Signal · indexed from World Bank Documents & Reports
World Bank
This paper develops an empirical model to predict episodes of debt servicing difficulties (“debt distress”) in low-income countries, with three main contributions to the existing literature. First, it develops more refined measures of external debt distress episodes that allow timing the onset of distress episodes with increased precision. Second, it develops a systematic algorithm to comprehensively assess the out-of-sample predictive performance of more than 550,000 candidate binary prediction models using J-K-fold cross-validation. Third, it tests whether more sophisticated machine learning algorithms can outperform simple probit models. The paper finds that simple single-equation probit models have better predictive power for debt distress than more sophisticated algorithms and are comparable in terms of predictive performance to important policy benchmarks such as the IMF and World Bank debt sustainability framework for low-income countries.
Provenance
The notice above is the public record. Signing in adds the analysis: saved views and alerts on searches like this one, workflow and pipeline, exports and full workspace tooling.
| Field | How we know | State | Confidence |
|---|---|---|---|
| Publication date | Official API or feed | Data quality: Source-backed | 100 |
States describe GLODA's confidence in its own fields, never a judgement on the notice or its parties. Methodology