Percental Change in Time Series Data

I just want to generate a variable that captures the relative decline of a variable (v2x_polyarchy) from year to year.This is what my data frame 'episodes_demreg' basically looks like:

country_name year   demreg_id  v2x_polyarchy
<chr>        <dbl>  <dbl>      <dbl>   
Suriname     1974   1          0.536 
Suriname     1975   1          0.375
Suriname     1979   2          0.713
Suriname     1980   2          0.226
Suriname     1990   3          0.743
Suriname     1991   3          0.455

Based on the suggestion on very similar thread (Percent change difference for time series data) I tried it like this:


episodes_demreg <- episodes_demreg %>%
  group_by(demreg_id) %>%
  mutate (decline_year_relative = ((v2x_polyarchy - lag(v2x_polyarchy)) / lag(v2x_polyarchy)) * 100)

However, this does not work, as it results in the value '0' for each observation. Going back and forth revealed that obviously 'lag' does not work and indeed it subtracts the observation of the same row and NOT the row before. Any ideas why this happens?

Solution

You are using lag from the stats package, while you want to use lagfrom the dplyr package.

You can either specify the package when calling lag

  group_by(demreg_id) %>%
  mutate (decline_year_relative = ((v2x_polyarchy - dplyr::lag(v2x_polyarchy)) / dplyr::lag(v2x_polyarchy)) * 100)

or set dplyr as the default package to use lag

lag <- dplyr::lag