Intro to using R and DBnomics
Political economy of finance in Europe: Week 3
Why do we do this?
Agenda
Why do we do this?
Current account balance
International investment position
Bilateral international investment position
Why R?
Much more efficient than Excel and much, much more fun
Python just as useful (but I can only teach you R)
Possibilities are endless: For example, all slides for this course are written using R and Quarto
How to get started in R
How to learn R
Start with R for Data Science (2e), by Hadley Wickham, father of the
tidyverse.Datacamp or similar for interactive learning.
Stack Overflow has answers to almost all your questions
AI models got trained on all of the above and are very good at answering coding questions.
DBnomics
- A uniquely helpful platform for official economic data
- Standardized access point to statistical agencies, central banks, and international orgs
- The R package {rdbnomics} makes querying their API easy; downloading data is efficient and reproducible
- This vignette introduces the various options to download data via {rdbnomics}
Install and load packages
Install packages (you need to do this only once; code can be deleted or commented out afterwards):
Load the packages (you need to do this every time you open this script):
── Attaching core tidyverse packages ───────── tidyverse 2.0.0 ──
✔ dplyr 1.1.4 ✔ readr 2.1.5
✔ forcats 1.0.0 ✔ stringr 1.5.1
✔ lubridate 1.9.3 ✔ tibble 3.2.1
✔ purrr 1.0.2 ✔ tidyr 1.3.1
── Conflicts ─────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag() masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorsVisit <https://db.nomics.world>.
Attaching package: 'janitor'
The following objects are masked from 'package:stats':
chisq.test, fisher.testCurrent account balance
Agenda
Why do we do this?
Current account balance
International investment position
Bilateral international investment position
Let’s say we want to download and plot a time series of the UK’s current account balance.
Loading the data
- On DBnomics, select provider (
IMF) and series (BOP) - Use filters to select the UK and, in the field ‘Indicator’, the relevant balance of payments item. Here, we’ll go with
Current Account, Total, Net, US Dollars
Loading the data
- Move to {rdbnomics}. We’ll write our download code using the most efficient method, called ‘masking’ (explained in this vignette):
- Provider:
IMF| Series:BOP - Frequency:
A(annual) - Country:
GB - Indicator:
BCA_BP6_USD
Standard steps
We’ll be working in the Tidyverse: A beginner-friendly ‘dialect’ of R supported by a collection of packages.
- In the Tidyverse, dataframes should be formatted as tibbles. We re-format what {rdbnomics} delivers into a tibble:
Standard steps
- Inspect variable names: Many are standard across DBnomics datasets. Unfortunately, the names are all over the place:
# A tibble: 54 × 5
Indicator period REF_AREA `Reference Area` value
<chr> <date> <chr> <chr> <dbl>
1 Current Account, T… 1970-01-01 GB United Kingdom 1970.
2 Current Account, T… 1971-01-01 GB United Kingdom 2717.
3 Current Account, T… 1972-01-01 GB United Kingdom 533.
4 Current Account, T… 1973-01-01 GB United Kingdom -2412.
5 Current Account, T… 1974-01-01 GB United Kingdom -7448.
6 Current Account, T… 1975-01-01 GB United Kingdom -3465.
# ℹ 48 more rowsStandard steps
clean_namesfrom the {janitor} package let’s us rename variables according to a consistent scheme:
# A tibble: 54 × 5
indicator period ref_area reference_area value
<chr> <date> <chr> <chr> <dbl>
1 Current Account, Tot… 1970-01-01 GB United Kingdom 1970.
2 Current Account, Tot… 1971-01-01 GB United Kingdom 2717.
3 Current Account, Tot… 1972-01-01 GB United Kingdom 533.
4 Current Account, Tot… 1973-01-01 GB United Kingdom -2412.
5 Current Account, Tot… 1974-01-01 GB United Kingdom -7448.
6 Current Account, Tot… 1975-01-01 GB United Kingdom -3465.
# ℹ 48 more rowsStandard steps
- We use {ggplot2} for plotting. Years go on the x-axis, values on the y-axis. The plot is still empty because we haven’t yet told ggplot what
geomto use to populate it.
Standard steps
- For a bar chart thats
geom_col. And let’s give the plot a title.
Now that’s nice, but this is a class about finance, not trade. So what about the UK’s international investment position?
International investment position:
Shows an economy’s stock of external financial assets and liabilities at a particular point
This stock is the result of past external transactions measured according to current market values
Components:
- Credit assets (and debt liabilities)
- Equity assets and liabilities
- Derivatives
- Monetary gold and special drawing rights
International investment position
Agenda
Why do we do this?
Current account balance
International investment position
Bilateral international investment position
Loading the data: IIP & GDP
#Load BOP data using mask method
bop_imf_raw <- rdb("IMF", "BOP", mask = "A.DE+BR+CN+GB+IN+MX.IADF_BP6_USD+IAD_BP6_USD+IAO_BP6_USD+IAPD_BP6_USD+IAPE_BP6_USD+IAR_BP6_USD+ILD_BP6_USD+ILF_BP6_USD+ILO_BP6_USD+ILPD_BP6_USD+ILPE_BP6_USD+ILRRL_NRES_S_BP6_USD")
#Load GDP data using mask method
weo_gdp_raw <- rdb("IMF", "WEO:2024-10", mask = "GER+BRA+CHN+GBR+IND+MEX+USA.NGDPD.us_dollars")
#Convert data frame to tibble format
bop_imf <- as_tibble(bop_imf_raw)
weo_gdp <- as_tibble(weo_gdp_raw)
#Remove rows with missing values
bop_imf <- bop_imf |>
filter(!is.na(value))
#Clean variable names via 'janitor' package
bop_imf <- clean_names(bop_imf)
weo_gdp <- clean_names(weo_gdp)
#Change confusing country variable name
bop_imf <- bop_imf |>
rename(country = "reference_area")
weo_gdp <- weo_gdp |>
rename(country = "weo_country") |>
rename(gdp_value = "value")
#Select gdp_value
weo_gdp <- weo_gdp |>
select(period, country, gdp_value)
###Join
bop_imf_gdp <- bop_imf |>
left_join(weo_gdp)Joining with `by = join_by(period, country)`To produce the plot on the next slide, we need corresponding assets and liabilities to have identical character values. So let’s create a variable bop_items that fulfils this requirement:
bop_imf_gdp <- bop_imf_gdp |>
mutate(bop_item = case_when(
str_detect(indicator, "Derivatives") ~ "Derivatives",
str_detect(indicator, "Direct Investment") ~ "Direct investment",
str_detect(indicator, "Other Investment") ~ "Other investment",
str_detect(indicator, "Debt Securities") ~ "PF: Debt securities",
str_detect(indicator, "Direct Investment") ~ "Direct investment",
str_detect(indicator, "Equity") ~ "PF: Equity and fund shares",
str_detect(indicator, "Reserve Assets") ~ "Reserve assets"))Since bop_item no longer contains any information about whether we’re looking at an asset or a liability, let’s add a separate asset/liability label variable:
It turns out the indicator column contained two variables—a no-go for tidy data. What we’ve done: We’ve separated the names of the financial instruments from the asset/liability label.
Now we can give a negative sign to all liability values:
Plot the UK’s IIP
Three imbalanced growth models
p <- bop_imf_gdp |>
filter(ref_area %in% c("BR", "CN", "MX")) |>
filter(period > "1999-01-01") |>
filter(!is.na(bop_item)) |>
ggplot(aes(x = period, y = value, fill = bop_item)) +
geom_col() +
geom_hline(yintercept=0) +
scale_fill_manual(values=okabe_ito()) +
labs(y = "% of GDP",
x = NULL,
fill = "BoP item") +
facet_wrap(vars(country), scales = "free_y") +
theme(legend.position = "bottom")Three imbalanced growth models
Now that’s nice, but I want to know about my country’s bilateral IIP.
Bilateral international investment position
Agenda
Why do we do this?
Current account balance
International investment position
Bilateral international investment position
Sourcing bilateral IIP data
Bilateral IIP data is not available from the IMF. For European countries, we can go to Eurostat, which luckily is also available via DBnomics.
Here I started selecting the same BOP items as above, but for Italy.
Loading the data
#Eurostat BoP data via rdbnomics package: Assets, liabilities, net-position
#To break down 'Other investment', use FA__O__F2+FA__O__F4+FA__O__F519+FA__O__F81
bop_bil_raw <- rdb("Eurostat", "bop_iip6_q", mask = "Q.MIO_EUR.FA__D__F+FA__F__F7+FA__O__F+FA__P__F3+FA__P__F5.S1.S1.A_LE+L_LE+N_LE.AT+BE+HR+CY+EE+FI+FR+GR+IT+LV+LT+MT+PT+SI+SK+ES+IE+LU+NL+CH+OFFSHO+CN_X_HK+UK+HK+JP+US+IN+RU+BR+WRL_REST.IT")
#Remove missing values
bop_bil <- bop_bil_raw[!is.na(value)]
#Transform data frame into tibble
bop_bil <- as_tibble(bop_bil)
#Clean variable names
bop_bil <- clean_names(bop_bil)
#Convert values to € million to € billion
bop_bil <- bop_bil |>
mutate(value = value/1000)Renaming
#Rename BOP items for plotting
bop_bil <- bop_bil |>
mutate(bop_item = case_when(
str_detect(bop_item_2, "direct investment") ~ "Direct investment",
str_detect(bop_item_2, "other investment") ~ "Other",
str_detect(bop_item_2, "debt securities") ~ "PF: Debt securities",
str_detect(bop_item_2, "equity") ~ "PF: Equity and fund shares",
str_detect(bop_item_2, "derivatives") ~ "Derivatives"))Further processing
#Select relevant variables
bop_bil <- bop_bil |>
select(geo, country, country_partner, period, stk_flow, bop_item, value)
#Rename partner countries
bop_bil <- bop_bil |>
pivot_wider(names_from = country_partner,
values_from = value) |>
rename("China" = "China except Hong Kong") |>
pivot_longer("Austria":"Rest of the world",
names_to = "country_partner",
values_to = "value")
#Add negative sign to liabilities
bop_bil <- bop_bil |>
mutate(value = if_else(stk_flow == "L_LE", -1*value, value)) Plot assets and liabilities for Italy
p <- bop_bil |>
filter(stk_flow != "N_LE") |>
filter(!country_partner %in% c(
"Hong Kong", "Austria", "Offshore financial centers",
"India", "Finland", "Croatia", "Portugal", "Belgium",
"Rest of the world", "Cyprus", "Malta", "Estonia",
"Slovakia", "Slovenia", "Latvia", "Lithuania")) |>
ggplot(aes(x = period, y = value, fill = bop_item)) +
geom_col(position = "stack") +
scale_x_date(date_minor_breaks = "10 years") +
scale_fill_manual(values= okabe_ito()) +
labs(y = "EUR billion",
x = NULL,
fill = "BoP item",
title = "Bilateral net IIP for Italy") +
guides(fill=guide_legend(nrow=2,byrow=TRUE)) +
facet_wrap(vars(country_partner),
scales = "free", nrow = 3) +
theme(legend.position = "bottom") Warning: Removed 6400 rows containing missing values or values outside
the scale range (`geom_col()`).
Aggregate countries into groups
bop_bil <- bop_bil |>
mutate("country_group" = case_when(
country_partner %in% c("Brazil", "China", "India", "Russia") ~ "BRICs",
country_partner %in% c("Hong Kong", "Offshore financial centers",
"Switzerland") ~ "Financial centers - non-EU",
country_partner %in% c("Ireland", "Luxembourg", "Netherlands", "Malta",
"Cyprus") ~ "Euro area - Financial centers",
country_partner %in% c("Austria", "Belgium", "Croatia", "Cyprus",
"Estonia", "Finland", "France", "Greece", "Italy",
"Latvia", "Lithuania", "Portugal", "Slovenia",
"Slovakia", "Spain") ~ "Euro area - rest",
.default = country_partner
))Plot assets and liabilities for Italy
p <- bop_bil |>
filter(stk_flow != "N_LE") |>
filter(period >= "2008-01-01") |>
filter(!country_group %in% c("Rest of the world", "Japan")) |>
group_by(period, country_group, bop_item, stk_flow) |>
summarise("value" = sum(value, na.rm = TRUE)) |>
ggplot(aes(x = period, y = value/1000, fill = bop_item)) +
geom_col(position = "stack") +
scale_fill_manual(values= okabe_ito()) +
labs(y = "EUR trillion",
x = NULL,
fill = "BoP item",
title = "IIP for Italy, selected counterparts") +
facet_wrap(vars(country_group), scales = "free_x") +
theme(legend.position = "bottom") +
guides(fill=guide_legend(nrow=2,byrow=TRUE))`summarise()` has grouped output by 'period', 'country_group',
'bop_item'. You can override using the `.groups` argument.
LSE | EU469 | Week 3