Financial Networks Key to Understanding Systemic Risk
By Camelia Minoiu and Sanjay Sharma
May 28, 2014
- Network perspective sees financial institutions as interrelated economic agents
- Interconnectedness in financial system part of the problem in the financial crisis
- Data gaps hinder translation of network analysis into policy
With financial markets around the world so interconnected, the analysis of “networks” in the financial system would help deepen understanding of systemic risk and is key to preventing future financial crises, say leading researchers and policymakers at a conference on Interconnectedness: Building Bridges between Research and Policy.
The global financial crisis has shed light on the importance of contagion and systemic risk. One source of risk is the interconnectedness of economic agents created through financial transactions. These transactions generate a “financial network.” Yet there is no clear understanding of how financial networks function and how robust they are. Understanding systemic risk in networks is critical to establish rules that will effectively manage it.
In his opening remarks, Olivier Blanchard, IMF Chief Economist and Director of its Research Department, admitted to being “a bit skeptical” when network theory techniques were introduced in macroeconomics. However, his work on East European transition economies revealed that breakdowns in production (supply chain) networks helped explain the output drop in the early 1990s. Blanchard remarked that such network effects, also prevalent in the financial sector, have important lessons for understanding the economic recession in advanced economies following the recent crisis.
The conference, held on May 8-9, 2014, was jointly organized by the IMF’s Research Department and the Monetary and Capital Markets Department in collaboration with the Institute for New Economic Thinking and the Deutsche Bundesbank.
‘Full interconnectedness’ is not optimal
In his keynote address, Professor Joseph Stiglitz of Columbia University said that before the 2007–2008 crisis, “the thrust of economic discussion was that diversification, or interconnectedness, was a great thing.” However, the belief that diversification enables risk to be spread was proven wrong during the crisis. The high degree of interconnectedness in the financial system “facilitated the breakdown” and became “part of the problem,” he noted.
Stiglitz: ‘Network theory provides a critique of standard wisdom on how to create a stable financial system’ (photo: IMF)
Banks can be “too interconnected, too central, and too correlated to fail”
Stiglitz highlighted several lessons from the financial networks literature.
There is no single optimal financial architecture—the nature of the systemic risk depends “on the interplay of the architecture with the nature of capital buffers, asset market liquidity, correlations, and the nature of the shocks” to the financial system. Different financial architectures (or network topologies) react differently to shocks. Stiglitz noted that “an architecture that might work well under one circumstance might not work well in another.” Furthermore, “there should be no presumption that a network that evolves on its own is efficient” from a social welfare point of view.
In addition to being too big to fail, banks can be too interconnected, too central, and too correlated to fail—this situation creates moral hazard problems. Focusing on the “too big to fail problem” is not sufficient for designing resilient financial systems. In addition, measuring and monitoring interconnectedness created by common ownership of assets as opposed to direct interbank exposures is very difficult.
Finally, interconnectedness involves not only linkages within the financial sector, but also linkages between the financial sector and the real economy—there are real costs associated with bankruptcies, as evidenced by the 1997–98 East Asian crisis, when shocks to a vast network of economic relationships among firms and between firms and the financial sector “led the whole system into paralysis.”
“A richer ecology of financial institutions”
Stiglitz argued for policies that limit interconnectedness and for a “richer diversity of financial institutions.” Such policies may imply restrictions on the kinds of activities that banks are allowed to undertake and limit the uniformity of business models and size. While “specialized institutions as opposed to universal banks can be better at information gathering, […] they may also be more subject to shocks in their particular area of specialization.” However, the benefits of a diverse financial system may well outweigh the costs. A “rich ecology of financial institutions would address the problem of too-correlated-to-fail financial structures and may result in a more robust and resilient financial system,” he added.
The conference featured paper presentations that answered questions such as: Why are some financial networks more stable than others? What is the link between the structure of a network and contagion? What are the advantages of a high degree of interconnectedness for individual financial institutions? Is there a trade-off between network resilience to shocks and social welfare?
The conference concluded with a high-level policy panel moderated by José Viñals, Director of the IMF’s Monetary and Capital Markets Department. Panelists from academia, banking industry, and policy institutions were asked to distill the main messages of the conference and to reflect on the main directions for future research.
Bridges between theory and empirics
Buch: ‘Lots of empirical work is not yet linked to theory’ (photo: IMF)
Claudia Buch, Deputy President of the Deutsche Bundesbank, said that “indirect financial linkages––in terms of common exposures and business models—matter a lot” but data challenges remain. “A more pragmatic approach is to use the datasets already available in different countries and a common methodology to analyze the data.” Buch added that the empirical research is not sufficiently guided by theoretical findings.
Sujit Kapadia, Senior Manager in the Prudential Policy Division at the Bank of England, emphasized that future research “should focus on real world network structures” as policy makers are mostly concerned with “core institutions at the center of the financial network.” He argued that that it is important to look at the more indirect channels that reflect “liquidity risk—where banks withdraw their funding from each other—as well as fire sales or pure fear-driven contagion linked to uncertainty.” Focusing on counterparty risk and defaults alone is insufficient “because a lot of the contagion and externalities occur before the point of default,” he added.
Time-varying, not static networks
Kapadia: ‘The challenge is trying to identify the next AIG’ (photo: IMF)
Several panelists agreed that static networks are a useful starting point, but future research should allow for time-varying risk in networks, that is, risk that varies over the business cycle.
Nellie Liang, Director of the Office of Financial Stability Policy and Research at the Federal Reserve Board, stated that systemic risk can generate real economy costs even before financial institutions default. Given that “systemic risk can be quite procyclical,” rising during good times and subsiding during bad times, she encouraged further research on “whether networks are procyclical, too.”
Liang: ‘How do we incorporate interconnectedness into stress tests?’ (photo: IMF)
Liang emphasized understanding leading vulnerabilities in the financial system. This includes (i) whether interconnectedness can generate systemic risk in a financial system without excessive leverage or maturity transformation; (ii) how to expand network analyses of the federal funds market to the multi-trillion dollar derivatives markets; (iii) how much information about individual institutions should be disclosed after stress tests; and (iv) how to identify non-bank systemically important financial institutions (SIFI).
Data gaps large but not insurmountable
Panelists agreed that a major obstacle in the analysis of financial networks and the translation of results into concrete policy recommendations is data gaps.
According to Kapadia, “it is very important to have better data on nonbanks, shadow banks, the linkages with the non-core financial system” to understand true exposures and “complex issues around netting, the treatment of derivatives, and balance sheet items that can often mask underlying exposures.”
Patricia Mosser, Deputy Director of the Office of Financial Research, argued that research should focus on the markets for wholesale funding. She stressed that it will take time before global SIFIs will be able to report counterparty risk data in a way that is comparable across financial institutions. Creating data platforms and utilizing “innovative technology solutions” to safeguard and analyze the data is very important.
Mossner: ‘We should not underestimate how hard it is going to be to get the data we would like to have to measure systemic risk’ (photo: IMF)
Sanjay Sharma, Chief Risk Officer of Global Arbitrage and Trading at RBC Capital Markets, discussed the lag between the reporting of financial statements and current conditions and the “difficulties in deciphering the risk of an institution from an opaque 500-page disclosure document.” He also argued that data collected during the crisis may not be useful for predicting the next crisis due to the unprecedented government interventions that backstopped the financial sector. Sharma cautioned that the simple agent behavior embedded in many theoretical models is not representative of true agent behavior during crises due to its complexity during stress, citing the fire sales that occurred in the aftermath of the Lehman Brothers failure.
Sharma: ‘Are the crisis data relevant for predicting what might be the next crisis? I think not’ (photo: IMF)
As a practical and cost-effective solution to the “data black hole problem,” Sharma proposed that a universal global codification framework be implemented. A barcode system for every financial transaction is technically feasible and “would go a long way” towards understanding financial flows and exposures.