EndGame Macro
When the Buffers Run Thin: Reading the Fed’s Balance Sheet in Real Time

If you look at the latest H.4.1 without getting lost in the line items, the message is pretty simple: the Fed has taken QT about as far as it can go without pressing directly on the parts of the system that actually matter. The asset side is still huge at roughly $6.58 trillion but the shift isn’t happening there. The interesting story is on the liability side, where you can see how QT, Treasury behavior, and foreign flows are all tightening the same bolts at the same time.

The Fed still holds an enormous pile of securities: just over $4.19 trillion in Treasuries and around $2.07 trillion in MBS. QT shaved a few hundred billion off those totals over the past year, but in the grand scheme it barely dents the mountain. Ending QT on December 1 simply freezes the portfolio where it is and lets it slowly tilt toward more Treasury bills as older MBS run off. It’s not a shift back to QE, it’s more like pressing pause and letting time shorten the duration profile.

Where the Real Tightness Is Coming From

The real pressure shows up on the liability side, the plumbing that determines how far QT can go before something breaks. Bank reserves are down almost $400 billion over the past year. Reverse repo balances fell by more than $200 billion. And now the Treasury General Account is sitting around $943 billion not quite a trillion, but still very high which means Treasury has rebuilt a large cash buffer. Every time they issue debt and leave the cash idle in the TGA instead of spending it, it quietly drains liquidity out of the banking system. Add to that the steady rise in physical currency in circulation, and you have another slow but constant pull on reserves.

Those pieces might look separate, but they all push in the same direction: the pool of easy excess liquidity has thinned out. Early in QT, most of the balance sheet shrinkage came from the reverse repo facility, a giant pool of cash money markets had parked at the Fed because they had nowhere else to put it. That buffer is mostly gone now. Any further QT would hit bank reserves directly, and the Fed remembers what happened in 2019 when reserves got too close to the floor. They have no desire to run that experiment again, especially now with larger deficits, heavier issuance, and more complicated global flows.

This is where the foreign side matters. The H.4.1 shows roughly $3.06 trillion in U.S. securities that foreign central banks and international institutions hold in custody at the Fed. That’s the official sector…reserve managers, sovereign wealth authorities, multilaterals, not traders. They still hold a massive amount of Treasuries, but over the past year they reduced those holdings by about $264 billion. That doesn’t mean a dump; it just means they weren’t adding to their positions during a period when the Treasury needed buyers and the Fed was letting its own portfolio shrink. Even foreign use of the Fed’s reverse repo facility has ticked slightly lower, which is another gentle sign that they’re not adding liquidity the way they did in past cycles.

Put all of that together and the picture comes into focus. The Fed still has a huge portfolio. Foreign official holdings remain enormous. But the buffers that made QT look effortless for a while…giant RRP balances, steady foreign accumulation, a smaller TGA have all moved in the opposite direction. So now, every additional step of QT would mostly come out of bank reserves at a time when nobody wants to find out where the new lower bound actually sits.

Ending QT here isn’t a dramatic pivot. It’s the Fed reading the room. The official sector isn’t stepping in to absorb more supply, Treasury is pulling liquidity by running a fat TGA, and the easy drains have been used up. Better to stop tightening the plumbing before the plumbing starts tightening them.

Interactive guide to our weekly #BalanceSheet report: https://t.co/75xiVY3BGu #FedData - Federal Reserve tweet

#BalanceSheet report: https://t.co/75xiVY33QW #FedData - Federal Reserve tweet

EndGame Macro
The Banking System Is Still Expanding But The Shape of That Expansion Is the Red Flag

The newest H.8 release is one of those reports that looks calm if you only skim the top line. Bank credit is growing, deposits are coming in, and nothing jumps out as an immediate crisis. But the moment you start looking at where the lending is happening and where it isn’t, the tone changes. This is a banking system still moving forward, but in a way that feels late cycle, not early cycle.

The Real Economy Is Losing Access to Credit

Commercial and industrial lending is soft across the board. A year ago, big banks were holding around $2.26 trillion in C&I loans. Now they’re closer to $2.15 trillion. At the system level, C&I barely moved, drifting from $2.784T to $2.692T. That is not a picture of businesses gearing up for expansion. It’s the opposite…businesses aren’t borrowing for inventories, for payroll, for growth. They’re holding back.

Construction and land development lending tells the same story. It has slipped from roughly $481B to the low $450B range. Developers are always early cycle barometers: they don’t borrow when they don’t like the demand outlook.

And consumer credit? Also down year over year, dropping from about $1.925T to $1.846T. Credit card balances inch up, but total consumer credit weakens. That usually means households aren’t comfortable stretching further, not because life is cheap, but because budgets are tight.

Put those together and you get a simple truth…households, businesses, and builders are all stepping back at the same time. That’s not what expanding cycles look like.

The Credit Boom Is Happening… Just Not In the Real Economy

Now the part that really matters: “All Other Loans and Leases.” This category jumps from about $2.20T to $2.91T in a year. Inside it, lending to non depository financial institutions goes from $1.12T to roughly $1.71T and keeps rising weekly.

That is a huge shift. Banks aren’t lending to firms that make things, build things, or hire people. They’re lending to the financial ecosystem of private credit funds, securities lenders, mortgage intermediaries, hedge funds, private equity structures.

This is what late cycle credit looks like. The real economy slows, bank appetite for risk migrates toward financial players, and leverage rises in places that don’t show up in traditional economic indicators. It keeps asset prices afloat longer than fundamentals justify, but it also creates the kind of conditions where a shock can travel fast.

Liquidity Buffers Are Thinning

Cash assets dropping from $3.28T to around $3.02T is not nothing. Banks are holding less liquidity while expanding credit especially to leveraged financial institutions. Deposits are up slightly, borrowings are drifting down, but the shrinking cash cushion makes the system more sensitive to funding stress.

It’s not a crisis signal, but it’s not the profile of a system preparing for stronger growth either. It’s a system optimizing liquidity because the real economy isn’t giving it enough profitable lending opportunities.

My Read

This isn’t a banking crisis report. It’s a slowdown is already underway report. And the pattern is incredibly clear that credit is still growing, but it’s flowing into the financial economy, not the real one.

C&I weak. Construction slipping. Consumer credit soft. Meanwhile loans to financial intermediaries surge.

That’s how you get a world where markets look stable while fundamentals fade. It’s how late cycle expansions turn into early cycle contractions…quietly, in the credit data, long before it hits the headlines.

The H.8 is telling you something simple and important…The real economy is cooling, and the credit system knows it.

Now available: Weekly data on the H.8 release, Assets and Liabilities of Commercial Banks in the United States #FedData https://t.co/WeJhdaAjBV - Federal Reserve tweet

EndGame Macro
H.8: Credit Is Expanding, Flexibility Is Shrinking

Total bank credit is still expanding at roughly 5–6% annualized. Lending hasn’t frozen. Banks are still making loans. On the surface, the system appears functional.

But the composition of that growth matters. Nearly all of it is coming from loans and leases, running close to 6% year over year, while securities holdings are flat to slightly negative. Banks are not building liquid buffers. They’re adding credit risk. That’s fine in a strong economy with easy funding. It’s much riskier when growth is slowing and funding is tightening.

Deposits are no longer doing the work

Deposit growth has essentially stalled. On a short term annualized basis, it’s near zero.

Inside that flat number, the mix is shifting in an uncomfortable way. Lower cost “other deposits” are declining, while large time deposits are rising sharply. That usually means banks are paying up to keep money from leaving. This isn’t a run. It’s a repricing. But repricing raises funding costs, compresses margins, and increases rollover risk if rates stay high or volatility picks up.

Liquidity cushions are thinner

Cash assets have dropped from roughly $3.23T to $2.92T over the past year, more than a $300B decline. Some of that reflects reserves leaving the system, but the practical effect is less immediately usable liquidity.

Banks have offset this by holding more fed funds sold and reverse repos, so liquidity hasn’t disappeared, but it’s become more conditional. At the same time, total bank assets are now shrinking on a short term annualized basis, which signals balance sheet defense, not expansion.

Borrowings are being cut back

Borrowings are falling sharply year over year. That can reflect prudent derisking of expensive funding, or simply fewer attractive funding options. Either way, it signals constraint, not abundance.

Where the risk is concentrated

Commercial real estate exposure remains skewed. Small banks hold about $2.08T in CRE loans, compared with roughly $0.85T at large banks. Office vacancy rates are around 18–19%, and CMBS office delinquencies are near 12%, both record highs. CRE doesn’t need a price collapse to hurt and just refinancing at higher rates with weak cash flow is enough, and that pressure lands where capital buffers are thinner.

At the same time, large banks’ lending to non depository financial institutions has jumped from about $770B to $1.13T, a roughly 45% increase year over year. In calm markets that looks like intermediation. In stress, it becomes a fast transmission channel.

The backdrop isn’t helping

Consumer stress is rising. Subprime auto 60 day delinquencies are around 6.5%, a record. Credit card delinquencies are near 3%, elevated. Student loan delinquencies are back, with 9–10% of balances 90+ days past due and some measures showing 30% of borrowers behind.

Labor and business data are weakening. Over 1.17M job cuts have been announced in 2025, the highest since 2020. Initial claims are near 236k, continuing claims around 1.8M, and business bankruptcies are up roughly 5–6% year over year.

The uncomfortable takeaway

Loan loss allowances are not rising meaningfully yet, which is typical late cycle behavior. Recognition usually lags reality.

The system right now is reallocating risk while losing flexibility with more loans, less cash, flat deposits, higher funding costs, CRE concentrated at small banks, and growing exposure to the nonbank system. It works as long as losses stay deferred. That’s what makes this phase dangerous…not because it looks bad, but because it still looks orderly.

Now available: Weekly data on the H.8 release, Assets and Liabilities of Commercial Banks in the United States #FedData https://t.co/WeJhdaAjBV - Federal Reserve
tweet

EndGame Macro
Banks Look Healthy Until You Ask Who They’re Lending To

The December 19 H.8 tells a pretty clean late cycle story. Banks are still expanding, but they’re doing it with less liquidity and more exposure to the parts of the system that matter most when conditions tighten. As of the week ending December 10, total commercial bank assets are about $24.5T, bank credit is $18.9T, and loans and leases sit around $13.3T all higher than mid year. The change is on the cushion side. Cash assets have slid from roughly $3.38T in June to about $2.9–$3.0T now. There is more lending but less cash on hand. That’s not a crisis by itself, but it absolutely changes how stress would spread if something breaks.

Where the Risk Is Quietly Migrating

The composition of that growth matters far more than the headline totals. Loans to non depository financial institutions (NDFIs) have jumped from about $1.57T in May to roughly $1.81T by mid December. These aren’t households or operating businesses they’re funds, mortgage credit intermediaries, private equity vehicles, hedge funds, insurance related entities, securitization vehicles, and other financial middlemen. In calm markets, this credit looks fine. When volatility hits, these are the borrowers that tend to all need liquidity at the same time, which is how funding stress accelerates.

By contrast, traditional C&I lending is basically flat around $2.7T, construction and land development loans have drifted lower to about $455B, and CRE overall is still inching higher near $3.07T. That mix says banks are cautious about new real economy activity, even as exposure to the existing asset stock keeps building.

Why This Lines Up With the Bankruptcy Data

That balance sheet picture matches what we’re seeing in business stress. Through September 2025, there were about 8,937 Chapter 11 filings including Subchapter V, with 767 commercial Chapter 11s and 210 Subchapter V elections in September alone. On the surface, that looks manageable because it’s below crisis era peaks. But in practice, it means a growing share of stress is being worked through smaller, faster restructurings that don’t look dramatic in aggregate data. Distress is rising it’s just being absorbed quietly.

Why This Setup Matters

Two things stand out to me. First, liquidity is being used up. Cash is shrinking while loan growth is running north of 6%, and the allowance for loan and lease losses is still hovering around $203B, basically flat, despite a larger and more complex loan book. Second, risk is drifting downstream away from plain vanilla lending and toward nonbank financial exposures and smaller business restructurings that don’t immediately trip system wide alarms.

My View

Banks are still extending credit, drawing down cash, and increasingly financing the nonbank financial system at the same time that business failures are rising, just not in a way that looks explosive. That works fine when markets are calm. It’s exactly the setup where pressure shows up first when volatility spikes, funding tightens, or confidence slips. The H.8 and the bankruptcy data are quietly telling you where the stress is being parked and where it’s most likely to surface if conditions turn.

Now available: Weekly data on the H.8 release, Assets and Liabilities of Commercial Banks in the United States (1/2) #FedData https://t.co/WeJhdaARrt - Federal Reserve
tweet

eSheet report: https://t.co/75xiVY33QW #FedData - Federal Reserve tweet