Rakesh Vohra: Can we make our financial systems more resilient?

One of the major research areas in the Warren Center for Network and Data Science is resilience. Whether it’s a trader working the stock market, a doctor combating infections in a hospital, or a quadrotor flying in a swarm, the more connections there are between agents in a system, the more likely it is that problems in one part of a network can spread and put everyone at risk. How can we maximize the benefits of these networks while minimizing the potential for real catastrophe?

Rakesh Vohra, a Penn Integrates Knowledge Professor with appointments in Engineering and Arts & Sciences, is trying to answer that question. As the co-director of the Warren Center, he and his colleagues are developing techniques for protecting networks from this kind of “endogenous” risk.

At Zilient.org, he writes about the 2008 financial meltdown and how protecting against future collapses means more than making individual institutions more resilient, but understand their interconnections as a network.

The chain of events that so upset the banking world in 2008 began with the collapse of Lehman Brothers. Panic spread, the dollar wavered and world markets fell. These events became an archetypal example of what is called systemic risk: instability so widespread it could lead to a catastrophic failure of the entire financial system.

How could Lehman’s fall threaten the whole banking system? Many blamed the system’s interconnectedness and the lack of oversight. Lehman became ‘infected’, and its financial links with other firms allowed the ‘infection’ to spread.

A focus on the regulation of individual firms meant no one was tasked with spotting the risks that could bring the system down. Unlike natural disasters, systemic risk in modern financial systems arises endogenously, or from within. It can’t be captured by individual institutions’ balance sheets, or specific market or asset price-based measures alone.

Forestalling a repetition of the events of 2008 requires examining the interactions between and across institutions. Moreover, policy must anticipate the side effects of one regulation or action on others.

Of the policy responses to the crisis, two kinds have been actively discussed. The first limits the links between financial institutions by restricting, for example, the contracts they may write with each other. The second is to increase the resilience of each institution, by heightened liquidity requirements, for example. Will these prescriptions have the desired effect?

Limiting links is costly. They represent loans, joint ventures and partnerships that exist to exploit potentially lucrative opportunities, although there is uncertainty about the returns from these opportunities. We are faced with a trade-off. Each link represents a joint opportunity but is also a channel through which ‘infection’ can spread.

In lightly regulated markets, this trade-off is delegated to individual firms. Will they, acting independently and in their own interest, get the trade-off right? If not, oversight is required.

Unfortunately, our intuitions about these can be mistaken because the risks are endogenous not exogenous. These intuitions are explored in a recent paper by Selman Erol and Rakesh Vohra, whose findings may be relevant to other networked systems where resilience of the entire system is the product of individual decisions.

Continue reading at Zilient.org.