Optimal cut-off for rare events and unbalanced misclassification costs

This paper develops a method for handling two-class classification problems with highly unbalanced class sizes and misclassification costs. When the class sizes are highly unbalanced and the minority class represents a rare event, conventional classification methods tend to strongly favour the majority class, resulting in very low detection of the minority class. A method is proposed to determine the optimal cut-off for asymmetric misclassification costs and for unbalanced class sizes. Monte Carlo simulations show that this proposal performs better than the method based on the notion of classification accuracy. Finally, the proposed method is applied to empirical data on Italian small and medium enterprises to classify them into default and non-default groups.     

Modelling small and medium enterprise loan defaults as rare events: the generalized extreme value regression model

A pivotal characteristic of credit defaults that is ignored by most credit scoring models is the rarity of the event. The most widely used model to estimate the probability of default is the logistic regression model. Since the dependent variable represents a rare event, the logistic regression model shows relevant drawbacks, for example, underestimation of the default probability, which could be very risky for banks. In order to overcome these drawbacks, we propose the generalized extreme value regression model. In particular, in a generalized linear model (GLM) with the binary-dependent variable we suggest the quantile function of the GEV distribution as link function, so our attention is focused on the tail of the response curve for values close to one. The estimation procedure used is the maximum-likelihood method. This model accommodates skewness and it presents a generalisation of GLMs with complementary log–log link function. We analyse its performance by simulation studies. Finally, we apply the proposed model to empirical data on Italian small and medium enterprises.     

Learning in and from projects: the learning modes and a learning capability model

Abstract

The notion of ‘project delivery’ is well embedded in and across the management and organizational sciences literature – generating a narrative that reflects and recognizes the instrumental nature of projects and programmes in strategy execution. Project management, as a distinct and well-established body of research enquiry, has increasingly sought to focus our attention on the impacts of complexity, risk and uncertainty in projects; the corollary being a desideratum to strengthen our theoretical understanding of how insight and learning from projects may influence improvements to organizational efficiency. The wider literature suggests that organizational learning remains a challenging proposition, particularly in the context of organizations operating in environments of high complexity. In this paper, we enhance the conversation on organizational learning through a series of case studies, generating evidence of thirteen ‘learning modes’. The paper proposes that mature organizations tend to exhibit a greater number of learning modes and that there is a tendency to capture and socialize knowledge with a greater emphasis on the context of the learning situation rather than the learning artefact in isolation. The empirical evidence gathered in this paper forms the basis of a capability model, characterized by the thirteen modes of learning. The model intimates that learning occurs, and is more effective, when knowledge and information are enacted in practice through the learning modes which form a nucleus of the organizational learning capability. The research concludes with a 'call to action' that emphasizes the strategic importance of learning practices and routines in project oriented-organizations.     

Firm internationalization and performance: Evidence for designing policies

The debate on the relationship between internationalization and performance is still open and is a hot topic for policymakers, but they ignore the impacts for most of the firms supported. The paper is based on a large database of seven years that considers quantitative-qualitative and control variables. To reduce industry and country heterogeneity we focused on the Italian automotive supply chain. We tested 5 modes of internationalization on firms’ profitability so to support the design of policies. Our results are partially in contrast with literature; therefore we identify specific environments where policies, namely for exporting, could be more effective.     

Genetic algorithms for the analysis of Bayesian hierarchical partition models

Hierarchical partition models (see Malec and Sedransk, 1992, Consonni and Veronese, 1995) aim at finding an optimal grouping (partition) of a set of experiments regarding a target variable. In this class of models the partition is regarded as an unknown parameter, and one of the main goals is computing the posterior distribution over the class of the possible partitions. This problem has been addressed in Sampietro and Veronese (1998), where a Metropolis-Hastings algorithm is applied. In this paper the performance of an alternative procedure, based on the logic of genetic algorithms, is evaluated. The results of the two approaches are compared, even if a conjoint use of them is to be advised.     

Hormesis: A Highly Generalizable and Reproducible Phenomenon With Important Implications for Risk Assessment

From a comprehensive search of the literature, the hormesis phenomenon was found to occur over a wide range of chemicals, taxonomic groups, and endpoints. By use of computer searches and extensive cross-referencing, nearly 3000 potentially relevant articles were identified. Evidence of chemical and radiation hormesis was judged to have occurred in approximately 1000 of these by use of a priori criteria. These criteria included study design features (e.g., number of doses, dose range), dose–response relationship, statistical analysis, and reproducibility of results. Numerous biological endpoints were assessed, with growth responses the most prevalent, followed by metabolic effects, reproductive responses, longevity, and cancer. Hormetic responses were generally observed to be of limited magnitude with an average maximum stimulation of 30 to 60 percent over that of the controls. This maximum usually occurred 4- to 5-fold below the NOAEL for a particular endpoint. The present analysis suggests that hormesis is a reproducible and generalizable biological phenomenon and is a fundamental component of many, if not most, dose–response relationships. The relatively infrequent observation of hormesis in the literature is believed to be due primarily to experimental design considerations, especially with respect to the number and range of doses and endpoint selection. Because of regulatory considerations, most toxicologic studies have been carried out at high doses above the low-dose region where the hormesis phenomenon occurs.     

Tests of Conditional Predictive Ability

We propose a framework for out‐of‐sample predictive ability testing and forecast selection designed for use in the realistic situation in which the forecasting model is possibly misspecified, due to unmodeled dynamics, unmodeled heterogeneity, incorrect functional form, or any combination of these. Relative to the existing literature (Diebold and Mariano (1995) and West (1996)), we introduce two main innovations: (i) We derive our tests in an environment where the finite sample properties of the estimators on which the forecasts may depend are preserved asymptotically. (ii) We accommodate conditional evaluation objectives (can we predict which forecast will be more accurate at a future date?), which nest unconditional objectives (which forecast was more accurate on average?), that have been the sole focus of previous literature. As a result of (i), our tests have several advantages: they capture the effect of estimation uncertainty on relative forecast performance, they can handle forecasts based on both nested and nonnested models, they allow the forecasts to be produced by general estimation methods, and they are easy to compute. Although both unconditional and conditional approaches are informative, conditioning can help fine‐tune the forecast selection to current economic conditions. To this end, we propose a two‐step decision rule that uses current information to select the best forecast for the future date of interest. We illustrate the usefulness of our approach by comparing forecasts from leading parameter‐reduction methods for macroeconomic forecasting using a large number of predictors.