Speaker:

Dr. Richard Gerlach

Affiliation:

University of Sydney

Venue:

Room 498, Merewether Building

Title:

Bayesian time-varying quantile forecasting for Value-at-Risk in financial markets

Abstract:

Recently, Bayesian solutions to the quantile regression problem, via the likelihood of a Skewed-Laplace distribution, have been proposed. These approaches are extended and applied to a family of dynamic conditional autoregressive quantile models. Popular Value at Risk models, used for risk management in finance, are extended to this fully nonlinear family. An adaptive Markov chain Monte Carlo scheme is designed for estimation and inference. Simulation studies illustrate favorable performance, compared to the standard numerical optimization of the usual non-parametric quantile criterion function. An empirical study employing ten major financial stock indices and Value at Risk forecasting finds significant nonlinearity in dynamic quantiles and evidence favouring the proposed model family, for lower level quantiles, compared to standard parametric volatility and risk models in the literature. Simpler more standard models are favoured at less extreme quantiles.

Speaker:

Dr. Erick Li

Affiliation:

University of Sydney

Venue:

Room 498, Merewether Building

Title:

On Managing Supply Chains with Cournot Competition and Supply Disruptions

Abstract:

We consider two competing supply chains with supply disruptions. Each supply chain consists of a supplier and a retailer. The suppliers offer supply chain contracts to their exclusive retailers and the retailers engage in Cournot competition. We analyse the equilibriums in three different scenarios: i) both chains are coordinated, ii) one chain is coordinated and the other is not; and iii) both chains are uncoordinated. We find that when the probability of disruption is high, both supply chains are better off under the coordinated scenario than under the uncoordinated scenario. But when the disruption is not eminent, both supply chains are often worse off under the coordinated scenario relative to the uncoordinated scenario, as in the prisoner's dilemma.

Speaker:

Ching-Hua Chen-Ritzo

Affiliation:

Venue:

Room 498, Merewether Building

Title:

Optimizing Call Center Performance With Respect to Outsourcing Contracts

Abstract:

We address the problem of optimizing the allocation of resources in a call/contact center outsourcing engagement comprising multiple call center vendors, a service provider and a client. Our optimization model captures non-linear penalty and bonus functions that may be included in the terms of a contract between the service provider and the client. Our model can be used to guide contract negotiations, and to optimize tactical and operational allocation of call center resources.

Speaker:

Prof. Song-Ping Zhu

Affiliation:

University of Wollongong

Venue:

Room 498, Merewether Building

Title:

An Exact and Explicit Solution for the Valuation of American Put Options

Abstract:

In this talk, an exact and explicit solution of the well-known Black-Scholes (1973) equation for the valuation of American put options is presented for the first time. To the author's best knowledge, never has an exact and explicit formula been found for the valuation of American options of finite maturity, although there have been quite a few approximate solutions and numerical approaches proposed. The exact solution presented here is written in the form of a Taylor's series expansion, which is constructed based on the homotopy-analysis method. The optimal exercise boundary, which forms the key difficulty of this highly nonlinear problem, has been elegantly and temporarily removed in the solution process, and consequently, the solution of a set of infinitely many linear problems can be analytically worked out at each order, resulting in a completely analytical and exact series-expansion solution for the optimal exercise boundary and the option price of American put options. The newly found analytical solution is also explicit as the optimal exercise boundary as well as the option price can be written as explicit functions of the risk-free interest rate, the volatility and the time to expiration.

Speaker:

Nuttanan(Nate) Wichitaksorn

Affiliation:

The University of Sydney

Venue:

Room 498, Merewether Building

Title:

A Bayesian Analysis of Tobit Model with Non-Gaussian Error: The Case of Three-Parameter Asymmetric Laplace Distribution

Abstract:

We extend an analysis of a standard Tobit model (in which the error term is usually assumed to follow a normal distribution) to the case that the error term follows the three-parameter Asymmetric Laplace Distribution (ALD). In the estimation, we employ the Markov Chain Monte Carlo (MCMC) methods including Metropolis-Hastings, Gibbs sampler, modified efficient jump, and griddy Gibbs, to obtain the estimates of the model. The MCMC estimation results are also compared to those of Maximum Likelihood Estimation (MLE). The results from the simulated data show that the MLE estimates are closer to the true values than those of MCMC but the MCMC estimates are more statistically significant than those of MLE. The MLE and MCMC methods are also applied to the wage income data of Thai males who live in rural areas. The estimation results show that the MCMC methods perform better than those of MLE.

Speaker:

Mike Smith

Affiliation:

University of Melbourne

Venue:

Room 498 Merewether Building

Title:

Modeling Multivariate Distributions Using Copulas: Applications in Marketing

Abstract:

In this research we introduce a new class of multivariate probability models to the marketing literature. Known as "copula models", they have a number of attractive features. First, they permit the combination of any univariate marginal distributions that need not come from the same distributional family. Second, a particular class of copula models, called "elliptical copula", have the property that they increase in complexity at a much slower rate than existing multivariate probability models as the number of dimensions increase. Third, they are very general, encompassing a number of existing multivariate models, and provide a framework for generating many more. These advantages give copula models a greater potential for use in empirical analysis than existing probability models used in marketing. We exploit and extend recent developments in Bayesian estimation to propose an approach that allows reliable estimation of elliptical copula models in high dimensions. Rather than focusing on a single marketing problem, we demonstrate the versatility and accuracy of copula models with four examples to show the flexibility of the method. In every case, the copula model either handles a situation that could not be modeled previously, or gives improved accuracy compared with prior models.

Speaker:

Ming Xi (Nicole) Huang

Affiliation:

University of Technology, Sydney

Venue:

Room 498 Merewether Building

Title:

Modelling Default Correlations in a Two-Firm Model by Dynamic Leverage Ratios Following Jump Diffusion Processes

Abstract:

The recent financial crisis that initiated in the United States and rapidly spread elsewhere was related to the large amount of correlated defaults indicating clearly the topicality and importance of this research topic. One of the main challenges in credit risk analysis is the estimation of correlation among defaults of firms. A number of approaches have been developed to tackle this problem. The Gaussian copula method has become a kind of market standard to estimate default correlations. It is easy to implement, but has the drawback that there is no easy way of knowing which copula to use. Another approach is the reduced-form approach, in which default is driven by surprises captured by some jump process. The probability of surprise depends on an intensity parameter which is estimated by calibration. Another common approach, the structural approach, relates the advent of default to the dynamics of the underlying structure of firm. This approach developed out of the work of Merton (1974) who developed a corporate bond pricing model depending on the firm asset value and the face value of debt. This paper provides a generalized two-firm model of default correlation, based on the structural approach incorporating interest rate and jump risks.

In this talk we review the key features of some major structural models in credit risk modelling, from the fundamental model of Merton (1974) to the stationary-leverage-ratio models of Collin-Dufresne & Goldstein (2001) and Hui et al. (2006), and then to a dynamic leverage ratio model of Hui et al. (2004). If structural models are modelled by continuous diffusion processes, firms can never default by surprise. However in many areas of finance, especially in the area of credit risk, it is sudden, abrupt changes that are of interest, for example the sudden default of a firm or a sovereign borrower. A very useful way of capturing such sudden changes in financial markets is to consider jump-diffusion processes. In the second part of this talk, we introduce a one-firm model with the dynamic leverage ratio following a jump-diffusion process that based on the Hui et al. (2004) to capture the surprise due to unexpected external shocks. Some numerical results showing the impact of the jump components on the single firm default probability will be discussed. Then we extend the one-firm model to the two-firm situation, thereby capturing the surprise risk of default in a group of firms.

At the end of the talk, we summarize the impact of jump components on default correlations and joint default probabilities, draw some conclusions and make some suggestions for future research arising out of the issues considered here.

Speaker:

Dr Tong Shilu

Affiliation:

School of Information Systems, Technology and Management at UNSW

Venue:

Room 498 Merewether Building

Title:

Sharing Imperfect Demand Information in Competing Supply Chains with Production Diseconomies

Abstract:

This paper studies the incentive for vertical information sharing in competing supply chains with production technologies that exhibit diseconomies of scale. We consider a model of two supply chains each consisting of one manufacturer selling to one retailer, with the retailers engaging in Cournot competition. The problem is analyzed using a multi-stage game. We fully characterize the information sharing, wholesale pricing and retail quantity decisions in equilibrium and show that information sharing benefits a supply chain when the production diseconomy is large, competition is less intense, and the information is less accurate. When a supply chain makes its information more accurate or production more efficient, it may be worse off if such an improvement induces the firms in the rival supply chain to stop sharing information. We also consider the model with Bertrand competition. When there is no production diseconomy, information sharing benefits a supply chain when competition is intense and the information is accurate. When there is production diseconomy, a manufacturer may be worse off by receiving information. Our results show that information sharing in one supply chain triggers a competitive reaction from the other that is negative under Cournot competition but may be positive under Bertrand competition.

This is joint work with Albert Ha and Hongtao Zhang.

Speaker:

Dr Steven Lu

Affiliation:

Senior lecturer in Marketing at the University of Sydney

Venue:

Room 498 Merewether Building

Title:

Loyalty Programs and Retail Chain Management

Abstract:

Nowadays loyalty programs have been commonly seen in retailing (e.g., Fly buys in Australia, Optimum by Shopper Drug Mart in Canada, Reward Zone by Best Buy in the USA). In spite of the popularity of loyalty programs in practice, the effectiveness of loyalty programs has been questioned (sharp and sharp 1997). Shugan (2005) argues many loyalty programs only 'lock' consumers in rather than create 'true' loyalty which most firms really need. In this study we investigate whether loyalty programs can actually create long term loyalty or not by using a newly-developed multicategory consumer choice model (Mehta 2007). Specifically, we study a loyalty program provided by a retail chain which owns both a supermarket and department store in one location. Simulated maximum likelihood is used to estimate the parameters of the model. Our results show that loyalty programs may actually 'damage' consumer loyalty in the long run although they may bring some short run benefits to firms.

Speaker:

Simon Jackman

Affiliation:

United States Studies Centre, Univ of Sydney and Political Science, Stanford.

Venue:

Room 498 Merewether Building

Title:

Recovering Estimates of Legislators' Preferences using Item-Response Models

Abstract:

The recorded votes of legislatures and other deliberative bodies ("rollcall" votes) are of great interest to students of political economy. Within political science and economics these (typically binary) votes are usually considered to have arisen from a random utility model known as the Euclidean spatial voting model: legislators are presumed to have preferences represented by a point in a low-dimensional Euclidean space, and consider proposals and status quo points that are also assumed to be points in the same space. We present this model, showing how it leads to a relatively simple operationalization as a two-parameter item-response model (widely used in psychometric modeling of educational testing data). Markov chain Monte Carlo methods are an attractive way to tackle the high dimensionality of the inferential problem.

I present recent work in this field, including analyses of historical and recent U.S. Congresses, the U.S. Supreme Court, and the Australian Senate. We consider some recent controversies in American politics, such as the claim that Barack Obama was the "most liberal senator" in the 110th U.S. Senate, inter alia. Interesting theoretical and statistical issues to be discussed include (a) assessing goodness of fit and the dimensionality of the preference space; (b) the sensitivity of the results to the prior densities we use in Bayesian inference for this problem; (c) hierarchical modeling of rollcall data; (d) points of contact and difference with the large literature in psychometrics using this model; (e) extending the model so as to permit estimating richer, structural models of legislative behavior.

Speaker:

Dr Cathy Chen

Affiliation:

Graduate Institute of Statistics and Actuarial Science, Feng Chia University, Taiwan

Venue:

Room 498 Merewether Building

Title:

Bayesian causal effects in quantiles: accounting for heteroscedasticity

Abstract:

Testing for Granger non-causality over varying quantile levels could be used to measure and infer dynamic linkages, enabling the identification of quantiles for which causality is relevant, or not. However, dynamic quantiles in financial application settings are clearly affected by heteroscedasticity, as well as the exogenous and endogenous variables under consideration. GARCH-type dynamics are added to the standard quantile regression model, so as to more robustly examine quantile causal relations between dynamic variables. An adaptive Bayesian Markov chain Monte Carlo scheme, exploiting the link between quantile regression and the skewed-Laplace distribution, is designed for estimation and inference of the quantile causal relations, simultaneously estimating and accounting for heteroscedasticity. Dynamic quantile linkages for the international stock markets in Taiwan and Hong Kong are considered over a range of quantile levels. Specifically, the hypothesis that these stock returns are Granger-caused by the US market and/or the Japanese market is examined. The US market is found to significantly and positively Granger-cause both markets at all quantile levels, while the Japanese market effect was also significant at most quantile levels, but with weaker effects.

Keywords: Bayesian; Granger non-causality in quantiles; Skewed-Laplace distribution; GARCH; Markov chain Monte Carlo method; Quantile regression.

Speaker:

Prof Moshe Haviv

Affiliation:

The Hebrew University of Jerusalem, Israel

Title:

Strategic behaviour in queues

Abstract:

The whereabouts of customers in a queueing system interact. For example, if more customers join the queue, one's waiting time may increase to such a level that one may prefer not to join at all. But who decides who will join and who will not? Many decision problems in queues can be modelled as non-cooperative games when the customers themselves are the decision makers. Hence, we look for Nash equilibrium strategy profile among them.

In the talk I will present a few such examples. Among them: To join or not to join the queue, which type of service to select, to pay or two a premium in order to become a high priority customer, when to abandon the queue, and when to retry and check if the server is busy or not.

For all of these examples equilibria will be looked at. I will distinguish between cases where one should follow the crowd and when one should avoid it. At times, comparison between the equilibrium and the social optimal policy will be made.

Speaker:

Dr John Lau

Affiliation:

School of Mathematics and Statistics, University of Western Australia

Venue:

Room 498 Merewether Building

Title:

A Monte Carlo Markov Chain Algorithm for a Class of Mixture Time Series Models

Abstract:

This article generalizes the Monte Carlo Markov Chain (MCMC) algorithm, based on the Gibbs weighted Chinese restaurant (gWCR) process algorithm, for a class of kernel mixture of time series models over the Dirichlet process. This class of models is an extension of Lo's (1984) kernel mixture model for independent observations. The kernel represents a known distribution of time series conditional on past time series and both present and past latent variables. The latent variables are independent samples from a Dirichlet process, which is a random discrete (almost surely) distribution. This class of models includes an infinite mixture of autoregressive processes and an infinite mixture of generalized autoregressive conditional heteroskedasticity (GARCH) processes.

Evaluating estimates of such models involves sampling partitions of n integers and sampling unique values of latent variables. Unfortunately, existing algorithms for the kernel mixture models are not applicable because of the dependencies among the latent variables through the likelihood of the time series models. We contribute by generalizing existing algorithms for mixture time series models using the reseating idea of the gWCR process, which was originally from the Polya Urn sampling scheme. Our methodology is illustrated by volatility estimations of ten financial indices fitted to an infinite mixture of GARCH models. An extension to more general random probability measures such as two-parameter Poisson-Dirichlet processes and normalized generalized Gamma processes is also discussed.

(This is a joint work with Mike K. P. So)

Speaker:

Professor Haya Kaspi,

Affiliation:

Industrial Engineering and Management, Technion

Venue:

Room 498 Merewether Building

Title:

Measure Valued Processes in the Asymptotic Approximation of Many Servers Queues

Abstract:

The lecture focuses on queueing systems with many servers serving in parallel, where the arrival process into the system is a quite general counting process, the service times of various customers are i.i.d. random variables with general distribution and are independent of the arrival process, and the number of servers N is large. A primary motivation for studying such systems is that they arise as models for telephone call centers. While most research to date on such systems assumes that the service time is exponentially distributed, a fact which makes the number of customers in the system a Markov process, statistical analysis of large service stations performed recently have shown that the service times are typically non exponential but rather Lognormal. An extension of the exponentially distributed service times to phase type service distribution by Puhalski and Reiman, lead to a Markov process with a finite dimensional state descriptor. The general service time assumption lead us to represent the Markovian dynamics of the system in terms of a process that describes the total number of customers in the system, as well as a measured valued process that keeps track of the 'ages' (the time in service) of the various customers in service. Fluid (first order)and diffusion (second order) approximations of the pair consisting of the number of customers in the system and the measure valued process described above, in heavy traffic as N ! 1 will be discussed in this lecture.

Speaker:

Dr. Eun-Seok Kim,

Affiliation:

Department of Industrial & Systems Engineering, Korea Advanced Institute of Science and Technology (KAIST), South Korea; Discipline of Operations Management & Econometrics, The University of Sydney

Venue:

Room 498 Merewether Building

Title:

Precedence constraints, and coordination of production and delivery in scheduling theory

Abstract:

In this talk, we discuss two interesting topics in the field of production scheduling. First we study a new concept of precedence constraint, called s-precedence constraints. For s-precedence constraints, job i cannot start processing until all jobs that precede j have started. This is different from the standard definition of a precedence relation where job i cannot start until all prior jobs are completed. While they are not discussed in the scheduling literature, the s-precedence constraints have a wide applicability in real world settings such as the first-come-first-served processing systems. Second, we study the coordination of classical scheduling models and other operations which exist in the supply chain. We focus on coordinating multi-location production and customer delivery where the jobs are processed on machines which are located at different sites, and delivered to a customer by a single vehicle.

Speaker:

John Geweke,

Affiliation:

The University of Iowa

Venue:

Room 498 Merewether Building

Title:

Optimal Prediction Pools

Abstract:

The lecture focuses on queueing systems with many servers serving in parallel, where the arrival process into the system is a quite general counting process, the service times of various customers are i.i.d. random variables with general distribution and are independent of the arrival process, and the number of servers N is large. A primary motivation for studying such systems is that they arise as models for telephone call centers. While most research to date on such systems assumes that the service time is exponentially distributed, a fact which makes the number of customers in the system a Markov process, statistical analysis of large service stations performed recently have shown that the service times are typically non exponential but rather Lognormal. An extension of the exponentially distributed service times to phase type service distribution by Puhalski and Reiman, lead to a Markov process with a finite dimensional state descriptor. The general service time assumption lead us to represent the Markovian dynamics of the system in terms of a process that describes the total number of customers in the system, as well as a measured valued process that keeps track of the 'ages' (the time in service) of the various customers in service. Fluid (first order)and diffusion (second order) approximations of the pair consisting of the number of customers in the system and the measure valued process described above, in heavy traffic as N ! 1 will be discussed in this lecture.

Speaker:

Professor Les Oxley,

Affiliation:

University of Canterbury, Christchurch, New Zealand

Venue:

Room 498 Merewether Building

Title:

Long memory or shifting means? A new approach and application to realised volatility

Abstract:

It is now recognised that long memory and structural change can be confused because the statistical properties of times series of lengths typical of financial and econometric series are similar for both models. We propose a new set of methods aimed at distinguishing between long memory and structural change. The approach, which utilises the computational efficient methods based upon Atheoretical Regression Trees (ART), establishes through simulation the bivariate distribution of the fractional integration parameter, d, with regime length for simulated fractionally integrated series. This bivariate distribution is then compared with the data for the time series. We also combine ART with the established goodness of fit test for long memory series due to Beran. We apply these methods to the realized volatility series of 16 stocks in the Dow Jones Industrial Average. We show that in these series the value of the fractional integration parameter is not constant with time. The mathematical consequence of this is that the definition of H self-similarity is violated. We present evidence that these series have structural breaks.

Speaker:

Professor Timo Terasvirta,

Affiliation:

CREATES, University of Aarhus, Denmark

Venue:

Room 498 Merewether Building

Title:

Conditional Correlations Models of Autoregressive Conditional Heteroskedasticity with Nonstationary GARCH Equations

Abstract:

We investigate the effects of careful modelling the long-run dynamics of the volatilities of stock market returns on the conditional correlation structure. To this end we allow the individual unconditional variances in Conditional Correlation GARCH models to change smoothly over time by incorporating a nonstationary component in the variance equations. The modelling technique to determine the parametric structure of this time-varying component is based on a sequence of specification Lagrange multiplier-type tests derived in Amado and Teräsvirta (2009). The variance equations combine the long-run and the short-run dynamic behaviour of the volatilities. The structure of the conditional correlation matrix is assumed to be either time independent or to vary over time. We apply our model to seven pairs of daily returns of stocks belonging to the S&P 500 stock index and traded at the New York Stock Exchange. The results suggest that accounting for deterministic changes in the unconditional variances considerably improves the fit of the multivariate Conditional Correlation GARCH models to the data. The effect of careful specification of the variance equations on the estimated correlations is variable: in some cases rather small, in others more discernible.

Speaker:

Professor Jiti Gao,

Affiliation:

The University of Adelaide

Venue:

Room 498 Merewether Building

Title:

A New Diagnostic Test for Cross-Section Uncorrelatedness in Nonparametric Panel Data Models

Abstract:

In this paper, we propose a new diagnostic test for residual cross-section uncorrelatedness in a nonparametric panel data model. The proposed nonparametric cross-section uncorrelatedness (CU) test is a nonparametric counterpart of an existing parametric cross-section dependence (CD) test proposed in Pesaran (2004) for the parametric case. We establish asymptotic distributions of the proposed test statistic for several different cases. Without assuming cross-section independence, we establish asymptotic distributions for the proposed test for the case where both the cross-section dimension and the time dimension go to infinity simultaneously. We then analyze the power function of the proposed test under a sequence of local alternatives that involve a nonlinear multi-factor model. We also provide several numerical examples. The small sample studies show that the nonparametric CU test associated with an asymptotic critical value works well numerically in each individual case. An empirical analysis of a set of CPI data in Australian capital cities is given to examine the applicability of the proposed nonparametric CU test.