2.00 - 2.05 Welcome/Chair's introduction
 
2.05 - 2.40 Dr Hao Ma (Lecturer in Economics and Finance, Queen Mary University of London) - Double Machine Learning Inference for Conditional Latent Factors (with Patrick Gagliardini)

2.40 - 3.15 Jorge Yslas (Lecturer in Actuarial Mathematics, University of Liverpool) - Assessing continuous common-shock risk through matrix distributions (with Martin Bladt and Oscar Peralta)
 
3.15 - 3.20 Refreshment break
 
3.20 - 3.55 Mattia Bevilacqua (Senior Lecturer in Finance, University of Liverpool) - How to Assess Crisis Risk: Developing a New Machine Learning Early Warning System

3.55 - 4.00 Close

Dr Hao Ma (Lecturer in Economics and Finance, Queen Mary University of London) - Double Machine Learning Inference for Conditional Latent Factors (with Patrick Gagliardini)

This paper deals with identification and asymptotically normal inference in conditional latent factor models for large, unbalanced panels of asset returns. The setting is nonparametric regarding the time-variation of risk exposures. For a general class of parameters of interest relevant for asset pricing with latent factors, we show identification via a semiparametric moment condition involving an unknown vector function containing the conditional moments of portfolio returns given (high-dimensional) predictors. We use a doubly-robust moment restriction for serially dependent data and establish feasible asymptotic normality of the estimators. For a panel of monthly returns of U.S. stocks, the number of conditional factors is large and counter-cyclical, and the spanning of the latent space involves dynamically e.g. liquidity, leverage and time-series reversal on top of Fama-French factors. The implied SDF outperforms competitors in pricing benchmark universes of test assets. Conditioning information plays an important role when estimating moments of conditional risk premia.


Jorge Yslas (Lecturer in Actuarial Mathematics, University of Liverpool) - Assessing continuous common-shock risk through matrix distributions (with Martin Bladt and Oscar Peralta)

We introduce a class of continuous-time bivariate phase-type distributions for modeling dependencies from common shocks. The construction uses continuous-time Markov processes that evolve identically until an internal common-shock event, after which they diverge into independent processes. We derive and analyze key risk measures for this new class, including joint cumulative distribution functions, dependence measures, and conditional risk measures. Theoretical results establish analytically tractable properties of the model. For parameter estimation, we employ efficient gradient-based methods. Applications to both simulated and real-world data illustrate the ability to capture common-shock dependencies effectively. Our analysis also demonstrates that common-shock continuous phase-type distributions may capture dependencies that extend beyond those explicitly triggered by common shocks.


Mattia Bevilacqua (Senior Lecturer in Finance, University of Liverpool) - How to Assess Crisis Risk: Developing a New Machine Learning Early Warning System

The early detection of vulnerabilities leading to the build-up of economic crises is key to preserving growth and prosperity. This study develops a more flexible and comprehensive machine learning (ML)-based Early Warning System (EWS) to predict a larger range of financial crises worldwide. The new toolkit predicts banking, currency, debt and fiscal crises, achieving up to a 23% higher forecasting accuracy than commonly adopted models. ML-based models are well-suited to identify vulnerabilities, especially in developing countries and low-income economies, given their scarcity of relevant data and their pronounced exposure to non-traditional economic risks. Our toolkit draws on a significantly broader range of predictor variables at mixed frequencies covering macroeconomic and financial data, including international spillovers and asset price volatility, as well as new challenges ranging from climate change to geopolitical tensions. We employ Shapley values and Shapley regressions to uncover the main determinants of the forecast, allowing to identify levers for preventive and mitigating actions, and to communicate actionable proposals to policy makers.

 

Contact Gareth Liu-Evans for Merseyside Local Group.