When FSOC announced earlier this week that it is studying the secondary mortgage market’s systemic risks, it reopened questions many thought the Trump Administration buried: who or what is systemic and what then is to be done about it? FSOC was as transparent as always – not – and provided absolutely no information about which activities are on the griddle or whether individual firms – most notably Fannie and Freddie – are in the dock for potential SIFI designation. However, it’s clear that a new regulatory construct for residential-mortgage finance will be wrought before the GSEs are allowed to leave conservatorship. What’s in store will determine the next generation of mortgage-market winners and losers, making it important at the outset to ensure that FSOC’s conclusions are founded on a sound understanding of what constitutes systemic risk and then how best to mitigate it.
Since the question of what’s systemic burst into our consciousness in 2008, academic and regulatory thinking has come up with an array of systemic-risk signifiers. As a 2018 meta-analysis of systemic-risk research describes, model-builders have had a field day with various network-effect and value-at-risk models with scant application to real-world finance on a forward-looking basis. However, regulators have heeded at least some of this advice, using proxies for 2008-style contagion risk to craft a set of systemic-risk indicators. In its own stab at defining systemic risk for the GSEs, FHFA proposes a systemic surcharge based on a market-share indicator. In contrast, the GSIB indicators are multi-faceted, with size – a sort of market-share proxy – buttressed by considerations such as the indispensable nature of a financial entity or infrastructure, complexity, and cross-border scope.
The problem with each of these size, structural, and network-effect approaches is that they are endogenous – that is, inward-focused. They focus on factors about one institution that purport to represent its impact on the financial system as a whole, but in fact only measure an entity in relation to the other entities being measured. Thus, academic research measures network effects only within the financial system. FHFA’s mortgage-debt measure of marketshare may well miss other systemic signifiers – perhaps GSEs are systemic not because they are big, but because they are indispensable or, less flattering, ill-capitalized and illiquid. The conservatorship was called in 2008 not because Fannie and Freddie were big, but because they couldn’t handle all the mortgages they purchased when the secondary market suddenly shut down.
GSIBs may or may not be systemic in their own right – all we know is how they stack up against other banks because they are designated in comparison to a constant list of big banks with no attention to systemically-important entities outside the scope of the analysis. If the world were constituted of entities just like GSIBs, then this structural approach might – just might – capture systemic risk before it’s upon us, but the financial world is of course very, very different than the bank-centric construct on which current systemic identifications are premised.
The activity-and-practice construct FSOC prefers to SIFI designation is a major advance over the approach taken in the past to banks and once-designated nonbank SIFIs. It attempts to identify business lines or financial products that pose systemic risk regardless of who offers them, turning not to a simple remedy such as a capital surcharge, but instead to a more structural risk-mitigation strategy to contain systemic risk. However, FSOC’s methodology could still be unduly endogenous, looking at inter-connections among financial institutions as sources of systemic risk unless it accounts for exogenous shocks.
Think of exogenous shocks as the equivalent of asteroid strikes – exogenous systemic risk arrives not from within the complex network effects academics love to model and regulators expect, but from outside the financial system. A German bank threatened systemic solvency in 1974 due not to its size or scope, but because it suffered an outage that wreaked havoc in the foreign-exchange market. A more recent exogenous systemic risk came on 9/11 when two planes struck at the heart of the global payment system in ways no surcharge could have mitigated. COVID-19 is of course another exogenous shock to the financial system, one that makes painfully clear how the macroeconomy instantly affects financial stability. Up until now, virtually all thinking about systemic risk assumes that it’s financial-market instability that causes macroeconomic risk, not the other way around.
Exceptions to the belly-button model of systemic-risk analytics are the Fed’s stress tests and the resolution plans large banks must live by. TLAC also figures here in that it requires banks to hold buffers at all times above and beyond those meant to be drawn down under stress, although these buffers are also systemic shock absorbers as we’ve learned of late. All of these requirements assume exogenous shocks to a bank’s balance sheet and require advance insulation against it. It is indeed this insulation that made banks so resilient even as nonbanks crumpled until the Fed stepped in to backstop them with the trillions deployed in mid-March and beyond.
Importantly, all banks under these rules proved resilient, not just the GSIBs subject to the toughest standards. It’s thus resilience and resolvability – not designation – that makes the difference.
Extrapolating these lessons to the secondary mortgage market, we thus know that naming names is far less meaningful than ensuring resilience under stress. Stress must be judged not just from within the financial system, but also from beyond due to operational, geopolitical, and of course macroeconomic risk. We also know that pockets within the financial system of seeming systemic insignificance based on existing designation criteria can be very, very systemic in the absence of risk-mitigation buffers. As I posited in 2012, resolvability – not size, complexity, and other indicators – is the best indicator of systemic risk. If a financial institution can absorb stress without resort to its federally-insured depository or the central bank, it is likely to be a bulwark against systemic risk, not a cause thereof.
Resolvability also provides an initial buffer against systemic activities and practices – if the entity can absorb stress without transmitting it to innocent bystanders, then activities offered by providers are likely to prove both robust under financial-market stress and as safeguards in macroeconomic debacles. Using the resolvability and resilience criterion, it is clear that credit-enhancement counterparties that can pay claims under even acute stress do not pose systemic risk; those that can’t, do. Similarly, originators and servicers who can sustain demand and handle delinquencies do not pose systemic risk; those who can’t, do. It doesn’t matter what undermines the provider’s ability to meet its obligations so much as its ability to carry on under acute strain. The activity-and-practice framework also captures products that transmit resilience and resolvability risk. For example, products that put borrowers at risk because loan terms are manageable under only the best of circumstances pose systemic risk if the number of affected borrowers is large enough to threaten macroeconomic or financial-market stability. A few abused borrowers are the ambit of the CFPB; many should be a spark for systemic designation.
On whom or what the systemic label is affixed will befall significant new strategic challenges. There is an array of methodological and analytical ways to judge resolvability and resilience on both an institutional and product basis. Hopefully, FSOC will use them – not tired, inward-looking criteria – to make its determinations. We have learned the very hard way that rearing financial entities only for financial-system instability leaves not just them, but also the broader economy at acute risk. A system independent of federal guarantees and bailouts requires a systemic construct recognizing the world as it is, not was.