Kwon 2007; Birdsall et al. 2013).

24 Incomes of PPP$ 2.5-10 are the omitted category. The estimates use the Arellano and Bond estimator for panel data. 25 The Asian crisis also affected Russia and parts of Latin America.

22

comparative analysis by Haggard and Kaufman (2012) suggests that distributive conflict,

defined broadly as elite responses to the mobilization of redistributive grievances on the part

of economically disadvantaged groups (p. 500) and backed by various measures of income

and asset inequality, was present in just over half of 84 country-events of regime changes

between 1980 and 2000. Contrary to expectations, a substantial number of transitions took

place at high levels of inequality, while reversions to autocracy were often driven by conflicts

that cut across class lines or by intra-elite or military disputes (see also Teorell 2010).

Social policy reforms that were implemented across developing countries after the

adjustment period also only partially coincide with expectations about the political-

institutional mediation of distributional conflicts. Quantitative and comparative analysis for

Latin America finds a robust link between democratic reforms and the region-wide shift to a

new post-adjustment consensus (Roberts 2012), including increases in social spending and

the introduction of poverty alleviation programmes like Conditional Cash Transfers (Birdsall

et al. 2011, Daz-Cayeros and Magaloni 2009; Haggard and Kaufman 2008). However, these

reforms were implemented by governments across the left-right spectrum, undermining

popular theories about the link between political partisanship and welfare state generosity

(Huber et al. 2008; Roberts 2012).

In other regions autocratic governments seem to have addressed social grievances of

lower middle income groups without explicit commitment to political power sharing. In

particular examples from fast growing Asian economies suggest that regimes with widely

different sources of political support and legitimacy have begun to pass on benefits of rapid

economic growth to expanding lower and middle income groups without engaging in

accompanying democratic reforms. This is best exemplified by governments in China and

Singapore whose authoritarian regimes engaged in relatively generous expansion to broaden

their political support base (Haggard and Kaufman 2008; see also Cook and Kabeer 2010;

Cook and Kwon 2007).

The only region where social policy outcomes appear to conform more closely to

predictions of institutionalist theories is Eastern Europe. In this region the legacy of

expansive socialist social protection frameworks appears to have created broad-based vested

interest groups that were able to resist excessive cuts to social expenditures during adjustment

and push for an expansion of safety nets during the recent recovery period (Haggard and

Kaufman 2008).

Notwithstanding these reforms it can be questioned whether current social protection

systems do enough to secure lower middle-income groups future commitment to existing

23

political arrangements. A well-known limit on government redistribution is that many

emerging nations collect very low levels of progressive tax (such as income or wealth taxes).

This means that even in countries where social safety nets are put in place, poorer populations

contribute disproportionally to the funding of these programmes (Lindert et al. 2006, Birdsall

et al. 2013).26 Another problem is the persistent dominance of regressive social spending on

state-subsidized contributory social insurance and pensions in many emerging economies

(Huber 2009: 652; see also de Ferranti et al. 2004, 26872; Lindert et al. 2006; Cook and

Kabeer 2010). In particular vulnerable households in self-employment or informal sector jobs

are less likely to benefit from these programmes that tend to serve formal sector employees

(Cook and Kabeer 2010; Haggard and Kaufman 2008). Moreover, vulnerable populations just

above official poverty lines often fall outside the targeting criteria of new social protection

programmes put in place to address the need of the least well-off. For instance Birdsall et al.

(2013) find that, compared to poor households, vulnerable populations in Latin America on

incomes between PPP$ 4 and 10 benefit relatively little from means-tested conditional cash

transfers (CCTs).27

Recent events in emerging economies suggest that these imbalances can be a source

of considerable tension along class and social lines. Examples like Venezuela or Thailand

indicate that grievances of vulnerable groups may be exploited by populist leaders, who opt

for unsustainable expansion of social expenditures to strengthen their support base. Violent

clashes between opponents and supporters of these programmes in both countries suggest that

this may exacerbate existing divides between classes, as more affluent groups resist the high

economic costs of these policies.

At the same time there are again indications that political transformations in the

context of rapid lower middle class growth will not be exclusively driven by distributional

conflicts. The growing incidence of public protests against high levels of perceived

corruption in many developing regions suggests that grievances of upwardly mobile groups

are as much about illegitimate demands of governments and bureaucrats on private property

26 Birdsall et al. 2013) find that vulnerable groups in Latin America on incomes from PPP$410/day are net payers into the fiscal system once indirect taxes are taken into account. The only area where this group benefits are health and education services, though no more or less than other income groups given its share of the overall population (p.17). However, the fact that even lower-income vulnerable parents increasingly opt for private education suggests that the poor quality of many public schools acts as a strong deterrent (Banerjee and Duflo 2008; Birdsall et al. 2013). 27 The exceptions are Uruguay and Brazil, whose higher living standards mean that vulnerable groups still fall under local targeting criteria. In these countries, as well as in Guatemala, vulnerable populations also receive more benefits from new non-contributory pension systems than more affluent middle classes (Birdsall et al. 2013).

24

as they are about social uncertainties caused by rapid growth and economic vulnerability.

This is consistent with our claim above that the growth in discretionary consumer spending

and asset accumulation among lower middle income groups increasingly results in new

demands for the protection economic property rights. However, the long history of autocratic

strong men in emerging economies, and the willingness of some lower middle income

respondents to trade-off political freedoms against greater political stability does not suggest

that these new demands will always result in a systematic shift to more democracy. At least

in the short term many of these demands may be exploited by populist autocratic leaders.

6. Conclusions and research questions arising This paper has reviewed consequences of recent lower middle class growth for social

and political development in emerging economies. We have illustrated how the objective and

subjective realities of emerging economies middle income groups differ from established

notions of middle and working class status in the welfare regime and political transition

literature. We have also considered recent political transformations that differ from the

predictions of standard political economy models of middle class development.

One of the worrying findings that emerges from our review is the growing divide over

social policies that appears to emerge between rapidly expanding population majorities on

lower middle incomes and more affluent middle classes. Despite recent expansion in

government social spending in many emerging economies, participation of lower middle

income groups in social protection systems is still held back by institutional barriers and the

largely informal nature of employment contracts among these populations. Evidence from

self-reported political preferences reviewed in this paper, and recent public protests over

social inequality, suggest that this is already changing political dynamics in emerging

economies. Political attempts to mitigate these problems will occur in environments where

middle classes become increasingly concerned about maintaining their advantaged status

position, shifting the burden on the ability of governments to respond effectively to shifting

population majorities and social conflict lines. Moreover, because lower middle income

groups are increasing vocal and able to spend higher shares of their income on physical assets

and human capital we predict more frequent incidences of political conflict around

government accountability and the protection of individual economic property rights.

Other results of our review require more careful investigation. There are some

indications that distributional changes associated with lower middle class growth interact

strongly, and in often unexpected ways, with other salient divides such as urban-rural

25

differences or ethnicity. One fruitful area for future research would be to combine analysis of

the evolution of lower middle class interests with these cultural categories. This work would

be consistent with recent attempts by political scientists and economists to move away from

purely social-identity based measures of political behaviour (e.g. indices of ethnic

polarisation) to more complex models that account for group-level economic inequalities

(Baldwin and Huber 2010, Huber 2013; Haggard et al. 2013; Stewart et al. 2008).

Lastly there is still considerable room for progress in improving the strength of

empirical evidence on the political consequences of recent lower middle income growth. One

striking trend in the political science literature, which we have aimed to redress with this

review, is the almost complete lack of consideration of newly available, detailed information

on the rapid evolution of new income groups in emerging economies We believe that this can

be rectified quite easily by incorporating new evidence from micro-surveys in the way we

have attempted in this article.

Another area for improvement, which cuts across disciplines, is the lack of hard

evidence on how individual mobility experiences influence behaviour of lower-income

groups. As noted before rigorous analysis of the effects of upward (or downward) mobility

experiences on political behaviours is still often constrained by the lack of detailed panel

surveys that permit tracing over time how attitudes of the individuals change with transitions

into new income classes. Again this shortcoming can be addressed through more systematic

collaboration across disciplines. For example indicators on political preferences and

behaviours could be relatively easily introduced into new expenditure panel surveys that are

increasingly used by economists to investigate economic determinants of transitions out of

poverty.

26

Bibliography Acemoglu, D. and Robinson, J.A. (2006). Economic Origins of Dictatorship and Democracy.

New York: Cambridge University Press.

Acemoglu, D., & Zilibotti, F. (1997). Was Prometheus Unbound by Chance? Risk,

Diversification, and Growth. Journal of Political Economy 105(4), 70951.

African Development Bank (AfDB) (2011). The Middle of the Pyramid: Dynamics of the

Middle Class in Africa, AfDB, Market Brief.

Alesina, A. and La Ferrara, E. (2005). Preferences for Rredistribution in the Lland of

Oopportunities. Journal of Public Economics, 89(5-6), 897-931.

Amoranto, G., Chun, N. and Deolalikar, A. (2010). Who are the Middle Classes and What

Values Do They Hold? Economics Working Paper Series No 229, Mandaluyong City:

Asian Development Bank.

Ansell, B. and Samuels, D. (2010). Inequality and Democratization: A Contractarian

Approach. Comparative Political Studies, 43 (12). 15431574.

Ansell, B. and Samuels, D. (2011). Inequality and Democratization: Individual-Level

Evidence of Preferences for Redistribution under Autocracy. Paper presented at the

Annual Political Science Association, Seattle.

Asian Development Bank (ADB) (2010). Key Indicators, Part I: Special chapter on The Rise

of Asias Middle Class, Mandaluyong City, Philippines.

Baldwin, K. and Huber, J. D (2010). Economic versus cultural differences: Forms of ethnic

diversity and public goods provision. American Political Science Review, 104(4),

644-662.

Banerjee, A. and Duflo, E. (2008). What is Middle Class about the Middle Classes around the

World? Journal of Economic Perspectives 22(2), 328.

Barro, J. R. (1999). Determinants of Democracy. Journal of Political Economy, 107(6), 158-

183.

Benabou, R. and Ok, E.A. (2001). Social Mobility and the Demand for Redistribution: The

POUM Hypothesis. The Quarterly Journal of Economics, 116, 44787.

Birdsall N., Graham, C., and Pettinato, S. (2000). Stuck in the Tunnel: Is Globalization

muddling the middle class? Brookings Institution Working paper 14. Washington DC:

Brookings Institution.

Birdsall, N., Lustig, N. and Meyer, C. (2013). The Strugglers: The New Poor in Latin

America? Center for Global Development Working Paper. Washington DC: CGD.

Birdsall, N., Lustig, N., and McLeod, D. (2011). Declining Inequality in Latin America:

27

Some Economics, Some Politics. Center for Global Development Working Paper 251.

Washington DC: CGD.

Blofield, M. & Luna, J.P. (2011). Public Opinion on Redistribution in Latin America; in

Blofield, M. (Ed.), The Great Gap: Inequality and the Politics of Redistribution in

Latin America. University Park: Pennsylvania State University Press, 147-183.

Boix, C. (2003). Democracy and Redistribution. New York, NY: Cambridge University

Press.

Castellani, F. and Parent, G. (2011). Being "Middle-Class" in Latin America, OECD

Development Centre Working Papers, No. 305, OECD: Paris

Chun, Nathalie (2010). Middle Class Size in the Past, Present, and Future: A Description of

Trends in Asia, ADB Economics Working Paper Series 217, Asian Development

Bank.

Cobham, A. and Sumner, A (2013a). Putting the Gini Back in the Bottle. Mimeo. Kings

College London.

Cobham, A. and Sumner, A (2013b). Is It All About the Tails? The Palma Measure of

Income Inequality. CGD Working Paper. CGD: Washington, DC

Cook, S. and Kabeer, N. (eds) (2010). Social Protection as Development Policy: Asian

Perspectives. London: Routledge.

Cook, S. and Kwon, H.C. (2007). Social Protection in East Asia, Global Social Policy, 7 (2),

223229.

Cramer B.D. and Kaufman R.R. (2011). Views of Economic Inequality in Latin America.

Comparative Political Studies, 44(9), 120637.

Davis, B, Winters, P, Carletto, G. et al. (2010). A Cross-Country Comparison of Rural

Income Generating Activities. World Development, 38 (1), 4863.

de Ferranti, David, Perry, Guillermo E., Ferreira, Francisco H.G. and Walton, Michael

(2004). Inequality in Latin America: Breaking with History? Washington, D.C.: The

World Bank.

Daz-Cayeros, A. and Magaloni, B. (2009). Aiding Latin Americas Poor, Journal of

Democracy, 20(4), 36-49.

Dion, M.L. and Birchfield, V. (2010). Economic Development, Income Inequality, and

Preferences for Redistribution, International Studies Quarterly 54, 315334.

Doepke, M., and Zilibotti, F. (2007). Occupational Choice and the Spirit of Capitalism.

CEPR Discussion Paper DP6405. London: CEPR.

28

Dutt, Pushan, and Devashish Mitra. 2008. Inequality and the Instability of Polity and

Policy. Economic Journal 118 (531): 12851314.

Easterlin, R. A. (1974). Does economic growth improve the human lot? Some empirical

evidence, in Nations and Households in Economic Growth: Essays in Honor of

Moses Abramowitz, by Paul A David and Melvin W. Reder. New York: Academic

Press, Inc.

Easterly, W. (2001). The Middle Class Consensus and Economic Development, Journal of

Economic Growth, 6, 317-335.

Easterly, W., Ritzen, J., Woolcock, M. (2006). Social Cohesion, Institutions and Growth.

Economics and Politics, 18(2), 103-120.

Edward, P., and Sumner, A. (2014a) Estimating the Scale and Geography of Global Poverty

Now and in the Future. World Development 58: 67-82.

Edward, P., and Sumner, A. (2014b) The Poor, the Prosperous and the inbetweeners: A

Fresh Perspective on Global Society, Inequality and Growth. UNDP IPC-IG Working

Paper 122. Brasilia: UNDP IPC-IG

Ferreira, F.H.G., J. Messina, J. Rigolini, L.-F., Lopez-Calva, M.A., Lugos and R. Vakis

(2013). Economic Mobility and the Rise of the Latin American Middle Class.

Washington, DC, World Bank.

Frank, R. H. (2007). Falling Behind: How Rising Inequality Harms the Middle Class.

California: University of California Press.

Gerring, J., Thacker, S.C. and Rodrigo, A. (2012). Democracy and Human Development, The

Journal of Politics, 74 (1), 117.

Goldman Sachs (2008). The Expanding Middle: The Exploding World Middle Class and

Falling Global Inequality.

Goldthorpe, J.H., McKnight, A. (2004). The Economic Basis of Social Class. CASE Paper

80. London School of Economics.

Graham, C. and Pettinato, S. (2002). Happiness and Hardship. Opportunity and Insecurity in

New Market Economies. Washington D.C: Brookings Institutions Press.

Grusky, David B. and Kanbur, Ravi (2006), 'Introduction: The Conceptual Foundations of

Poverty and Inequality Measurement', in David B. Grusky and Ravi Kanbur (eds.),

Poverty and Inequality (Stanford CA: Stanford University Press), 1-29.Grusky and

Kanbur 2006.

29

Habyarimana, J., Humphreys, M., Posner, D., & Weinstein, J. (2007). Why does ethnic

diversity undermine public goods provision? American Political Science Review, 101,

709725.

Haggard, S. and Kaufman, R.R. (2008). Development, Democracy, and Welfare States: Latin

America, East Asia, and Eastern Europe, Princeton: Princeton University Press.

Haggard, S., Kaufman, R.R. and Long, J.D. (2013). Income, Occupation, and Preferences for

Redistribution in the Developing World, Studies in Comparative International

Development, 48, 113140.

Hirschman, A.O. (1973). The Changing Tolerance for Income Inequality inthe Course of Economic Development, The Quarterly Journal of Economics, 87, 544-66.

Houle, Christian 2009. Inequality and Democracy: Why Inequality Harms Consolidation but

Does Not Affect Democratization, World Politics 61 (4): 589622.

Huber E., Mustillo, T. and Stephens, J. (2008). Politics and Social Spendingin L atin America. The Journal of Politics, 70 (2), 420-436.

Huber, E. (2009). Politics and Inequality in Latin America, Political Science and Politics

42(4): 651655.

Huber, E. and Stephens, J. (2012). Democracy and the Left. Chicago: Chicago University

Press.

Huber, J.D. (2013). Ethnicity or Class? Identity Choice in Elections and its Implications for

Redistribution. Working Paper.

Kaufman, R. R. (2009b). Inequality and Redistribution: Some Continuing Puzzles. Political

Science and Politics, 42 (2), 657-660.

Kaufman, R.R. (2009). The Political Effects of Inequality in Latin America: Some

Inconvenient Facts. Comparative Politics, 41: 35979.

Kenworthy, L. and McCall, L. (2007). Inequality, Public Opinion, and Redistribution. Socio-

Economic Review, 6 (1), 35-68.

Kharas, H. (2010). The Emerging Middle Class in Developing Countries. OECD

Development Center Working Paper 285. Washington DC: Brookings Institution.

Kharas, H. and Gertz, G. (2010). The New Global Middle Class: A Cross-Over from West to

East. Wolfensohn Center for Development.

Korpi, W. (1983). The Democratic Class Struggle. London: Routledge

Landes, D. (1998). The Wealth and Poverty of Nations. New York: Norton

30

Leventoglu, Bahar (2013). Social Mobility, Middle Class, and Political Transitions, Journal

of Conflict Resolution published online 1 April 2013, Doi.

0.1177/0022002713478563

Lieberman, E. S., & McClendon, G. H., (2012). The ethnicitypolicy preference link in sub-

Saharan Africa. Comparative Political Studies, 46, 574602.

Lindert, Kathy, Emmanuel Skoufias and Joseph Shapiro. 2006. Redistributing Income to the

Poor and the Rich: PublicTransfers in Latin America and the Caribbean. Washington,

DC: World Bank.

Lindert, P. (2004). Growing Public, Cambridge: Cambridge University Press.

Lipset, S.M. (1959). Some Social Requisites of Democracy: Economic Development and

Political Legitimacy. American Political Science Review 53(1): 69105.

Lipset, S.M. (1960). Political Man. The Social Bases of Politics. New York: Doubleday.

Loayza N., Rigolini, J. and Llorente, G. (2012). Do Middle Classes Bring About Institutional

Reforms? Economic Letters 116: 440-444.

Lpez-Calva, L. F. and E. Ortiz-Juarez (2014). A Vulnerability Approach to the Definition of

the Middle Class. Journal of Economic Inequality 12(1): 23-47.

Lopez-Calva, L.F., Rigolini, J. and Torche, F. (2011). Is There Such Thing as Middle Class

Values? Class Differences, Values and Political Orientations in Latin America. Policy

Research Working Paper no 5874, Washington DC: World Bank.

Lbker, M. (2007). Inequality and the Demand for Redistribution: Are theAssumptions of the New Growth Theory Valid? Socio-Economic Review, 5(1 ), 1 17 -48.

Mallet, R. and Sumner, A. (2013). The Future of Foreign Aid: Development Cooperation and

the New Geography of Global Poverty. Basingstoke: Palgrave Pivot.

McGuire, J.W. (1994). Development Policy and Its Determinants in East Asia and Latin

America. Journal of Public Policy 14 (2), 205-242.

McGuire, J.W. (1999). Labor Union Strength and Human Development in East Asia and

Latin America. Studies in Comparative International Development 33 (4), 3-34.

Meltzer, A.H. and Richard, S.F. (1981). A Rational Theory of the Size of Government.

Journal of Political Economy, 89 (5), 914927.

Milanovic, B., and Yitzhaki, S. (2002). Decomposing World Income Distribution: Does the

World have a Middle Class? World Bank Policy Research Working Paper 2562,

Washington DC: World Bank (also published in Review of Income and Wealth 48(2),

15578). Moore, Barrington. (1966). The Social Origins of Dictatorship and Democracy. Boston: Beacon Press.

31

Palma, J.G., (2006) Globalizing inequality: Centrifugal and centripetal forces at work

DESA Working Paper 35, New York: UN Department of Economic and Social

Affairs.

Palma, J.G., (2011). Homogeneous middles vs. heterogeneous tails, and the end of the

Inverted-U: The share of the rich is what its all about, Cambridge Working Papers

in Economics 1111, Cambridge: University of Cambridge Department of Economics

(later published in Development and Change, 42, 1, 87-153).

Piketty, T (1995). Social Mobility and Redistributive Politics. The Quarterly Journal of

Economics, 110(3), 551-584.

Przeworski, A. (1985). Capitalism and Social Democracy. New York: Cambridge University

Press.

Ravallion, M. (2010). The Developing World's Bulging (But Vulnerable) 'Middle Class'.

World Development 38 (2), 445454.

Ravallion, M. (2012). Why Don't We See Poverty Convergence? American Economic

Review 102(1): 50423.

Ravallion, M. and Lokshin, M. (2000). Who Wants to Redistribute? The TunnelEffect in 1990s Russia, Journal of Public Economics, 76(1), 87104.

Reenock, C., Bernhard, M. and Sobek, D. (2007). Regressive Socioeconomic Distribution

and Democratic Survival. International Studies Quarterly 5: 67799.

Roberts, K.M. (2012). The Politics of Inequality and Redistribution in Latin Americas Post-

Adjustment Era. Working Paper 2012/008 Helsinki, UN WIDER.

Ross, M. (2006): Is Democracy Good for the Poor? American Journal ofPolitical Science, 50(4), 860-874.

Rueschemeyer, Dietrich, Stephens, Evelyn H. and Stephens, John D. (1992). Capitalist

Development and Democracy. Chicago: University of Chicago Press.

Stewart, F., G. Brown, et al. (2008). Policies towards horizontal inequalities. Horizontal

Inequalities and Conflict: Understanding Group Violence in Multiethnic Societies.

London, Palgrave: 301-326.

Sumner, A. (2012). Poverty, Vulnerability and Class: The Expanding Non-Polar Groups and

Development, Public Administration and Development 32 (45): 444454.

Sumner, A., Yusuf. A, Suara, Y. 2014, forthcoming. Prospects for the Poor Poverty Lines: A

new set of poverty lines based on the probability of remaining poor (or not) in

Indonesia. Mimeo

32

Teorell, Jan. 2010. Determinants of Democratization: Explaining Regime Change in the

World, 19722006. Cambridge: Cambridge University Press.

Therborn, Goran. (1977). The Role of Capital and the Rise of Democracy. New Left

Review 103 (MayJune): 341.

Therborn, Goran. (1979). The Travail of Latin American Democracy. New Left Review

113114 (JanuaryApril): 71109.

Torche, F. and Lpez-Calva, L. F. (2012). Stability and Vulnerability of the Latin American

Middle Class. In Katherine Newman (Ed.) Dilemmas of the Middle Class around the

World. New York: Oxford University Press.

Wietzke, F.B. (2014). Kicking Away the Ladder? Poverty Reduction and Public Support for

Redistribution in Emerging Economies. Paper presented at the Annual meeting of the

European Political Science Association, Barcelona 2013.

Wilson, William T. (2013). Hitting the Sweet Spot: The Growth of the Middle Class in

Emerging Markets. London: Ernst and Young.

Yuan, Z., Wan, G. and Khor, N. (2010). The Rise of the Middle Class in the Peoples

Republic of China. Economics Working Paper Series No 247, Mandaluyong City:

Asian Development Bank.

International Development Institute Working Paper 2014-03/King's International Development InstituteRoom 8C, Chesham Building