Working Paper No. 658
By Jamie Coen, William B Francis and May Rostom
This paper examines the determinants of credit union failure in the United Kingdom. Using regulatory data on credit unions, we estimate several discrete-time logit models and evaluate their predictive ability at one, two and three-year time horizons. We find that a small set of financial attributes related to capital adequacy, asset quality, earnings performance and liquidity is useful for early identification of troubled credit unions. Both in and out-of-sample results indicate that this parsimonious set of firm-level characteristics, augmented with national and regional unemployment rates, reliably identifies failures while keeping false alarm rates at modest levels. The results provide support for establishing early-warning criteria for supervisory use in monitoring credit unions.