Score-driven models provide a general framework for modeling time-varying parameters.
Time-varying parameters can be volatilities, correlations, default probabilities, loss-given-default, rating transition intensities,
the speed at which a central bank buys assets, macro or term structure factors, etc.
A brief review of the key idea is here.
Code is available from the www.gasmodel.com website.
Dynamic dependence models
CLSZ (2020, JME) use a high-dimensional model for dependent defaults among many bank and sovereign counterprties.
The statistical model is a minor extension of CKL (2011, JBES), whose code is here.
Clustering multivariate panel data
LSS (2019, JBES) group a large- and multi-dimensional array of bank accounting data into different bank business model groups.
Code illustrating the approach is here.
Time-varying extreme tail parameters
SZL (2020) study time-variation in the upper extreme tail of changes in euro area sovereign bond yields.
Code is here.
Mixed-measurement dynamic factor models
Occasionally one may be interested in studying the joint variation across panel data observations for which different families of
conditional distributions are appropriate. For example, CSKL (2014, REStat) consider the joint modeling of firm rating and default
transitions (dynamic logit), macro-financial observations (normal), and loss-given-defaults (beta distribution). In bad times, defaults
and downgrades are systematically up, macros are down, and losses given default are high.
Ox code for the CSKL observation-driven mixed-measurement dynamic factor model is here.
KLS (2012, JBES) introduce parameter-driven mixed-measurement dynamic factor models; see also SKL (2014, IJF) and SKL (2017, JAE).
This Ox code refers to SKL (2017, JAE).
Non-Gaussian credit risk models in state space form
Credit risk conditions can vary substantially over time, and up to an order of magnitude.
Standard portfolio credit risk and stress testing models relate the variation in pd's to ratings and easily-observed macro-financial
time series data. Unfortunately, they can fit and forecast badly, particularly in times of stress when they are needed most.
The addition of a latent factor is a practical way to capture the previously-documented excess clustering in non-Gaussian data.
This Ox code replicates the simulation results in KLS (2011, JoE).