RUGGEDNESS: THE BLESSING OF BAD GEOGRAPHY IN AFRICA
A replication and extension
1. Introduction
and Replication[1]
Under
normal conditions, rugged terrain negatively affects income because it can hamper
trade and other productive activities like agriculture. Nonetheless, through
quantitative analysis Nunn and Puga (2012) demonstrate that, in the case of
Africa, there is a positive differential effect of ruggedness on income.
Moreover, they explain that the reason of this positive effect is because ruggedness
limited the long-term negative effects of slave trades.
The purpose
of replicating this paper is to deepen the understanding on how geography, via
its effect on history, can impact economic development. The following replication
and extension paper is organized as follows section 2 analyses the effect of
ruggedness on economic growth after year 2000. Section 3 analyses the effect of
ruggedness on ethnic fractionalization (EF). Section 4 analyses the effect of
ruggedness on corruption perception. Section 5 analyses the impact of ruggedness
in both EF and CPI when including slave export intensity (SEI). Finally,
section 6 presents the conclusions.
Nunn and
Puga (2012) first use a basic regression model to explore the impact of
ruggedness on GDP per capita in year 2000, this model includes the effect of
ruggedness for the entire world, a control variable for African countries, and
the interaction effect of ruggedness and African countries. In order to address
for omitted variable differential effects, Nunn and Puga (2012) also include
other controls like diamonds extracted per square kilometre, percentage of
fertile soil, percentage of land with tropical climate, and distance to coast. After
performing their initial regressions, they check for robustness of their model.
They show, in all iterations, that their results are very robust.
On subsequent analysis they check for the effects of ruggedness by prevalent characteristics in Africa and the effects of ruggedness in different regions within Africa. On another model, they include SEI and conclude that SEI “fully accounts for the differential effect of ruggedness within Africa”. Finally, they analyse the effects of ruggedness, and SEI on rule of law, from these last set of regressions they conclude that SEI “adversely [affects] domestic institutions today”. All the replications are shown on Appendix 1. The replication has the same results as those shown on Nunn and Puga (2012). Therefore, the replications present no further implications.
2. Extension
One – The effect of ruggedness on
economic growth after year 2000
With the
purpose of addressing for compression of history, Nunn and Puga (2012) also perform
regressions using average annual income between years 1950 and 2000. Across all
regressions they find a very robust, positive, and statistically significant
differential effect of ruggedness in GDP. The first extension of this paper aims
to analyse if the effect of ruggedness persists after year 2000. Therefore,
regression analysis having GDP for years 2005, 2010 and 2015 as dependent
variables is performed. GDP data is obtained from Maddison (2020).
All performed
regressions have a positive and statistically significant effect of ruggedness
in Africa and a negative effect of ruggedness in the world. However, in years
2005 and 2010 the differential effect is slightly negative, while in year 2015
is positive but close to zero. Moreover, the differential effect for all years
gets progressively closer to zero. The results of the differential effect for
years 2005, 2010, and 2015 could suggest that, as years pass by, the continuous
negative impact of ruggedness in productive activities could eventually
overcome the historical benefits of ruggedness in Africa. It should also be
noted that between years the significance levels vary slightly. (See Table 1).
3. Extension
two – The effect of ruggedness on ethnic fractionalization
Enslavement
in Africa was usually performed through raids and kidnapping by people from one
ethnicity against another, sometimes even between people of the same ethnicity.
(Northrup, 1978; Lovejoy, 2000). Research has suggested that EF is a significant
factor of underdevelopment in Africa (Easterly, W. & Levine R., 1997; Lea,
A., 2014). Concerning slave exports, Whatley
and Gillezeau
(2010) investigated the transatlantic slave trade and found a positive relation
between ethnic heterogeneity and slave exports. Moreover, Nunn (2008) explains
that slave procurement caused EF and state collapse.
The second
extension of this paper analyses the effect of ruggedness on EF. Since
ruggedness reduced SEI and SEI increased EF, “hypothesis 1 (H1)” is that there is
a negative correlation between ruggedness and EF. To perform this extension, EF
for year 2010[2]
will be computed using the approach of Alesina, Devleeschauwer, Easterly,
Kurlat, and Wacziarg (2003) (See Appendix 2). Moreover, shares of ethnic groups
will be retrieved from Drazanova (2019). Data on Drazanova (2019) had several
ethnic group share duplicates, these duplicates were removed before computing
EF.
The
regression results are not significant when including all controls (See Table 2).
However, in all the models that do not include the variable measuring the
percentage of land surface that has tropical climates show significant results
(See Table G in Appendix 3). It is plausible that results are not significant
because of the reduced number of observations (observations dropped from 170 to
155). Nevertheless, as proposed in H1, the magnitude of the effect of
ruggedness in Africa is negative.
4. Extension
three – The effect of ruggedness on corruption perception
Through
quantitative research, Pak Hung (2001), showed that corruption has a negative
impact on economic growth; as well he finds that corruption affects growth
through political instability, that it reduces the share of private investment,
and that if lowers human capital levels. Concerning slavery trades, Nunn and
Wantchekon (2011) show that the descendants of people who were heavily raided
experience lower levels of trust. Trust is directly related to corruption. Li
and Wu (2010) argue that corruption tends to be more destructive and
inefficient in countries with a low level of trust. Moreover, interpersonal and political trust can
be “both a cause and a consequence of corruption” (Morris and Klesner, 2010).
Extension three aims to analyse the effect of ruggedness on perception of corruption. To perform the regression Corruption Perception Index (CPI) data for year 2010 is retrieved from Transparency International (2010). Since ruggedness prevented raids, heavy raided areas have lower levels of trust, and there is a strong relation between trust and corruption, “hypothesis 2 (H2)” is that there is a positive correlation between ruggedness and the corruption perception index. Regression results are significant and confirm H2 (See Table 2). Moreover, the magnitude of the effect of ruggedness on CPI for the rest world is statistically significant to the 10% level but close to 0. It would be interesting to analyse the geographical effects in other regions and see if due to the harder conditions rugged terrain could have detrimental effects on perception of corruption.
5. Extension
Four – The impact of ruggedness in EF and CPI when including SEI
Extension
four is performed with the in order to analyse what is the impact of ruggedness
in EF and CPI once SEI is included in the model. “Hypothesis 3 (H3)” is that,
regardless of the dependent variable (EF or CPI), the effect of ruggedness in
Africa is mainly accounted by SEI . The performed regressions include all
controls. Since EF and CPI data was gathered for year 2010, column 1 shows the
results for a regression that has 2010 GDP as the dependent variable. Column 2
has EF as the dependent variable. Column 3 show the results for a regression
with CPI as the dependent variable. The results are shown in table 3. As in the
basic model, the results for the regression with EF as a dependent variable are
not significant.
Columns 1
to 3 in table 3 show that the common effect of ruggedness almost remains
unchanged. However, the magnitude of the effect of ruggedness in African
countries gets very close to zero and is no longer statistically significant. These
results confirm H3 and provide further support for explaining that the effect
of ruggedness arises because of slave trades. Moreover, Columns 4 to 6 of table
3 report that there is an unconditional, significant, and negative relationship
between ruggedness and slave exports. As well, the product of both the
coefficient of ruggedness and the coefficient of SEI can be used to compute an
alternative estimate of the indirect historic effect of ruggedness in CPI.
6. Conclusions
Regressions
using GDP of years after 2000, show that ruggedness in African countries
maintains a positive correlation with economic growth. However, years 2005 and
2010 show a negative differential effect. Additionally, year 2015 shows a
positive differential effect but the effect is minimal and close to zero. These
results could suggest that, in the future, the persistent common negative
effect of ruggedness in productive activities could overweight the previous
benefits from ruggedness due to the restriction of slave trades in Africa.
Extensions two,
three, and four are performed with the intention to answer three hypotheses. H1
states that there will be a negative correlation between ruggedness and EF. H2
states that there is a positive correlation between ruggedness and the
corruption perception index. H3 states that regardless of the dependent
variable (EF or CPI), the effect of ruggedness in Africa is mainly accounted by
SEI. All three hypotheses are confirmed. It should be noted that the results of
regressions with EF as a dependent variable are not significant.
To
conclude, Nunn and Puga (2012) research not only helps to have a better
understanding of the causes and consequences of slave trades, but it also
provides a ground-breaking approach for analysing geographical conditions. As
with ruggedness, several geographical conditions are assumed to have unfavourable
deterministic effects on economic growth. Nevertheless, Nunn and Puga (2012) show
the importance of considering historical context when exploring the impact of
geography in development.
References
Alesina,
A., Devleeschauwer, A., Easterly, W., Kurlat, S. & Wacziarg, R. (2003).
Fractionalization, Journal of Economic growth, vol. 8, no. 2, pp.155–194.
Drazanova,
Lenka. (2019). Historical Index of Ethnic Fractionalization Dataset
(HIEF). Harvard Dataverse, V2. Available Online: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/4JQRCL
[Accessed 4 March 2021]
Easterly,
W. & Levine R. (1997). Africa's Growth Tragedy: Policies and Ethnic
Divisions, The Quarterly Journal of Economics, Volume 112, Issue 4. Pages
1203–1250.
Harvard University (n.d.). Publications. Ruggedness:
The Blessing of Bad Geography in Africa. Nathan Nunn. Available Online: https://scholar.harvard.edu/nunn/publications/ruggedness-blessing-bad-geography-africa
[Accessed 27 February 2021]
Lea, A.
(2014). National Versus Ethnic Identification in Africa: Modernization,
Colonial Legacy, and the Origins of territorial Nationalism Research Note. World
Politics 66 World Pol.
Li S. &
Wu J. (2010). Why some countries thrive despite corruption: The role of trust
in the corruption–efficiency relationship. Review of International Political
Economy, Volume 17, Issue 1. Pages 129-154.
Lovejoy, P. (2000). Transformations in Slavery: A History of Slavery in Africa, 2nd ed. Cambridge University Press.
Maddison
Project Database, version 2020. Bolt, J. & van Zanden. J. (2020), Maddison
style estimates
of the
evolution of the world economy. A new 2020 update
Morris S.
and Klesner J. (2010). Corruption and Trust: Theoretical Considerations and
Evidence From Mexico. Comparative Political Studies. 2010;43(10):1258-1285
Northrup, D. (1978). Trade without Rulers: Pre-Colonial
Economic Development in
South-Eastern Nigeria. Oxford,
UK,
Clarendon Press.
Nunn,
N. (2008). The
Long-Term
Effects of Africa’s Slave Trades. Quarterly Journal of Economics 123:1,
139–176.
Nunn,
N. & Wantchekon
L. (2011). The Slave
Trade and the Origins of
Mistrust in Africa. American
Economic Review.
Nunn, N. & Puga D. (2012). Ruggedness: The
Blessing of Bad Geography in Africa. Review of Economics and Statistics. 2012;
94 (1): 20-36.
Pak Hung
(2001). Corruption and Economic Growth. Journal of Comparative Economics. Volume
29, Issue 1, March 2001, Pages 66-79.
Transparency International (2010). Corruption
Perception Index. Available Online: https://www.transparency.org/en/cpi/2010 [Accessed 6 March 2021]
Whatley, W. & Gillezeau, R. (2010). The Impact of the Transatlantic Slave Trade on Ethnic Stratification in Africa. American Economic Review 101(3):571-76.
Appendix 1
– Replication tables and figures from Nunn and Puga
(2012)
Figure A – Ruggedness and GDP in African Countries.
Figure B – Ruggedness in
Non-African Countries
Appendix 2 – Ethnic Fractionalization computation
Following Alesina,
Devleeschauwer, Easterly, Kurlat, and Wacziarg (2003) approach, Ethnic
Fractionalization is computed using the following formula:
where sij is
the ethnic share group i (i = 1… N) in country j
Appendix 3
– Differential effect of ruggedness on Ethnic
Fractionalization
[1] The data and do files were retrieved
from Nunn’s Harvard University website (Harvard University, n.d.). However,
there was no coding for the scatter plot graphs or the tables output. As well,
the coding of the regressions that included ‘standard controls’ had to be
modified to work properly.
[2] Year 2010 is chosen because it is
the year that has more observations for EF and CPI.
