J. Samuel Baixauli

The bias of unhealthy SMEs in bankruptcy prediction models

Purpose – This paper aims to construct a financial health indicator to define the degree of financial health in order to decontaminate the estimation sample and to make predictions that are not biased by unhealthy firms.Design/methodology/approach – The binomial logit model is used to examine the likelihood that a firm will go bankrupt. In order to evaluate the accuracy of the estimated models, measures proposed by the Basel Committee on Banking Supervision are applied: cumulative accuracy profile (CAP) and the receiver operating characteristics (ROC).Findings – The proposed financial health indicator permits the heterogeneity of the firms to be reduced as well as identifying a strong firm sample to estimate the bankruptcy probability accurately.Originality/value – A drawback of all bankruptcy prediction models comes from the fact that bankruptcy is an example of a homogeneous observable qualitative response while non‐bankruptcy would be expected to be represented by a healthy firm. However, the non‐bankr...

Analysis of the conditional stock-return distribution under incomplete specification

Abstract The analysis of the distribution of the stock-returns plays an important role in financial theory since a distributional assumption is required for mean–variance portfolio theory, theoretical models of capital asset prices, determining the price of derivative products, efficient estimation by maximum likelihood procedure and establishing forecasting confidence intervals. Up to now, most of the studies analyze the unconditional distribution of the stock-returns using the likelihood ratio test, the Schwarz Information Criterion or by graphical analysis. However, the analysis of the conditional distribution is very important since most of the empirical financial studies use high-frequency data and these data present dynamic structure. In this paper, we focus on the conditional distribution. We design different bootstrap mechanisms for comparing the goodness-of-fit of several functional forms postulated under the null hypothesis for individual Spanish stocks and for the General Index of the Madrid Stock Market. We use the Cramer–von Mises test statistic, based on the standardized residuals and assuming that the null distribution can depend on some unknown parameters.The analysis of the distribution of the stock-returns plays an important role in financial theory since a distributional assumption is required for mean–variance portfolio theory, theoretical models of capital asset prices, determining the price of derivative products, efficient estimation by maximum likelihood procedure and establishing forecasting confidence intervals. Up to now, most of the studies analyze the unconditional distribution of the stock-returns using the likelihood ratio test, the Schwarz Information Criterion or by graphical analysis. However, the analysis of the conditional distribution is very important since most of the empirical financial studies use high-frequency data and these data present dynamic structure. In this paper, we focus on the conditional distribution. We design different bootstrap mechanisms for comparing the goodness-of-fit of several functional forms postulated under the null hypothesis for individual Spanish stocks and for the General Index of the Madrid Stock Market. We use the Cr amer–von Mises test statistic, based on the standardized residuals and assuming that the null distribution can depend on some unknown parameters. 2003 Elsevier B.V. All rights reserved.

Combining structural models and accounting-based models for measuring credit risk in real estate companies

Purpose - The purpose of this paper is to, first, analyse to what extent the default probability based on structural models provides additional information and that accounting ratios do not contemplate. Second, to design hybrid models by including the default probability from structural models as explanatory variable, in addition to accounting ratios, in order to evaluate the differences in the accuracy of default predictions using an accounting-based model and a hybrid model. Design/methodology/approach - The authors calculated the scores from the accounting models annually during the period from 2003 to 2007 and estimated several structural models. Findings - The results show that the market information obtained from the structural models includes additional information not reflected in the accounting information. Also, it can be concluded that including default probability from structural models as an explanatory variable allows the out-sample predictive capacity of accounting-based models to be improved. Practical implications - The study highlights the importance of combining a structural model with an accounting model rather than expending energy on determining which of the two provides a greater predictive capacity. In fact, recent literature demonstrates no superiority of one approach over the other because both approaches capture different aspects related to the risk of bankruptcy in companies and they should be combined to improve credit risk management. Originality/value - This study expands on the existing literature on the probability of business failure in the real estate sector. The authors present a comparative analysis of the accuracy of default predictions using accounting-based models and hybrid models which will consider the default probability implicit in market information.

Toeholds and takeover probability: implications for investment strategies

Purpose - The purpose of this paper is to propose various toehold indicators and analyse whether the models incorporating these indicators can be used to establish investment strategies. Design/methodology/approach - Logistic regression is used to test toehold indicator significance. Findings - The results reflect that the designed measures are positively correlated to the likelihood of launching a takeover, although the power of the models to predict out-sample takeovers is moderate, between 60.71 percent and 71.59 percent. The indicators allow us to design strategies which offer positive abnormal returns. In particular, abnormal return over the Fama-French factors is 0.5 percent. Originality/value - Toeholds are used to initiate takeover processes. As previous studies have indicated, a toehold increases the likelihood of success in a tender offer. Nevertheless, the studies on takeover prediction do not include the toehold since it is a variable which is unobservable prior to the announcement of a takeover bid.

On the accuracy of loss-given-default prediction intervals

Purpose - The purpose of this paper is to critically analyze the common assumption, made by many credit risk models such as the Moodys KMV Loss-Calc model, of a Design/methodology/approach - Simulation experiments were conducted to highlight the potential problems associated with this distributional assumption in constructing prediction intervals for LGD. Findings - The simulation experiments show that, when starting from a different assumption concerning the shape of the population distribution, the beta distribution does not perform well in constructing prediction intervals for LGD. Originality/value - The analysis performed in this study addresses a relevant subject. Indeed, a correct estimate of a credit exposure LGD is particularly relevant not only for internal risk management and management purposes, but also for regulatory reasons within the context of the internal ratings based approach of the recently approved capital regulation framework (Basel II).

Coverage properties of beta estimated prediction intervals for multimodal recovery rates

In this paper we use bootstrap methodology to achieve accurate estimated prediction intervals for recovery rates. In the framework of the LossCalc model, which is the Moodys KMV model to predict loss given default, a single beta distribution is usually assumed to model the behaviour of recovery rates and, hence, to construct prediction intervals. We evaluate the coverage properties of beta estimated prediction intervals for multimodal recovery rates. We carry out a simulation study, and our results show that bootstrap versions of beta mixture prediction intervals exhibit the best coverage properties.

On the performance of recovery rate modeling

To ensure accurate predictions of loss given default it is necessary to test the goodness-of-fit of the recovery rate data to the Beta distribution, assuming that its parameters are unknown. In the presence of unknown parameters, the Cramer-von Mises test statistic is neither asymptotically distribution free nor parameter free. In this paper, we propose to compute approximated critical values with a parametric bootstrap procedure. Some simulations show that the bootstrap procedure works well in practice.

Hypothesis of Normality in the Context of the Market Model

This paper discusses the normality assumption of the market model errors, convention- ally accepted. Some other possible specifications are proposed and their performance is testing using a test statistic based on the empirical distribution function of the residuals of the model and assuming that the null distribution can depend on some unknown parameters. The parametric boot- strap method is used. Empirical evidence is provided using a sample of thirty companies of the Spanish Stock Market.