Marco Plazzogna - Endogenous Dynamic Networks for Heteroscedastic Outcomes

vendredi 4 déc. 2026, 16:00 → 17:00 Europe/Paris

Salle Olga Ladyjenskaïa (IHP - Bâtiment Borel)

Many economic and financial systems are naturally represented as networks: the outcome of one unit may depend on those of other units through relationships that evolve over time. A central difficulty is that these relationships need not be exogenous. The network itself may react to the same shocks that affect the variables of interest, creating an endogeneity problem that cannot be addressed by treating network weights as fixed or predetermined.

We study a dynamic spatial model in which the interaction network is time-varying and generated by an auxiliary formation process. Dependence between innovations driving network formation and the structural outcome shock is explicitly incorporated through a control-function approach.

The framework is further extended to allow for conditional heteroskedasticity. In particular, we distinguish between two potentially different dynamic networks: one governing the transmission of outcomes in the conditional mean, and another governing the propagation of past shocks in the conditional variance. This makes it possible to separate network effects on the level of the process from network effects on its risk and volatility.

The resulting model provides a framework for studying systems in which both interactions and uncertainty evolve endogenously over time.