Home > Department of Statistics > Events > Special Events and Conferences > Nonlinear time series analysis

 

Conference organiser

 

Qiwei Yao

Professor in Statistics

LSE

Email: q.yao@lse.ac.uk|

 

 

Administration support and general enquiries

 

Ian Marshall

Research Administrator

LSE

Tel: +44 (0)20 7955 7511

Email: i.marshall@lse.ac.uk|

  

Nonlinear time series analysis - thresholding and beyond: a conference in honour of Professor Howell Tong to celebrate his 70th birthday

HowellTongOrganised by the Department of Statistics, London School of Economics and Political Science

Friday 19 and Saturday 20 September 2014

On the occasion of Professor Howell Tong's 70th birthday, LSE hosted this conference to celebrate the research achievement and the applications in nonlinear time series and related areas by bringing together experts, scholars and young researchers from around the world.

Professor Tong has made pioneering contributions in nonlinear time series analysis. His work on threshold models has had lasting influence, both on theory and application.

Conference programme here|

Abstracts here|

Conference photos here|

For all conference enquiries please email i.marshall@lse.ac.uk|


Location

New Theatre, East Building, LSE (London)

A conference dinner was held on the evening of 19 September 2014, with a conference dinner speech by Professor Bernard Silverman| (FRS), Chief Scientific Advisor to the Home Office.

Plenary speech

Kung-Sik Chan| (University of Iowa)
Threshold modelling and Howell Tong

Invited speakers

John Aston| (University of Cambridge)
Change-points in high dimensional settings
Matteo Barigozzi| (LSE)
Dynamic factor models, cointegration and error correction mechanisms
Peter J. Brockwell| (Colorado State University)
Prediction of Lévy-driven CARMA processes
Ngai-Hang Chan| (Chinese University of Hong Kong)
LASSO estimation of threshold autoregressive models
Haeran Cho| (University of Bristol)
High-dimensional panel data segmentation
Rainer Dahlhaus| (University of Heidelberg)
Volatility decomposition and online volatility-estimation with nonlinear market microstructure noise models
Cees Diks| (University of Amsterdam)
Comparing the accuracy of copula-based multivariate density forecasts in selected regions of support
Bärbel Finkenstädt| (University of Warwick)
Switch time modelling for gene expression: an overview
Piotr Fryzlewicz| (LSE)
The use of randomness in time series analysis
Simone Giannerini| (University of Bologna)
The mathematical structure of genetic information: a (nonlinear) time series perspective
Liudas Giraitis| (Queen Mary, University of London)
Integrated ARCH and AR models: origins of long memory
Shaojun Guo| (LSE and Chinese Academy of Sciences)
High-dimensional and banded vector autoregression
Marc Hallin| (Free University of Brussels)
Quantile spectral processes, asymptotic analysis and inference
Javier Hidalgo| (LSE)
Specification in time series regression models
Kostas Kalogeropoulos| (LSE)
Bayesian inference for partially observed SDEs driven by fractional Brownian motion
Jens-Peter Kreiss| (Technical University of Braunschweig)
Baxter's inequality and sieve bootstrap for random fields
Clifford Lam| (LSE)
Nonparametric eigenvalue-regularised precision or covariance matrix estimator
Michele La Rocca| (University of Salerno)
Neural network sieve bootstrap for nonlinear time series
Tony Lawrance| (University of Warwick)
Analysis of a laser-chaos communication experiment
Degui Li| (University of York)
Inference on structural breaks in panel data models with interactive fixed effects
Wai Keung Li| (University of Hong Kong)
Some results on the buffered time series models
Shiqing Ling| (Hong Kong University of Science and Technology)
Self-weighted LAD estimation for infinite variance threshold autoregressive models
Zudi Lu| (University of Southampton)
Semiparametric nonlinear regression models for irregularly located spatial time-series data
Guy Nason| (University of Bristol)
Analysis and forecasting of locally stationary time series
Sofia Olhede| (UCL)
Estimating multivariate nonstationary time series models in the fourier domain
Jiazhu Pan| (University of Strathclyde)
Threshold models for count time series
Rainer von Sachs| (Catholic University of Louvain)
Shrinkage estimation of the dependence structure of high dimensional time series
Mysung Hwan Seo| (LSE)
Extending the scope of the cube root asymptotics
Nils Christian Stenseth| (University of Oslo)
Thresholding and beyond in ecology
Dag Bjarne Tjostheim| (University of Bergen)
Nonstationary processes with a threshold
Ruey S. Tsay| (University of Chicago)
Threshold models for functional time series with applications
Andrew Walden| (Imperial College)
Advances in shrinkage methods for spectral matrices
Wei Biao Wu| (University of Chicago)
Estimation of high-dimensional vector auto-regressive processes
Yingcun Xia| (National University of Singapore)
Whittle likelihood estimation on nonlinear autoregressive models with moving average errors
Rongmao Zhang| (Zhejiang University)
Identifying cointegration by eigenanalysis
Wenyang Zhang| (University of York)
An iterative estimation procedure for generalised varying-coefficient models with unspecified link functions

Session chairs

Ngai-Han Chan| (Chinese University of Hong Kong)
Session A: Further thresholding
Kung-Sik Chan| (University of Iowa)
Session B1: More thresholding
Shaojun Guo| (LSE and Chinese Academy of Sciences)
Session B2: Inference on change-points
Yingcun Xia| (National University of Singapore)
Session C1: Non/semi-parametric inference
Ruey S Tsay| (University of Chicago)
Session C2: High-dimensional time series
Piotr Fryzlewicz| (LSE)
Session D1: Inference for volatility
Clifford Lam| (LSE)
Session D2: Nonstationary processes
Javier Hidalgo| (LSE)
Session E1: Inference for large matrices
Haeran Cho| (University of Bristol)
Session E2: Convergence rates
Wai Keung Li| (University of Hong Kong)
Session F1: Spectral methods
Myung Hwan Seo| (LSE)
Session F2: Continuous time processes and long memory
Dag Bjarne Tjostheim| University of Bergen)
Session G1: Chaos and nonlinearity
Matteo Barigozzi| (LSE)
Session G2: Bootstrap and randomness

Research posters

The following research posters were on display throughout the conference:

Mark Fiecas| (University of Warwick)
Introducing the evolving evolutionary spectrum, with applications to an associative learning study
Abstract: Our goal is to use local field potentials (LFPs) to rigorously study changes in neuronal activity in the hippocampus and the nucleus accumbens over the course of an associative learning experiment. We show that the spectral properties of the LFPs changed during the experiment. While many statistical models take into account nonstationarity within a single trial of the experiment, the evolution of brain dynamics across trials is often ignored. In this work, we developed a novel time series model that captures both sources of nonstationarity. Under the proposed model we rigorously define the spectral density matrix so that it evolves over time within a trial and also across trials. To estimate the evolving evolutionary spectral density matrix, we used a two-stage procedure. In the first stage, we computed the within-trial time-localized periodogram matrix. In the second stage, we developed a data-driven approach for combining information across trials from the local periodogram matrices. We assessed the performance of our proposed method using simulated data. Finally, we used the proposed method to study how the spectral properties of the hippocampus and the nucleus accumbens evolves over the course of an associative learning experiment.

Ali Habibnia| (LSE)
Nonlinear forecasting using many predictors with neural network factor models
Abstract: This study proposes a nonlinear forecasting technique based on an improved factor model with two neural network extensions, which would be able to capture both non-linearity and non-normality of a high-dimensional dataset. This proposed model has been developed on the basis of statistical inference and special emphasis is given to data-driven specification. It has also been proved that a linear factor model is a special case of this neural network factor model.

Timothy J Heaton| (University of Sheffield)
Towards denoising shapes using local geometry
Abstract: Suppose that you have observed, subject to noise, a mesh of points lying on a manifold combined with their neighbourhood structure. We present ongoing work aiming to recover the underlying manifold by denoising. Our approach consists of creating localised wavelet-type bases on the noisy manifold using the local geometry and the lifting scheme (Jansen et al. 2009). These bases are designed to provide sparse expansions of smooth functions defined on the mesh using local predictions based on the Laplace-Beltrami operator. To denoise the manifold we use these shape-adaptive bases to expand the separate co-ordinate functions before thresholding and inversion. We also indicate how rotational invariance can be obtained through selection of an appropriate thresholding rule.
(With Matthew A Nunes|, Lancaster University)

Na Huang| (LSE)
NOVELIST estimator of covariance matrix
Abstract: We propose a NOVEL integration of the sample and thresholded co-variance estimators (NOVELIST) to estimate large covariance matrix. It is shrinkage of the sample covariance towards a general thresholding target, especially soft or hard thresholding estimators. The benefits of NOVELIST include simplicity, ease of implementation, and the fact that its application avoids eigenanalysis, which is unfamiliar to many practitioners. The simulation results will be presented and comparison with other competing methods will also be given. 
(With Piotr Fryzlewicz)

Emmanouil Karimalis| (Cass Business School)
Modelling temporal and cross-country dependence of European sovereign yield curves using liquidity and credit quality factors
In times of economic uncertainty, investors tend to prefer more liquid and less risky assets. This phenomenon is commonly referred to as a flight-to-liquidity and flight-to-quality, respectively. In this study, we focus on the sovereign yield curves of several European countries and model their temporal and cross-country dependence using broader European and country-specific measures of liquidity and credit quality factors. We assess the impact of those factors both unconditionally in time, as well as conditional on times of heightened market volatility. The analysis is split into two main parts. In the first part, the macro-finance dynamic Nelson-Siegel model of Diebold, Rudebusch and Aruoba (2006) is employed to study the interaction of broader European liquidity and credit quality measures and the yield curve. In the second part, we employ the covariance regression model of Hoff and Niu (2012) to relate the covariation of sovereign yields to country-specific liquidity and credit quality measures. Our findings suggest that liquidity and credit factors are important in explaining changes in the yields and covariance structure, while their importance varies across maturity and time. Furthermore, sensitivity analysis results reveal significant “flights” and spill-over effects among European countries.
(With Ioannis Kosmidis and Gareth W Peters)

Nikolaos Karouzakis| (LSE)
Modelling the Libor-OIS Spread: The roles of default and liquidity risks
Abstract: In this paper we estimate a no-arbitrage model of the term structure of money market spreads. We suggest a framework in which the spreads are functions of interest rates, credit and liquidity factors. In a first set of estimations, the identification of the last two factors relies on credit and liquidity proxies. In a second set of estimations, these factors are latent and the model is estimated using a Kalman Filter. Our approach allows us to identify credit and liquidity effects and helps to delineate interest rate, credit and liquidity factors in the evaluation of spreads.  Our empirical results indicate that prior to Lehman’s default, credit risk is less volatile and constitutes a relatively small part of the spread, with liquidity risk accounting for more of the spreads’ level. By early 2009, spreads are mainly driven by the credit component.

Weining Wang| (Humboldt University of Berlin)
Hidden Markov structures for dynamic copulae
Understanding the dynamics of a high dimensional non-normal dependency structure is a challenging task. A multivariate Gaussian or mixed normal time varying models are limited in capturing important types of data features such as heavy tails, asymmetry, and nonlinear dependencies. This research aims at tackling this problem by building up a hidden Markov model (HMM) for hierarchical Archimedean copulae (HAC). The HAC constitute a wide class of models for high dimensional dependencies, and HMM is a statistical technique for describing regime switching dynamics. HMM applied to HAC flexibly models high dimensional non-Gaussian time series. In this paper we apply the expectation maximization (EM) algorithm for parameter estimation. Consistency results for both parameters and HAC structures are established in an HMM framework. The model is calibrated to exchange rate data with a VaR application. This example is motivated by a local adaptive analysis that yields a time varying HAC model. We compare the forecasting performance with other classical dynamic models. In another, second, application we model a rainfall process. This task is of particular theoretical and practical interest because of the specific structure and required untypical treatment of precipitation data.
(With Ostap Okhrin and Wolfgang Karl Härdle)

Scientific committee

Kung-Sik Chan| (University of Iowa)
Piotr Fryzlewicz| (LSE)
Javier Hidalgo| (LSE)
Wai Keung Li| (University of Hong Kong) 
Qiwei Yao| (LSE)

Share:Facebook|Twitter|LinkedIn|