Wednesday May 6 2015
Scuola Normale Superiore
Department of Economics and Management “Marco Fanno” – Università di Padova
The need for understanding the propagation mechanisms behind the recent financial crises lead the increased interest for works associated with asset interconnections. In this framework, network-based methods have been used to infer from data the linkages between institutions. In this paper, we elaborate on this and make a step forward by introducing network linkages into linear factor models. Networks are used to infer the exogenous and contemporaneous links across assets, and impacts on several dimensions: network exposures act as in inflating factor for systematic exposure to common factors with implications for pricing; the power of diversication is reduced by the presence of network connections; in the presence of network links a misspecied traditional linear factor model provides residuals that are correlated and heteroskedastic. We support our claims with an extensive simulation experiment. Joint work with Monica Billio, Roberto Panzica, and Loriana Pelizzon.