Friday, November 20, 2009

Lit in Review: Easterly on Growth

Just clearing out my paper files and extending my memory.

Easterly and Rebelo, 1993 J Monet Ec "Fiscal policy and economic growth: An empirical investigation"
Type: Descriptive, Cross-section (change in 1970-1988) and panel (1870-1988). Lots of bivariate regressions, some multi.
Findings: Very poor and small countries rely heavily on int'l trade taxes while income taxes are important in developed (>$2500 GDP/cap) and large countries; the share of investment in transport and communication and the budget surplus are correlated with growth, but little else is; taxation results are fragile and difficult; government revenue increases with income (Wagner's Law), mostly on health and social security. No significant differences between democracies and non-democracies.

Easterly and Levine, 1997 QJE "Africa's Growth Tragedy: Policies and Ethnic Divisions
Type: Cross-country (1960s, 1970s, 1980s), SURE, country random effects. [Picture of official languages in Africa]
Findings: 40% of the difference between Africa and East Asia is explained by: low schooling, instability, underdeveloped financial systems, distorted foreign exchange markets, high deficits, and insufficient infrastructure. Ethnolinguistic diversity (the probability that two randomly selected people belong to different e.l. groups) is linked with black market premiums, poor financial development, infrastructure, and education. They use other measures of ethnic diversity also, and find much the same thing. Even after accounting for diversity's effects on these variables, diversity still explains 28% of the growth difference (Afr/EAsia). In bivariate, going fdrom complete homogeneity to complete heterogeneity reduces growth 2.3 percentage points. If Nigeria had only the average level of diversity, it would have doubled its growth.
Examples: 40 different groups in Kenya, with ~90% of the vote going to your tribe's candidate(s). Ghana conflict over cocoa tax rates and overvaluing the exchange rates.
Botswana has one of the most homoegenous societies and highest growth rates, also overcoming resource curse.
Literature recommendations: Pakenham (1991) for a popular history of how Africa's ethnic groups were split and lumped into countries by the European powers.
Best Table: 3: Ethnoliguistic Fractionalization Index, 15 most fractionalized countries in 1960: Tanzania, Uganda, Zaire (DRC), Cameroon, India, S. Africa, Nigeria, Ivory Coast, Central Afr Rep, Kenya, Liberia, Zambia, Angola, Mali, Sierra Leone all have a greater than 75% chance that two randomly selected people are not from the same ethnic group.

Easterly, 2007 J Dev Econ, Inequality does cause underdevelopment: Insights from a new instrument
Type: IV, Cross-country
Discussion: An important differentiation is made between strucutral inequality (S. America) and market inequality (China). Market inequality may be good for growth or bad, but structural is bad. He instruments for structural inequality using the percent of land suitable for wheat farming (low returns to scale, favors family farms and equal land distribution) vs. sugar farming (high returns to scale, favors plantations with high inequality). Land suitability for rainfed sugarcane pictures on the right. Structural inequality then causes bad institutions - which he is unable to identify beyond 'bad for democracy' - that reduce growth and mobility. There are about 50 ways the chain of assumptions could go wrong, but it adds to the emerging consensus that strucutral inequality is bad for growth.
Findings: Take a wild guess. Yes, everything is significant except the link to specific institutions.

No comments:

Post a Comment