Ashraf and Galor, AER, 2011

Ashraf and Galor, AER, 2011
International Development
L. Pascali
University of Warwick
July 2016
Ashraf and Galor, AER, 2011
Introduction
Sustained long-term economic growth is a recent phenomenon
in human history.
Ashraf and Galor, AER, 2011
Introduction
Facts:
GDP per capita in the world economy was no higher in the
year 1000 than in the year 1.
It was only 53% higher in 1820 than in 1000 (1000- 1820:
g = 0.0019).
Around 1820, the world growth rate started to rise (1820-1870:
g = 0.005).
In 2000 world per capita GDP has rises more than 8.5 times its
1820 value.
Uni…ed growth theory attempts to build a growth model that
applies not only to modern era of sustained growth but also to
the much longer period before the Industrial Revolution when
growth was negligible.
Ashraf and Galor, AER, 2011
Malthusian economy
Malthus (1798)
Long-run growth in living standards is impossible. The
problem comes from population growth and diminishing
returns to labor.
If per capita income were to rise substantially, then people
would live longer and have larger families
As population rose, per capita income would fall because more
people would be working with a …xed amount of land.
In the end, per capita income would fall back to where it was
in the …rst place.
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
This paper exploits exogenous sources of cross-country
variation in land productivity and technological levels to
examine their di¤erential e¤ects on population density and
income per-capita during 1-1500 CE.
DATA:
Maddison Database:
Income per capita in 1 CE, 1000 CE, 1500 CE
McEvedy and Jones (1978)
Population Density in 1 CE, 1000 CE, 1500 CE
Atlas of Cultural Evolution
Technology data on four sectors (communications, industry,
transportation, agriculture)
Putternam (2008)
Year elapsed since Neolithic transition
Olsson and Hibbs (2005)
Number of domesticable species of plants and animals
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Neolithic Transition and technological development
According to the authors, the Neolithic Revolution triggered a
cumulative process of economic development, conferring a
developmental head start to societies that experienced the
transition earlier.
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Basic Regression Model
ln Pit = α0 + α1 ln Ti + α2 ln Xi + ε1
ln yit = β0 + β1 ln Ti + β2 ln Xi + ε2
where:
Pit is the population density
Ti is the number of years since the onset of agriculture (proxy
for technology)
Xi is a set of covariates
This model is estimated by OLS and by 2SLS. In the second
case, the instrument for Ti is the number of domesticable
plants and animals (see Diamond, 1997)
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Basic Regression Model
∆ ln Pit = α0 + α1 ∆ ln Ti + ε1
∆ ln yit = β0 + β1 ∆ ln Ti + ε2
where:
Pit is the population density
Ti is the number of years since the onset of agriculture (proxy
for technology)
This model is estimated by OLS. Di¤erentiating with respect
to time allows to control for country …xed e¤ects.
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Ashraf and Galor, AER, 2011
Conclusions
Consistent with the Malthusian predictions, the analysis
uncovers statistically signi…cant positive e¤ects of land
productivity and technological level on population density in
1500 CE, 1000 CE and 1 CE
In contrast, the e¤ects of land productivity and technology on
income per capita in these periods are not signi…cantly
di¤erent from zero.
These qualitative results remains robust to controls for the
confounding e¤ects of a large number of geographical factors.
Moreover, they are robust on the cross-section as on the
time-series dimension.