A direct Monte Carlo approach for Bayesian analysis of the seemingly unrelated regression model

Graduate School of Business Administration, Keio University, Yokohama, Kanagawa, 223-8523, Japan
Journal of Econometrics (Impact Factor: 1.6). 11/2010; 159(1):33-45. DOI: 10.1016/j.jeconom.2010.04.005


Computationally efficient methods for Bayesian analysis of seemingly unrelated regression (SUR) models are described and applied that involve the use of a direct Monte Carlo (DMC) approach to calculate Bayesian estimation and prediction results using diffuse or informative priors. This DMC approach is employed to compute Bayesian marginal posterior densities, moments, intervals and other quantities, using data simulated from known models and also using data from an empirical example involving firms’ sales. The results obtained by the DMC approach are compared to those yielded by the use of a Markov Chain Monte Carlo (MCMC) approach. It is concluded from these comparisons that the DMC approach is worthwhile and applicable to many SUR and other problems.

1 Follower
12 Reads
  • [Show abstract] [Hide abstract]
    ABSTRACT: A description of computationally efficient methods for the Bayesian analysis of Student-t seemingly unrelated regression (SUR) models with unknown degrees of freedom is given. The method combines a direct Monte Carlo (DMC) approach with an importance sampling procedure to calculate Bayesian estimation and prediction results using a diffuse prior. This approach is employed to compute Bayesian posterior densities for the parameters, as well as predictive densities for future values of variables and the associated moments, intervals and other quantities that are useful to both forecasters and others. The results obtained using our approach are compared to those yielded by the use of DMC for a standard normal SUR model.
    No preview · Article · Apr 2010 · International Journal of Forecasting
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Cost-effectiveness analyses (CEAs) may use data from cluster randomized trials (CRTs), where the unit of randomization is the cluster, not the individual. However, most studies use analytical methods that ignore clustering. This article compares alternative statistical methods for accommodating clustering in CEAs of CRTs. Our simulation study compared the performance of statistical methods for CEAs of CRTs with 2 treatment arms. The study considered a method that ignored clustering--seemingly unrelated regression (SUR) without a robust standard error (SE)--and 4 methods that recognized clustering--SUR and generalized estimating equations (GEEs), both with robust SE, a "2-stage" nonparametric bootstrap (TSB) with shrinkage correction, and a multilevel model (MLM). The base case assumed CRTs with moderate numbers of balanced clusters (20 per arm) and normally distributed costs. Other scenarios included CRTs with few clusters, imbalanced cluster sizes, and skewed costs. Performance was reported as bias, root mean squared error (rMSE), and confidence interval (CI) coverage for estimating incremental net benefits (INBs). We also compared the methods in a case study. Each method reported low levels of bias. Without the robust SE, SUR gave poor CI coverage (base case: 0.89 v. nominal level: 0.95). The MLM and TSB performed well in each scenario (CI coverage, 0.92-0.95). With few clusters, the GEE and SUR (with robust SE) had coverage below 0.90. In the case study, the mean INBs were similar across all methods, but ignoring clustering underestimated statistical uncertainty and the value of further research. MLMs and the TSB are appropriate analytical methods for CEAs of CRTs with the characteristics described. SUR and GEE are not recommended for studies with few clusters.
    Full-text · Article · Oct 2011 · Medical Decision Making
  • [Show abstract] [Hide abstract]
    ABSTRACT: A process or system under study often requires the measurement of multiple responses. The optimization of multiple response variables has received considerable attention in the literature with the majority focusing on locating optimal operating conditions within the current experimental region and thus often occurs in the later stages of experimentation. This article focuses instead on the initial experiment and the location of additional experimental runs if the region of interest shifts. Considerable trade-off is often required in the multiple response context. In order to account for uncertainty in the model parameters and correlations among the responses, we propose the computation of Bayesian reliabilities to determine optimal factor settings for future experimental runs. The approach will be described in detail for two common design follow-up strategies: steepest ascent (descent) and shifting factor levels. Illustrative examples are provided for each application. Copyright © 2011 John Wiley & Sons, Ltd.
    No preview · Article · Dec 2011 · Quality and Reliability Engineering
Show more