Contagion in Financial Networks
Saturday, Jan. 6, 2018 8:00 AM - 10:00 AM
- Chair: Cecilia Parlatore, New York University
Fire-sale Spillovers in Debt Markets
AbstractWe assess fire-sale spillovers empirically using a new approach to measure network linkages across financial institutions and rich micro data for the universe of open-end fixed-income mutual funds. We find evidence that flows are interdependent across funds with asset class overlap, consistent with the hypothesis that one fund’s redemptions may spill-over on to those of other funds by leading to distressed sales that adversely impact other funds’ performance. We use several strategies to identify the causal link between any given fund’s flows and those of its peers, including a regression discontinuity (RD) design that exploits sharp changes in peer flows around Morningstar 5-star ratings. The source of identification of our RD approach is quasi-random variation in peer flows around the arbitrary performance cutoffs used by Morningstar to assign their 5-star ratings, which is plausibly unrelated to changes in common industry fundamentals. Consistent with a fire-sale mechanism, not just fund flows, but also fund performance and liquidity, as well as the pricing of corporate bonds sold by funds under peer pressure, are adversely affected. Our approach yields simple measures of vulnerability of a fund family to system-wide flow pressures, which can be used for policy evaluation of alternative financial stability tools.
Identifying Contagion in a Banking Network
AbstractThis paper studies the impact of trading profits and losses on bank counterparty borrowing costs using data from a derivatives trade depositary. We use the network of credit default swap (CDS) transactions between banks to identify bank CDS returns attributable to counterparty losses. Any bank’s exposure to corporate default increases whenever counterparties from whom it has purchased default protection themselves experience losses. In line with this statement, we document an increase in the own CDS spread of such a bank. We find no such effect from losses of non-counterparties, nor from counterparties who have bought protection from, rather than sold protection to, the bank. We also find that the effect on bank CDS returns through this counterparty loss channel is large relative to the direct effect on a bank’s CDS returns from its own trading losses.
University of Michigan
Marco Di Maggio,
Harvard Business School & NBER
London School of Economics
- G2 - Financial Institutions and Services