The synthetic representation of the variation measured by the indicators is obtained by calculating their variations over the long-term reference period (through the already available time series):

• Long-term variation (latest available year to \(t_0\), where \(t_0\) is the starting point of the available time series).

If the available time series do not allow adequate estimates, they are not processed.

The variations are then classified according to the values of a Compound Annual Growth Rate (CAGR), which is calculated in the following way \[CAGR = { { \left( \frac{y_t}{y_{t_0}} \right)}^\frac{1}{t - t_0} -1 } \] where \(t_0\) is the reference year, \(t\) is the most recent year and \(y\) the value of the indicator over the two years.

For indicators with a positive polarity (i.e. those whose increase indicates compliance with the goals) the following categorisation applies:

• Clear improvement for CAGR values above 0,05 (i.e. for annual growth rates above 5%);
• Slight improvement for CAGR values lower than or equal to 0,05 and higher than 0,01;
• Stable for CAGR values lower than or equal to 0,01 and higher than -0,01;
• Slight worsening for CAGR values lower than or equal to -0,01 and higher than -0,05;
• Clear worsening for CAGR values lower than or equal to -0,05.

The scale applies in the opposite way for indicators with a negative polarity.

The procedure is based on the methodology described in the SDG Report 2019 of the National Institute of Statistics (ISTAT) on pages 37 and 38.