Creating gravity datasets
In gravity: Estimation Methods for Gravity Models

knitr::opts_chunk$set(eval = FALSE)

Description

Gravity dataset with zero trade flows is an edited version of the full gravity dataset that is used in @Head2010.

Dataset gravity_no_zeros corresponds to the dataset without zero trade flows, gravity_zeros, on the other hand, includes zero trade flows.

Details

In order to have a dataset suited for all functions, a cross-sectional dataset is chosen. All incomplete rows and observations with missing trade flows are therefore excluded from the dataset.

As some of the functions in the package are capable of handling zero values in trade flows and some are not, two datasets, gravity_zeros and gravity_no_zeros, are provided.

The original dataset was downloaded from SciencesPo but the original link is not available anymore. It was edited in the following way:

# 1: Import and read the dataset

# As of 2022-09-25 the original link from Sciences Po is broken
# I kept the zip on GitHub back in 2017, fortunately
# url <- "http://econ.sciences-po.fr/sites/default/files/file/tmayer/data/col_regfile09.zip"
url <- "https://github.com/pachadotdev/gravity/blob/master/vignettes/col_regfile09.zip?raw=true"
zip <- "col_regfile09.zip"

if (!file.exists(zip)) {
  try(download.file(url, zip))
}
try(system("7z e -aos col_regfile09.zip"))

library(haven)
col_regfile09 <- read_dta("col_regfile09.dta")

# 2: Isolation of one year

library(dplyr)
data06 <- col_regfile09 %>%
  filter(year == 2006)

# 3: Choosing variables

data06 <- data06 %>%
  select(iso_o, iso_d, distw, gdp_o, gdp_d, rta, flow, contig, comlang_off, comcur)

# 4: Isolation of complete cases

library(tidyr)
gravity_zeros <- data06 %>%
  drop_na()

# 5: Exclusion of trade flows equal to 0

gravity_no_zeros <- gravity_zeros %>%
  filter(flow > 0)

# 6: Export the data

save(gravity_zeros, file = "gravity_zeros.rdata", compress = "xz")
save(gravity_no_zeros, file = "gravity_no_zeros.rdata", compress = "xz")