Hierarchical Generalized Linear Models and Frailty Models with Bayesian Nonparametric Mixing

This paper proposes Bayesian nonparametric mixing for some well-known and popular models. The distribution of the observations is assumed to contain an unknown mixed effects term which includes a fixed effects term, a function of the observed covariates, and an additive or multiplicative random effects term. Typically these random effects are assumed to be independent of the observed covariates and independent and identically distributed from a distribution from some known parametric family. This assumption may be suspect if either there is interaction between observed covariates and unobserved covariates or the fixed effects predictor of observed covariates is misspecified. Another cause for concern might be simply that the covariates affect more than just the location of the mixed effects distribution. As a consequence the distribution of the random effects could be highly irregular in modality and skewness leaving parametric families unable to model the distribution adequately. This paper therefore proposes a Bayesian nonparametric prior for the random effects to capture possible deviances in modality and skewness and to explore the observed covariates' effect on the distribution of the mixed effects.     

Bayesian semiparametric inference for the accelerated failure-time model

Bayesian semiparametric inference is considered for a loglinear model. This model consists of a parametric component for the regression coefficients and a nonparametric component for the unknown error distribution. Bayesian analysis is studied for the case of a parametric prior on the regression coefficients and a mixture-of-Dirichlet-processes prior on the unknown error distribution. A Markov-chain Monte Carlo (MCMC) method is developed to compute the features of the posterior distribution. A model selection method for obtaining a more parsimonious set of predictors is studied. The method adds indicator variables to the regression equation. The set of indicator variables represents all the possible subsets to be considered. A MCMC method is developed to search stochastically for the best subset. These procedures are applied to two examples, one with censored data.     

Modeling interest rate cycles in India

The present study tries to examine the behaviour of various Indian interest rates such as call money rate, and yields on secondary market securities with maturity periods of 15–91 days, 1-year, 5-years and 10-years. In the first stage, the study investigates the determinants of interest rates and finds that although the interest rates depend on some domestic macroeconomic variables such as yield spread and expected exchange rate, they are mainly affected by the movements of international interest rates, although with some lags. The policy variables such as Bank Rate and Federal Funds Rate did not show any significant impact on any of the interest rates. Further, it was found that the interest rates in the very recent period show some cyclical movements similar to that of the developed countries. Future behaviour of interest rates show that the present cycle of each interest rate would peak at different time points. This expected behaviour in domestic interest rates could be due to the integration of the domestic economy with the international money and financial market. This trend may be same in most of the emerging economies of Asia.     

KNOWLEDGE SPILLOVERS AND OUTPUT PER WORKER: AN INDUSTRY‐LEVEL ANALYSIS FOR OECD COUNTRIES

This study analyzes the impact of knowledge spillovers on output per worker at the industry level using a primal production function approach. The article makes three different contributions to the international spillovers literature: (1) it identifies trade‐related spillovers under alternative assumptions regarding the information transferred through imports; (2) it explores the importance of horizontal and vertical foreign direct investment (FDI) in knowledge spillovers; and (3) it looks at how institutional factors determine the impact of FDI‐related spillovers on productivity. The main findings of the study are: (1) international knowledge spillover is an important driver of industry output per worker, and the magnitude of this spillover effect varies with alternative assumptions about the information content embodied in imports, while high technology industries benefit significantly more from import‐related knowledge spillovers; and (2) the gains from FDI spillovers are primarily horizontal, but when institutional factors are considered, countries with stronger protection of intellectual property rights and a high “ease of doing business” tend to experience a substantial increase in the effectiveness of both horizontal and vertical FDI‐related spillovers. (JEL E24, F1, F6, O3, O4)     

A Bayesian prediction using the skew Gaussian distribution

A model based on the skew Gaussian distribution is presented to handle skewed spatial data. It extends the results of popular Gaussian process models. Markov chain Monte Carlo techniques are used to generate samples from the posterior distributions of the parameters. Finally, this model is applied in the spatial prediction of weekly rainfall. Cross-validation shows that the predictive performance of our model compares favorably with several kriging variants.     

Spiralling up and spiralling down: implications of a long‐term study of temperament and conduct disorder for social work with children

A model of social work grounded in the integration of the social sciences is presented. In social work with children this is exemplified by the importance of difficult childhood temperament and the goodness of fit between parenting practices in the evolution of conduct disorders. A professional model in which the social worker is an equal status member of a cooperative team of professionals engaged in treatment and research is stressed. The basis for this model of social work is that of the person‐in‐environment in which loopbacks from the child’s environment reinforce or diminish the evolution of conduct disorder in the process of spiralling down or spiralling up. This process is illustrated by two extensive case histories. In one the spiralling down process has been checked through intensive social work inputs. Nevertheless, in a longitudinal study it is shown that without such interventions difficult temperament observed at age 3 has significant links to conduct disorders, and movement into care or institutional life at age 17. Severe conduct disorders which lead to permanent school exclusion are likely to be expensive.     

Bayesian bridge regression

Classical bridge regression is known to possess many desirable statistical properties such as oracle, sparsity, and unbiasedness. One outstanding disadvantage of bridge regularization, however, is that it lacks a systematic approach to inference, reducing its flexibility in practical applications. In this study, we propose bridge regression from a Bayesian perspective. Unlike classical bridge regression that summarizes inference using a single point estimate, the proposed Bayesian method provides uncertainty estimates of the regression parameters, allowing coherent inference through the posterior distribution. Under a sparsity assumption on the high-dimensional parameter, we provide sufficient conditions for strong posterior consistency of the Bayesian bridge prior. On simulated datasets, we show that the proposed method performs well compared to several competing methods across a wide range of scenarios. Application to two real datasets further revealed that the proposed method performs as well as or better than published methods while offering the advantage of posterior inference.     

REVISITING ALTERNATIVE THEORETICAL PARADIGMS IN MANUFACTURING STRATEGY

Testing and cross‐validation of theories and paradigms are necessary to advance the field of manufacturing strategy. When the findings of one study are also obtained in other studies, using entirely different databases, we become more confident in the results. Replication alleviates concerns about spurious results and is one motivation for this study. We examine aspects of the tradeoffs concept, production competence paradigm, and a manufacturing strategy taxonomy framework. In regard to the tradeoffs concept, we found evidence of tradeoffs between some, but certainly not all, manufacturing capabilities of quality, cost, delivery, and customization. The relationships get sharper when controlling for process choice. For example, the tradeoff between cost and customization is particularly strong between plants that have different process choices. We find that such tradeoffs can change, or even disappear, however, once the process choice is in place. With respect to the production competence paradigm, our analysis shows a statistically significant correlation between production competence and operations performance in batch shops, but not in plants with other process choices. Finally, using variables similar to those of Miller and Roth, our data produced three similar clusters even though their unit of analysis was much more macro than ours. Controlling for process choice is consistent with the current manufacturing strategy literature that emphasizes dynamic development of capabilities within the context of path dependencies. A major argument of this strand of research is that operations decisions not only affect current capabilities, but also set the framework for development of capabilities in the future. That being the case, controlling for process choice (or other factors such as industry or markets) should contribute to the understanding of capability‐development paths adopted by different manufacturing plants. In short, we found at least partial support for each of the theories examined here, even though the theories seem on the surface to be contradictory and mutually exclusive. Controlling for process choice or other measures of dependency goes a long way in uncovering consistency across different theories and empirical studies in operations management.     

Hierarchical Bayesian models for predicting spatially correlated curves

Functional data analysis has emerged as a new area of statistical research with a wide range of applications. In this paper, we propose novel models based on wavelets for spatially correlated functional data. These models enable one to regularize curves observed over space and predict curves at unobserved sites. We compare the performance of these Bayesian models with several priors on the wavelet coefficients using the posterior predictive criterion. The proposed models are illustrated in the analysis of porosity data.