Skip to content

Latest commit

 

History

History
104 lines (73 loc) · 3.7 KB

README.md

File metadata and controls

104 lines (73 loc) · 3.7 KB
Raw

Replication files for Reed, Tristan; Trubetskoy, Alexandr. 2019. Assessing the Value of Market Access from Belt and Road Projects (English).

The files in this folder produce the tables and figures in the paper.

Required Packages

All code is in python. The instructions assume that Python 3.4 is default, being run from Terminal on MacOS High Sierra 10.13. Some versions of python later than 3.4 are known to cause problems.

The following non built-in python packages are required:

Package Version used Purpose
NumPy 1.14.2 Mathematical operations
Pandas 0.22 Read and manipulate CSV and Excel data
Xlrd 1.1.0 Required for Pandas to read Excel
Unidecode 1.0.22 Convert Unicode strings with accents into simple ASCII characters
GeoPy 1.14.0 Geographical distance calculation
SciPy 1.0.1 Nearest neighbor calculations
NetworkX 2.1 Network analysis, shortest path
Tqdm 4.23.0 Draws progress bars
PyProj 1.9.5.1 Converts coordinates to X, Y
Matplotlib 2.2.3 Generates graphs
Statsmodels 0.9.0 Runs regressions
Seaborn 0.9.0 Runs nonparametric estimation

The above packages can all be installed using pip. For convenience, a script has been provided, which can be run from the bri_market_access directory:

./install_packages.sh

Replication Procedure

Due to the large computation time required to produce the results, the replication procedure is split into 6 steps.

It is important that the steps be followed in order, since each requires the results from the previous step.

Step 0

Make sure you have done this before starting the replication procedure.

First, download the raw data by clicking here. Unzip the file, then move the folder data into the top-level, bri_market_access directory.

Recommended:
Create virtual environment and install needed packages:

mkdir env
python3 -m venv env/bri-market-access
source env/bri-market-access/bin/activate

Step 1

⏳ This step takes approximately 30 minutes to run.

Step 1 generates the following tables and figures:

  • Table: Transport cost sources
  • Table: Baseline assumptions
  • Table: Market acccess deciles
  • Figure: Trade vs. size
  • Figure: Trade cost asymmetry
  • Figure: Added value by country
  • Figure: Trade cost vs. distance
  • Figure: Pop. growth vs MA change

These tables and figures do not require significant calculations beyond running the baseline scenario. From the bri_ma_replication directory, run

./step_1.sh

Step 2

⏳ This step takes approximately 4.5 hours to run.

Step 2 generates the following tables and figures:

  • Table: Sensitivity
  • Table: Trade cost distributoins

These tables require running the various sensitivity scenarios. From the bri_ma_replication directory, run

./step_2.sh

Step 3

⏳ This step takes approximately 50 hours to run.

Step 3 generates the following figure:

  • Table: Top project for cities
  • Table: Top projects

These require calculating the effects of 71 projects under two different scenarios each.

./step_3.sh

Remaining tables and figures

Table "Representation" is not generated from data.

Tables "Current status of top projects" and "Projects reduced or scaled down" contain discussion of results from Table "Top projects" and are created manually.

Figures "BRI overview", "Change in log market access" and "Change in land value" are maps that can be created from the shapefiles and data provided.