Why the Global Balance Sheet Grew

In this issue:

• Why the Global Balance Sheet Grew—Global financial obligations are rising faster than GDP. If you frame that as the world borrowing money, it looks unsustainable. But it's also the world lending money—a borrower's interest cost is a lender's interest income. We should expect balance sheets to grow in comparison to output, and it's a bad sign if they don't, even if it's also true that individual borrowers can go too far.
• The Fed—In practice, central bank independence consists of conflicts between central banks and other parts of the government, in which the bank comes out ahead.
• Walmart as Google's Alt-Azon—A big retailer is an indirect complement to a big ad, logistics, and cloud computing business. And if you're missing the retail piece, you can get it via partnerships.
• Two Convertible Bond Stories—The volatility monetization continues.
• FX Trades—The US scores a win on an emerging market bet, but it's an asymmetric one.
• Cheap Oil—It's hard to keep an easy arbitrage going forever.

The Diff January 12th 2026 14:31 • Click to play audio

Why the Global Balance Sheet Grew

In the early 2000s, Warren Buffett and Carol Loomis wrote about what's now known as the Buffett Indicator, the ratio of market value to GDP, warning that valuations remained stretched and that forward returns would be low.^[1] As a macro call, this was pretty solid; the S&P's total return over the next decade worked out to 2.9% compounded, and at two separate points in that period, late 2002 and early 2009, an investor could have sold out with a 30% loss. The timing was even better if you start the clock in the summer of 1999, when Buffett made the same point at the Allen & Co. Sun Valley conference—over the next decade, the S&P compounded at negative 2.9%, and an investor who had happened to buy just then would be treated to one of the rare decades of negative nominal returns from US stocks.

Then, in late 2019, the indicator hit a new all-time high, but if you'd bought then you would have made 16.4% compounded since. Oh well.

If you compare market capitalization to GDP, you're making some reasonable assumptions and some unreasonable ones. A simple intuition is that markets are pricing in current earnings and future growth, and that corporate profits can't be more than 100% of GDP and will probably mean-revert over sufficiently long periods. So, for the market to go up, you probably want 1) GDP to go up, and 2) perceptions of future GDP growth to be high.

On the other hand, every time a big company accepts a private equity bid, or a tech company raises a late-stage round instead of going public, the indicator goes down. To put a finer point on it: if someone does a big, levered transaction like RJR Nabisco in 1988 and TXU in 2007, the indicator treats these as bullish signs. Similarly, if a tech company stays private specifically because VCs will give it a higher valuation than public markets, that, too, gets translated into a bullish signal.

You can take a more comprehensive view, as McKinsey does in their annual global wealth report: they look at the total size of the global balance sheet, including equities and debt, compare the growth of this balance sheet to total GDP growth, and then note that for lenders to get their money back and for equity holders to get a satisfactory return, we need more productivity growth (and should borrow a bit less).

But is that actually true? Suppose we start out with a simple economy that has zero leverage; everyone transacts with bearer instruments, so their financial wealth just consists of however many coins they have. Let's suppose this society invents the mortgage: an older person with lots of savings and few expenses decides to lend money to a young person so they can buy a house and form a family. How much more output do we need?

That question is clearly incomplete as stated. The borrower's balance sheet is now more stressed, since they have a new nominal obligation. But the lender now has a new source of income. If both sides have similar spending behaviors, the only place to look for an output impact is in what happens with the money, and in this case the effect is stimulative: there's more demand for housing, so an additional unit of housing gets built, and previously inert savings turn into additional economic activity. (We'll assume that this society has invented the mortgage but not yet invented zoning, permitting, etc.)

In fact, in one important sense the introduction of this new leverage, and the expansion of the collective balance sheet, has reduced risk. Our older saver may worry that he'll lose the ability to work well before he dies, and that those finite savings will run out. Earning a return on them extends the financial runway. The younger saver runs that risk, too, but it's just a lot less likely to happen to them. So in this case, what more leverage actually signifies is a successful risk transfer that shifts financial risk to someone with less income risk, and vice-versa.

If you view credit entirely through the lens of the borrowers, it's worrisome to see the number rise. But a loan is a bilateral transaction: every borrower needs a lender, and even if the banks that intermediate loans are creating money by changing numbers in a database rather than by taking cash deposits out of a vault and delivering them to the borrower, the economic impact of making an additional long-term loan is to shift slightly more economic activity towards things with a longer-term payoff.^[2]

What about government borrowing? This, too, partly fulfills an intermediation function. Savers, particularly older ones, shouldn't be taking much equity market risk and also shouldn't be taking much credit risk, and it's good for the rest of the financial system to have some kind of low-risk asset to keep money in temporarily. In some cases, savers are doing this indirectly; an annuity, pension, or life insurance policy will also be backed by government bonds, in part because these bonds are the cleanest source of duration, and any promise to deliver a fixed return over some future period is shorting duration.^[3] If you look at who has the most money and where that money goes, deficits are just a way to ensure that extreme wealth inequality among the 65+ cohort leads to smaller consumption inequality in the same: richer old people buy the treasury bonds that pay for the Medicare and Social Security of the poorer ones.

But that's a single-country picture. Globally, one of the functions of sovereign debt is to do the same intergenerational intermediation between countries with different demographics, rather than within those countries. Aging countries like China or Korea need to accumulate financial claims on the output of countries that are younger, or at least getting older more slowly. You can read this in an incredibly literal but fairly useful way: the big categories of long-duration physical assets that get funded with long-duration debt are houses and infrastructure. A country with a growing population needs more houses, and, to the extent that home costs limit family formation, will grow even faster if it can build more of them. In aging countries, there's potentially a surplus of housing stock, except in Japan, where physical houses are designed to depreciate faster and get torn down more often ($, Diff). So an older country with a higher demand for fixed income assets is, directly or indirectly, providing mortgage financing to a younger country that needs it.

And that housing isn't equally valuable everywhere. The richer a country is, and the more efficient its labor markets, the more expensive housing should get relative to incomes. Every home's value is partly the physical dwelling, but mostly the local labor market and amenities e.g. high quality public schools, entertainment, a better dating pool, etc. (Things like a good view matter, but there are a lot more million-dollar condos in downtown-wherever than there are million-dollar vacation homes in beautiful-but-empty places that you'd love to spend a few weeks a year in but go crazy living in full-time.) The more predictable that labor market—the better people can assess which superstar city is the right one for them—the higher housing prices will get because more people will make moves that both increase their housing costs as a share of disposable income and increase their purchasing power-adjusted disposable income ($, Diff).

All that is to say that in the aggregate, a bigger global balance sheet relative to output means two things:

1. Markets are getting less efficient, and everyone's bidding up growth, or
2. Markets are getting more efficient, meaning that we find the right high bidder for every person, house, share of stock, and bond, and that this means there are more real assets that it makes sense to convert into financial obligations.

Or both; it's complicated. But in general, expect the size of the global balance sheet to expand relative to GDP indefinitely, with brief retreats during large-country recessions and global crises. These retracements will incorporate some mean-reversion as stretched valuations return to sanity, but a lot of what they'll measure is that richer civilizations can make more promises and keep more of them. And that, in the end, is what the global balance sheet relative to GDP really measures: how many credible promises are worth making, and how many of them are worth keeping?

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• A frontier investment firm is looking for someone with exceptional judgement and energy to produce a constant feed of interesting humans who should be on their radar. This person should find themselves in communities of brilliant people hacking on technologies (e.g. post-quantum cryptography, optical computing, frontier open source AI etc.) that are still well outside the technological Overton window. You will be responsible for identifying the 50–100 people globally who are obsessed with these nascent categories before they are on-market, then facilitating the high-bandwidth IRL environments (dinners, retreats, small meetups) that turn those connections into a community. (Austin, NYC, SF)
• A startup is automating the highest tier of scientific evidence and building the HuggingFace for humans + machines to read/write scientific research to. They’re hiring engineers and academics to help index the world’s scientific corpus, design interfaces at the right level of abstraction for users to verify results, and launch new initiatives to grow into academia and the pharma industry. A background in systematic reviews or medicine/biology is a plus, along with a strong interest in LLMs, EU4, Factorio, and the humanities. (Remote, Toronto)
• A blockchain company that’s building solutions at the physical limits of distributed systems and driving 10x performance improvements over other widely adopted L1s is looking for an engineer with expertise in systems programming languages (e.g C, C++, Rust, etc.) HFT, embedded systems, high performance computing or operating systems experience a plus. (SF)
• A well-funded, Series C startup building the platform and agent primitives to drive operational transformation at large, complex institutions (starting with higher education) is hiring platform engineers. The work spans distributed systems, applied AI, and full-stack infrastructure, focused on deploying reliable agents that meaningfully bend institutional cost curves. (Remote)
• A hyper-growth startup that’s turning the fastest growing unicorns’ sales and marketing data into revenue (driven $XXXM incremental customer revenue the last year alone) is looking for a senior/staff-level software engineer with a track record of building large, performant distributed systems and owning customer delivery at high velocity. Experience with AI agents, orchestration frameworks, and contributing to open source AI a plus. (NYC)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.

Elsewhere

The Fed

It would not be an especially big deal if Jerome Powell had wasted some money renovating the Fed's buildings. Compared to the potential cost of other mistakes he could make, it would be a rounding error. That's not to say that he should be entitled to take kickbacks from contractors or install a solid gold hot tub in his office, just that if he did that, the impact would pale in comparison to getting the timing of the next rate hike or cut wrong. But he's being investigated for this, and he's gone ahead and said that this investigation is completely political and that he does not intend to change the Fed's policy in response. There are plenty of examples of American presidents leaning on the Fed, with varying degrees of success. But there's also a norm that they shouldn't do this, and that if they do, it ought to be subtle and deniable. Truman, Johnson, Nixon, Reagan, George H.W. Bush, and Clinton all did this to varying degrees, so it's not universal, but pretty common. But the more aggressive pushes tend to be the ones that the Fed successfully pushes back on—central bank independence in the US is a long history of the President wishing rates were lower and the Fed successfully saying no. Which is one reason markets don't react too strongly to this kind of thing: the modal outcome is that Powell will do his job and that future Fed chairs will have even more discretion, though that has to be balanced against the risk that this will be the one big exception.

Walmart as Google's Alt-Azon

It's easy to overstate the extent to which AWS was a direct strategic complement to Amazon's retail business. It's a lot closer to a successful attempt to extract the platonic essence of that business, by identifying a missing market that had enough characteristics in common with Amazon's retail business that they could justify working on it, but that was different enough that it gave them exposure to a different addressable market. That said, it is convenient for AWS that Amazon's retail business is a big anchor customer that reliably runs into scaling problems and feature needs ahead of the median AWS user, and is thus a valuable source of feedback. And for Amazon's retail business, it's nice to have a cloud partner whose product roadmap will tend to reflect what they need. This is something other hyperscalers don't quite have, but Google has been deepening its Walmart partnerships in some interesting ways: adding more drone delivery and allowing Gemini users to do Walmart shopping directly through Gemini.

What's interesting about all of these maneuvers is that every company involved has a positive-sum reason to do the deal—Google Wing or Gemini as standalone companies would be happy to work with a big customer like Walmart, and Walmart is always delighted to find a new way to reach customers and get products to them. But in the increasingly mature retail market, Walmart is very happy to take share from its biggest competitor, and in AI Google is very excited to start collecting more first-party purchase data and to refine its monetization approach. Ads remain an incredibly economically compelling way to monetize LLMs, but it takes a lot of work to go from noticing that to actually having LLMs that can make good product recommendations.

Disclosure: Long AMZN, GOOGL (they're both good at what they do).

Two Convertible Bond Stories

Convertible bond issuance hit a 24-year high last year. The thing to remember with convertible bonds is that for individual companies, they monetize the volatility of the stock, and that in the aggregate, their popularity is basically a measure of how many companies that are big enough to be worth the time of a major investment bank are also volatile enough that converts are the cheapest source of capital. And it turns out that there are a lot of them. But this also leads to a demand-side story; at one point early last year, one company, the one formerly known as MicroStrategy, was 30% of new convertible bond issuance (though mostly because it was a slow start to the year. Convertible bond funds don't want a single company to dominate their portfolios, even if they aren't making a directional bet on it. So when one company with a temporary infinite money machine finds out that spiking its volatility by constantly marketing itself to retail investors creates another infinite money machine, that ends up encouraging other potential convertible bond issuers to follow suit.

It's not clear how many companies, if any, are deliberately trying to make their stock volatile specifically so they can raise cheap capital with convertible bonds. But it doesn't have to be a deliberate strategy: if the companies that can engineer an unusually bouncy stock price tend to stick around longer and grow bigger because they have cheap capital, that'll be a more popular investor relations approach even if they don't quite understand the mechanics of how it works.

Disclosure: Short MSTR (small amount).

FX Trades

In October, the US gave Argentina access to an emergency swap line ahead of an election, and after that election went well for Milei, the money was repaid ($, FT). Milei is basically right about the direction Argentina's economy should go, and skeptics are also right that it's very hard to unwind an interventionist economic policy without risking a crisis—price controls and limitations on competition create an economic asset for the beneficiaries, so undoing them has the same kind of inflationary effect as a typical credit contraction, but less visible because the assets in question aren't marked to market. Providing emergency liquidity to a country in this situation can work, but it's a picking-up-pennies-in-front-of-a-steamroller kind of trade, where the upside is a series of small wins and the downside is being the newest creditor to a country that runs into a financial crisis. The US's balance sheet is big enough that this alone isn't a meaningful threat, but it is a cost, and one that has to be weighed against the upside.

Cheap Oil

In one sense, oil is the most global market imaginable—it's extracted in many different places, cheap to transport relative to its cost, and useful just about everywhere. But various sanctions regimes mean that there are really multiple oil markets, where buyers can pay the full price for oil imports that the US approves of, or can trade a discount for the risk that America will get mad at them. China, for example, was paying $9/barrel less for Venezuelan oil than for comparable oil from Canada, but now at least some of that oil will be going to the US ($, WSJ). In the short term, geopolitical actors just aren't going to wave around copies of Free to Choose and insist that everyone interact in a standard market. But in the long run, it's hard for the same molecules to have different prices for different people without someone having an incentive to close the arbitrage.