ABSTRACT
Up to this point I have insisted on the fact that direct interaction between agents plays a crucial role in determining aggregate outcomes. However, to justify this assertion, as I said in the first chapter, I have to say more about how this interaction is organised and, in particular, who interacts with whom. To analyse this, we need to know about the network of links between the individuals, whether these are consumers, firms or other collective entities. Networks and network analysis play a central role in many disciplines and for a long time their role in economics was ambiguous. For many economists the study of networks was limited to analysis of the functioning of physical networks such as the railway, the telephone system or the internet for example. Yet, more recently it has been recognised that networks, in fact, are much more fundamental and pervasive than this, and this is well illustrated by Goyal’s (2007) and Jackson’s (2008) recent books on economic and social networks. Almost any serious consideration of economic organisation leads to the conclusion that network structures both within and between organisations are important. Let us go back for a moment to the crisis of confidence in the world economy. Here we see the role of networks. Bad risks from the American mortgage market, had been bundled with good risks into derivatives and these had diffused through the international banking system. Up to that point the financial system, thought of as a network of banks, had become larger and more connected and it was argued that the resultant diversification of risk was a stabilising influence. Yet, there was little analysis of the effect of the increasing connectivity of the network on the probability of an epidemic of negative impacts. The problem is that as risk is diversified into different instruments those who buy it lose track of the underlying asset. Thus, while the risk is diversified the information is not. When this happens, an epidemic of mistrust can develop as each bank in the network is wary of lending to another which may have inherited the risks that turned out to be bad. Worse, not only may banks find themselves with assets which may turn out to be ‘toxic’ but also the market may revise its valuation of these assets. Thus, the fact that
various banks have been obliged to reassess their losses as a result of the subprime episode and its consequences was due to not only their discovering the true nature of their assets but also to the downward revaluation of these by the market. The resultant losses of the banks enhanced the epidemic of mistrust. It is possible that it is simply the increased connectivity of the network that has favoured the development of such an epidemic. But, in fact, the problem is more subtle than this. Propagation of information or of shocks may be more likely and the effect may be bigger in networks that are much less than fully connected.
