-
Notifications
You must be signed in to change notification settings - Fork 1
/
README.Rmd
121 lines (76 loc) · 6.77 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
---
title: "State Reach Data for Africa"
author: "Carl Müller-Crepon"
output:
md_document
---
```{r setup, include=FALSE, results = FALSE, message=FALSE, warning=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(raster)
library(rgeos)
library(rgdal)
library(sp)
```
**Attention:** This is only a Beta version. Please email any feedback to [carl.muller-crepon at politics.ox.ac.uk](mailto:carl.muller-crepon@politics.ox.ac.uk).
## Description
Prominent arguments hold that African states' geography limits state capacity, impedes public service provision, and slows economic development. This repository contains comprehensive panel data on a proxy of local state capacity, travel times to national and regional capitals. These are computed on a yearly 5 x 5km grid using time-varying data on roads (since 1966) and administrative units (1945–2016). The road data has been digitized using a convolutional neural network from a collection of Michelin maps that cover the African continent at 23 points in time since 1966. Data on administrative units, their capitals, and borders has been self-collected, drawing on public sources such as statoids.org, FAO's GAUL database, and the GADM data.
The data is also available for download from this Dropbox: [https://www.dropbox.com/sh/xsxhow5aslr5tnv/AADeSxh_9NC8cvlw-zKtXvh9a?dl=0](https://www.dropbox.com/sh/xsxhow5aslr5tnv/AADeSxh_9NC8cvlw-zKtXvh9a?dl=0)
See README.Rmd for details on how to load the data (in R).
When using the data, please cite the following publications and refer to them for further details:
[Müller-Crepon, Carl. (2021) State Reach and Development in Africa since the 1960s: New Data and Analysis. *Political Science Research and Methods:* Conditionally accepted for publication.](http://www.carlmueller-crepon.org/publication/state_reach_development/) (presents the main data)
[Müller-Crepon, Carl, Philipp Hunziker & Lars-Erik Cederman (2021). Roads to Rule, Roads to Rebellion: Relational State Capacity and Conflict in Africa. *Journal of Conflict Resolution, 65*(2-3), 563-590.](https://journals.sagepub.com/doi/full/10.1177/0022002720963674) (presents the original road digitization)
## Data
### 1st Level Administrative Units (a.k.a. Regions)
GIS data for 1st level administrative units of independent African countries until 2016 are stored under data/admin_units/africa_regions_panel.GeoJSON . Each geometry corresponds to one administrative unit which existed for a period indicated by the start and end years.
```{r adminunits, echo=FALSE, results = FALSE, message=FALSE, warning=FALSE}
## Load data
admin.shp <- readOGR("data/admin_units/africa_regions_panel.GeoJSON",
"africa_regions_panel")
## Plot for year 2000
year <- 2000
plot(admin.shp[admin.shp$startyr <= year & admin.shp$endyr >= year,],
main = "Regions in Africa, 2000")
points(admin.shp@data[admin.shp$startyr <= year & admin.shp$endyr >= year,c("caplon", "caplat")],
main = "Regions in Africa, 2000", pch = ".", col = "red")
```
### State reach measure
Because physical accessibility is a necessary (but not sufficient) condition for state capacity, this data proxies local state reach via the travel times to a locations' national and regional capitals. Using time-varying data on roads, administrative borders, and capitals, these are computed yearly, for every ell on a 5 x 5km raster for all years between 1945 and 2016. Note that all years before 1966 use the road network as observed in 1966 for computing travel times.
Travel times based on time-varying road data are stored under data/admin_units/time2regcap_1945_2016_dynroads.tif and data/admin_units/time2natcap_1945_2016_dynroads.tif for times to regional and national capitals, respectively. Data of the same format but computed with time-invariant road networks as observed in 1966 are stored here: data/admin_units/time2regcap_1945_2016_1966roads.tif and data/admin_units/time2natcap_1945_2016_dynroads.tif. These are multi-band .tif files where each band corresponds to one year, in chronological order since 1945.
```{r statereach, echo=FALSE, results = FALSE, message=FALSE, warning=FALSE, out.width="50%"}
## Year to plot
year <- 2000
## Load data
regcap <- raster("data/state_reach/time2regcap_1945_2016_dynroads.tif",
band = year-1945+1)
natcap <- raster("data/state_reach/time2natcap_1945_2016_dynroads.tif",
band = year-1945+1)
## Plot for year 2000
par(mar = c(4, 4, 1, 1))
plot(regcap,
main = "Time to regional capitals, 2000 (hrs)")
plot(natcap,
main = "Time to national capitals, 2000 (hrs)")
```
### Market Access Measures
Because these proxies for state reach correlate with (but are not the same as) access to economic markets, gridded access measure to national and Africa-wide markets are included under data/market_access for all years since 1940. These data are computed using time-variant road networks (again, using those from 1966 for all years prior) and data from Africapolis on the location and population of the 1530 biggest cities and towns in Africa. These are all cities that ever reached more then 50'000 inhabitants since 1950. Market access is computed using all cities in the same country as a grid cell (national market access) or all cities on the continent (international market access) using the following formula from the economic literature:
$MA_{p,t}=\sum_{m=1}^{M}c_{p,m,t}^{-\theta}*P_{m,t},$
where the market access of point $p$ in year $t$ is the sum of the market potential $P$ of a market $m$ in year $t$ multiplied by the travel time between $p$ and $m$ calculated on the road network and discounted by a trade elasticity $\theta$. Because Donaldson (2018) and Eaton & Kortum (2002) estimate different trade elasticity measures ($\theta = $ 8.28 and 3.2 respectively), I construct the market access measure for both parameters.
The data come as multi-band .tif files where each band corresponds to one year, in chronological order since 1940. Again, I use the 1966 road data for all years before that year.
```{r market, echo=FALSE, results = FALSE, message=FALSE, warning=FALSE, out.width="50%"}
## Year to plot?
year <- 2000
## Load data
ma.nat <- raster("data/market_access/ma_nat_donal_1940_2017_dynroads.tif",
band = year-1940+1)
ma.int <- raster("data/market_access/ma_int_donal_1940_2017_dynroads.tif",
band = year-1940+1)
## Plot for year 2000
par(mar = c(4, 4, 1, 1))
plot(log(ma.int),
main = "International Market Access, 2000 (log)")
plot(log(ma.nat),
main = "National Market Access, 2000 (log)")
```
## References
Donaldson, Dave. 2018. “Railroads of the Raj: Estimating the Impact of Transportation Infrastructure.” American Economic Review 108(4-5):899–934.
Eaton, Jonathan and Samuel Kortum. 2002. “Technology, Gravity, and Trade.” Econometrica 70(5):1741–1779.