Dipartimento di Ingegneria informatica, automatica e gestionale

Relatore: Prof. Sergio Currarini (Università Cà Foscari di Venezia &  Bristol University) 
Venerdi 18 maggio, Aula Magna DIS,  ore 11.00-12.00 - Serie DIAG Lunch Seminars.

Abstract
Many types of risky behavior involve significant behavioral complementarities (peer effects), and exert externalities on others. In this paper we adapt the linear-quadratic framework of Ballester et al. (2006) for the analysis of peer effects on networks to allow for the presence of external effects and altruism. We find that the network density affects the amount of risky behavior in a non monotonic fashion, increasing for small densities and decreasing afterwards. This qualifies the usual conclusions that risky behavior increases in denser networks due to peer effects alone. We also find that clustering agents according to their preferences for the risky behavior has non monotonic impacts on the polarization of risk and on aggregate risk, with maximum risk occurring for moderate levels of clustering. Finally, we show that in non regular network architectures, more central agents do not necessarily adopt more risky behaviour due to altruism. In terms of policy implications, we find that policies that target the degree of altruism are more efficient in densely connected networks, where the non linear effect of altruism takes over the linear peer effects.

&;