Volume 98 | Issue 2 | May 2016 p.268-284
Remi Jedwab and Alexander Moradi
The Review of Economics and Statistics https://doi.org/10.1162/REST_a_00540
To study the impact of transportation (railroads) investments in the poor countries of Africa while focusing on one of the countries Ghana. The paper exploits the construction and later demise of the colonial railroads in sub-Saharan Africa. The main aim of the paper is to study both the long and short-run effects. The effects are also differentiated as pre and post-independence periods.
The focus is on Ghana while the results also apply to 39 African countries with similar data for 194,000 cells with the same degree of precision for 1890-2010. There are 5 models used in the paper as follows-
Cc,31 = α + Railc,18β + λ˜Rc,01 + ρ˜Uc,01+ Xc,01γ + uc, (1)
Pc,31 = α’ + Railc,18β’ + λ’˜Rc,01 + ρ’˜Uc,01+ Xc,01γ’ + u’c, (2)
Pc,100 = α’’ + Railc,18β’’ + λ’’˜Rc,01 + ρ’’˜Uc,01+ Xc,01γ’’ + u’’c, (3)
Uc,s,60 = αssa + Railc,s,60βssa + ρssa˜Uc,s,00 + πs + Xc,s,00γssa + vc,s, (4)
Uc,s,110 = α’ssa + Railc,s,60β’ssa + ρ’ssa˜Uc,s,00 + π’s + Xc,s,00γ’ssa + v’c,s, (5)
IV model 1&2: Railc,18 are cell dummies capturing rail connectivity of being 0 to 10, 10 to 20, 20 to 30, or 30 to 40 kilometers away from a line.
DV model 1&2: z-scores of cocoa production (Cc,31)and Rural and Urban Pop (Pc,31)
c= cells, s= countries 31= 31years (1901-1931) Rc,01 and Uc,01-control for populations in 1901, Xc,01 is the geography variable for Ghana. It has been included to account for contaminating factors.
IV model 3: Railc,18 – four rail dummies
DV model 3: z-scores of the population variable (Pc,100) 1901-2000, 100 years
IV model 4&5: Railc,s,60– four rail dummies
DV model 4&5: z-scores of the urban population of cell c in 60 (1901-1960) and 110 (1901-2010) years for country s.
Uc,s,00-Dummy for population control vc,s-39 country fixed effects
Xc,s,00 are the control variables for Africa included capturing physical and economic activity.
Result & Discussion:
The cells connected earlier became large cities in 1960. The pre- 1918 effects remain higher than the post-1946 effect. Ghana’s urbanization rate increased from about 20% in 1901 to 40% in1931 and to 70% in 2000. The result shows the emergence of the spatial equilibrium between 1901 and 1931 and shows its persistence over 1931-2000. The number of cities increased by 150 towns and the urbanization rate of the forest by 10 percent between 1901-1931. 75% of forest urbanization and 30% of national change in urbanization is caused by rail. The mean effect of the population increase in the countries is 0.35 and 0.65 for Ghana. The long term effects also give an increase of 0.34 approximately at the mean level. The effects of the railroads are 0.83 are four times larger for the roads in the case of Ghana.
Railroads constituted a transportation revolution and affected hugely spatial distribution. These effects are persisting up to date but the railroad system has collapsed as post-independence gave rise to the road system. Although, the development effects of now follow the same old context of a century ago.