The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.

Programmatic access to the European Central Bank’s Statistical Data Warehouse (SDW)

Eric Persson

2023-04-20

Introduction

The ecb package package provides an R interface to the European Central Bank’s Statistical Data Warehouse.

Example usage

The following example extracts the last twelve observations of headline and “core” HICP inflation for a number of countries available in the ICP database. See details below on how to use the filter parameter and how to find and use the SDW series keys.

library(ecb)
library(ggplot2)
key <- "ICP.M.DE+FR+ES+IT+NL+U2.N.000000+XEF000.4.ANR"
filter <- list(lastNObservations = 12, detail = "full")

hicp <- get_data(key, filter)

hicp$obstime <- convert_dates(hicp$obstime)

ggplot(hicp, aes(x = obstime, y = obsvalue, color = title)) +
  geom_line() +
  facet_wrap(~ref_area, ncol = 3) +
  theme_bw(8) +
  theme(legend.position = "bottom") +
  labs(x = NULL, y = "Percent per annum\n", color = NULL,
       title = "HICP - headline and core\n")

Details

The filter option

The filter option of get_data() takes a named list of key-value pairs. If left blank, it returns all data for the current version.

Available filter parameters:

See the SDW API for more details.

Using SDW keys

The easiest way to find and learn more about SDW series key is to browse the SDW website. After finding the series one is interested in, and applying the relevant filters (frequency, geographic area, etc), one can just copy the key:

screenshot

The SDW website also has provides all the necessary metadata, so it is much easier to explore data availability (in terms of available breakdowns, time periods, etc) directly on the website than it is to do it directly through the ecb package.

The ecb package supports using wildcards in the series key, which takes the form of simply leaving the wildcard dimension empty. For example, the key ICP.M.DE.N.000000.4.ANR retrieves HICP data for Germany (DE), while leaving the third dimension empty - ICP.M..N.000000.4.ANR - retrieves the same data for all available countries and country groups.

Instead of wildcarding, one can use the + operator to specify multiple values for a dimension. For example, ICP.M.DE.N.000000+XEF000.4.ANR retrieves both headline inflation (000000) and core inflation (XEF000). Learning that e.g. XEF000 corresponds to core inflation would be done by browsing the SDW website:

screenshot2

To remind oneself of what different values for different dimensions mean, one can use the get_dimensions) function, which returns a list of dataframes:

dims <- get_dimensions("ICP.M.DE.N.000000+XEF000.4.ANR")
lapply(dims, head)

Extended example

As a more extended example, we will retrieve data to plot the annual change in wages against the annual change in unemployment. Economic theory suggests a negative relationship between these two variables.

We start by retrieving the two series, using wildcards for the geographic area dimension:

unemp <- get_data("LFSI.M..S.UNEHRT.TOTAL0.15_74.T", 
                 filter = list(startPeriod = "2000"))

wages <- get_data("MNA.A.N..W2.S1.S1._Z.COM_HW._Z._T._Z.IX.V.N", 
                 filter = list(startPeriod = "2000"))

head(unemp)
head(wages)

To get a human-readable description of a series:

desc <- head(get_description("LFSI.M..S.UNEHRT.TOTAL0.15_74.T"), 3)
strwrap(desc, width = 80)

We now join together the two data sets:

suppressPackageStartupMessages(library(dplyr))
suppressPackageStartupMessages(library(lubridate))

unemp <- unemp %>% 
  mutate(obstime = convert_dates(obstime)) %>% 
  group_by(ref_area, obstime = year(obstime)) %>% 
  summarise(obsvalue = mean(obsvalue)) %>%
  ungroup() %>% 
  select(ref_area, obstime, "unemp" = obsvalue)

wages <- wages %>% 
  mutate(obstime = as.numeric(obstime)) %>% 
  select(ref_area, obstime, "wage" = obsvalue)

df <- left_join(unemp, wages)
head(df)

Finally, we plot the annual change in wages against the annual change in unemployment for all countries:

library(ggplot2)

df %>% 
  filter(complete.cases(.)) %>% 
  group_by(ref_area) %>% 
  mutate(d_wage = c(NA, diff(wage)) / lag(wage),
         d_unemp = c(NA, diff(unemp))) %>% 
  ggplot(aes(x = d_unemp, y = d_wage)) +
  geom_point() +
  facet_wrap(~ref_area, scales = "free") +
  theme_bw(8) +
  theme(strip.background = element_blank()) +
  geom_smooth(method = "lm") +
  labs(x = "\nAnnual change in unemployment", y = "Annual change in wages\n",
       title = "Relationship between wages and unemployment\n")

Disclaimer

This package is in no way officially related to, or endorsed by, the ECB.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.