George A. Overstreet

A Bayesian framework for the combination of classifier outputs

We explore a Bayesian framework for constructing combinations of classifier outputs, as a means to improving overall classification results. We propose a sequential Bayesian framework to estimate the posterior probability of being in a certain class given multiple classifiers. This framework, which employs meta-Gaussian modelling but makes no assumptions about the distribution of classifier outputs, allows us to capture nonlinear dependencies between the combined classifiers and individuals. An important property of our method is that it produces a combined classifier that dominates the individuals upon which it is based in terms of Bayes risk, error rate, and receiver operating characteristic (ROC) curve. To illustrate the method, we show empirical results from the combination of credit scores generated from four different scoring models.

A study in the combination of two consumer credit scores

Credit scores measure the creditworthiness of individuals in a population of interest. In this paper, we employ the concepts of sufficiency and extraneousness to study the conditions under which scoring results can be improved through the combination of individual scores. The concept of sufficiency is used to identify scores that are dominant. Extraneousness is used to determine whether a particular score provides additional useful information relative to other scores. In addition, we employ a profit-based utility measure to evaluate the performance of different scores. We investigate the performance of a regression-based combination of a bureau credit score and an application credit score on a large historical data set. Our results show that the bureau score is dominated by the application score, but the bureau score is not extraneous to the combination. Thus, both scores contribute to the combined score, which indeed outperforms both of the scores upon which it is based.

Planning for a non-profit service: A study of U.S. credit unions☆

Abstract This article focuses on formal planning within U.S. credit unions. It provides an overview of credit unions within the U.S. financial services industry, and the result of a recent study of 283 such institutions. Results indicate that the propensity to plan is related to key organizational processes, structural configurations, administrative procedures, managerial perceptions of environmental predictability, and multiple performance measures. It was found that formal planning has different roles among various types of organizations and that managers should bear this in mind as they engage in planning processes within non-profit institutions. The results also suggest that formal planning facilitates communication between management and the board, directs the organization to focus on critical success factors, and results in superior performance along important criteria.

The power to borrow and lend: investigating the cultural context as part of the lending decision

This article addresses social contextual aspects of the lending decision not as irrational dimensions to be improved upon, but as important information that adds value to the rational decision-making process. It advocates a broader conceptualization of behavioural finance and the adoption of behavioural science theory and methods into credit scoring research. The article illustrates the use of behavioural sciences by presenting the results of an experimental study on the role of financial and social-cultural information on the lending decision process of experienced lenders in the United States. The study found that social-cultural information influenced the decision process in patterned and predictable ways. A future research agenda is outlined, and implications for credit scoring research and practice are discussed.

Energy-efficient homes and mortgage risk: crossing the chasm at last?

A topic of recent interest is energy-efficient houses. In the last twenty years, the US market has witnessed growth in the adoption of energy efficiency via various building technologies, design schemes, and eco-labels by developers, homebuilders, and homebuyers. Although housing and real estate scholars have begun explorations of the factors associated with the diffusion of energy efficiency, the subject has yet to be explored from the perspective of a critical stakeholder—the mortgage lender. This paper distinguishes itself from the extant literature in several ways. First, using a diffusion of innovation intellectual framework, the authors explore the role of the lender as an innovation gatekeeper. Second, as a review article, the authors summarize several key papers from the underwriting, credit scoring, and finance literature to help frame a number of market frictions that could mute further diffusion of energy efficiency in housing. Finally, the authors analyze these frictions and offer both market and public policy suggestions that may help overcome these frictions and increase opportunities for both private competitive advantage and market transparency.

Creating a value in Oversupplied Markets: The Case of Charlotesville, Vinia, Hotels

The hotel industry remains haunted by the room supply resulting from the 1980s building boom. Demand has recovered somewhat from the 1991 recession, but not enough to absorb the supply overhang. This mature industry, mired in overcapacity, is characterized by high exit barriers, low entry barriers, and numerous, diverse competitors. As a result the past four years have been a period of wholesale destruction of value, with many hotels selling at distressed prices. However, even under such adverse conditions, a few well-positioned players managed to achieve competitive advantage and create value. The Charlottesville, Virginia, market offers a microcosm of those conditions. Dissecting that market with an analysis of strategies shows that a differentiation strategy with careful cost containment was successful in creating value. In particular, limited-service entrants were able to sustain their competitive advantage. On the other hand, a differentiation strategy that depended on premium pricing failed to create value for three different properties. Finally, the analysis makes clear that although overhanging debt damaged value, even debt-free properties were not able to create value absent a differentiating strategy.

Simulating the impact of expanded delegation of dental procedures

This paper discusses a project underway at the University of Alabama in Birmingham Dental School. The study objective is to investigate the productivity of dental teams using therapists. Therapists, sometimes called technotherapists or expanded duty dental auxiliaries, are dental assistants who perform under the supervision of dentists some of the reversible procedures previously performed only by dentists. The study involved both an actual practice using varying numbers of therapists and a GPSS model which simulates practices with varying numbers of therapists. Some of the reasons are discussed for using the simulation model to confirm and expand the actual practice findings. The paper also presents the model and discusses some of its possible future applications. Productivity data from the actual practive are compared to data from simulations. Simulation data are presented which compares the services provided by a dental team using a therapist to the services provided under identical conditions when the therapist and her assistant are excluded from the team.