Forecasting with single source of randomness state space models. The framework developed in this project, for identifying and
extrapolating trends, seasonal patterns and economic cycles in time
series, has a large and diverse range of useful applications in
Australia. Some examples include its potential use in the
development of appropriate monetary policy, its use to better inform
finance markets of risk levels associated with shares, its use to
forecast demand in supply chains to provide ....Forecasting with single source of randomness state space models. The framework developed in this project, for identifying and
extrapolating trends, seasonal patterns and economic cycles in time
series, has a large and diverse range of useful applications in
Australia. Some examples include its potential use in the
development of appropriate monetary policy, its use to better inform
finance markets of risk levels associated with shares, its use to
forecast demand in supply chains to provide a better service to
customers, and its use in call centres to better tailor staff
schedules to meet customer calls.Read moreRead less
Bayesian Inference for Flexible Parametric Multivariate Econometric Modelling. The anticipated outcomes include the development of enhanced multivariate econometric models and innovative computationally intensive methods for their estimation. These models are used in numerous and diverse applications which are data-intensive and where more complete models will greatly enhance data-based decision-making. Results include improved information use in the wholesale electricity markets, in financial m ....Bayesian Inference for Flexible Parametric Multivariate Econometric Modelling. The anticipated outcomes include the development of enhanced multivariate econometric models and innovative computationally intensive methods for their estimation. These models are used in numerous and diverse applications which are data-intensive and where more complete models will greatly enhance data-based decision-making. Results include improved information use in the wholesale electricity markets, in financial market investment decision-making and for the assessment of the impact of internet advertising.Read moreRead less
A Bayesian State Space Methodology for Forecasting Stock Market Volatility and Associated Time-varying Risk Premia. Accurate prediction of stock market volatility is critical for effective financial risk management. Along with information on volatility embedded in historical stock market returns, the prices of options written on the underlying stocks also reflect the option market's assessment of future volatility. This project will exploit this dual data source in a completely new way, using it ....A Bayesian State Space Methodology for Forecasting Stock Market Volatility and Associated Time-varying Risk Premia. Accurate prediction of stock market volatility is critical for effective financial risk management. Along with information on volatility embedded in historical stock market returns, the prices of options written on the underlying stocks also reflect the option market's assessment of future volatility. This project will exploit this dual data source in a completely new way, using it to produce forecasts of both volatility itself and the premia factored into asset prices as a result of traders' perceptions of volatility risk. State-of-the-art statistical methods will be used to produce up-dates of the probability of extreme volatility and/or extreme risk aversion, as new market data becomes available each trading day.Read moreRead less
New Statistical Procedures for Analysing Dependence in Non-Gaussian Time Series Data. In the economic, finance and business spheres, statistical data is often discrete, binary, strictly positive, or characterized by an uneven distribution of values above and below the average. Prominent examples are the high frequency financial data that have become accessible with the computerization of financial markets, including the number of trades in successive time intervals, the direction of price change ....New Statistical Procedures for Analysing Dependence in Non-Gaussian Time Series Data. In the economic, finance and business spheres, statistical data is often discrete, binary, strictly positive, or characterized by an uneven distribution of values above and below the average. Prominent examples are the high frequency financial data that have become accessible with the computerization of financial markets, including the number of trades in successive time intervals, the direction of price changes, the time between trades and the return on a financial asset over short periods. This project develops a range of new statistical tools that will enable both researchers and practitioners to analyze the dynamic behaviour in such data and thereby validate and implement a range of financial models.Read moreRead less
Building flexible multivariate models and their application in Finance. The project will develop methods for analyzing the properties of dependent measurements that may evolve through time. The new methods will significantly improve on current best statistical practice and will be applied to important problems in the financial sector such as asset allocation and risk management. The financial sector is a vital part of the Australian economy and it is important to understand the joint behavior of ....Building flexible multivariate models and their application in Finance. The project will develop methods for analyzing the properties of dependent measurements that may evolve through time. The new methods will significantly improve on current best statistical practice and will be applied to important problems in the financial sector such as asset allocation and risk management. The financial sector is a vital part of the Australian economy and it is important to understand the joint behavior of financial assets in order to understand and allow for risk. The methods will have immediate application in other disciplines such as medicine, engineering and the environmental sciences. The project will train a postdoctoral student and three PhD students in cutting edge financial econometrics. Read moreRead less
Non-parametric estimation of forecast distributions in non-Gaussian state space models. The production of accurate forecasts is arguably one of the most challenging tasks in economics, business and finance, where data often assume strictly positive, integer or binary values, or are characterized by many extreme values far from the average. This project will produce new, state-of-the-art statistical methods for generating accurate estimates of the probabilities attached to different possible futu ....Non-parametric estimation of forecast distributions in non-Gaussian state space models. The production of accurate forecasts is arguably one of the most challenging tasks in economics, business and finance, where data often assume strictly positive, integer or binary values, or are characterized by many extreme values far from the average. This project will produce new, state-of-the-art statistical methods for generating accurate estimates of the probabilities attached to different possible future values of such variables. Although far-ranging in scope, the techniques advocated will have particular impact in the financial sphere, where the concept of future risk is inextricably linked to the probability of occurrence of extreme values and, hence, to the future probability distribution of the financial variable. Read moreRead less
Censored Regression Techniques for Credit Scoring. This project will apply censored regression techniques to a loans database from the industry partner, the ANZ bank. We will accurately estimate the actual time to loan repayment, rather than simply the risk of default. In a novel approach for credit scoring we will build a model using current, right-censored, rather than historic data, incorporating loans that are not yet repaid but are underway and clearly have a length of loan longer than obse ....Censored Regression Techniques for Credit Scoring. This project will apply censored regression techniques to a loans database from the industry partner, the ANZ bank. We will accurately estimate the actual time to loan repayment, rather than simply the risk of default. In a novel approach for credit scoring we will build a model using current, right-censored, rather than historic data, incorporating loans that are not yet repaid but are underway and clearly have a length of loan longer than observed. This approach has the immense advantage of being able to reflect contemporary borrowing patterns in the model, rather than relying on historic trends.
Read moreRead less