3.1 Companies Registration Trend

We have the year column ready for quick charts. So first we can quickly see in which year industries are registering.

df.companies %>% 
  group_by(register_year, register_month) %>% 
  count() %>% 
  ungroup() %>% 
  mutate(register_month = as.character(register_month)) %>% 
  bind_rows(group_by(., register_year) %>% 
              summarise(register_month = "Total", 
                        n = sum(n)))%>%   
  mutate(register_month = factor(register_month, levels = c(1:12,"Total"))) %>% 
  spread(register_month, n) %>% 
  gt() %>% 
  
  fmt_missing(columns = 1:13,
              missing_text = "") %>% 
  cols_label(
    register_year = "Registration Year (Nepali)",
  ) %>% 
  tab_spanner(
    label = "Month (Nepali)",
    columns = vars(1,2,3,4,5,6,7,8,9,10,11,12)
  ) %>% 
  tab_style(
    style = cells_styles(
      text_size = px(12)
    ),
    locations = cells_data(
      columns = everything()
    )
  )

registration_years <- unique(df.companies$register_year)
df.companies %>% 
  group_by(register_year) %>% 
  count() %>% 
  modelr::add_predictions(loess(n ~ register_year, data = .), "prediction") %>% 
  mutate(prediction = ifelse(register_year<2046,NA,prediction)) %>% 
  ggplot(aes(x=register_year, y=n)) +
  geom_bar(stat="identity") +
  scale_x_continuous(labels = registration_years, breaks = registration_years) +
  geom_line(aes(y=prediction), linetype = "longdash", color = "blue", size = 0.3) +
  labs(x="Registration Year",y="") +
  theme(axis.text.x = element_text(angle = 75, vjust = 0.5, size = 7.5))

3.2 Registratration Trend with FDI

Our plan is to explore FDI. We have already segregated the FDI share for each industry. Lets see the trend of FDI share in the registered industries.

df.companies %>% 
  filter(total_capital > 0) %>% 
  mutate(has_fdi = ifelse(foreign_percent>0,"Yes","No")) %>% 
  group_by(has_fdi, register_year) %>% 
  count() %>% 
  ggplot(aes(x = register_year, y = n, fill = has_fdi)) +
  geom_bar(stat= "identity") +
  scale_x_continuous(labels = registration_years, breaks = registration_years) +
  labs(x="Registration Year",y="",fill="Has FDI?") +
  theme(axis.text.x = element_text(angle = 75, vjust = 0.5, size = 7.5),
        legend.position = c(0.92,0.85))

We see that FDI started flowing in from 2069 only and significant number of industries are FDI invested. It’s not clear from the data whether Nepal received FDI before that or not.

But we will limit our discussion to this data set only. It doesn’t give full picture of FDI, but we want to see whether the Lawmakers thoughts on banning FDI justifies or not.

So we will limit the analysis from 2069 onwards only.

3.3 FDI and Non-FDI industry exploration

The following chart considers the industry data from 2069 onwards. Here we are trying to see where each industry (of either small, medium or large scale) resides in the chart of employment against total capital.

df.companies %>% 
  filter(register_year >=2069) %>% 
  filter(total_capital > 0) %>% 
  mutate(has_fdi = ifelse(foreign_percent>0,"Yes","No")) %>% 
  ggplot(aes(x=total_capital, y=employment, color=scale)) +
  geom_point(alpha = 0.2) +
  scale_x_log10(label = comma) +
  scale_y_log10() +
  facet_grid(has_fdi ~ .) +
  labs(x = "Total Capital", y = "Employment")

At quick glance, we see that most of the FDI industries are of small scale. Though some of the small scaled industries (blue dots) are also on the region of large scale industries (red dots). This raises the question on data authenticity, on the definition of scale.

3.4 FDI Share breakdown

There are industries with as little as 2% FDI as well with 100% FDI. Lets see the breakdown based on FDI percentage share.

df.companies %>% 
  filter(foreign_percent > 0) %>% 
  mutate(foreign_percent_range = cut(foreign_percent, 
                                     breaks = c(0,11,49,99.99,100),
                                     labels = c("10 or less than 10%","11 to 49%", "50 to less than 100%", "100%"),
                                     include.lowest = TRUE)) %>% 
  group_by(foreign_percent_range) %>% 
  count() %>% 
  mutate(percent = n/sum(.$n)) %>% 
  ungroup() %>% 
  gt() %>% 
  cols_label(
    foreign_percent_range = "FDI % share ",
    n = "Number of Registered Industries",
    percent = "Percentage of Industries"
  ) %>% 
  fmt_percent(
    columns = vars(percent)
  )

82% of registered industries are 100% FDI. So we can further narrow our analysis to 100% FDI invested data only.

3.5 Exploring 100% FDI industries

Further analysis will be carried to those industries with 100% FDI.

df.companies.mean <- df.companies %>% 
  filter(foreign_percent == 100) %>% 
  group_by(scale) %>% 
  summarise(
    employment = mean(employment, na.rm = TRUE),
    total_capital = mean(total_capital, na.rm = TRUE)
  )

df.companies %>% 
  filter(foreign_percent == 100) %>% 
  ggplot(aes(x = total_capital, y = employment, color = scale)) +
  geom_point(alpha = 0.2, size = 1) +
  geom_point(data = df.companies.mean, size = 3, alpha = 1) +
  scale_x_log10(label = comma) +
  scale_y_log10() +
  labs(x = "Total Capital", y = "Employment")

df.companies %>% 
  filter(foreign_percent == 100) %>% 
  ggplot(aes(x=scale, y=employment)) +
  geom_boxplot(alpha=0.5) +
  geom_jitter(alpha=0.05) +
  stat_summary(fun.y=mean, colour="darkred", geom="point", 
               shape=18, size=3,show_guide = FALSE) +
  scale_y_log10()

df.companies %>% 
  filter(foreign_percent == 100) %>% 
  ggplot(aes(x=scale, y=total_capital)) +
  geom_boxplot(alpha=0.5) +
  geom_jitter(alpha=0.05) +
  stat_summary(fun.y=mean, colour="darkred", geom="point", 
               shape=18, size=3,show_guide = FALSE) +
  scale_y_log10(label = comma)

3.6 Sector-wise analysis

Again we are limiting the analysis to 100% FDI industries. Lets see the summary of sector wise breakdown of 100% FDI industries.

df.companies %>% 
  filter(foreign_percent == 100) %>% 
  group_by(category) %>% 
  summarise(industry = n(), 
            employment = sum(employment, na.rm = TRUE), 
            capital = sum(total_capital, na.rm = TRUE)) %>% 
  mutate(category = tolower(category)) %>% 
  bind_rows(group_by(.) %>% 
              summarise(category = "Total",
                        industry = sum(industry),
                        employment = sum(employment),
                        capital = sum(capital))
            ) %>% 
  mutate(capital_per_capita = capital/industry,
         employment_per_capita = employment/industry) %>% 
  gt() %>% 
  cols_label(
    category = "Sector",
    industry = "No of Industry",
    employment = "Total",
    employment_per_capita = "Per industry",
    capital = "Total",
    capital_per_capita = "Per industry"
  ) %>% 
  cols_move(
    columns = vars(employment_per_capita),
    after = vars(employment)
  ) %>% 
  tab_spanner(
    label = "Employment",
    columns = vars(employment, employment_per_capita)
  ) %>% 
  tab_spanner(
    label = "Capital",
    columns = vars(capital, capital_per_capita)
  ) %>% 
  fmt_number(
    columns = vars(capital_per_capita, capital, employment, employment_per_capita)
  ) %>% 
  tab_style(
    style = cells_styles(
      text_size = px(12)
    ),
    locations = cells_data(
      columns = everything()
    )
  )  

3.7 Sector-wise scatterplot with mean

The big dot is the mean for each category.

df.companies.category.mean <- df.companies %>% 
  filter(foreign_percent == 100) %>% 
  group_by(category) %>% 
  summarise(
    employment = mean(employment, na.rm = TRUE),
    total_capital = mean(total_capital, na.rm = TRUE)
  )

df.companies %>% 
  filter(foreign_percent == 100) %>% 
  ggplot(aes(x = total_capital, y = employment, color = tolower(category))) +
  geom_point(alpha = 0.1, size = 1) +
  geom_point(data = df.companies.category.mean, size = 3) +
  scale_x_log10(label = comma) +
  scale_y_log10() +
  labs(x = "Total Capital", y = "Employment", color = "Sector")

3.8 District-wise summary

Here’s district-wise summary for 100% FDI industries.

df.companies %>% 
  filter(foreign_percent == 100) %>% 
  group_by(district) %>% 
  summarise(industries = n(), 
            employment = sum(employment, na.rm = TRUE), 
            capital = sum(total_capital, na.rm = TRUE)) %>% 
  arrange(desc(employment), desc(capital)) %>% 
  gt() %>% 
  cols_label(
    district = "District",
    industries = "No of Industries",
    employment = "Total Employment",
    capital = "Total Capital"
  ) %>% 
  fmt_number(
    columns = vars(capital)
  ) %>% 
  tab_style(
    style = cells_styles(
      text_size = px(12)
    ),
    locations = cells_data(
      columns = everything()
    )
  )

df.companies %>% 
  filter(foreign_percent == 100) %>% 
  mutate(district_group = ifelse(tolower(district) %in% c("kathmandu","lalitpur","bhaktapur"), "Kathmandu Valley", "Rest")) %>% 
  group_by(district_group) %>% 
  summarise(industries = n(), 
            employment = sum(employment, na.rm = TRUE), 
            capital = sum(total_capital, na.rm = TRUE)) %>% 
  arrange(desc(employment), desc(capital)) %>% 
  mutate(capital_percent = capital/sum(capital),
         employment_percent = employment/sum(employment),) %>% 
  gt() %>% 
  cols_label(
    district_group = "Valley and Rest",
    industries = "No of Industries",
    employment = "Total",
    employment_percent = "% Share",
    capital = "Total",
    capital_percent = "% Share"
  ) %>% 
  cols_move(
    columns = vars(employment_percent),
    after = vars(employment)
  ) %>% 
  tab_spanner(
    label = "Employment",
    columns = vars(employment, employment_percent)
  ) %>% 
  tab_spanner(
    label = "Capital",
    columns = vars(capital, capital_percent)
  ) %>% 
  fmt_number(
    columns = vars(capital)
  ) %>% 
  fmt_percent(
    columns = vars(capital_percent, employment_percent)
  ) %>% 
  tab_style(
    style = cells_styles(
      text_size = px(12)
    ),
    locations = cells_data(
      columns = everything()
    )
  )