Status: Published in International Journal of Financial Studies
Methodology: Deep Learning (LSTM), Ensemble Methods (XGBoost, Random Forest), Regularized Regression.
Abstract: Constructed a high-dimensional Canadian macroeconomic database (analogous to McCracken & Ng) by aggregating and harmonizing hundreds of series from Statistics Canada. This paper benchmarks a comprehensive model suite, including Deep Learning and Ensemble Methods, against standard ARIMA/OLS baselines. The findings demonstrate that non-linear ML ensembles significantly outperform affine term structure models (ATSM) in out-of-sample yield forecasting, particularly for capturing term premia dynamics.
Firm Innovation and the Transmission of Monetary Policy
Status: Working Paper
Methodology: Event-Study Analysis, Local Projections (LP).
Abstract: This research establishes a direct channel between monetary policy and the valuation of intangible assets. By utilizing Local Projections to trace the dynamic response of real variables, the study finds that firms with "exploratory" innovation strategies exhibit significantly higher sensitivity to policy shocks. An Event-Study constructed around FOMC announcements captures minute-by-minute market reactions, isolating the causal effect of policy news on the cross-section of returns.
Work-from-Home and Employee Performance: Evidence from the Sell-Side Analyst Industry
Status: Working Paper
Methodology: Difference-in-Differences (DiD), Staggered Treatment.
Abstract: Investigated the causal impact of Work-from-Home (WFH) on labor productivity by exploiting the exogenous pandemic shock in a Difference-in-Differences framework. The study discovers that coordination costs rise significantly with team size in remote settings, turning the traditional "team size premium" into a penalty. It further documents an asymmetric negative impact on teams led by female analysts, providing empirical evidence of the unequal burden of household responsibilities during structural economic shifts.