Search

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.jedc.2010.05.008 Byline: Shirley J. Huang (a), Jun Yu (b) Abstract: In this paper a Markov chain Monte Carlo (MCMC) technique is developed for the Bayesian analysis of structural credit risk models with microstructure noises. The technique is based on the general Bayesian approach with posterior computations performed by Gibbs sampling. Simulations from the Markov chain, whose stationary distribution converges to the posterior distribution, enable exact finite sample inferences of model parameters. The exact inferences can easily be extended to latent state variables and any nonlinear transformation of state variables and parameters, facilitating practical credit risk applications. In addition, the comparison of alternative models can be based on deviance information criterion (DIC) which is straightforwardly obtained from the MCMC output. The method is implemented on the basic structural credit risk model with pure microstructure noises and some more general specifications using daily equity data from US and emerging markets. We find empirical evidence that microstructure noises are positively correlated with the firm values in emerging markets. Author Affiliation: (a) Lee Kong Chian School of Business, Singapore Management University, 50 Stamford Road, Singapore 178899, Singapore (b) School of Economics and Sim Kee Boon Institute for Financial Economics, Singapore Management University, 90 Stamford Road, Singapore 178903, Singapore

Empirical modeling of Bayes equilibrium for decentralized trading mechanism of Web-based second price auctions is presented.

The unobserved preference heterogeneity within an omitted variable framework, which provides a theoretical rationale for more continuous preference distributions, multivariate normal in the limit, is examined. A comparison of the random coefficients model (RCM) and the latent class model (LCM) using simulated data illustrates that the RCM dominates the LCM if the underlying distribution is strictly continuous.

This paper uses a Litterman Bayesian vector autoregression model (LBVAR) to make production forecasts for technology industries. The LBVAR model utilizes interacting clusters of related industries for greater accuracy of predictions, and it proves superior to Bayesian statistics for making small sample forecasts.

This model suggests an answer to the question regarding price formation during disequilibrium. Sellers project demand and monopolistic competition in order to set their prices.

A model of learning about rival's behavior shows that its predictio ns support the Bayesian learning model. Decision-making exhibits a Nash equilibrium after sufficient choices.

Two red deer populations in Scotland are used to test culling rates based on a Bayesian calibration of a wildlife management model. The two samples demonstrate the need to incorporate different variables in management models to maintain stability for dissimilar populations.

The signalling subgame was considered to assess the design of contract and verification under an investor-entrepreneur contract. The game indicated that truth-telling and deterministic verification are not part of an equilibrium behavior as shown by the presence of a unique Perfect Bayesian Equilibrium in mixed strategies. Consideration of the entire game also revealed that less verification expenditure is achieved under a contract formulated by an informed trader.

An information partition strategy was employed to confirm the presence of a Bayesian Nash equilibrium in a Bayesian game with infinitely many players. The strategy permits the action set of an individual to rely on the condition of nature and to become a subset of an infinite dimensional space. Existence theorems with set of agents that can either be categorized as infinite set or a measure space were also utilized.

An a posteriori gradient estimation scheme based on Gibbs sampler is used in a Bayesian net approach for defining and solving a stochastic optimization problem. The use of the Bayesian net approach allows for the description of joint probability distributions in the optimization model while the Gibbs sampler allows for gradient estimation when random variables are discrete. A validation of the model in a semiconductor environment proved the robustness of the technique.

A model for predicting the failure of hydraulic structures is derived from observed degradation characteristics over a time period. The model employs a Bayesian approach in that it calculates the average deterioration for a fixed unit of time based on a prior density. This allows the probabilities of preventive repair and failure to be expressed as a function of the average deterioration. The model is applied in computing the lifetime probability function of the Netherlands' Eastern Scheldt Storm-Surge Barrier.

A proposed general model for supervised learning suitable for all classification or function approximation methods is presented. Also, probably almost Bayesian (PAB) algorithms are outlined to derive adequate solutions. Results show that the proposed model can approximate the Bayesian optimum in polynomial time. It also unifies classical and connectionist approaches to approximation problems.

An analysis of foreign exchange rates is presented. The analysis considers a multivariate Bayesian time-varying coefficients (TVC) technique for examining exchange rate data. Results show that an application of TVC as data generating mechanism generates nonlinear conditional moments and leptokurtosis in unconditional series distribution. In addition, time aggregation reduces leptokurtic behavior.

Research is presented concerning the healthy fiscal conditions and lower inflation which have been achieved by southern European countries adopting the Maastricht criteria for joining the euro.