T Evidence of intergenerational educational persistence and the F - - PowerPoint PPT Presentation
The geography of mobility T Evidence of intergenerational educational persistence and the F Great Gatsby Curve in Brazil A R Tharcisio Leone D Free University of Berlin School of Business and Economics Institute for Latin American
Evidence of intergenerational educational persistence and the ”Great Gatsby Curve” in Brazil Tharcisio Leone
Free University of Berlin School of Business and Economics Institute for Latin American Studies
June 11, 2018
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 1 / 35
1 Introduction 2 Data 3 Estimating intergenerational mobility 4 The Great Gatsby Curve 5 Linking Inequality and school dropouts 6 Conclusion
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 2 / 35
1 Introduction 2 Data 3 Estimating intergenerational mobility 4 The Great Gatsby Curve 5 Linking Inequality and school dropouts 6 Conclusion
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 3 / 35
Source: Corak (2012) Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 4 / 35
.577 .543 .529 .527 .526 .523 .512 .510 .505 .503 .501 .501 .497 .495 .489 .488 .487 .483 .482 .480 .475 .470 .469 .457 .449 .444 .416 .515
Gini
Distrito Federal Acre Amazonas Maranhão Bahia Rio de Janeiro Tocantins Paraíba Pernambuco Ceará Roraima Piauí Alagoas Rio Grande do Norte São Paulo Espírito Santo Pará Sergipe Minas Gerais Mato Grosso do Sul Rio Grande do Sul Amapá Rondônia Mato Grosso Paraná Goiás Santa Catarina Brazil
.1 .2 .3 .4 .5 .6 .7 .8 .9 1 Outcome share (Proportion)
Bottom 50% Middle 40% Top 10% Note: Estimations based on per capita household income. Source: PNAD-2014, own estimates.
(a) Income distribution
States
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 5 / 35
1 There is a variation in intergenerational educational mobility
across Brazilian states?
2 What socioeconomic indicators are correlated with the mobility at
state level?
3 Does the “Great Gatsby Curve” also hold true within a single
country?
4 How higher income inequality leads to lower rate of mobility?
Investigation of one specific mechanism behind the correlation between inequality and mobility: School dropout rate.
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 6 / 35
1 Introduction 2 Data 3 Estimating intergenerational mobility 4 The Great Gatsby Curve 5 Linking Inequality and school dropouts 6 Conclusion
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 7 / 35
1 Brazilian National Household Sample Survey (PNAD)
Mobility supplement 2014 (46, 051 individuals)
2 Main variables for the investigation
Number of years of schooling Level of education School Dropout rate Income inequality
3 Variable construction
Only individuals born between 1940 and 1989 The most educated parent 25/10 Ratio for income inequality Economic Marginalization
4 Further variables
Individual characteristics (gender, birth cohort, race, locality of residence and living with both parents at age 15)
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 8 / 35
2 4 6 8 10 12 2 4 6 8 10 12 2 4 6 8 10 12 2 4 6 8 10 12 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 1985-1989 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 1985-1989 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 1985-1989 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 1985-1989 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 1985-1989 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 1985-1989 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 1985-1989 North, Acre North, Amapá North, Amazonas North, Pará North, Rondônia North, Roraima North, Tocantins Northeast, Alagoas Northeast, Bahia Northeast, Ceará Northeast, Maranhão Northeast, Paraíba Northeast, Pernambuco Northeast, Piauí Northeast, Rio Grande do Norte Northeast, Sergipe South, Paraná South, Rio Grande do Sul South, Santa Catarina Southeast, Espírito Santo Southeast, Minas Gerais Southeast, Rio de Janeiro Southeast, São Paulo West, Distrito Federal West, Goiás West, Mato Grosso West, Mato Grosso do Sul Brazil
Children Parents Years of Schooling Birth cohorts
(a) Average years of schooling
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 9 / 35
Mean 1 2 3 4 5 6 7 8 9 10 11 Years of schooling
M i d w e s t S
t h S
t h e a s t N
t h e a s t N
t h Mato Grosso do Sul Mato Grosso Goiás Distrito Federal Santa Catarina Rio Grande do Sul Paraná São Paulo Rio de Janeiro Minas Gerais Espírito Santo Sergipe Rio Grande do Norte Piauí Pernambuco Paraíba Maranhão Ceará Bahia Alagoas Tocantins Roraima Rondônia Pará Amazonas Amapá Acre Source: PNAD-2014, own estimates.
by region and states in Brazil
Average Education
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 10 / 35
1 Introduction 2 Data 3 Estimating intergenerational mobility 4 The Great Gatsby Curve 5 Linking Inequality and school dropouts 6 Conclusion
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 11 / 35
Probability of children from parents with the educational attainment j to achieve the education level i. Provide a overview about the direction of the mobility.
Summarize the grade of persistence between parents’ and children’s educational attainment. Take into account the changes over time in the education inequality.
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 12 / 35
1 Classification of educational outcomes Education of children
(generation t) and parents (generation t + 1) into four categories: no school certificate, primary, secondary and tertiary education.
2 Estimation of matrices
A trasition matrix is a process {X0, X1, X2, ...} with number of states S, where S has size R (possibly infinite) such that:
pij = P(Xt+1 = j | Xt = i) for i, j ∈ S, t = 0, 1, 2, ... (1)
with two important properties:
∀ i, j ∈ R, P(i, j) ≥ 0, and
N
pij =
N
P(Xt+1 = j | Xt = i) =
N
P{Xt=i}(Xt+1 = j) = 1
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 13 / 35
3 Measure of intergenerational mobility
Immobility Ratio: ImR = Tr(P) S = N
i=1 ρij
S Upward and Downward Mobility: UpM = Pr (Xt > l | Xt+1 = l) and DoM = Pr (Xt < l | Xt+1 = l) Prais–Shorrocks-Indicator: MP S(P) = S − Tr(P) S − 1 with MP S ∈ [0, 1]
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 14 / 35
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990
Year of Birth Parents with no school certificate
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990
Year of Birth Parents with primary education
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990
Year of Birth Parents with secondary education
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990
Year of Birth
Tertiary Secondary Primary No school
Parents with tertiary education Source: PNAD-2014, own estimates.
Descendants predicted propabilities of education attainment
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 15 / 35
.1 .2 .3 .4 .5 .6 .7 .8 .9 1
Probability
M i d w e s t S
t h e a s t S
t h N
t h e a s t N
t h B r a z i l Mato Grosso do Sul Goiás Distrito Federal Mato Grosso São Paulo Rio de Janeiro Minas Gerais Espírito Santo Santa Catarina Rio Grande do Sul Paraná Sergipe Rio Grande do Norte Pernambuco Paraíba Maranhão Ceará Bahia Alagoas Piauí Tocantins Roraima Rondônia Pará Amazonas Amapá Acre Brazil Source: PNAD-2014, own estimates.
by regions and states in Brazil
Intergenerational Education Mobility
Downward Immobility Upward Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 16 / 35
.1 .2 .3 .4 .5 .6 .7 .8 .9
Probability
Paraíba
Alagoas Maranhão Sergipe Rio de Janeiro Pernambuco Mato Grosso
Pará Brazil Minas Gerais Ceará Bahia São Paulo Paraná Santa Catarina Acre Distrito Federal Espírito Santo Roraima Goiás Amazonas Rondônia Amapá Piauí Tocantins
Prais-Shorrocks Index Botton to Top Bottom Persistence Top Persistence
Source: PNAD-2014, own estimates.
Intergenerational Mobility Indexes
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 17 / 35
The association between the education of children and parents is given by:
educc
is = α + β educp is + ǫi
for i = 1, 2, ...N (2)
Normalisation of coefficient β by the corresponding standard deviation:
ˆ β = ρcp
s
σp
s
σc
s
, with σ =
N
N
(x1 − µ)2 (3)
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 18 / 35
From equations (2) and (3) the resulting model can be summarized as:
educc
is
σc
s
= δ + ρ educp
is
σp
s
with ρ ∈ [0, 1] (4)
Inclusion of a vector X with controls variables and some interaction terms:
educc
is
σc
s
= δ + ρ educp
is
σp
s
+ η educp
is
σp
s
× UFi
with: Control variables: Gender, year of birth and race Dummy variables UF present the state of residence
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 19 / 35
(0.500 - 0.525] (0.475 - 0.500] (0.450 - 0.475] (0.425 - 0.450] (0.400 - 0.425] (0.375 - 0.400] [0.350 - 0.375] [0.250 - 0.275] Source: PNAD-2014, own estimates.
Intergenerational persistence in education
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 20 / 35
1 Introduction 2 Data 3 Estimating intergenerational mobility 4 The Great Gatsby Curve 5 Linking Inequality and school dropouts 6 Conclusion
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 21 / 35
Interim result
The chances of attaining intergenerational educational mobility vary a very great deal from one state to another in Brazil.
Next steps
Investigation of why the mobility ranges so widely within a single country. Correlation of mobility with income inequality at state level.
Model of Solon (2004)
Family (parents and children) as an intergenerational decision maker. Higher-income parents have (1) a higher capacity to invest in the human capital of their children, and (2) a higher incentive for this investment.
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 22 / 35
RO AC AM RR PA AP TO MA PI CE RN PB PE AL SE BA MGES RJ SP PR SC RS MS MT GO DF
r = 0.3612*
.25 .35 .45 .30 .40 .50
Intergenerational persistence
.42 .44 .46 .48 .52 .54 .56 .58 .50
GINI-Coefficient
Note: r = Pearson´s Correlation. Asterisk be printed for correlation coefficients with p-values of .1 or lower. Source: PNAD-2014, own estimates.
Great Gatsby Curve
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 23 / 35
1 Introduction 2 Data 3 Estimating intergenerational mobility 4 The Great Gatsby Curve 5 Linking Inequality and school dropouts 6 Conclusion
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 24 / 35
Interim results
The level of intergenerational mobility ranges across Brazilian states The ”Great Gatsby Curve” holds true within a single country (States with greater income disparity tend to have lower levels of education mobility between generations)
Next steps
Change from an analysis of intergenerational mobility via correlation hypothesis to an investigation of the determinats The link between inequality and school dropout rate Economic marginalization: Children with very low expected earnings premium Marginalization arises as a consequence of higher income inequality
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 25 / 35
Research question
Have children from low socioeconomic backgrounds who live in states with high income inequality levels a greater chance to drop out the school system? The empirical probit model can be write as: EduOutcomeisc = π0 + π1 (IllitePis × ratios) + π2 (NoEducPis × ratios) + π3 IllitePis + π4 NoEducPis + π5 ratios + γ1 maleis + γ2 ruralis + γ3 bothPis + γ4 raceis + γ5 birthcis + ǫis
where: EduOutcome: Primary school dropout, secondary school dropout and concluded (tertiary) education system. Ratio: Ranking in income inequality (measured by the 25 to 10 ratio). Proxy for Marginalization: IlliteP for illiterate parents and NoEducP for parents with no primary education.
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 26 / 35
High [5.2 - 7.5] Middle [4.8 - 5.1] Low [3.8 - 4.7] Level of inequality Notes: (i) Estimations based on per capita household income. (ii) Ratio represents the relation between average income at the 75th and 10th percentiles. Source: PNAD-2014, own estimates.
(b) Ratio 25/10 of income distribution
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 27 / 35
.1 .2 .3 .4 .5 .6 .7
School Dropout
Illiterate No education Primary Secondary Tertiary Level of parent's education
Low Middle High Low Middle High Low Middle High Low Middle High Low Middle High Source: PNAD-2014, own estimates.
by education of parents and ratio 25/10 of income distribution
Dropout rate in primary education
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 28 / 35
Table 1: The impact of Inequality on School Dropout.
(PSD) (SSD) (CTE) VARIABLES Primary school dropout Secondary school dropout Concluded tertiary education Illiterate Parents * Ratio 25/10 0.0321 0.0206 0.0809 Parents with no school certificate * Ratio 25/10 0.0609** 0.0676**
Illiterate Parents 0.319 0.367*
Parents with no school certificate 0.468*** 0.458***
Ratio 25/10 0.00416
0.122*** Male 0.176*** 0.197***
Rural 0.691*** 0.748***
Living with both parent
0.143*** White (reference)
0.164*** 0.239***
Mixed (white/black) 0.320*** 0.316***
Asian
0.654*** Indigenous 0.297** 0.164
1980 - 1989 (reference)
0.415*** 0.351***
1960 - 1969 0.489*** 0.512***
1950 - 1959 0.463*** 0.819***
1940 - 1949 0.724*** 1.121***
Constant
Observations 35,555 35,305 34,109 Note: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Source: PNAD-2014, own estimates. Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 29 / 35
Focus on the marginal effects at a point ˜ x:
How change in primary school dropout is related to change in the ratio 25/10 ? ∂E(EduOutcome|x) ∂x
x
= ∂F(xβ) ∂x
x
= f(˜ xβ)β
.1 .2 .3 .4 .5
Dropout probability
4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5
Ratio 25/10 of income distribution Yes No
Parents with primary education?
Source: PNAD-2014, own estimates.
Primary education
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 30 / 35
Focus on the marginal effects at a point ˜ x:
How change in educational outcome is related to change in the proxy for economic marginalization ? ∂E(EduOutcome|x) ∂x
x
= ∂F(xβ) ∂x
x
= f(˜ xβ)β
Table 2: Contrasts on School Dropout, by education of parents.
(PSD) (SSD) (CTE) VARIABLES Primary school dropout Secondary school dropout Concluded tertiary education Illiterate Parents (1 vs 0) 0.172*** 0.180***
(0.0104) (0.0114) (0.00650) Parents with no education (1 vs 0) 0.253*** 0.305***
(0.00651) (0.00719) (0.00523) Notes: (i) Robust standard errors in parentheses; (ii) All other predictors at their mean value; (iii) ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 Source: PNAD-2014, own estimates. Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 31 / 35
Contrasts of Linear Prediction
4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5
Ratio 25/10 of income distribution
Source: PNAD-2014, own estimates.
Contrasts of Adjusted Predictions of NoEducP with 95% CIs
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 32 / 35
1 Introduction 2 Data 3 Estimating intergenerational mobility 4 The Great Gatsby Curve 5 Linking Inequality and school dropouts 6 Conclusion
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 33 / 35
Intergenerational persistence in education varies substantially across Brazilian states. Statistically signifficant association between intergenerational mobility and income inequality and expected earnings return to human capital. Confirmation of the ”Great Gatsby curve” at a national level. Children born in families with no education are more likely to leave the school early if they are living in states where the gap between the bottom and middle of income distribution is wider.
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 34 / 35
More Information? Questions? Suggestions? Please contact me: Tharcisio Leone ttleone13@zedat.fu-berlin.de
Tharcisio Leone (FU-Berlin) The geography of mobility June 11, 2018 35 / 35