Unemployment rates and trends

raw_metric definition change (edited to fix a unit)

The forecasting model uses three-year historical data to project five years forward, which assumes that patterns will continue. But if the historical period includes an anomaly (e.g., a one-time pandemic effect), the forecast will perpetuate that anomaly. Should we either exclude anomalies or should we use explicit assumptions to override the historical pattern?

South Africa vs Mexico: is retention comparable?