Publications

List of Publications

Leon-Gonzalez R, Majoni B. (2025) Approximate Factor Models with a Common Multiplicative Factor for Stochastic Volatility.  Studies in Nonlinear Dynamics & Econometrics. doi.org/10.1515/snde-2024-0103  

Leon-Gonzalez, R., & Majoni, B. (2024). Exact Likelihood for Inverse Gamma Stochastic Volatility Models.  Journal of Time Series Analysis  2024 Dec 2, doi.org/10.1111/jtsa.12795.

Majoni, B. (2021). VAT Withholding Tax and its Impact on VAT Compliance: Evidence from the Zimbabwe Revenue Authority. African Multidisciplinary Tax Journal, 2021(1), 228-243.  doi.org/10.47348/AMTJ/2021/i1a13   

Job Market Paper

Majoni, B. (2024) Generalized Common Inverse Gamma Stochastic Volatility Factor Models in Vector Autoregressions.

The most recent version of the paper can be accessed from here

Working Papers

Majoni, B. (2025).  Time Varying Volatility with Discrete Jumps in an Inverse Gamma Stochastic Volatility Model

Abstract: 

We extend the inverse gamma stochastic volatility model by introducing discrete volatility jumps whose probabilities vary with scheduled macroeconomic news announcements. We estimate the parameters of the model using a delayed-acceptance particle marginal Metropolis--Hastings algorithm with an auxiliary particle filter. Monte Carlo experiments show accurate recovery of parameters. An application to FTSE 100 daily returns shows that the model identifies higher jump probabilities on announcement days. An out-of-sample exercise shows that the jump model has a higher cumulative log predictive score than the baseline, with larger predictive gains on announcement days.

The most recent working paper can be accessed from here

Majoni, B. (2025). Integrating Deep Learning and Inverse Gamma Stochastic Volatility Models in Forecasting Climate-Induced Losses. Work in Progress.

Abstract: 

This study proposes a hybrid modeling framework that integrates deep learning techniques with inverse gamma stochastic volatility (SV) models to forecast disaster-related economic losses. The SV models explicitly capture time varying volatility and uncertainty in loss generating processes, while deep neural networks handle complex nonlinear pattern recognition and enhance predictive performance. Root Mean Squared Error (RMSE) and multivariate regression comparisons will be used to evaluate the model’s performance. The envisioned outcome is a forecasting system that is both accurate and interpretable, supporting improved climate adaptation strategies in vulnerable regions.

Work In Progress

1) Multi Server Hospital Phlebotomy Operations Research – with Professor Takashi Tsuchiya (GRIPS) and Professor Yoshifumi Uwamino (Keio University).

2) Asymmetric volatility and Climate Shocks: Modeling Non-Gaussian Dependencies.