Economic Mobility and the Rise of the Latin American ... · Note: * denotes final year between 2004-2006 Source: SEDLAC ... one, there is great mobility and change so that the position

Economic Mobility and the Rise of the yLatin American Middle Class

Preliminary Findings

A l l hLAC Regional Flagship 2012

This presentation summarizes preliminary findings from work in progress by a team that includes Francisco H. G. Ferreira, Luis Felipe López-Calva, María Ana Lugo, Julián Messina, Jamele Rigolini, and Renos Vakis. The team has benefited enormously from substantial inputs from JoãoP d A d M i i B l G ill C P L j N L L d L h i Willi M l Ed d O iPedro Azevedo, Maurizio Bussolo, Guillermo Cruces, Peter Lanjouw, Norman Loayza, Leonardo Lucchetti, William Maloney, Eduardo Ortiz-Juarez, Elizaveta Perova, Roby Senderowitsch, Florencia Torche and Mariana Viollaz. Guidance was gratefully received from Nancy Birdsall,François Bourguignon, Gary Fields and James Foster (as well as Augusto de la Torre, Louise Cord, Phil Keefer and Ana Revenga), but the usualdisclaimer applies. We are grateful to Manuel Fernandez Sierra, Gonzalo Llorente, Owen McCarthy and Nathaly Rivera Casanova for excellentresearch assistance.

Structure of the Reportp

1 Introduction and Motivation1. Introduction and Motivation

2. Concepts and Measurement

3 I t ti l M bilit3. Inter-generational Mobility

4. Intra-generational Mobility

5. The Rise of the Latin American Middle Class

6. Implications of a Larger Middle Class:

Political economy and economic policy

7. Preliminary Conclusions

Poverty and inequality fell in the 2000s…

• The 2000s were a good decade for Latin America. Growth in GDP per capita averaged 2.0% per annum between 2000 and 2009, despite the global recession of 2008-09.

I ddi i i li f ll i 13 f h 18 i f hi h d il bl• In addition, inequality fell in 13 of the 18 countries for which data are available.

• As a result, (moderate) poverty fell by about a quarter, from 39.8% in 2000 to 29.5% in 2009. In net terms, 32 million Latin Americans left poverty over the period. p y p

4.4

4 00

6.00Change in the Gini Coefficient (points), C.2000‐2009

-1.8-1.0* -0.6* -0.4

0.5 0.71.2

2.0*

-2.1-2.00

0.00

2.00

4.00

-7.0-6.1

-5.6 -5.4-4.9 -4.5

-3.3 -3.3 -3.12.1

-3.6

-8.00

-6.00

-4.00

Note: * denotes final year between 2004-2006Source: SEDLAC

… but who escaped poverty, and where did they go?… but who escaped poverty, and where did they go?

• But 160 million people are still poor in the region.

1. How and why were some Latin Americans capable of seizing opportunities to escape poverty, while so many others were not?

i. How much does family background matter for individual achievement?

ii. What are the correlates of successful income growth – and escaping poverty?

iii Wh t li i b d i i tit ti f ilit t d bilit ?iii. What policies, programs or broader economic institutions facilitate upward mobility?

2. How are Latin American middle classes changing as a result of th f i bilit ?these processes of economic mobility?

i. Who belongs to the continent’s middle classes?

ii. Are the composition, profile, beliefs and behaviors of the middle classes changing?

iii Will this matter for the region’s future development path? If so how?iii. Will this matter for the region s future development path? If so, how?










Mobility and the middle class have long been iseen as important

“Consider two societies that have the same distribution of annual income. In one, there is great mobility and change so that the position of particular families in the income hierarchy varies widely from year to year. In the h h i i idi h h f il i h i i other, there is great rigidity so that each family stays in the same position

year after year. Clearly, in any meaningful sense, the second would be the more unequal society”

Milton Friedman (1962): Capitalism and Freedom (cited by Fields 2009)Milton Friedman (1962): Capitalism and Freedom (cited by Fields, 2009)

“It is possible for those states to be well governed that are of the kind in It is possible for those states to be well governed that are of the kind in which the middle class is numerous, and preferably stronger than both the

other two classes, or at all events than one of them…” Aristotle (c .350 BC [1932]): Politics( [ ])

But mobility means different things to different peoplepeople

• Sociological vs. economic approaches.

• Mobility of what? • Mobility of what?

– Spaces: educational attainment, achievement, incomes, consumption…

– Domains: inter-generational vs. intra-generational

– Concepts: movement (incomes, shares, positions); origin independence; equalization of long-term incomes

Concept \ Domain Inter-generational Intra-generational

Origin independence

Basic idea: equality of opportunity. Lifetime transitions.

Does your parents’ place in the distribution determine your own?

Directional income movement

Absolute progress across generations. Basic idea: individual income growth.

Who moves in and out of poverty; and in and out of the middle class. And why?

So does “middle class”So does middle class

• Large sociological literature

l f l b l• Recent literature in economics uses fairly arbitrary lines.

• We draw on an “economic security” approach, applied specifically to LAC:

Figure 2.2: Vulnerability to poverty g y p y

14

16

18

20

ome

ChileMexicoPeru

6

8

10

12

old pe

r capita in

co

0

2

4

tial daily hou

seho

Source: Lopez‐Calva and Ortiz‐Juarez (2011).

Init

Probability to fall into poverty (below 4 USD PPP/day)

So does “middle class”So does middle class

• Large sociological literature

l f l b l• Recent literature in economics uses fairly arbitrary lines.

• We draw on an “economic security” approach, applied specifically to LAC:

Figure 2.2: Vulnerability to poverty g y p y

14

16

18

20

ome

ChileMexicoPeru

6

8

10

12

old pe

r capita in

co

0

2

4

tial daily hou

seho

Source: Lopez‐Calva and Ortiz‐Juarez (2011).

Init

Probability to fall into poverty (below 4 USD PPP/day)

A meaningful and robust definition?A meaningful and robust definition?

• Possible concern about arbitrariness of the vulnerability threshold

• Independent validation from a subjective approach, using five Ecosocialsurveys (and corresponding household surveys):

Figure 2 3: Distribution of self reported class status for Mexico (2007)Figure 2.3: Distribution of self‐reported class status for Mexico (2007).0

3.0

401

.02

Den

sity

0.0

0 20 40 60 80Permanent Income (per capita daily income in USD PPP)

Note: Densities are weighted by class size. Source: Authors’ calculations based on Ecosocial (2007) and the Encuesta Nacional de Ingreso y Gastos de los Hogares (2008).

Middle to Upper Middle lower & Lower

A meaningful and robust definition?A meaningful and robust definition?

• Possible concern about arbitrariness of the vulnerability threshold

• Independent validation from a subjective approach, using five Ecosocialsurveys (and corresponding household surveys):

Figure 2 3: Distribution of self reported class status for Mexico (2007)Figure 2.3: Distribution of self‐reported class status for Mexico (2007).0

3.0

401

.02

Den

sity

0.0

0 20 40 60 80Permanent Income (per capita daily income in USD PPP)

Note: Densities are weighted by class size. Source: Authors’ calculations based on Ecosocial (2007) and the Encuesta Nacional de Ingreso y Gastos de los Hogares (2008).

Middle to Upper Middle lower & Lower

A meaningful and robust definition?

The upper bound of the middle class, and the resulting four classes in the LAC-wide income distribution

.05

classes in the LAC wide income distribution

Average per capita income in the US, 2009

03.0

4.0

2.0

Den

sity

0.0

1 Proposed upper threshold for LAC (percentile 97.8)

4 10 50 100Per capita daily income (USD PPP)










Mobility in educational attainment between generations is low in LACgenerations is low in LAC…

Impact of one sd. of parental education on educational attainment of the childrenImpact of one sd. of parental education on educational attainment

of the children 3

2

of E

duca

tion

1

Yer

as o

0

hiopia

Chin

a zs

tan

eland

gd

om

aland

rw

ay

publi

c mark

kra

ine

aysia

va

kia

nland

ton

ia Tim

or lgi

um

oland

USA Nep

al ed

en

atlab

) ela

nd

Africa

lands

pin

es

rland

tna

m ve

nia

ngary

La

nka

kistan

Ita

ly ha

na

nesia

ag

ua

mbia

Chile

Egypt

Brazil

uado

r na

ma Peru

Rural E

thio

Rural C

hKyrg

yzs

Northe

rn Ire

la

United

Kingd

New Ze

ala Norw

Czech

Rep

uDen

mUkraMala

ySlov

aFin

laEsto

East T

imBelg Pola U NeSwed

Bangla

desh

(Matl Ire

laSou

th Af

Netherl

aPhil

ippi

Switzerl

aViet

nSlov

eHun

gSri L

aPak

is IGha

Indon

eNica

ragColo

m C Eg BrEcu

aPan

a P

Note: The graph shows the average impact of a standard deviation in the years of education of the parents on children’s years of education. One regression is run separately for each country and birth cohort, using survey weights and each bar represents the average across birth cohorts. Source: Own calculations based on Hertz (2007)

…and if we look at achievement, things do not look much better

50co

remuch better…

40re

n's

Test

S30

SE

S o

n C

hild

1020

of P

aren

tal S

01

Imac

t o

china

nesia

china

baija

nns

tein

qatar

stonia

nland

eland

unisi

ako

realatvia

orway

ailan

dord

anna

daerb

iaex

icojap

anmarkba

niaroa

tiakh

stan

ration

zstan

spain

lands

uaniaita

lyurk

eyerl

and

reece

weden

mania

publi

cob

ago

publi

cve

niaola

ndrtu

gal

chile

strali

aela

ndbra

zilgd

omom

biaus

triaman

yna

maap

orebo

urgsta

tesala

ndlgi

umisr

ael

ugua

yng

aryperu

entin

alga

ria

macao

-chind

one

hong

kong

-chaz

erba

liech

tens q esto finl

icel

tun ko la norw

thail jor can se mex jap

denmalb

acro

kaza

khs

russia

n fed

eraky

rgyzs sp

nethe

rlalith

ua tursw

itzerl gresw

erom

a

czec

h rep

u

trinida

d and

tob

slova

k rep

uslo

ve pol

portu c

austr ire

l br

unite

d king

dco

lom aus

germpana

singa

plux

embo

unite

d sta

new ze

albe

lg isuru

ghu

ng parg

enbulg

Note: The graph shows the impact of one standard deviation in the socioeconomic status (SES)of the parents on children’s reading test scores. One regression is run separately for each country using survey weights. Each regression includes SES, gender, first and second generation immigration status, country dummies and a small town (less than 1,500 inhabitants) dummy. Source: LCRCE calculations based on PISA 2009

…especially if one considers that the average performance is fairly poor

HK G FIN KOR

550

performance is fairly poor

MAC

LI E

ES T

IS L

LV A

NO R

CANJP N

DN K

HR V

ES P

NL D

IT A

CH E

G RC

SWE

CZ E

SV K

SV N

PO L

PR T

AU S

IR L

G BR

AU T

DEU

SG P

LU X

USA

NZ L

BE L

IS R

HU N500

e

TH A

SR B

MEX

RU S

LT U

TU R

RO U

TTO

CHL

BRACOL

URUBG R

045

0ge

Tes

t Sco

re

IDN

AZ E

Q AT

TU N

JO R

AL B

KA Z

BRA

PAN PER

ARG

350

400

Ave

rag

KG Z

300

3

10 20 30 40 5010 20 30 40 50Impact of Parental SES on Children's Test Score

Note: The graph shows the impact of one standard deviation in the socioeconomic status (SES)of the parents on children’s reading test scores against the average test score in the country. To determine the impact, one regression is run separately for each country using survey weights. Each regression includes SES, gender, first and second generation immigration status, country dummies and a small town (less than 1,500 inhabitants) dummy. Source: PISA 2009

But mobility has improved somewhat, especially during the last two decadesduring the last two decades

Y f Ed ti

The impact of one sd. in parental education on years of completed education of the children

2 42.62.8

3

Brazil

2.5

3

3.5

Chile

2.5

3

Colombia

Years of Education

2.22.4

1930 1940 1950 1960 19701.5

2

1930 1940 1950 1960 1970 19801.5

2

1930 1940 1950 1960 1970 1980

on

Ecuador Nicaragua Panama

22.22.42.62.8

3

Yea

rs o

f Edu

cati

1930 1940 1950 1960 19701.5

2

2.5

3

1930 1940 1950 1960 1970 1980

2

2.5

3

3.5

1940 1950 1960 1970 1980

2.5

3

3.5

Peru

1.11.21.31.4

High Income

1.21.31.41.51.6

Low-Middle Income

1920 1930 1940 1950 1960

.91

1940 1950 1960 1970 1980 1930 1940 1950 1960 1970 1980

Note: Own calculations based on Hertz (2008)

An example of policy interventions: the case of student loans and tertiary education in Chile (I)student loans and tertiary education in Chile (I)

Source: Solis (2011)

An example of policy interventions: the case of student loans and tertiary education in Chile (II)student loans and tertiary education in Chile (II)


An example of policy interventions: the case of student loans and tertiary education in Chile (III)student loans and tertiary education in Chile (III)











High levels of upward long term mobility last 15 years in the region…

Circa 2009 Circa 2009 c1995

MarginsPoor Vulnerable Middle Class

Circa 1995

Poor 51.1 43.9 4.9 47.0

Vulnerable 3 1 46 50 9 34 0Circa 1995 Vulnerable 3.1 46 50.9 34.0

Middle Class 0.5 3.5 96 19.0

c2009 Margins 25.2 37.0 37.8

Source: Latin American Household Surveys. Lower bound estimates of mobility. 1995-2009

g

Education is a key driver for mobility across countries


…as it is access to formal jobs, especially to enter the middle class


Living in urban areas is an important factor promoting upward mobilitypromoting upward mobility








6. Implications of a Larger Middle Class: Political

economy and economic policy


Latin America is becoming a middle class society…

50%

40%

45%

25%

30%

35%

of p

opul

atio

n

15%

20%

25%

Per

cent

age

o

Middle Class (10$ a day line)

5%

10%

Middle Class (10$ a day line)

Poverty (4$ a day line)

Vulnerable (4$ - 10$ a day)

0%1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

Source: Authors’ own calculations.

…although strong heterogeneities persistg g g p

90%

100%

60%

70%

80%

90%

Upper Class

30%

40%

50%

60% Upper Class

Middle Class

Vulnerable

P

0%

10%

20%

30% Poor

0%


The emergence of China

Micro-linked CGE forecasts0

100

Central asia andE t E

6080

nt

Eastern Europe

China

Sub saharian Africa

40pe

rcen

Middle East andNorth Africa

South Asia

20

East Asia

Latin America andthe Caribbean

0

2005 2030

Source: Bussolo and Murard (2011), “The evolution of the middle class in Latin America: 2005-2030,” mimeo.

Who are middle class households? (1/2)

6Household size

5

6

3

4Poor

Vulnerable

2

3 Middle Class

0

1


0Brazil Chile Colombia Honduras Mexico Peru Dom. Rep. Costa Rica

Who are middle class households? (2/2)

Median Years of Education

14161820

468

1012

024

2009 2008 2009 2008 2006 2009 2009 2009

d l b l h lHonduras Dominincan Republic

Peru Mexico Colombia Costa Rica Brazil Chile

Poorer Vulnerable Middle Class











The middle classes have been attributed many i tvirtues…

• These virtues include values leading to greater social cohesion,These virtues include values leading to greater social cohesion,political stability and growth

• But there is little empirical evidence

Health Education

Social Policy Economic Structure Governance

Cross-country GMM regressions (World)

Health Expenditures

/ GDP

Education Expenditures

/ GDPMean Applied

TariffCredit Mkt

Liberalization Polity Score Corruption

Middle Class 2.054*** 2.918** -10.945*** 1.357*** 6.431*** -1.764*** (% of population with income above 10 USD) [3.849] [2.337] [-3.072] [2.799] [4.068] [-4.767]Poverty 0 019** 0 042** 0 203*** 0 047*** 0 042** 0 011*Poverty -0.019 -0.042 0.203 0.047 -0.042 -0.011 (2.5 USD a day Poverty Headcount) [-2.411] [-2.472] [2.874] [3.383] [-2.345] [-1.825]Inequality -3.716** 3.028 20.736*** -0.209 9.262** 4.083*** (Gini Index) [-2.456] [1.360] [3.373] [-0.165] [2.164] [3.866]Output per capita 0.121 -0.922 5.485** 1.292*** 0.280 -0.470** (ln of GDP per capita) [0.416] [-1.310] [2.439] [3.009] [0.334] [-2.218]

Observations (5 year averages) 269 192 265 294 318 285Number of countries 107 97 103 100 106 92Hansen Test - p value 0.174 0.640 0.934 0.469 0.701 0.451

Notes:1 z statistics in brackets

…but does it hold for LAC?

• We investigate how LAC middle classes think, and how the socialcontract in LAC may foster or hinder mobility out of povertycontract in LAC may foster or hinder mobility out of poverty

• Looking at values, we find little evidence of a “middle classparticularism”p

• We group values into Trust in institutions; Political alienation; Perception of mobilityand opportunity; Support for individual rights; Legitimization of political violence; Voted;Social tolerance; Nationalism; Political ideology; Interpersonal trust; and Interpersonalalienation

• We find little evidence of particularism – middle class values stand betweenthe ones of the rich and the poorthe ones of the rich and the poor

• Variation across countries is much larger than the one dictated by income• Overall, class and individual characteristics explain little of the variation in

values (low R squared).

Middle class “particularisms”p

0.30

Support of individual rights under any circumstances

0.053

0 10

0.00

0.10

0.20

0.30

-0.120 -0.120-0.065

-0.30

-0.20

-0.10

Education Vulnerable (4-10)

Middle class (10-20)

Middle class (20-50)(4 10) (10 20) (20 50)

0 067 0.0990 10

0.20

0.30

Social tolerance

0.0060.067

-0.050

-0.20

-0.10

0.00

0.10

-0.30Education Vulnerable

(4-10)Middle class

(10-20)Middle class

(20-50)

Source: Lopez-Calva, Rigolini & Torche (2011), “Is There Such Thing as Middle Class Values?,” Policy Research Working Paper, No. 5874.

The social contract in LAC (1/3)( )

• In some countries social policies may have become moreIn some countries social policies may have become moreprogressive…

120

150

r. P

oor

= 100

)

ArgentinaFlagship cash transfer programs

Nontertiary public education spending 120

150

Poo

r =

100)

BoliviaFlagship cash transfer programs

Nontertiary public education spending

30

60

90

ing

per

capi

ta (

Ext

r

30

60

90

g pe

r ca

pita

(E

xtr.

0

30

Extreme Poor (0-2.5)

Poor (2.5-4)

Vulnerable (4-10)

Middle Class (10-50)

Spen

di

0

30

Extreme Poor (0-2.5)

Poor (2.5-4)

Vulnerable (4-10)

Middle Class (10-50)

Spen

din

g

Source: Lustig (2011), “Fiscal Policy, Fiscal Mobility, the Poor, the Vulnerable and the Middle Class in Latin America,” mimeo.


• …but overall taxation and social policies seem to affect little…but overall taxation and social policies seem to affect littleclass transitions – or even to foster downwards transitions

BrazilBrazil

Market Income y < 2.5 2.5 < y < 4 4 < y < 10 10 < y < 50 y > 50

y < 2.5 88% 8% 4% 0% 0%

2 5 < y < 4 18% 72% 9% 1% 0%

Post‐fiscal Income

Post‐fiscal Income

Peru

2.5 < y < 4 18% 72% 9% 1% 0%

4 < y < 10 0% 13% 84% 3% 0%

10 < y < 50 0% 0% 18% 82% 0%

y > 50 0% 0% 0% 35% 65% Peru

Market Income y < 2.5 2.5 < y < 4 4 < y < 10 10 < y < 50 y > 50

y < 2.5 92% 8% 0% 0% 0%

2.5 < y < 4 1% 94% 5% 0% 0%

4 < y < 10 0% 2% 98% 0% 0%

Post fiscal Income

Source: Lustig (2011), “Fiscal Policy, Fiscal Mobility, the Poor, the Vulnerable and the Middle Class in Latin America,” mimeo.

y

10 < y < 50 0% 0% 8% 92% 0%

y > 50 0% 0% 0% 16% 84%


• Middle classes also appear to opt out from the social contractbecause of poor quality of services

Percentage of 6-12 years old students enrolled in private schools

because of poor quality of services

40%

60%

80%

100%

0 -2.5

2.5 - 4

4 -10 Percentage of households with

0%

20%

40% 4 10

10 - 50

50 + 5.9%

Percentage of households with inverter or generator (Dom. Rep.)

18 9%

54.8%

1 1% 1 3%

2.1%2.0%

Inverter Generator

Source: Authors’ calculations and Sánchez and Senderowitsch (2011), “The Political Economy of the middle class in the Dominican Republic.”

.1% 1.9% 5.7%18.9% 12.9%1.0% 1.1% 1.3%

Poorest Poorer Middle Upper Middle Richer Total

What changes may growing middle classes bring in LAC?What changes may growing middle classes bring in LAC?

• Middle classes may bring change, but, at least in LAC, they doMiddle classes may bring change, but, at least in LAC, they donot appear to hold particular values – are they different?

• LAC middle classes also appear to opt out from the social• LAC middle classes also appear to opt out from the socialcontract

• A more inclusive social contract with higher taxation andbetter services may be needed to break the intergenerationalcycle of povertyy p y

• But will LAC middle classes buy into it? Or is there a dangerthat more and more the remaining poor may be left alone?that, more and more, the remaining poor may be left alone?










Preliminary ConclusionsPreliminary Conclusions

1. Inter-generational mobility is low in LAC– Family background is highly associated with educational attainment and achievementFamily background is highly associated with educational attainment and achievement– Though there is some evidence that it is becoming a little less so…– …and that school-level and educational credit interventions, for example, may help

2 Withi ti th h b b t ti l d i t2. Within generations, there has been substantial upward income movement– About half of those who were originally poor circa 1995 have moved out of poverty– But only 10% of those made it to the middle class. – Key correlates of upward movement: more education and better jobs.

3. From 2002 to 2009, the size of the Latin America middle class grew by 52%– Although substantial heterogeneity persists across countries– Globally, larger middle classes are correlated with “better” policiesb y, g b p– But in LAC, there is less evidence of a “middle-class exceptionalism”– In the prevailing social contract, LAC middle classes often opt out of public services (and

taxation…)– Challenge is to better anchor the new, larger middle class into a progressive social contract for g , g p g

LAC

Th k Thank you

Documents

Economic Mobility and the Rise of the Latin American ... · Note: * denotes final year between 2004-2006 Source: SEDLAC ... one, there is great mobility and change so that the position