Lecture 13

02 103 04 195 no 50 68 lnexports lnarea absolute

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ) −.00 (−.33) −.02 (1.21) .03 (2.00) .00 (.19) .05 (2.12) No 50 .48 −.13 (−4.43) .02 (.28) .01 (.59) .24 (.43) −.01 (−.07) .00 (.09) −.01 (−1.19) .01 (.56) .06 (3.56) .02 (1.03) .04 (1.95) No 50 .68 ln(exports) ln(area) Absolute latitude Ethnic fractionalization Legal origin=French % of pop Islamic % of pop Christian ln(avg diamond prod/pop) ln(avg oil prod/pop) ln(avg gold prod/pop) ln(coastline/area) Colonial dummies Number obs. R2 (3) (4) .02 (1.66) −1.34 (−2.75) .51 (1.13) −.00 (−.42) .01 (.81) .02 (1.19) .02 (1.33) .02 (.93) .05 (2.18) Yes 50 .68 −.10 (−3.68) −.02 (−.32) .01 (.86) −.05 (−.09) .56 (1.49) .00 (.27) −.00 (−.07) .01 (.66) .05 (3.07) .03 (1.74) .03 (1.29) Yes 50 .80 Econometric Issues 1. Unobserved/omitted country characteristics • Areas that initially had poor institutions or domestic slavery may have selected into the slave trade and these characteristics persist today. ˆ • The effect of slave exports may be overestimated: βols will be biased away from zero. 2. Measurement error in slave export estimates ˆ • Overall noise in the data, i.e. attenuation bias: βols will be biased towards zero. ˆ • Under-representation of slaves from the interior: βols will be biased towards zero. Strategies • What does the descriptive evidence suggest? • Instrumental variables. • Controlling for observable characteristics. Log Total Slave Exports / Avg Population (in thousands) between 1400 and 1800 −12 −4 2 Initial Population Density and Slave Exports (beta coef = .33, t−stat = 2.51) COG AGO LBY MRT BEN GHA GIN SEN GNB LBR DJI MLI ZAR SDN TGO NER MOZ SOM TCD GAB TZA ETH MDG MWI CMR COM CIV BFA MAR EGY GNQ ZMB GMB NGA SLE CAF DZA NAM KEN UGA ZWE ZAF MUS SYC STP CPV −3 BWA SWZ LSO 0 Log Initial Population Density TUN RWA BDI 3 10 Relationship Between Current Income and Past Population Density (beta coef = −.51, t−stat = −4.15) MUS SYC GAB ZAF Log 1998 Real per Capita GDP 7.5 BWA TUN SWZ DZA NAM EGY MAR CPV GNQ GIN ZWE MRT CIV CMR COG DJI COM LSO LBR SEN SOM SDN KEN GHA BEN GO NGA GMB CAF AMOZ BFA TGO UGA TCD SLE ZMB NER MLI MDG MWI GNB ETH TZA RWA BDI 5 ZAR −4 0 Log Pre−Slave Trade Population Density Data source: PWT and author’s calculations 4 Instrumental Variables • Instruments – Must be uncorrelated with unobservable country characteristics ε, but correlated with slave exports S. – I use the distance from each country’s interior to the closest major slave market in each of the four slave trades. • Crucial Assumption – Location of demand influenced the location of supply. – Location of supply did not influence the location of demand. What determined the location of demand? • Climate and soil conditions suitable for plantation agriculture. – West Indies, Mauritius and Zanzibar • Existence of natural resource...
View Full Document

This document was uploaded on 02/28/2014 for the course ECON 2328 at Harvard.

Ask a homework question - tutors are online