Analyzing Remote Work in European Countries

We’re making descriptive analytics about the Remote Work and European UbiWorkers data (Eurostat). Finland, Luxemburg, the Netherlands and Germany have more than 30% of his workforce in some kind of remote work modality. Norway surprising!

Carlos Vecina

2021-06-16 R, Post

1. Data downloading

As we always do, we are going to connect and download the desired data. In this case, our data source is the Eurostat. We download and read the data file.

library(tidyverse)
download.file("https://ec.europa.eu/eurostat/api/dissemination/sdmx/2.1/data/LFSA_EHOMP/?format=SDMX-CSV&compressed=false", "data.csv")

As a first analytical step, we are going to check the TOP 5 countries with employees in any kind of Remote Work / Work From Home (WFH) mode.

df_european_data <- read_csv(file = 'data.csv',
                             col_types = cols(sex = col_character()))  # Sex columns type

df_european_data_filtered <-df_european_data %>% 
  filter(freq == "A" & unit == "PC" & wstatus == "EMP" &
        sex == "T"  & age == "Y20-64" & geo != "SE") %>% 
  select(geo, TIME_PERIOD, OBS_VALUE, frequenc) %>% 
  rename(remote_perc=OBS_VALUE,
         country=geo)

df_european_data_filtered %>% 
  filter(frequenc == "NVR") %>% 
  mutate(remote_perc=if_else(frequenc == "NVR",
                             round(1-remote_perc/100, 3)*100,
                             round(remote_perc/100, 3)*100)) %>% 
  select(-frequenc) %>% 
  arrange(country, TIME_PERIOD) %>% 
  filter(TIME_PERIOD == "2020") %>% 
  slice_max(n=6, order_by=remote_perc) %>% 
  kable()

country	TIME_PERIOD	remote_perc
LU	2020	47.8
NL	2020	42.7
CH	2020	41.5
FI	2020	40.4
IS	2020	39.8
DK	2020	36.9

df_european_data_filtered_bar_plot <- df_european_data_filtered %>% 
  filter(frequenc == "NVR" & TIME_PERIOD =="2020") %>% 
  mutate(highlight = if_else(country == "EU27_2020","t","f"),
         remote_perc = if_else(frequenc == "NVR",
                               round(1-remote_perc/100,3),
                               round(remote_perc/100,3))) %>% 
  arrange(remote_perc) 
df_european_data_filtered_bar_plot %>% 
  ggplot(aes(country, remote_perc, fill=highlight)) +
  geom_bar(stat="identity", position="dodge") +
  geom_text(aes(label=paste0(remote_perc*100,"%")), size=2) +
  coord_flip() +
  scale_x_discrete(limits=df_european_data_filtered_bar_plot$country) +
  scale_fill_manual(values=c( "t"="tomato", "f"="paleturquoise3" ),
                    guide=FALSE) + 
  scale_y_continuous(labels=scales::percent) +
  labs(title="Employee Percentage(%) with any Remote Work modality by Country",
       caption = "Source: Eurostat")

And the TOP 10 with more relative increment?

df_european_data_filtered %>%  
  filter(frequenc == "NVR") %>% 
  mutate( remote_perc = if_else(frequenc == "NVR",
                                round(1-remote_perc/100,3)*100,
                                round(remote_perc/100,3)*100)) %>% 
  select(-frequenc) %>% 
  arrange(country, TIME_PERIOD) %>% 
  group_by(country) %>% 
  mutate(delta = (remote_perc - lag(remote_perc))/lag(remote_perc)*100) %>% 
  ungroup() %>% 
  filter(TIME_PERIOD=="2020") %>% 
  slice_max( n=10,order_by=delta) %>% 
  kable()

country	TIME_PERIOD	remote_perc	delta
CY	2020	7.4	196.00000
IT	2020	13.7	191.48936
BG	2020	3.0	172.72727
HU	2020	11.0	139.13043
RO	2020	3.2	128.57143
MT	2020	26.0	122.22222
EL	2020	10.4	100.00000
LT	2020	8.4	86.66667
ES	2020	15.2	80.95238
DE	2020	21.1	63.56589

2. Remote worker profile in Europe

(Usually vs Sometimes vs Never)

A good way to summarize the country labour market profile regarding remote work is to make a donut plot. It’s true that the human brain is not very good at comparing areas, but this plot will allow us to perceive the Remote Work adoption in each modality at a glance.

list_countries <- c("AT","BE","CH","CY","CZ",
                    "DE","DK","ES","EU15","EU27_2020","FI",
                    "FR","IS","IT","LU","MK",
                    "NL","NO","PL","PT","SK")
ggplot(df_european_data_filtered %>%
         filter(TIME_PERIOD=="2020" & (country %in% list_countries )) %>% 
         group_by(country) %>% 
         mutate(ymax=cumsum(remote_perc),
                ymin=if_else(row_number()==1,0,lag(ymax)),
                labelPosition=(ymax+ymin)/2,
                label=paste0(remote_perc, " %")) %>% 
         rename(freq=frequenc),
       aes(ymax=ymax, ymin=ymin, xmax=4, xmin=3, fill=freq)) +
  geom_rect() +
  geom_text(x=1.5,
            aes(y=labelPosition, label=label, color=freq),
            size=2.2,
            check_overlap = T)+
  scale_fill_brewer(palette=3) +  # donut color
  scale_color_brewer(palette=3) + # labs colour
  facet_wrap(~country) +
  coord_polar(theta="y") +
  theme_void() +
  xlim(c(-1, 4)) +
  labs(title="Remote Work composition by country",
       subtitle=" ",
       caption = "Source: Eurostat")

Remote Work temporal evolution

In order to bring more context, we are going to deep into the temporal evolution of each country. We are tracking the employees in each country and year that have been working in some Remote mode. We could see some clusters, with countries with steeper slope than others during the last year in the data reported.

df_european_data_filtered %>%
  filter(frequenc == "NVR" & (country %in% list_countries[list_countries != "CY"])) %>% 
  mutate( remote_perc = if_else( frequenc == "NVR",round(1-remote_perc/100,3),round(remote_perc/100,3)),) %>% 
  ggplot(aes(TIME_PERIOD, remote_perc, colour=remote_perc, group=country)) +
  geom_line() +
  facet_wrap(~country) +
  scale_colour_gradient(low = "yellow", high = "red", na.value = NA)+ 
  scale_y_continuous(labels=scales::percent)

3. Remote Work ~ Country GDP correlation

As a final thought, the data is pointing out to us that Northern Europe countries are doing more Remote Work. We are expecting that higher GDP countries are more prone to define a Work From Home model or hybrid ones, while the lower GDP ones have more difficulties or are not willing to enhance and embrace change.

The industrial sector composition, kind and size of national companies, politic ideology and government seem such a high impact reasons and probably act as a causality of the heterogenity across the European States.

In the next blog post, we are going to talk and go deeper analyzing the relation between the GDP and the WFH ratios, and discourse about it’s causes and effects.

You can find valuable Remote Work information and community at UbiWork. Work remotely from Spain & Portugal

Share this post using…