No Thanks for Moving the Firewood: Notes on Macro Musings with Bill Nelson

Bending the chimney and moving the firewood earns no gratitude; the ones with scorched heads and singed brows are seated as honored guests.

— Ban Gu, Book of Han, “Biography of Huo Guang” (Eastern Han, c. 1st century CE; author’s translation)

A man notices his neighbor’s chimney runs straight up with firewood stacked beside it, and suggests bending the flue and moving the wood. The neighbor ignores him. The house burns. The neighborhood puts it out, the owner throws a banquet, and the guests of honor are the ones who got burned fighting the fire. The man who gave the warning isn’t given a seat.

Two thousand years later it still describes the same thing: preparation earns no applause; firefighting does. In finance, that’s where the most expensive tuition gets paid.

What This Episode Is About

On the August 17, 2026 episode of Macro Musings, David Beckworth brought back Bill Nelson, now chief economist at the Bank Policy Institute, previously deputy director of the Fed Board’s Division of Monetary Affairs — where he helped redesign the discount window and worked on the first round of quantitative tightening.

The subject sounds technical and is actually elementary: how big should a central bank’s balance sheet be, and where should bank liquidity come from?

For the past fifteen years the answer has been “from the central bank.” Banks park large reserve balances at the Fed, the Fed pays interest, and that’s a floor system. Nelson’s argument is that this arrangement was supposed to be temporary, got rationalized after the fact into something permanent, and carries a cost that rarely makes it into the ledger: it crushed the market where banks lend to each other.

The interesting part is that he’s no longer a lone voice. The ECB, the Riksbank, Norges Bank, the Bank of Canada, the Bank of England, and the Reserve Bank of Australia are all shrinking, and one stated reason is that they want the interbank market back. This episode is about how that turn happened, how it might be done, and — most usefully — where it could go wrong.

Key Takeaways

1. Borrowing from the central bank isn’t a crisis act. Treating it that way is a recent invention. Nelson points out that the Fed was created in 1913 precisely to convert loans into reserves or currency, so banks would have confidence the liquidity would be there. Reading the discount window as “only troubled banks go there” is a very contemporary view, and the view itself manufactures the stigma. He dug up an examiner report from the late 1970s in which a small bank was witheringly criticized for not bothering to manage its reserve account and simply keeping extra money there. In that era, parking cash at the central bank wasn’t prudent — it was lazy.

2. He dismantles “self-insurance” using his own house. The conventional framing says a bank only self-insures when it holds highly liquid assets; a borrowing arrangement doesn’t count. Nelson notes that he recently had a sewage problem serious enough that his family had to move out for major construction. He keeps some money in his checking account for exactly that. He also did the paperwork to pledge his home and establish a home equity line of credit. Why does the checking account count as self-insurance while the credit line he took the trouble to arrange does not? His point: once you let the other side define the discussion in those terms, you’ve already lost — because who could argue against self-insurance?

3. The floor system was rejected once, and the reason was interbank markets. In April 2008, when the Fed first got authority to pay interest on reserves, staff evaluated a floor system and advised against it: too radical a departure, and it would crush the interbank market. Then the crisis arrived, it became temporary policy, and the temporary got rationalized into doctrine. Nelson thinks the detail is worth remembering — this was a path, not a choice.

4. What actually moved foreign central banks may be that their losses were visible. Holding long-duration assets funded by interest-bearing liabilities means real losses when rates rise. The difference abroad is that they couldn’t be hidden. The Bank of England had to go to the finance ministry for a capital injection. The Riksbank built a facility where banks earn 0% on a slice of their deposits until the central bank is recapitalized — specifically so it wouldn’t have to ask the government. A central bank designing a mechanism to protect its own independence tells you what a large balance sheet does politically. The Fed, by contrast, carries its losses as a deferred asset, quietly, and there has been correspondingly little domestic attention.

5. The Quarles line is worth writing down. At the November 2018 meeting, then-Vice Chair Randy Quarles said having the FOMC control such a large stock of assets presents what lawyers call an attractive nuisance — an object a property owner leaves on his land that is obviously dangerous if misused and irresistibly appealing to passersby of impulsive and immature judgment, “such as children and congressmen.” Past the laugh, Nelson adds that government interest in using the Fed’s balance sheet to fund things is not an abstract worry.

6. A ceiling system doesn’t automatically revive interbank markets. The spread does. Many people, the host included, assume switching to a ceiling system solves it: the central bank becomes the marginal lender rather than the marginal deposit-taker. Nelson is cooler on this. There has to be daylight between the bid and the ask for there to be a market in between. That’s precisely the Fed’s current problem — the discount window was originally designed with a substantial spread, so it was used as a backstop and not constantly; today there’s very little daylight, and it’s unclear what the facility is supposed to be. Likewise, shrinking the balance sheet purely by adjusting liquidity regulations gets you a smaller balance sheet without getting you a market-first structure.

7. The SVB / Silvergate contrast is the hardest evidence in the episode. About a year before it failed, SVB called Nelson to ask whether signing up for the Fed’s standing repo facility would let it demonstrate it could monetize its Treasuries and agency MBS for internal liquidity stress tests. He raised it with the Fed; the answer was that they weren’t ready to go there. SVB didn’t sign up — it didn’t use repo, and building the relationships would have been expensive. The rest is known: sell available-for-sale securities on Wednesday, put the losses in plain view, run on Thursday, doors closed midday Friday, followed by a series of extraordinarily moral-hazard-laden rescues. He is emphatic: he is not claiming SVB could have survived — its losses roughly equaled its capital — only that it could have failed in an orderly way.

The control case is Silvergate. Same crypto exposure, same run (he recalls perhaps 60% of deposits), but it was prepared: it borrowed against collateral at the discount window to meet the run, sold the collateral down over time, repaid the loan, and wound itself up. No FDIC cost, no panic. “That’s how the system is supposed to work.”

Going Further

”The news says a bank borrowed from the central bank. Is it about to fail?”

This is the headline most likely to frighten a retail investor, and the episode offers a good way to read it: the information content of that signal depends entirely on what the regime looks like.

In a system where borrowing is an ordinary tool that everyone uses and nobody explains, “Bank X borrowed” carries almost no information — no more than a company drawing on a revolver it already negotiated. In a system where borrowing is widely read as a mark of desperation, the same headline suddenly carries a great deal. Not because borrowing became dangerous, but because stigma turned willingness-to-borrow into a filter: only those with no alternative go.

There’s a counterintuitive step here. The heavier the stigma, the stronger the signal — and the more fragile the system. Stigma makes banks that aren’t yet desperate wait until they are, and by then an orderly resolution is usually off the table. SVB is the full demonstration: it wasn’t even unwilling. It asked, was told not yet, and took the path that put its losses on public display instead.

So next time, ask first: under today’s arrangements, is going to the window routine or humiliating? The same act means two entirely different things in the two regimes. Which is why Nelson keeps saying the only way to make people comfortable with the window is to make it ordinary.

”What does any of this institutional debate have to do with my positions?”

Honestly, not “therefore buy X.” More like “therefore stop reading certain moves as direction.”

Nelson describes a practical habit: he ran a small cottage industry writing notes warning that in two weeks coupon settlements would create a reserve shortage, that an open market operation was needed, and that otherwise there would be volatility — followed by “there’s volatility, we have to stop shrinking.” That’s a textbook case of a predictable technical pothole being misread as a structural signal.

There are structural signals worth watching, though. He notes JPMorgan’s Fed balance fell from roughly $400 billion at end-2023 to $100 billion at end-2025, and one driver was repo rates moving up relative to the rate on reserves. When the central bank’s account stops being the best place for the money, banks move it themselves. That’s the test of whether the engine is running. He also offers a sense of scale: pre-2019 Fed staff told the committee structural reserve demand was about $1 trillion; today it’s around $3 trillion, and the economy has not tripled.

Put together, the portable judgment is this: assume short-term money market moves are a calendar problem (settlements, quarter-end, tax dates) before assuming they’re a system problem; to see whether the system is genuinely changing, watch whether price relationships have flipped. That habit of separating noise from structure is worth more than any specific rate forecast.

”I prepared and nothing happened. Was it wasted?”

This is the one worth taking home.

The difference between Silvergate and SVB wasn’t asset quality — it was whether the channel had been walked. SVB was awash in high-quality liquid assets and still failed its internal stress test, because it had no path for turning them into liquidity. Unpack that and it’s deeply counterintuitive: high asset quality is not liquidity. Liquidity is a route that works on the day you need it, not a number on a page.

The difficulty with this kind of preparation is that its return is negative right up until it isn’t. You sign documents, build relationships, complete formalities, and in 999 days out of 1,000 nothing happens. That’s exactly the situation the opening line describes — the value of preparation is invisible precisely when it isn’t triggered.

For a portfolio, the analogue is that your trim plan, your cash allocation, and your stop discipline aren’t there to raise returns. They’re there so that in the worst week you can still make decisions in an orderly fashion. The best decision anyone makes in a panic is usually the one they wrote down while calm.

Worth a Look

• The episode itself: Macro Musings with David Beckworth, August 17, 2026, with guest Bill Nelson. If you want the full argument, go listen.
• Nelson’s remarks in Europe (ECB, BIS, Riksbank) are publicly available and combine the two threads: liquidity requirements should recognize the capacity to borrow from the central bank, and here’s how a smaller balance sheet could work.
• On why interbank markets matter, Claudio Borio made the case on the same program: an arrangement where a bank looks first to the interbank market and to the central bank only as a backstop beats one where all liquidity problems are solved by the central bank.
• Norway’s tiered reserve system (adopted 2011) and the Fed’s 2008 memo on voluntary reserve targets are two concrete templates for “what if not a floor system.”
• Recent operating-framework papers referenced in the episode — Lorie Logan, Steven Miran, Darrell Duffie — are part of the same wave.

The One Thing to Take Away

Liquidity isn’t how much you hold. It’s how many routes you’ve actually walked.

That’s the single idea worth keeping. SVB was full of quality assets and couldn’t pass its own stress test; Silvergate lost most of its deposits and wound down calmly. The only difference was that one had walked the route in advance. A channel you’ve never actually used is a channel you only think you have.

Here’s a practice for today, and it works well outside investing:

Take a piece of paper and write down your real emergency channels — not assets, channels. The phone number you’d call if something went badly wrong. The card you assume still works. The drawer with the important document. The paperwork you keep meaning to complete. The place you could stay for a few days if you truly had to. Three is enough.

Then pick one and walk it all the way today. Make the call and say “I’ve been getting some things in order and wanted to check that I can still reach out if something comes up.” Put a small charge on the card. Take the document out and look at the date on it. Finish the form you’ve been postponing for six months.

You’ll learn one of two things. Either the route works, and you can genuinely count it from now on — or it quietly stopped working some time ago, and you found out in good weather instead of on the day you needed it.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.