The use of collateral has become one of the most widespread risk mitigation techniques. While it brings stabilizing effects to the individual cash lender, it may exacerbate systemic risk by accelerating bank deleveraging under funding stress. We show how a liquidity shock to the cash lender may propagate as a solvency shock via liquidity hoarding even if the cash lender remains solvent in all states of nature. Albeit a privately optimal response of the cash lender to a liquidity shock, bank deleveraging may lead to excessive bankruptcy among its borrowing counterparties while, at the same time, triggering contagion across asset classes. To buttress financial system resilience, we lay out a menu of macroprudential policies that deactivate this channel of financial contagion.

To date, an operational measure of systemic risk capturing nonlinear tail-comovements between system-wide and individual bank returns has not yet been developed. This paper proposes an extension of the CoVaR methodology in Adrian and Brunnermeier (2011) to capture the asymmetric response of the banking system to positive and negative shocks to the market-valued balance sheets of individual banks. Building on a comprehensive sample of U.S. banks in the period 1990–2010, the evidence in this paper shows that ignoring asymmetries that feature tail-interdependences may lead to a severe underestimation of systemic risk. On average, the relative impact on the system of a fall in individual market value is sevenfold that of an increase. Moreover, the downward bias in systemic-risk measuring from ignoring this asymmetric pattern increases with bank size. In particular, the conditional tail-comovement between the banking system and a bank that is losing market value belonging to the top size-sorted decile is nearly 5.5 times larger than the unconditional tail-comovement versus 3.3 times for banks in the bottom decile. The asymmetric model also produces much better fitting, with the restriction that gives rise to the standard symmetric model being rejected for most firms in the sample, particularly, in the segment of large-scale banks. This result is important from a regulatory and supervisory perspective, since the asymmetric generalization enhances the capacity to monitor systemic interdependences.