Investigating Yield Curve Inversions and Recession Forecasting: Evidence from 32 Economies
Brooklyn Pak
*
Social Sciences Division, STEM Science Center, 111 Charlotte Place, Suite 100, Englewood Cliffs, NJ 07632, USA.
*Author to whom correspondence should be addressed.
Abstract
Background: Yield curve inversions have long been considered an important indicator of future economic downturns and financial instability. This study investigates the relationship between yield curve inversions and recession forecasting across 32 economies, examining whether inverted yield spreads consistently predict economic contractions. The research contributes to the understanding of Macroeconomics and financial market behavior by assessing the reliability of yield curves as a global recession forecasting tool.
Aims: This study examined whether the 10-year/3-month government bond yield spread predicts recessions across a broad international sample and whether forecasting performance depends on monetary-policy structure, recession base rates, and country-level institutional regimes.
Study Design: The paper used a quantitative, cross-country forecasting design based on monthly macro-financial panel data and country-level logistic regression models.
Place and Duration of Study: The analysis covered 32 advanced and emerging market economies from January 2000 to December 2025, using publicly available data from the Federal Reserve Economic Data database, the OECD, the IMF World Economic Outlook, and national recession chronologies.
Methodology: The main predictor was the 10-year minus 3-month yield spread. For each country and for forecast horizons from 1 to 24 months, recession probability was modeled using logistic regression. Predictive performance was evaluated using the receiver operating characteristic area under the curve (ROC-AUC), with country-level and grouped comparisons used to assess heterogeneity.
Results: Predictive power was highly uneven. The United Kingdom (AUC = 0.918) and the United States (AUC = 0.737 at the 12-month horizon and up to 0.870 at longer horizons) showed strong recession-prediction performance. In contrast, many eurozone economies clustered near the no-skill benchmark, including France (AUC = 0.488), Spain (AUC = 0.487), Greece (AUC = 0.449), and Ireland (AUC = 0.429). The pooled global panel reached only modest performance, with AUC peaking at approximately 0.633 around 7 to 9 months ahead.
Conclusion: The yield curve is not a universal recession indicator. It is most informative in countries with independent monetary policy and market-determined bond yields, while its usefulness is limited in monetary unions, policy-suppressed regimes, and countries with high unconditional recession frequency. The findings support the use of regime-aware models rather than a single global yield-curve rule.
Keywords: Yield curve, recession forecasting, term spread, ROC-AUC, logistic regression, eurozone, monetary policy, financial economics, cross-country predictability