No abstract

This paper examines firm-level determinants of mature firm exits after economic distress. Using nested logit models and a sample of 6,118 distress-related exits in Belgium, we analyze the type of exit that distressed firms experience. We show that 41% of the firms in our sample exit through a court driven exit procedure (mainly bankruptcy), 44% are voluntarily liquidated and 14% are acquired, merged or split (hereafter M&A). Distressed firm exit follows two distinct stages. First, a firm either decides to exit voluntarily or is forced into bankruptcy, which is the least efficient exit strategy. Compared to bankruptcy, the probability of a voluntary exit increases with higher levels of cash, lower leverage, holding no secured debt and being embedded in a group. If a firm exits voluntarily, it enters a second stage and decides either to exit through voluntary liquidation or through a M&A. Conditional on not going bankrupt, the likelihood of voluntary liquidation compared to M&A increases with higher levels of cash or secured debt, with smaller size and with an absence of group relations. We contribute to the firm exit literature by jointly analyzing three exit types and showing that bankruptcy and voluntary liquidation are fundamentally different exit routes. While voluntary liquidation is an important exit route for distressed firms, most previous studies have failed to distinguish between bankruptcy and liquidation. We hence contribute to the exit literature by showing that bankruptcy, voluntary liquidation and M&A are fundamentally distinct exit routes for distressed firms, driven by different firm level characteristics and following a two-stage process

Firm exit, Distress, Court driven exit, Voluntary liquidation, Slack resources, Stakeholders,

Faced with the question as to whether failure prediction models can easily be transferred and applied to a new data setting, this study examines the performance of seven models on a dataset of Belgian company failures after re-estimation of the coefficients. The validation results indicate that some models are widely usable: they are strongly predictive when applied to the new data set. The Gloubos-Grammatikos models and Keasey-McGuinness appear among the best performing models, and also Ooghe-Joos-De Vos and Zavgren seem to be widely usable, respectively for failure prediction 1 and 3 years prior to failure. At the same time, the Altman and Bilderbeek models show very poor results when applied to the Belgian dataset. The best performing models seem to combine the right variables in an intuitively right sense and it appears that the combination of some types of variables generally leads to good predictive results. On the contrary, the estimation technique, complexity and number of variables do not explain the predictive performances (JEL: G33,M49).

No abstract