Working with Korean Diplomatic Datasets

In my research, I often work with country-year data from Korean sources, including data on diplomatic visits, trade, aid and so on. One of the fundamental difficulties I have had is the lack of universal country codes across different datasets. Further complicating matters is the inconsistency of country names in these datasets. For example, Democratic Republic of the Congo has five different spellings across different official sources that I could find: 콩고 민주공화국, 자이르, 콩고민주공화국, 콩고 민주 공화국, 콩고민주공화국(DR콩고).

To address this issue, I have created a function in my kdiplo package that converts Korean country names into ISO 3166-1 alpha-3 (iso3c) country codes. This function, iso3c_kr, is designed to assign universal iso3c country codes to Korean-language country names that will make it easier to join different kinds of data.

One still needs to check if the output is correct, especially for countries that have gone through political transitions such as Germany, Yugoslavia, Russia, Vietnam, Yemen and so on.

Sometimes the Korean government sources have overlapping data for Yugoslavia and Serbia, for example. In such cases, one needs to check the data and make sure that the data is correct.

For example, the following is sample Korean trade data from Korean Statistical Information Service (KOSIS):

trade <- readxl::read_xlsx("../../../../korea_visits/data/kosis_trade_240330.xlsx")

trade[533:538,c(1,57:62)] %>% gt::gt()

국가별	2018 년	2019 년	2020 년	2021 년	2022 년	2023 년
잠비아	26241	16087	17619	28356	14068	15459
잠비아	108344	54542	15164	100606	82198	53867
자이르	NA	NA	NA	NA	NA	NA
자이르	618	8	113	4	NA	NA
짐바브웨	25964	14088	15514	20404	16083	19563
짐바브웨	4909	13098	11377	9627	10415	20862

And, the following is sample Korean aid data from Korea’s ODA portal:

aid <- readxl::read_xlsx("../../../../covid determinants 220818/data/korea_total_aid_2019_230709.xlsx")

aid %<>% select(1:5)

aid[c(50, 150, 250, 350, 450),] %>% gt::gt()

country_kr	sector	사업개수	약정액[USD]	약정액[KRW]
베트남	통신정책, 계획 및 행정(voluntary code)	2	232334	270736486
캄보디아	11321	1	85815	99999361
미얀마	사회보호/보장	1	103460	120560903
라오스	비정규 농업훈련	1	107958	125802378
몽골	의료서비스	5	511824	596423389

Wide format is quite common in official Korean data sources. Trade data is in wide format. Before using the iso3c_kr function, let’s first transform the trade data into a long (country-year) format to make it in the same format as the aid data. This will make joining the two datasets more feasible.

export <- trade
import <- trade

export %<>% select(-`...63`)
export_long <- export %>% pivot_longer(4:62, names_to = "year", values_to = "export_kosis")

export_long %<>% set_names(c("country_kr", "type", "unit", "year", "export_kosis"))


export_long %<>% filter(type == "수출액[천달러]") %>% 
  mutate(export_kosis = as.numeric(export_kosis) * 1000,
         year = parse_number(year)) %>%
  select(-type, -unit)





import %<>% select(-`...63`)
import_long <- import %>% pivot_longer(4:62, names_to = "year", values_to = "import_kosis")

import_long %<>% set_names(c("country_kr", "type", "unit", "year", "import_kosis"))


import_long %<>% filter(type == "수입액[천달러]") %>% 
  mutate(import_kosis = as.numeric(import_kosis) * 1000,
         year = parse_number(year)) %>%
  select(-type, -unit)

trade <- export_long %>% left_join(import_long, by = c("country_kr", "year"))

Using the iso3c_kr function, we can simply convert Korean country names into iso3c country codes. For example, the following is the output of the iso3c_kr function for the Korean trade data:

trade <- iso3c_kr(trade, "country_kr") #you copy paste the column name that has the Korean country names.


trade[c(50, 150, 250, 350, 450, 550), c(1,5, 2:4)] %>% gt::gt()

country_kr	iso3c	year	export_kosis	import_kosis
계	NA	2014	572664607000	525514506000
아랍에미리트 연합	ARE	1996	1377933000	2259205000
앤티가바부다	ATG	1978	NA	NA
앵귈라	AIA	2019	817000	1000
아르메니아	ARM	2001	1255000	43000
앙골라	AGO	1983	235000	NA

We see that in this example, “계” (gyae) did not get any iso3c country code. This is because the iso3c_kr function could not find the iso3c country code for this entry. This is because, it is not a country name. “계” means total. It is best to check the data to see which entries did not get an iso3c code.

missing_iso3c <- trade %>% filter(is.na(iso3c)) %>% pull(country_kr) %>% unique()

paste(missing_iso3c, collapse = ", ")

[1] "계, 국제통화기금, 기타, 기타국"

They mean “total”, “IMF”, “other”, and “other countries” in Korean. In other words, we are not missing any countries, which is good.

Now let’s convert the Korean country names in the aid data into iso3c country codes:

aid %<>% set_names(c("country_kr", "sector", "no_of_projects", "aid_usd", "aid_krw"))

aid <- iso3c_kr(aid, "country_kr") #you copy paste the column name that has the Korean country names.


aid[c(50, 150, 250, 350, 450, 550),c(1, 6, 2:5)] %>% gt::gt()

country_kr	iso3c	sector	no_of_projects	aid_usd	aid_krw
베트남	VNM	통신정책, 계획 및 행정(voluntary code)	2	232334	270736486
캄보디아	KHM	11321	1	85815	99999361
미얀마	MMR	사회보호/보장	1	103460	120560903
라오스	LAO	비정규 농업훈련	1	107958	125802378
몽골	MNG	의료서비스	5	511824	596423389
필리핀	PHL	농업용수자원	2	0	0

Once you know the iso3c country codes, you can get the English country names, or other country codes (such as Correlates of War country codes) using the countrycode package, for example.

trade <- trade %>% mutate(country_name = countrycode::countrycode(iso3c, origin = "iso3c", destination = "country.name"))

trade[c(50, 150, 250, 350, 450, 550),c(1, 5, 6, 2:4)] %>% gt::gt()

country_kr	iso3c	country_name	year	export_kosis	import_kosis
계	NA	NA	2014	572664607000	525514506000
아랍에미리트 연합	ARE	United Arab Emirates	1996	1377933000	2259205000
앤티가바부다	ATG	Antigua & Barbuda	1978	NA	NA
앵귈라	AIA	Anguilla	2019	817000	1000
아르메니아	ARM	Armenia	2001	1255000	43000
앙골라	AGO	Angola	1983	235000	NA

More importantly, this function allows users to be able to join different datasets that have Korean country names. For example, one can join the trade data with the aid data using the iso3c country codes. In this example, I will join the trade data with the aid data using the iso3c country codes.

# now that I think about it, this sample data is only 2019.

aid$year <- 2019

trade_aid <- trade %>% left_join(aid, by = c("iso3c", "year"), suffix = c("", "_aid"))

trade_aid %>%
  filter(year == 2019 & !is.na(iso3c)) %>%
  slice(c(50, 150, 250, 350, 450, 550)) %>%
  select(c(1, 5, 6, 2:4, 8, 10)) %>%
  gt::gt()

country_kr	iso3c	country_name	year	export_kosis	import_kosis	sector	aid_usd
아르메니아	ARM	Armenia	2019	12729000	16743000	전문대,대학(원) 교육	119069
방글라데시	BGD	Bangladesh	2019	1282342000	404703000	건설정책 및 행정관리	46251
볼리비아	BOL	Bolivia	2019	30434000	450576000	환경정책 및 행정관리	80969
코트디부아르	CIV	Côte d’Ivoire	2019	136494000	5264000	교육정책 및 행정관리	30096
콜롬비아	COL	Colombia	2019	1143075000	718214000	성인 기초생활교육	62976
알제리	DZA	Algeria	2019	700918000	1746239000	레크리에이션 및 스포츠(voluntary code)	22312

Voilà! Now we have a dataset that has both trade and aid data, both of which originally did not have consistent country names or country codes. I plan to add warning messages to the iso3c_kr function to make it easier to spot potential issues with the conversion of Korean country names. I will continue to update the Korean country name dataset in the kdiplo package as I come across new data sources. Feel free to report unavailable country names in the iso3c_kr function to me using the issue tracker on Github.