Deadline for application : December 15, 2013
Context:
Under the framework of the 5-year research project "Stochastic Modelling of Dependence: Systems under Stress", sponsored by the Académie Louvain, the Université catholique de Louvain opens a PhD position in statistics or econometrics.
The overall goal of the research project is to construct new ways of measuring and modelling risks in systems with intricate dependence structures, moving towards model assumptions that better reflect real life complexity. Particular attention is to be paid to the behaviour of such systems in periods of distress, that is, upon the arrival of shocks, after structural breaks, or when comovements between risk factors are higher than usual.
The PhD candidate is supposed to work on one of the following specific topics:
1. Nonparametric modelling and estimation of the dependence structure of high-dimensional time series panels.
Economic and financial data are increasingly of high dimensionality, hence estimation of its serial and cross-sectional dependence, e.g. for subsequent prediction purposes, is a difficult task. In this PhD project we will address these problems by combining recently developed methodology on: regularisation (shrinkage) of covariance matrices, locally stationary modelling (abrupt regime switching or smoothly varying over time), dimension reduction for time series panels. Good quantitative knowledge of parametric (and ideally nonparametric) estimation and inference for multivariate data and some experience with data simulation/analysis (in R or Matlab) are required.
2. Prediction and nonparametric estimation in the presence of dependent functional regressors
Driven by the idea that economic or financial variables observed with a high frequency in time can often be seen as discretizations of a data generating process in continuous time, the prediction and nonparametric estimation based on continuous measurements of economic indicators is a demanding task. Take for instance the prediction of the German DAX industry index based on (ideally) continuous measurements of energy market prices. Typically, a collection of weekly measurements of energy prices is obtained by splitting up a record over the last decade. The assumption of independence does not reflect the data structure and the assumption of stationarity is at least questionable over the whole decade. In this PhD project we will study the influence of dependencies within the functional regressors on the attainable accuracy of a nonparametric prediction or estimation procedure. This PhD thesis will provide an opportunity to extend the student's knowledge of statistical tools including nonparametric smoothing, out of sample evaluation, unbiased risk estimation or regularised estimation, as well as mathematical tools mainly from approximation theory, functional and numerical analysis.
Profile:
• You have completed (or you are near completion) of a five-year Master in Statistics, Actuarial Sciences, Econometrics or related fields (e.g. applied mathematics, probability, physics, engineering, ...) with honours.
• Strong interest in quantitative and mathematical modelling.
• Decent knowledge of written and spoken English. Knowledge of French is not required.
• Final-year master undergraduate students are especially encouraged to apply.