Essays on energy prices and time-varying volatility models

dc.contributor.authorZhu, Beili
dc.date.accessioned2020-12-10T02:13:00Z
dc.date.available2020-12-10T02:13:00Z
dc.date.issued2020
dc.description.abstractThis thesis contains three Chapters that use Bayesian econometrics to develop and apply time-varying models, especially stochastic volatility models, for forecasting, modelling and structurally analysing energy prices. Both univariate and multivariate models with stochastic volatility are applied in this thesis. Chapter 2 constructs a monthly real-time oil price dataset using backcasting and compares the forecasting performance of different models of constant and time-varying volatility based on the accuracy of point and density forecasts of real oil prices of both real-time and ex-post revised data. This Chapter uses Bayesian autoregressive and autoregressive moving average models with respectively, constant volatility and two forms of time-varying volatility: GARCH and stochastic volatility. In addition to the standard time-varying models, more flexible models including volatility in mean and moving average innovations are applied to predict the real oil price. The results show that time-varying volatility models dominate their counterparts with constant volatility with respect to point forecasts at longer horizons and density forecasts at all horizons. The inclusion of a moving average component provides a substantial improvement in the point and density forecasting performance for both types of time-varying models while stochastic volatility in mean is superfluous for forecasting oil prices. Overall, the time-varying volatility models with moving average components yield the best forecasting performance for the oil price forecast. Chapter 3 develops a moving average stochastic volatility model with leverage and heavy-tailedness using the scale mixture of the normal distribution. In terms of parameter estimation, an efficient method based on the Markov chain Monte Carlo (MCMC) algorithm is developed in this Chapter. To test the modelling performance of the newly developed model, this Chapter compares the new model with seven existing stochastic volatility models for their statistical properties using simulated data and three types of time series data. The logarithm of the marginal likelihood of each model is calculated as the model selection criterion. The empirical results reveal that the newly proposed model is highly competitive among the class of stochastic volatility models for modelling the returns of energy prices and financial data. It provides better model fit than the other seven stochastic volatility alternatives for the returns of the NY Harbor No.2 Heating Oil and Equity Hedge and can beat most of the stochastic volatility models when dealing with the weekly returns of the U.S. Gulf Coast Conventional Gasoline Regular. Chapter 4 employs a large BVAR model with common stochastic volatility to analyse the effects of oil shocks including an oil supply shock, a global oil demand shock and a precautionary oil shock on 17 U.S. macroeconomic and financial market variables from 1986Q1 to 2019Q2. Stochastic volatility means that generalised impulse responses can provide a time-varying account of the effects of the shocks occurring in each quarter. Standard impulse response functions are also computed for the shocks in 2019Q2 and 2008Q4. The results show the sizes of the impulse responses vary over time, but the fluctuations are not dramatic with the exception during the GFC. Four findings from the standard impulse responses are observed. First, the responses to the oil shocks are strikingly different. Second, there are permanent inflationary effects for all the shocks. Third, although the global oil demand shock is associated with increased GDP, there are some long-run negative effects on several macroeconomic variables. Fourth, while all the oil shocks negatively affect the U.S. stock and currency market in the long run, the effects on the U.S. bond market.
dc.identifier.otherb71500315
dc.identifier.urihttp://hdl.handle.net/1885/216796
dc.language.isoen_AU
dc.titleEssays on energy prices and time-varying volatility models
dc.typeThesis (PhD)
local.contributor.authoremailu4753366@anu.edu.au
local.contributor.supervisorMcKibbin, Renee
local.contributor.supervisorcontactu4036214@anu.edu.au
local.identifier.doi10.25911/S0NG-AN95
local.identifier.proquestYes
local.mintdoimint
local.thesisANUonly.author26cc5cd7-7967-4f8d-8aee-b76db13974bf
local.thesisANUonly.keyf3ceffbc-12af-71e3-def4-56be291acfe4
local.thesisANUonly.title000000013039_TC_1

Downloads

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Beili ZHU Thesis 2020.pdf
Size:
5.95 MB
Format:
Adobe Portable Document Format
Description:
Thesis Material