Predicting Corporate Failure and Turnaround: A Profitable Investment Strategy

Rico Voigt Hansen & Sebastian Skylv

Student thesis: Master thesis


This paper aims to answer the following question: Can predicted corporate failure probabilities be utilized in an investment strategy to generate an abnormal return, and can this return be further improved by adjusting the strategy according to predicted turnaround probabilities? The paper builds on the research by Campbell, Hilscher and Szilyi (2008) and starts out by predicting probabilities of corporate failure using Shumway’s (2001) dynamic logit specification. Distressed firms are defined in the paper as those belonging to the highest decile when firms are ranked according to their predicted probabilities of failure. Seven investment portfolios are then generated, experimenting with longing and shorting the distressed firms. Moreover, probabilities of corporate turnaround are predicted using a dynamic logit model. The analysis makes the unique contribution to the literature, that time spent in distress is a key determinant of the occurrence of corporate turnarounds. Finally, the predicted probabilities of turnaround are used as an attempt to enhance the returns of the optimal investment portfolio, by investing strategically in distressed firms with high probabilities of turnaround, while shorting those with low probabilities of turnaround. Though the turnaround prediction model is found to accurately predict the occurrence of turnarounds, the relating adjustments are found to have no significant impact on the portfolio returns. Interestingly, the two portfolios proposed by Campbell et al. perform most poorly. The optimal investment portfolio is found to be that which only goes long in distressed firms. The excess return is computed as the portfolio alpha in a Fama and French five-factor model. This portfolio generates an average excess return on the American market of 5.5% over the period 1973-2004. The authors suggest two main explanations of these findings: First, companies that are relatively distressed compared to the rest of the market are in general undervalued, and an abnormal return can be gained from investing in them. In contrast, Campbell et al. found that distressed companies were overvalued. Second, the underlying assumptions of the evaluation of excess return can be contested. Campbell et al. did not enter into a discussion of these assumptions. Aside from methodological biases arising from defining the concepts of failure, distress and turnaround and from modelling probabilities of failure and turnaround through a set of subjectively deemed relevant covariates, the authors of this paper highlight the importance of the assumption of a friction-free market. After accounting for such market frictions, the authors conclude that the premium on distressed stocks will diminish if not vanish altogether. These findings also challenge the abnormal returns reported by Campbell et al.

EducationsMSc in Finance and Investments, (Graduate Programme) Final Thesis
Publication date2020
Number of pages121