• Recurrent Neural Network GO-GARCH Model for Portfolio Selection ( Journal of Time Series Econometrics, vol. 16(2), 2024, pp. 67-81.) with Martin Burda

Awarded Best Student Poster Award at the 39th Canadian Econometrics Study Group Meeting (York University, 2024)
Short Article

We develop a hybrid model of multivariate volatility that uses Recurrent Neural Networks to capture the conditional variances of latent orthogonal factors in a GO-GARCH framework. Our approach seeks to balance model flexibility with ease of estimation and can be used to model conditional covariances of a large number of assets. The model performs favourably in comparison with relevant benchmark models in a Minimum Variance Portfolio (MVP) scenario.

• When are People More Open to Cheating? Economic Inequality Makes People Expect More Everyday Unethical Behavior (PLoS ONE, February 2024) with Anita Schmalor and Steven Heine
  
  Short Article
  Economic inequality has been found to be associated with increased unethical behavior and an increased acceptance of unethical behavior. In this paper we explored whether higher amounts of perceived inequality lead to an increase in the expectation of unethical behavior. We tested whether people would say that they themselves would engage in more unethical behavior in a context of high compared to low inequality. We find evidence for this hypothesis in 3 of 4 studies (n = 3,038). An internal meta-analysis shows a small but significant effect. Such increased expectations that oneself will behave unethically likely has consequences for societal trust and functioning.

• Dynamic Factor Binary Panels -  Identification and Particle Filter Estimation [SUBMITTED]

This paper extends binary outcome models by introducing a threshold latent variable with an unobserved and serially correlated factor structure. I derive conditions for point identification and sharp set identification under relatively weak model assumptions. I further provide consistency and unimodality proofs for a proposed simulated maximum likelihood estimator, employing a novel EM algorithm. Monte Carlo simulations demonstrate favorable MSE and bias properties in finite samples. An application to Klette & Kortum (2004) R&D innovation highlights a viable structural interpretation of the econometric theory of factors.

Presentations: University of Toronto (Econometrics Seminar), CIREQ Poster Session (upcoming)

• Dynamic Investment Games: Social Value of Information

Analysis of dynamic game involving uncertain investment value. Elements of a social learning process. Main question: is investment breakdown after a singe period of no investment by any party avoidable by introducing a public signal that is sufficiently accurate? With real option value capuring a linear equilibrium in the private and the common posterior