Tommaso Di Francesco

About Me

Welcome! I am a PhD student in Quantitative Economics at the Center for Nonlinear Dynamics in Economics and Finance (CeNDEF) and the University Ca' Foscari of Venice. I am pursuing a double degree as part of the EU-funded ITN EPOC. My research focuses on Behavioral Macro and Financial models, with a particular interest in expectation formation and heterogeneous agent modelling.

Research

Fields: Macroeconomics, Behavioral Economics, Non-linear Dynamics

Publications

1. Sentiment-Driven Speculation in Financial Markets with Heterogeneous Beliefs: A Machine Learning Approach

   (Journal of Economic Dynamics and Control, 2025)

   with Cars Hommes | Code | Data

   Abstract

   We study a heterogeneous asset pricing model in which different classes of investors coexist and evolve, switching among strategies over time according to a fitness measure. In the presence of boundedly rational agents, with biased forecasts and trend-following rules, we study the effect of two types of speculation: one based on fundamentalist and the other on rational expectations. While the first is only based on knowledge of the asset's underlying dynamics, the second takes also into account the behavior of other investors. We bring the model to data by estimating it on the Bitcoin Market with two contributions. First, we construct the Bitcoin Twitter Sentiment Index (BiTSI) to proxy a time-varying bias. Second, we propose a new method based on a Neural Network for the estimation of the resulting heterogeneous agent model with rational speculators. We show that the switching finds support in the data and that while fundamentalist speculation amplifies volatility, rational speculation has a stabilizing effect on the market.

Working Papers

1. (Mis)information Diffusion and the Financial Market

   with Daniel Torren Peraire | Code

   Abstract

   This paper investigates the interplay between information diffusion in social net- works and its impact on financial markets with an agent based model (ABM). Agents receive and exchange information about an observable stochastic component of the dividend process of a risky asset `a la Grossman and Stiglitz (1980). A small propor- tion of the network has access to a private signal about the component, which can be clean (information) or distorted (misinformation). Other agents are uninformed and can receive information only from their peers. All agents are Bayesian in updating their beliefs, but they are so in a behavioural way, so that in the construction of the likelihood function, they replace true precision with an individual parameter which depends on an endogenous and time evolving measure of the agent confidence in the source of the information. We examine, by means of simulations, how information diffuses in the network and provide a framework to account for delayed absorption of shocks, that are not immediately priced as predicted by classical financial models. We show the effect of the network topology on the resulting asset price and offer an inter- pretation for excess volatility with respect to fundamentals, persistence amplification and lepto-kurtosis of returns

Work in Progress

1. Sticky Information across the Wealth Distribution
   Abstract

   This paper investigates the role of wealth-dependent information stickiness in the transmission of monetary policy in a Heterogeneous Agent New Keynesian (HANK) model. Using survey data, I provide empirical evidence that households do not form expectations according to the full-information rational expectations (FIRE) hypothesis but instead exhibit stickiness in updating their information, with wealthier households updating more frequently. I evaluate the effect of this evidence on macroeconomic dynamics using a quantitative model. My findings reveal that inequality significantly affects the aggregate responses to monetary shocks. Specifically, models that neglect heterogeneity in information updating underestimate both the magnitude and the delay of the peak response to monetary policy shocks. Estimating the model by matching simulated impulse response functions (IRFs) to empirical ones shows that stickiness is crucial for accurately capturing the dynamics observed in the data.