Simulation-based inference for financial econometrics models

About the Project

In modern statistical applications, many complicated models have two common features. First the likelihood functions are often difficult to evaluate; second the model is generative. In particular, financial time series data pose the following challenges. First, when latent stochastic dynamics are considered, e.g. volatilities and regime switching, the likelihood is intractable. Second, in the big data era, the more sophisticated model is required for high-frequency data and their microstructure. The class of simulation-based methods is often used for statistical inference of intractable likelihood models by using model simulations. The inference is usually conducted under the Bayesian framework, providing uncertainty quantifications for both parameter estimation and prediction. It has seen successful applications and become increasingly popular in a wide range of areas, including population genetics, ecology, astronomy, etc. This project aims to develop new simulation-based statistical computing algorithms with emphasis on financial econometrics models. The underpinning convergence theory will be developed. Building blocks of the new algorithms include approximate Bayesian computation, sequential Monte Carlo, Markov chain Monte Carlo, Bayesian synthetic likelihood, their synergies, etc.

Eligibility

Applicants should have, or expect to achieve, at least a 2.1 honours degree or a master’s (or international equivalent) in a relevant science or engineering related discipline.

Funding

This 3.5-year PhD is for self-funded students. Exceptional candidates will be considered for Faculty funding (this will include an annual tax-free stipend of £20,780 and tuition fees will be paid. We expect the stipend to increase each year). This project will start in October 2026.

At The University of Manchester, we offer a range of scholarships, studentships and awards at university, faculty and department level, to support both UK and overseas postgraduate researchers.