As Higher Education Expands, Is It Contributing to Greater Inequality?

LLAKES Conference

University of London

July 5-6, 2010

The Debate Longstanding tradition in human capital theory that there is

positive relation between distribution of education & distribution of earnings, and that policies that increase education in the labor force are a “good” way to reduce economic and social inequality.

The technical debate is about what happens to inequality as the average level of schooling in the labor force rises.

A second debate is about whether labor market analysis can be separated from State incomes policies; that is, whether the State can or should be neutral in the way society distributes income, and whether this better explains how income is distributed than does educational expansion or education distribution.

The Technical Debate Possible positive relation between rising level of

education and variance of education in the labor force.

The “effect” of increasing average level of education may increase income equality up to a certain level and then decrease.

Demand for education in the labor force may also change as schooling expands and the variance of education declines. This changes the payoff, or ROR, to various levels of schooling.

The Political Debate Incomes policies versus the “natural” forces of the

market and technological change in influencing income distribution.

Political debate is related to the technical debate through the returns to education.

If RORs are essentially a market phenomenon, State is just an educational investor searching for efficient strategies for increasing HC. If RORs are considered to be influenced by incomes policies, then the State’s fiscal, spending, and other incomes policies are key and educational expansion less important.

The Case for Incomes Policies

Freeman on Canada versus US in 1980-2000. Same technological change, but much greater increase in income inequality in the US. Britain versus Scandinavia.

Pre- and post-tax and spending policies of States in LA and Europe (Figure 1).

OECD Data

0

0.1

0.2

0.3

0.4

0.5

0.6

Argen

tina

Brazil

Chile

Colom

bia

Mˇx

icoPe

r

Austri

a

Belgium

Denm

ark

Finland

Fran

ce

Germ

any

Greec

e

Ireland

Italy

Luxe

mbu

rg

Nethe

rland

s

Poland

Portu

gal

Spain

Swed

en

Unite

d King

dom

Gin

i Index

Inequality before taxes and transfers Inequality after taxes and transfers

The Relation of Investment in Education to Income Inequality This relation needs to be viewed in the larger context of

State incomes and investment policies. Also part of a more complex relationship between

education and economic growth and between economic growth and income distribution. Though growth has traditionally been linked (positively) to more unequal Y distribution, now many economists think that high inequality may hinder growth.

Even if the State’s role is to invest optimally in education for growth, the pattern of RORs may be such that such investment may contribute to more unequal income distribution; for example, if ROR to HE is higher than to primary schooling.

The Standard Human Capital Model

log Yi = Yi + rSi + e If we take the variance of both sides of

the equation, Var (log Yi) = = r2Var (S) + S2Var (r ) + 2rS Cov (r,S)

Findings and Biases Findings that Y distribution is negatively related to S as

measured by years of schooling and positively related to Var (S), although others show no relation. Do not include ROR.

But RORs have changed over time, rising for HE relative to lower levels even as HE expands. Positive relation between r and S contributes to increasing inequality as S increases.

Measuring S by years of schooling underestimates S differences between high inequality countries, which generally spend less per student, so coefficient of S is biased upward.

High inequality countries spend relatively more on HE, so coefficient of Var (S) is also biased upward.

Changing Distribution of Education and Relation to Y Distribution All countries in our sample increased average years

of education in labor force. Std. deviation from mean increased as countries

expand education--dispersion increases, then levels off and eventually declines as countries reach average of upper secondary school.

Observing the Y distribution changes over time, we observe little relation between changes in the distribution of education and in Y distribution.

Need to add 9-10 points to Ginis in the MENA region in 1996-2000 because they are measured in consumption distribution.

Income Distribution, 1960-2006 (Gini)

Country

1960

1970

1980 1985-89 1990-95 1996-2000 2001-03

2005-06 Algeria + 40.2 38.7 35.3 Egypt+ 42 (44)* 38** 32.1 32 28.9 34.4 32.1 Iran+ 44 (56)** 47.7 43.0 38.3 Jordan+ 40.8 36.1 40.7 36.4 37.7 Morocco 50 49 39+ (52) 39.2+ 39.5+ 40.9 Tunisia+ 42 (51) 44 (53) 42.7 43 40.2 41.7 39.8 Yemen+ 33.6 33.4 37.7 Argentina 47 44 52.2 49.5 Brazil 60 61 60 60 59.1 59.2 56.1 Chile 46 53 53 56.5 57.5 57.1 52.0 Colombia 52 57 55 53.7 57.1 58.5 Mexico 53 54 51 55 50.3 51.9 49.7 48.1 Peru 60 57 49 44.9+ 46.2 49.8 50.8 Uruguay (u) 42 42 42 44.6 45.5 China 30 32 38 40.3 41.5 Korea^ 32 33 (41) 38 34 31.6 31.6 Indonesia+ 33 31 (46)** 34 (51) 32 33 34.3 39.4 Malaysia 50 48.4 48.5 49.2 37.9 Philippines 50 49 45 45 46.2 46.1 44.0 Thailand 41 42 47 48 46+ (49) 41.4+ 43.2+ 42.4 Israel 31 31 33 38 35.5 35.5 39.2

Distribution of Years of Education, 1970-2000

Country

1970

1975

1980

1985

1990

1995

2000

Algeria 3.11 3.46 3.89 4.38 4.78 4.95 5.03 Egypt 3.42 4.24 4.67 5.00 5.13 5.24 Iran 3.43 3.92 4.28 4.49 4.66 4.90 5.08 Jordan 4.14 4.37 4.93 5.21 5.35 5.37 5.41 Morocco Tunisia 3.09 3.93 4.34 4.65 4.82 5.01 5.15 Syria 3.23 3.84 4.32 4.65 4.80 4.76 4.77 Yemen 0.90 1.55 2.55 3.29 Argentina 3.54 3.78 3.72 4.02 3.94 4.04 4.14 Brazil 3.55 3.22 3.41 3.56 3.65 3.73 3.87 Chile 4.04 4.15 4.35 4.43 4.56 4.76 4.90 Colombia 3.04 3.65 3.81 3.95 4.17 4.35 4.50 Mexico 3.67 3.80 4.40 4.51 4.62 4.65 4.64 Peru 3.67 3.80 4.40 4.51 4.62 4.65 4.64 Uruguay 3.98 3.86 4.00 4.05 4.26 4.40 4.53 Korea 4.53 4.55 4.68 4.42 4.03 4.04 4.03 Malaysia 4.00 4.18 4.30 4.44 4.49 4.51 4.55 Philippines 3.81 3.83 3.94 3.93 3.78 3.84 3.71 Thailand 3.30 3.39 3.62 4.01 4.29 4.53 4.71 Indonesia 3.22 3.34 3.47 3.29 4.33 4.45 4.53 China 4.43 4.36 4.37 4.36 4.36 4.34 Israel 4.60 4.45 4.36 4.51 4.58 4.70 4.77

Spending on Education, 1980-2006 One reason it is difficult to get close relation between Var (S)

and Var (Y) is that countries may change how much they spend on a given year of schooling.

If HC is measured by investment in each worker rather than years of schooling, cost per year has to be taken into account.

Measures of relative spending per pupil (private plus public) suggest that many countries have greatly reduced relative spending per pupil in tertiary education as tertiary education has expanded.

This suggests that the investment pattern in education when combined with the leveling off of increases in Var (S) as measured in years of schooling should have contributed to greater equality.

Costs of Education/Pupil

0

5

10

15

20

25

30

Iran

Kuwai

t

Mor

occo

Saudi

Ara

bia

Tuni

sia

Argen

tina

Chile

Colom

bia

Mex

ico

Peru

Urugu

ay

Korea

, Rep

.

Mal

aysia

Philip

pine

s

Thai

land

Rati

o U

niv

ers

ity/P

rim

ary

Sp

en

din

g P

er

Pu

pil

Univ/Primary 1980 Univ/Primary 2000 Univ/Primary 2006

The Role of Changing RORs

One of the features of the 1980s and 1990s has been the worldwide rise in RORs to investment in university education relative to RORs to investment primary and secondary education.

This change coincided with the worldwide increase in primary and secondary school completion that began in the 1970s.

The rise in RORs to HE vary from region to region, in part related to incomes policies.

Changing RORs over time Private Rate of Return Social Rate of Return

Country Primary Secondary Tertiary Primary Secondary Tertiary

Egypt 1988* 5 6 9 Egypt 1998* 5 6 8 Jordan 1997* 3 9 7 Jordan 2004* 2 4 9 Morocco 1991* 8 9 12 9 10 Morocco 1999* 5 8 9 8 9 Yeme n 1997* 3 2 4 Indonesia 1978 22 16 15 Indonesia 1989 11 5 Korea 1974 20 19 16 12 Korea 1979 14 19 11 12 Korea 1986 10 19 8 12 Philippines 1971 9 6 10 7 6 8 Philippines 1977 16 8 Philippines 1988 18 10 12 13 9 10 Argentina 1987 14 12 12 11 Argentina 1989 10 14 15 8 7 8 Argentina 1996 16 16 12 12 Brazil 1970 25 14 24 13 Brazil 1989 37 5 28 36 5 21 Brazil 2008 2 25 Chile 1985 28 11 10 12 9 7 Chile 1989 10 13 21 8 11 14 Chile 1996 16 20 11 17 Colombia 1973 15 15 21 Colombia 1989 28 15 22 20 11 14 Mexico 1984 22 15 22 19 10 13 Peru 1980 41 3 16 Peru 1990 13 7 Peru 1997 8 12 7 11 Uruguay 1987 19 18 19 16 Uruguay 1989 10 13 8 12 Uruguay 1996 36 12 30 10

Changing RORs and Y Distribution Regions vary. In Latin America, RORs to HE have generally

risen since the 1980s, and this has contributed to greater inequality, probably offsetting at least some of the decline in relative spending/student in HE (except in Peru, where both ROR and spending rose).

In East Asia, RORs are lower than in LA, and have remained fairly stable, suggesting that Var (S) and the declining relative investment/student in HE may have contributed to greater equality in Y distribution.

In the MENA, RORs to HE are even lower, and stable (Morocco declining). With declining investment/student in HE but rising Var (S), this should all end up with little impact of educational changes on Var (Y).

How Expanding HE Relates to Y Distribution In countries with high RORs to HE, expanding HE should lower

payoff and should contribute to more equal Y distribution. Especially true for more developed countries.

But since HE has always been highly differentiated, with different SES groups (on average) attending different types of institutions, and spending/student varies, Y distribution could become more unequal as HE expands.

This would be especially true if spending on those institutions attended by the mass of new (lower SES students) falls relative to spending on the “elite” institutions.

We see little of this trend in Europe and the US up to 2005, but if increasing differentiation occurs in the future, even if RORs are equal across groups, HE expansion could contribute to greater Y inequality.

Cost/Student in Sample of US Universities

0

5

10

15

20

25

30

2000-2999

3000-3999

4000-4999

5000-5999

6000-6999

7000-7999

8000-8999

9000-9999

10000-10999

11000-14999

17000-29999

30000-70000

Instructional Expenses per Student ($)

Fre

qu

en

cy (

nu

mb

er)

Spending/Student in Tertiary Education, 1998 & 2005

0

2000

4000

6000

8000

10000

12000

14000

16000

USA AUS FRA ITA NLD ESP SWE GBR JPN

Spendin

g/S

tudent

(2005 P

PP

$)

Spending/Student 1998 Spending/Student 2005