chitrab-github/Replication-Package-Morocco

# Morocco Replication Package ## Overview The code in this replication package constructs the analysis file from 5 external publicly available data sources (GADM, VIIRS, OSM, Inland Water Bodies, and NDVI) using R. Two master files run all of the code to generate the data for the 4 figures and 2 tables in the paper. ## Project Description The project uses nighttime lights imagery in conjunction with a combination of externally available datasets mentioned above to evaluate the impact of road upgrades of two major highways in Morocco. The sample consists of all the area within 20kms of each of the project highways. The two project roads are : 1. El-Jadida – Safi Highway 2. Khouribga – Benimellal Highway The final analysis is conducted on two datasets : 1. El-Jadida-Safi Grid: Pixel-level (resolution 750m) data around the El-Jadida-Safi highway within a 20km buffer of the road 2. Khouribga – Benimellal Grid : Pixel-level (resolution 750m) data around the Khouribga – Benimellal highway within a 20km buffer of the road Each of these roads underwent road upgrades. El-Jadida Safi opened to the public as of August 1, 2016 . Khouribga – Benimellal opened on May 17, 2014. Using these two road openings, we create a variable “road_improvement”, and determine the correlation between road openings and nighttime lights within the 20km buffer, using simple panel regression models. The econometric model used for the regression is as follows: 𝑁𝑇𝐿 𝑖𝑡 = 𝛼 + 𝛽𝐻𝑖𝑔ℎ𝑤𝑎𝑦𝑂𝑝𝑒𝑛𝑖𝑛𝑔 𝑖𝑡 + 𝐷 𝑖 + 𝐷 𝑡 + 𝜖 1. 𝑁𝑇𝐿 𝑖𝑡 is the light intensity of the pixel i in the course of month t. 2. 𝐻𝑖𝑔ℎ𝑤𝑎𝑦𝑂𝑝𝑒𝑛𝑖𝑛𝑔 𝑖𝑡 indicator that takes value 1 for all nearby pixels as of month t when the highway first opens to traffic and zero otherwise. 3. 𝐷 𝑖 + 𝐷 𝑡 represent the fixed effects of each pixel and the monthly trends relating to light intensity, respectively. The panel nature of the data allows to mitigate the omitted variable bias problem, control for country−wide shocks, and possible time−invariant unobs …