Search

Corporate bankruptcy prediction has attracted significant research attention from business academics, regulators and financial economists over the past five decades. However, much of this literature has relied on quite simplistic classifiers such as logistic regression and linear discriminant analysis (LDA). Based on a large sample of US corporate bankruptcies, we examine the predictive performance of 16 classifiers, ranging from the most restrictive classifiers (such as logit, probit and linear discriminant analysis) to more advanced techniques such as neural networks, support vector machines (SVMs) and "new age" statistical learning models including generalised boosting, AdaBoost and random forests. Consistent with the findings of Jones et al. (), we show that quite simple classifiers such as logit and LDA perform reasonably well in bankruptcy prediction. However, we recommend the use of "new age" classifiers in corporate bankruptcy modelling because: (1) they predict significantly better than all other classifiers on both the cross-sectional and longitudinal test samples; (2) the models may have considerable practical appeal because they are relatively easy to estimate and implement (for instance, they require minimal researcher intervention for data preparation, variable selection and model architecture specification); and (3) while the underlying model structures can be very complex, we demonstrate that "new age" classifiers have a reasonably good level of interpretability through such metrics as relative variable importances (RVIs).

In this study, we examine the predictive performance of a wide class of binary classifiers using a large sample of international credit ratings changes from the period 1983-2013. Using a number of financial, market, corporate governance, macro-economic and other indicators as explanatory variables, we compare classifiers ranging from conventional techniques (such as logit/probit and LDA) to fully nonlinear classifiers, including neural networks, support vector machines and more recent statistical learning techniques such as generalised boosting, AdaBoost and random forests. We find that the newer classifiers significantly outperform all other classifiers on both the cross sectional and longitudinal test samples; and prove remarkably robust to different data structures and assumptions. Simple linear classifiers such as logit/probit and LDA are found nonetheless to predict quite accurately on the test samples, in some cases performing comparably well to more flexible model structures. We conclude that simpler classifiers can be viable alternatives to more sophisticated approaches, particularly if interpretability is an important objective of the modelling exercise. We also suggest effective ways to enhance the predictive performance of many of the binary classifiers examined in this study.

The dissertation is directed toward a simulation study of social influence in small, task-oriented, groups composed of education professionals of differing status who hold differing beliefs about what constitutes the reportable physical abuse of elementary students by parents. It is asserted on philosophical grounds that simulation allows the development and refinement of computational, process-based, models that reflect the stratified nature of social and educational reality. More practically, simulation makes it possible to trace socio-cultural processes over time rather than simply settle for an input-output analysis. The possibility of more adequate explanation, and transformation, of social and educational systems makes simulation relatively superior to other research methods.Persons are modeled as computational objects that participate in social relations. The simulation is based on the social-psychological theory of expectation states and is expressly framed to allow comparison of: (1) a static model and process model of social influence and (2) the social theories of Giddens and Archer regarding the timing of social tie formation. National data are used to formulate models of professional belief concerning N = 111 cases of reported physical abuse. The chief findings are: (1) as applied to the group task of child abuse reporting, the collective decision outcomes obtained from the two models of social influence disagree by a number greater than that expected by chance alone, and (2) as modeled, there is essentially no difference in the simulated collective decision outcomes generated under each theory concerning the timing of social tie formation.Several refinements are needed. First, it is important to construct a more adequate characterization of the relevant beliefs of education professionals, best done via the construction of an ethnographic decision model for each professional type. Second, given the importance of social influence, the simulation should b