Ambar Narayan (The World Bank)

Transcription

1 Opportunity and Development Ezequiel Molina (Princeton) Ambar Narayan (The World Bank) Jaime Saavedra (The World Bank) 2nd World Bank Conference on Equity 2nd World Bank Conference on Equity, June 27-28, 2012

2 The Big Question Does Unequal Opportunities Hinder Economic Development? If so, how? The evidence at the individual level is clear. People that face better opportunities in life are able to develop their human potential at full, achieving better outcomes in terms of income, education, etc. However, there are no systematic study at the aggregate level.

3 How does Inequality of Opportunity Affect Economic Growth Today? 3 Broad Mechanisms Incomplete Markets leads to inefficiency Unequal Non-pecuniary Opportunities Unequal Political Influence Engerman and Sokoloff (1997) Factor Endowments at the outset of history generate inequalities among individuals. These inequalities are persistent through time and have an impact on development outcomes, namely, level of economic development and institutional quality. Easterly (2007) Empirically test ES (1997) using agricultural endowments as an instrument for inequality of outcomes.

4 Our Measure of Opportunities

5 PISA DATA DATA Opportunity: Educational Quality Circumstances: a measure of wealth, calculated using principal p component methodology based on asset ownership, gender, parental education, school location and head of household occupation Household Surveys Opportunity I: Being enroll on six grade on time Opportunity II: Attendance for children Circumstances: gender of the child, location, head of household education, per capita family income, number of children in the household, family structure (both parents present).

6 Identification Strategy Instrumental Variables Approach Log of Wheat Sugar Suitability Ratio: Abundance of Land Suitable for Growing Wheat Relative e to that Suitable for Growing Sugarcane (Easterly 2007). Share of Topical Land: Percentage of Land area in geographical tropics (Sachs, 1997).

7 o Relation between Agricultural Endowments and Equality of Opportunity ugar Suita ability Rati Log of Wheat S r = 0.576*** Human Opportunity Index - PISA Fitted values East Asia, South Asia & Pacific Middle East and Africa Americas Europe & Central Asia

8 Wheat Su ugar Suitab bility Ratio Relation between Agricultural Endowments and Inequality of Opportunity r = *** Log of D-Index - PISA Fitted values East Asia, South Asia & Pacific Middle East and Africa Americas Europe & Central Asia

9 Historical Data

10 Persistance

11 Opportunities and Economic Development Using PISA Test Scores Logarit tm of GDP pc at Current USD ( ) r = 0.766*** The Yes or No Question r = *** Logarit tm of GDP pc at Current USD ( ) Human Opportunity Index - PISA Test Scores Dissimilarity Index - PISA Test Scores Fitted values Americas East Asia, South Asia & Pacific Europe & Central Asia Middle East and Africa These graphs show the strong economical and statistical correlation (p-value is equal to zero in both cases) between our measures of equality of opportunity and economic development

12 Opportunities and Economic Development - Completing Six Years on Time ( ) 12 r = 0.695*** The Yes or No Question ( ) 12 r = *** GDP pc at Current USD ( GDP pc at Current USD ( KGZZ Logaritm of Logaritm of HOI - Completing Six Years of Education On Time D-Index - Completing Six Years of Education On Time Fitted values Americas East Asia, South Asia & Pacific Europe & Central Asia Middle East and Africa

13 Opportunities and Economic Development - School Attendance ( ) 12 r = 0.505*** The Yes or No Question ( ) 12 r = *** Logaritm of GDP pc at Current USD Logaritm of GDP pc at Current USD KGZZ HOI - Attendance for Years Old D-Index - Attendance for Years Old Fitted values Americas East Asia, South Asia & Pacific Europe & Central Asia Middle East and Africa

14

15

16

17

18 Economic Magnitude The effect of equality of opportunities in economic growth is not The effect of equality of opportunities in economic growth is not only statistically significant but also economically significant.

19

20

21

22

23

24

25 Overidentification Tests For IV approach to work we must satisfy two requirements: a) It must be uncorrelated with the error b) Correlated with the endogenous explanatory variable. We have seen that b) is satisfied. In the case of the first requirement we know it cannot be tested because it involves a correlation between the IV and an unobserved error. However, if we have more than one instrumental variable, we can effectively test t whether they are uncorrelated with the structural error. In this test under the null hypothesis all instruments are uncorrelated with the error term. Since for both cases, the p-value is considerable larger that 0.05 we cannot reject the null.

26 Overidentification ifi i Tests

27 Unequal Opportunity and Institutional Arena

28 Final Remarks Conflicting results in the literature on inequality of outcomes and economic growth are missing i an important tvariable. ibl Consistent with the hypothesis of ES (1997) this paper supports the prediction that agricultural endowments predict unequal educational opportunities and this, in turn, predicts development outcomes. Our innovation is that we develop a measure of educational opportunities that is a better fit for structural inequality than the GiniIndex. In fact, after controlling for any unequal opportunity measure we constructed, the gini index has no explanatory power. Agenda for future research: Disentangle whether what matters is poverty of opportunity and/or inequality for economic development.