The paper
This article uses machine learning and widely available country data to extend the coverage of the World Bank’s Logistics Performance Index. It also identifies groups of country characteristics that help predict logistics performance.
Research findings
- The best-performing model explains nearly 90% of variation in unseen observations, with average prediction error of about 6%.
- The extended series provides a broader picture across countries and years; improvements are generally gradual, with different patterns across income groups.
Policy takeaways
- Use predicted indicators to broaden benchmarking and track longer-term logistics trends where survey observations are sparse.
- Investigate the business environment, institutions, and other predictive factors alongside transport infrastructure when diagnosing logistics performance; prediction provides a starting point for country analysis.