TY - UNPB

T1 - Ambiguous Networks

AU - Pelliccia, Marco

PY - 2013

Y1 - 2013

N2 - We model the formation of cross-holding network structures. Each agent caninvest in an individual primitive asset or hold shares of others whose value is conditionalon their respective investment decisions. When an agent holds shares ofmany distinct and independent players, she is able to diversify her asset portfoliorisk. However, in a chain of cross-exposures, she could indirectly find herselfexposed to third agents whose expected payoff is perceived as uncertain in thesense of Knightian-uncertainty. As a direct consequence of this, the optimal holdingsand cross-exposure level of each agent will be in part conditional on her riskand ambiguity aversion attitude. We characterize the optimal and the stable network structures which arise in this environment; respectively, these are the structures which maximize a utilitarian welfare function and the ones which emergeendogenously. Sparse networks are optimal and stable when the cost per-link ishigh and the players are more sensitive to diversification than uncertainty, whilethe opposite is true for dense networks. In general, we find that the architecture ofthe stable networks is similar to the optimal ones but not necessarily of the samedensity; individual agents fail to internalize the positive/negative externality dueto the activation of a link toward another peer. The model proposes a new alternative interpretation for some of the structural changes observed in financialmarkets pre and post-crisis.

AB - We model the formation of cross-holding network structures. Each agent caninvest in an individual primitive asset or hold shares of others whose value is conditionalon their respective investment decisions. When an agent holds shares ofmany distinct and independent players, she is able to diversify her asset portfoliorisk. However, in a chain of cross-exposures, she could indirectly find herselfexposed to third agents whose expected payoff is perceived as uncertain in thesense of Knightian-uncertainty. As a direct consequence of this, the optimal holdingsand cross-exposure level of each agent will be in part conditional on her riskand ambiguity aversion attitude. We characterize the optimal and the stable network structures which arise in this environment; respectively, these are the structures which maximize a utilitarian welfare function and the ones which emergeendogenously. Sparse networks are optimal and stable when the cost per-link ishigh and the players are more sensitive to diversification than uncertainty, whilethe opposite is true for dense networks. In general, we find that the architecture ofthe stable networks is similar to the optimal ones but not necessarily of the samedensity; individual agents fail to internalize the positive/negative externality dueto the activation of a link toward another peer. The model proposes a new alternative interpretation for some of the structural changes observed in financialmarkets pre and post-crisis.

KW - Diversification

KW - Stable Network

KW - Financial Network

KW - Ambiguity

KW - Risk

KW - Integration

M3 - Working paper

BT - Ambiguous Networks

ER -