PUBLICATIONS
Abstract:
In this paper, we first consider the pseudo maximum likelihood estimation of the univariate GARCH (2,2) model and derive the underlying estimator. Then, we make use of the technique of martingales to establish the asymptotic normality of the pseudo-maximum likelihood estimator (PMLE) of the univariate GARCH (2,2) model. Contrary to previous approaches encountered in the statistical literature, the pseudo-likelihood function uses the general form of the density laws of the quadratic exponential family.
Keywords: Asymptotic normality, Pseudo-maximum likelihood estimator, Quadratic exponential family, Univariate GARCH (2,2) model.
Abstract:
One provides in this paper the pseudo-likelihood estimator (PMLE) and asymptotic theory for the GARCH (1,1) process. Strong consistency of the pseudo-maximum-likelihood estimator (MLE) is established by appealing to conditions given in Jeantheau (1998) concerning the existence of a stationary and ergodic solution to the multivariate GARCH (p, q) process. One proves the asymptotic normality of the PMLE by appealing to martingales' techniques.
Keywords: Pseudo-maximum-likelihood, estimation-strong-consistency-GARCH (1,1), asymptotic normality, martingales' techniques.
Completed paper and work in progress
I- Completed papers
1- "A Regime-Switching Approach for Detecting Shock Transmission Types across Asset Markets, with an Application to Sovereign Bond Markets" (Job market paper)
Abstract:
This paper proposes a modeling strategy for investigating the nature of the transmission of shocks—wether interdependence, contagion, or decoupling—between two asset markets during periods of high volatility (crises). Our regime-switching model captures co-movements in both the mean and volatility processes of asset returns. The mean dynamics incorporate PCA-derived factors that reflect global and market-specific influences. Meanwhile, the variance-covariance structure accounts for common and idiosyncratic shocks, each governed by an independent Markov-switching process. We define contagion or decoupling as occurring when a high-volatility idiosyncratic shock originating in one market significantly alters the interdependence of the mean returns of the considered assets. To statistically detect these phenomena, we employ bootstrap-based Student’s t-tests. Applying this methodology to the different pair combinations of the sovereign bond markets of three Latin American countries, we find evidence that—on average—decoupling has occurred in the Brazil/Mexico and Argentina/Mexico market pairs. For the remaining pairs, our findings suggest that the nature of the shock transmission is characterized primarily by interdependence.
2- Estimating Industry-level Equity Risk Premia using Information on Volatility of the Fundamentals: A Supervised Dynamic Orthogonal Component (sDOC) Approach (with Ba Chu).
Abstract:
This paper introduces the Supervised Dynamic Orthogonal Components (sDOC) method as a novel framework for forecasting excess stock returns out-of-sample. sDOC enhances traditional linear dimension reduction techniques—most notably Principal Component Analysis (PCA)—by integrating machine learning-based feature selection with the construction of dynamic, volatility-sensitive orthogonal factors. Unlike PCA, which fails to account for nonlinear relationships and volatility interactions among predictors, sDOC explicitly models these elements, resulting in superior forecasting accuracy. The method is applied to 63 U.S. monthly industry equity portfolios and evaluated against several benchmark models: PCA, univariate GARCH(1,1), random forest (RF), and a three-layer neural network (NN3). Empirical evidence shows that sDOC consistently outperforms both PCA and univariate GARCH(1,1), while demonstrating comparable predictive power to NN3. In particular, both sDOC and NN3 surpass RF in performance. This superiority is reflected across multiple evaluation criteria, including out-of-sample R squared, Mincer-Zarnowitz R squared, average excess portfolio return gains, and a range of risk-adjusted metrics such as Sharpe ratio, Sortino ratio, Calmar ratio, and gain-to-pain ratio. Furthermore, the rolling out-of-sample R squared highlights sDOC’s adaptability under both calm and turbulent market conditions, positioning it as a robust and interpretable tool for asset return prediction.
This paper has been presented at the 39th Canadian Econometric Study Group (CESG) conference, York University, Toronto (Fall 2024).
3- Commodity-Equity Linkages: A Markov-Switching Contagion Framework (with Maral Kichian)
Abstract:
This paper examines how country-specific major export commodity shocks affect that country’s equity returns when controlling for the influence of global risk factors. Using a Markov-switching framework, we jointly model and estimate commodity–equity return pairs, capturing co-movements in the mean and the volatility across endogenously identified regimes of crisis and stability. We then study the transmission of shocks during crises, where the shocks originate in the country’s commodity market and spill over to its equity market. We test for contagion, defined as an amplification of the commodity-specific shocks in the equity market. Analyzing data from six major commodity-exporting countries between 2006 and 2024, and applying likelihood tests, we find evidence of contagion during the Global Financial Crisis, notably in the oil–equity linkages of Norway and the United States (U.S.). Out-of-sample forecasts demonstrate that our proposed model outperforms various other popular benchmarks—especially for equity returns. These findings underscore the model’s practical relevance for both policymakers and market practitioners.
II- Work in progress
1- Wild Bootstrap test of Contagion in Multivariate Asset Markets.
2- Panel Data Models with Spatially Correlated Two-Way Error Components.