Following several recent posts about consumer spending, it’s high time I got back to writing about NDFI lending. I’ll resume writing about consumer spending in short order; James Hardie, a company that sells into the North American housing market and which recently bought a U.S. deck maker, is down 34% this morning (its worst performance in several decades) on a worse-than-expected outlook (related to weak single-family new home construction, distributors’ and dealers’ destocking, etc.). I’ve written many times about the surge in bank lending to nondepository financial institutions (NDFIs) since the Great Financial Crisis (GFC), likely owing in large measure to regulatory arbitrage. NDFI loans have quadrupled since the beginning of 2015, such that they account for 10% of all bank loans vs. just 4% then. As NDFI loans have boomed, commercial and industrial (C&I) loans have effectively flatlined since late 2022 (though have picked up in recent months perhaps because of uncertainty related to President Trump’s tariffs). NDFI loans are up to $1.32 trillion, nearly half the level of C&I loans ($2.87 trillion).
Post-GFC regulations enhanced the attractiveness of bank lending to NDFIs vs. lending to nonfinancial entities such as businesses and households. A recent economic bulletin from the Kansas City Fed compared the profitability of both types of lending and found that the loan-level return on equity on loans to private credit funds (a type of NDFI) is far higher than that on C&I loans (bearing in mind that loan-level profitability doesn’t capture the cross-selling benefits to banks of making C&I loans). Indeed, the growth of private credit—direct lending to businesses by NDFIs—has likely been driven by regulatory and supervisory actions along with the growth of private equity, per a recent FEDS Notes piece.
Up to now, bank lending to NDFIs such as broker-dealers, private credit/equity firms and hedge funds appears to have been a largely virtuous cycle: (1) banks earn a high return on equity on NDFI loans given the low risk weights attached to those loans; (2) private credit firms and hedge funds, among other NDFIs, grow rapidly aided by those loans; (3) the substantial growth of private credit firms and hedge funds pushes up asset prices, which increases the amount of equity that can be leveraged; and (4) higher asset prices and therefore equity creates more demand for NDFI loans. I wrote “largely” because the 2021 collapse of the large family office Archegos caused a great deal of damage among banks and broker-dealers (see Credit Suisse), and that episode highlights what can and does happen amid periods of high financial leverage. And the risks that have built up in recent years stemming from mushrooming NDFI lending are apparent to regulators, even if regulators aren’t doing much about them.
The FDIC noted in its 2025 risk review (link) that “while bank lending to NDFIs on the surface presents relatively low credit risk, several trends suggest vulnerabilities in the market.” And as the Office of Financial Research (OFR) put it in its 2024 annual report (link), “Some NBFIs, such as many private credit funds, specialize in high-risk strategies, increasing the risk to banks of losses on credit extended to NBFIs.” The OFR also noted that large banks also have substantial exposure to NDFIs via the provision of prime brokerage services. In other words, via their affiliated broker-dealers lending to highly leveraged hedge funds. OFR cited the example of Archegos, whose collapse in 2021 cost broker-dealers a cool $10 billion. For perspective, primary dealers’ equity hit a high of $150 billion in 2021 but fell by around $20 billion over the next two years, likely partly the result of Archegos.
So what steps are U.S. regulators taking to better understand the risks associated with NDFI loans in light of these vulnerabilities? Last year they began requiring a breakdown of bank loans to NDFIs into five broad categories: loans to mortgage credit intermediaries, loans to business credit intermediaries, loans to private equity funds, loans to consumer credit intermediaries, and “other.” How is this supposed to help regulators understand the risks that banks are taking on? 🤷♀️
What trends are occurring that may warrant regulators’ attention? To give just one example, banks lend to their affiliated broker-dealers, which in turn lend to hedge funds via repurchase agreements (repo). The repo market has grown dramatically post-pandemic; more on that below. As the OFR explained in a recent blog post (link), lenders reduce risk in the repo market by requiring collateral from a borrower; the collateral provides the cash lender with an asset they can sell if the cash borrower doesn’t pay back the loan. Along with requiring collateral, cash lenders will sometimes request a “haircut” on the collateral, meaning they’ll lend less cash than the value of the collateral; doing so protects them in the event the collateral loses value. However, in a 2022 pilot study of the non-centrally cleared bilateral repo (NCCBR) market—which a recent FEDS Notes piece estimated accounts for ~40% of the U.S. repo market—the OFR found that 70% of all repo outstanding had no haircut. In an updated study published earlier this month, the OFR noted that the proportion of hedge fund repo with zero haircuts remains high at 65%, “indicating that this is a predominant feature among leveraged firms utilizing repo.” As the OFR wrote, the amount of zero-haircut repo is “significant.”
Both the Bank for International Settlements (BIS) and European Central Bank (ECB) recently noted the prevalence of zero or even negative haircuts in large sections of the repo market, “meaning that creditors have stopped imposing any meaningful restraint on hedge fund leverage” in the BIS’s words (link). Indeed, the Fed has reported in recent months that hedge fund leverage was at or near a record high.
What are the potential implications of this activity? The more “significant” (to use the OFR’s characterization) the amount of zero-haircut repo lending, the more risk broker-dealers are taking in lending to hedge funds. And the more risk that broker-dealers expose themselves to, the more risk that the banks lending to them are exposed to, as became clear in the Archegos episode.
So how substantial is the NDFI lending and associated leverage among broker-dealers, hedge funds and others? A few charts tell the tale. And these charts don’t illustrate the apparently generous terms on which this leverage is frequently being extended, though the aforementioned OFR blog post did exactly that.
(1) The chart above highlighted the dramatic growth of bank lending to NDFIs, which include bank-affiliated broker-dealers, private equity/credit funds and hedge funds. (2) Primary dealers’ repo lending has reached $3.9 trillion, a record high and more than double what it was pre-pandemic; the majority of this lending is collateralized by U.S. Treasurys and a large chunk of it is used to finance the hedge fund Treasury cash-futures basis trade, a “very large, leveraged Treasury trade” according to the Dallas Fed. A proxy for the size of the Treasury cash-futures basis trade is (3) hedge funds’ net short Treasury positions, which have reached a record-high $1.1 trillion in notional exposure. (4) Margin debt extended by broker-dealers hit $1 trillion in June, a record high (this debt is at both retail and institutional accounts). (5) Banks’ notional amount of derivatives hit $210.4 trillion in 1Q, near a multi-year high. Of note is that 87% of this amount was at just four banks. Derivatives are an off-balance-sheet item for banks, as are the next item I’m about to mention. (6) Banks’ unfunded commitments—meaning money they’ve committed to lend on demand but which isn’t on their balance sheets—were $3.4 trillion in 2024 per the OCC’s Shared National Credit Report (link), which equate to 180% of bank equity (tier 1 capital) after adjusting for tax-effected unrealized losses. For perspective, such commitments were just 110% of adjusted bank equity in 2010.
What makes these unfunded commitments so important? They’re contractual obligations for a bank, meaning a bank has to provide this cash on request. In the event of a sudden surge of demand for cash as happened in March 2020, banks will have to find a way to fund those commitments. In March 2020, the Fed embarked on a huge quantitative easing (QE) program that created bank reserves and deposits, the latter of which funded the commitments. There’s no guarantee the Fed will do so again in the next crisis. In fact, the Fed continues to slowly reduce its assets via quantitative tightening (QT) as bank reserves remain abundant (link), and banks have prepositioned more collateral at the discount window in recent years at the Fed’s urging (link). As I mentioned last week, the consumer price index (CPI) growth rate has exceeded the Fed’s 2% inflation target for 53 straight months; the Fed still has an inflation problem, among others (such as large unrealized losses on its securities holdings from its most recent QE program).
Given how high these unfunded commitments are as a percentage of adjusted bank equity, how would banks respond to a sudden, sizable cash call from their clients (which would result in a surge in C&I loans)? Many may not have the ability to extend other types of loans such as consumer, real estate and NDFI in such a scenario, which I believe would lead to a contraction in repo lending, bearing in mind that repo lending has ballooned over the past three years. Those repo loans are partly what have supported the substantial asset price inflation seen over the past three years. In other words, the asset price inflation in recent years is based at least in part on a (shaky) foundation of bank and broker-dealer lending to NDFIs that could change quickly and dramatically, with margin calls and fire sales. Lest one think such an occurrence would be highly unusual, it just happened in April amid the Liberation Day tariffs episode, and has happened many times before that.