Modelling spatial effects for mortgage loans with transformed Gaussian Markov Random Fields

The Centre for Statistics awarded funding to an interdisciplinary team across Statistics and the Roslin Institute.

Research Team: Raffaella Calabrese, Finn Lindgren, and Victor Medina Olivares

Project Summary: The main contributions of this project are threefold. From a methodological point of view, we will propose a transformed GRF with spatial dependence for large samples of unbalanced binary data. From a software point of view, the INLABRU package will be adapt to estimate the proposed model. Finally, from an empirical point of view, the software will be applied to a large dataset of US mortgages. In the medium term, the research team will submit a proposal for a ESRC PhD scholarship supervised by Ra aella and Finn with starting date September 2020. Expected outputs from the research will be targeted at top-tier statistics and operational research journals and talks at conference (Conference on Computational and Financial Econometrics 2019). Seminars at Bayes Centre and Edinburgh Future Institute will be organised. In the long term, a research grant application will be submitted to EPSRC to recruit a PostDoc Researcher for 3 years to develop an inter-disciplinary project that further develop the methodological proposal and its applications to credit risk.