Country code differences to look out for

Andreas Beger

2018-02-23

Notes on common country code differences when comparing and merging different data sources.

Between COW and G&W lists

How many COW country names are in G&W country names?

## 
## FALSE  TRUE 
##    36   207

How many G&W country names are in COW country names?

## 
## FALSE  TRUE 
##    46   193

There is more apparent overlap when we focus on country codes:

Country codes in COW but not in G&W
cowcode cow3c country_name start end
679 YEM Yemen 1990-05-22 2016-12-31
946 KIR Kiribati 1999-09-14 2016-12-31
947 TUV Tuvalu 2000-09-05 2016-12-31
955 TON Tonga 1999-09-14 2016-12-31
Country codes in G&W but not in COW
gwcode iso3c country_name start end microstate
89 UPC United Provinces of Central America 1823-07-01 1839-12-31 FALSE
99 GCL Great Colombia 1821-08-30 1830-09-22 FALSE
340 SER Serbia 1878-07-13 1915-10-01 FALSE
340 SER Serbia 2006-06-05 2016-12-31 FALSE
563 TRA Transvaal 1852-01-01 1910-05-30 FALSE
564 OFS Orange Free State 1854-03-28 1910-05-30 FALSE
711 TBT Tibet 1913-01-01 1950-10-01 FALSE
815 VNM Vietnam (Annam/Cochin China/Tonkin) 1816-01-01 1893-01-01 FALSE
396 ABK Abkhazia 2008-08-26 2016-12-31 TRUE
397 SOT South Ossetia 2008-08-26 2016-12-31 TRUE
971 NAU Nauru 1968-12-31 2016-12-31 TRUE
972 TON Tonga 1970-06-04 2016-12-31 TRUE
973 TUV Tuvalu 1978-10-01 2016-12-31 TRUE

This is deceiving however. There are several country codes that refer to different countries in COW and G&W. Also, which code is retained for states after a split or merge is different in several cases.

G&W and WDI

Some example code for converting WDI data to G&W:

WDI(...) %>%
  mutate(gwcode = countrycode(iso2c, "iso2c", "cown")) %>%
  mutate(gwcode = case_when(
           iso2c=="RS" ~ 340L,
           iso2c=="XK" ~ 347L,
           iso2c=="SS" ~ 626L,
           gwcode==255L & year >= 1949 ~ 260L,
           gwcode==679L ~ 678L,
           TRUE ~ gwcode
         ))