Who Gets the Money First in America? A Horizontal-Vertical Analysis of Credit Allocation and Capitalization

A street vendor earns a markup. A person who builds a company earns a valuation. Between the two sits one question: who gets other people’s money first.

Author: Koutian Wu; GitHub: ktwu01

A Chinese essay I read recently describes a very “Chinese” view of business. Its gist: from the start, the goal of doing business is to get “the imperial court” to give you money; earning money from ordinary people is a side effect. The court releases money through policy money houses, ordinary people deposit their money in the money houses, and if you fit the policy, the money reaches you. You use it to turn a cornfield into a mall, a factory or an office tower; once there is cash flow, there is stock, and “the Earth God has nearly caught up with the God of Wealth.”

Parts of this account are exaggerated, and parts are accurate. My question is: does this logic hold in the United States? America has no “court-designated money houses,” but it has the Federal Reserve, $49.1 trillion in retirement assets, and hundreds of billions of dollars of venture capital each year. Money has to come out of somewhere and flow into someone’s hands.

So this post takes the question apart with horizontal-vertical analysis. The vertical axis traces how “who gets the money first” changed over more than two hundred years of American history. The horizontal axis looks at the channels that hand out money in the United States in 2026 and compares them with China, Japan and Germany. At the end I return to the Chinese essay and say where it is right and where it is wrong.

The full report is long. The PDF version (in Chinese) is here: 美国信用分配与资本化_横纵分析报告.pdf.

1. A one-sentence definition

I define the subject this way: newly created credit and other people’s savings pass through a gatekeeper and flow first to a small group of people; these people turn low-yield assets into cash-flowing assets, and the market then discounts the future cash flow into a valuation today.

There are four parts here: where the money comes from, who the gatekeeper is, what assets the money becomes, and how the valuation gets amplified. In the Chinese version, the state directly holds three of the four parts. The question for the American version is who holds each of the four.

2. Vertical analysis: 230 years, seven handoffs

Stage one (1790s): government debt becomes private capital

Hamilton installed the first part of the American machine.

In 1790 the new federal government held a pile of bad debts left over from the Revolutionary War. State debts and Continental Congress debts were worth something on paper but traded at deep discounts, and many veterans had long since sold their bonds to speculators at a fraction of face value. Hamilton’s plan: the federal government would assume all of these debts at face value, state debts included. Congress rejected it twice. Southern members then traded their votes for a capital on the Potomac (the Residence Act of 1790), and the assumption of state debts passed. The Funding Act of August 1790 authorized the assumption of up to $21.5 million in state debt; $18.3 million was actually assumed, and the new federal bonds paid 6% interest. By early 1791 total federal debt was about $75.46 million. (Funding Act of 1790)

The effect of this step: a pile of paper that nobody trusted became an asset backed by national tax revenue that paid steady interest. Whoever held the paper had “capital” overnight.

The First Bank of the United States came next. It received its charter in February 1791 with $10 million of capital: $2 million from the government and $8 million from private investors. The shares went on sale in July and sold out the same day, followed by frantic bidding for subscription rights. (Federal Reserve History) According to the common account, private subscribers could pay most of the price of their shares in government bonds, so the national debt turned directly into bank capital. I have not verified this point in a primary source.

Then trouble came. William Duer, Hamilton’s friend and former assistant secretary of the Treasury, borrowed money to try to corner the 6% bonds. Pushed by his buying and by the Bank of the United States’ credit expansion, the 6% bonds rose from 110 to 125 between December 1791 and January 1792. On March 9, 1792, Duer stopped payment, and Wall Street had its first crash. Hamilton responded by buying bonds in the open market and having the Bank of New York lend against bond collateral at a penalty rate of 7%. The panic subsided in April. (Liberty Street Economics) The authors of the New York Fed blog consider this a “lender of last resort” operation eighty years ahead of Bagehot’s theory.

In my view this episode already lays out the skeleton of the whole logic: state credit creates a new asset; the people closest to power learn of it first and buy first; someone bets with leverage; the bet collapses, and the state steps in to hold up the floor. More than two hundred years later the names have changed and the structure has not.

Stage two (1862–1913): cornfields become railroads, railroads become Wall Street

The most vivid line in the Chinese essay is “this land can’t keep growing corn; it has to grow buildings.” America’s largest nineteenth-century conversion of land into capital was the railroad.

On July 1, 1862, Lincoln signed the Pacific Railway Act, chartering the Union Pacific and the Central Pacific to build a transcontinental railroad. The government gave two things. The first was land: the odd-numbered sections within 10 miles of either side of the track, doubled by an 1864 amendment. (National Archives) The second was money: federal 30-year bonds lent by the mile, $16,000 per mile on the plains, $32,000 in the foothills and $48,000 in the mountains, a total of more than $60 million. (CPRR) From 1850 to 1871 the federal government gave railroads about 130 million acres, roughly one tenth of the public lands.

I think what the railroad companies did with the land closely resembles today’s Chinese local government financing vehicles: they borrowed against the land, attracted investors with the expectation of rising land prices along the line, and earned the least from building the railroad itself.

The most profitable link was the “construction company.” The Union Pacific’s builder was Crédit Mobilier, later led by Congressman Oakes Ames. Starting in 1867, Ames sold the company’s stock to members of Congress at par, far below market price; he lent some of them the money to buy it or let them pay with future dividends. The New York Sun exposed the scheme in September 1872. Secondary sources estimate the inflated construction charges at between $19 million and $44 million. (Library of Congress)

The Homestead Act, signed the same year, gave ordinary people up to 160 acres each for a small filing fee; after five years of residence the land was theirs. In the end about 270 million acres went to individuals. (National Archives) That sounds egalitarian. One frequently cited estimate, though, holds that of the roughly 500 million acres the General Land Office disposed of from 1862 to 1904, only about 80 million reached homesteading farmers; the rest went to railroads, state governments and land speculators. I could not verify this figure in a primary source.

Once the land was handed out, money began to concentrate. In the early twentieth century America had no central bank, so Morgan served as one. In the Panic of 1907 depositors ran on the Knickerbocker Trust in New York, and Morgan gathered the bankers in his own library to organize a rescue. (Federal Reserve History) After the rescue, Congress grew afraid: a private individual who can save the whole financial system can also decide who lives and who dies.

The Pujo Committee investigated the “Money Trust” in 1912 and 1913. It found that 180 men held 341 directorships in 112 corporations controlling $22.2 billion in resources. At the time the total market value of the New York Stock Exchange was only about $26.5 billion. (Pujo Committee)

Louis Brandeis, later a Supreme Court justice, wrote a book based on these findings. Its title states the theme of the Chinese essay outright: Other People’s Money and How the Bankers Use It (1914). Two sentences from the book:

“The power and the growth of power of our financial oligarchs comes from wielding the savings and quick capital of others.”

“They control the people through the people’s own money.” (Project Gutenberg)

The Chinese essay says that “ordinary people save their money into the purse, and you take the money back out of the purse.” Brandeis described the same thing in 1914. In his eyes, though, this system had to be dismantled, and nobody should copy it.

Stage three (1913–1945): the state hands out money in person for the first time

In December 1913 the Federal Reserve Act was signed, and the United States had a central bank. (Federal Reserve History) The Fed of the 1920s was still far from today’s. In that decade the real money machine was brokers’ loans: from September 1926 to September 1929 they rose from $3.2 billion to $8.5 billion. Ordinary people bought stock on margins of 10% to a few tens of percent. By spring 1929 more than half of brokers’ loans came from nonbank lenders: large companies lent their spare cash as margin loans because the interest beat the return on their core business.

Then came 1929. By July 1932 the Dow Jones index had fallen 89%, and it did not regain its 1929 peak until November 1954.

After the Great Depression, America did something at odds with its self-image: the state itself became the largest lender in the country.

In February 1932 the Hoover administration created the Reconstruction Finance Corporation (RFC). From 1933 to 1945 it was run by the Texas banker Jesse Jones. According to the Economic History Association’s encyclopedia, the RFC authorized about $38.5 billion in loans and investments in total and disbursed $33.3 billion, of which $20.9 billion went to its wartime subsidiaries. (EH.net) Jones later wrote a memoir titled Fifty Billion Dollars. By today’s standards the RFC was a policy bank: it rescued banks and railroads, lent to businesses, and in wartime invested directly in building factories.

The housing line ran deeper. The Home Owners’ Loan Corporation (HOLC), founded in 1933, refinanced about $3.09 billion for 1.01 million households over three years. It replaced “balloon loans,” which came due in 3 to 5 years with the full principal payable at maturity, with 15-year amortizing loans at 5% a year. (FRASER) The Federal Housing Administration (FHA), founded in 1934, insured amortizing mortgages. In 1938 the RFC put up $11 million to create a subsidiary that bought FHA-insured mortgages, so that banks could sell their loans and free up money to lend again. That subsidiary was Fannie Mae. (Fannie Mae)

A common claim needs correcting along the way: the 30-year mortgage did not exist in 1934. The FHA opened 30-year terms for new homes only in 1948 and extended them to existing homes in 1954. (AEI)

The other side of this mortgage system was redlining. From 1935 to 1940 the HOLC drew “residential security maps” for 239 cities, grading neighborhoods from A (green) to D (red). (Mapping Inequality) Historians dispute whether the HOLC itself used the maps to deny loans, but the FHA had its own rating standards, and the result was the same: Black neighborhoods could hardly get federally backed mortgages. The 1944 GI Bill guaranteed nearly 2.4 million home loans; in New York and northern New Jersey, fewer than 100 of 67,000 such loans went to nonwhite buyers.

The Chinese essay says that “you can only make money where ordinary people are willing to save.” 1930s America supplied the other half: the state decided which ordinary people could borrow to buy homes and which could not. The generation that bought homes then enjoyed the largest round of asset appreciation in American history over the next forty years.

Financial regulation also took its shape in this stage. The Glass-Steagall Act of 1933 separated commercial and investment banking and created the Federal Deposit Insurance Corporation (FDIC); the Securities and Exchange Commission (SEC) was founded in 1934. The first SEC chairman was Joseph Kennedy, a man who had made his fortune speculating in stocks. The appointment is usually summed up as “set a thief to catch a thief.” (Library of Congress)

Stage four (1945–1979): Cold War orders and the birth of venture capital

America’s most important postwar “new story” was told in Silicon Valley. The first customer to pay for the story was the Pentagon.

In 1946 Harvard Business School professor Georges Doriot, MIT president Karl Compton and Boston Fed president Ralph Flanders founded American Research and Development Corporation (ARD), the first publicly traded venture capital firm. (MIT Museum) In 1957 ARD paid $70,000 for 70% of Digital Equipment Corporation (DEC); after DEC went public in 1968, that stake was worth $355 million.

The same year, eight young men who had walked out of Shockley’s lab went looking for money to start a company. Thirty or forty companies turned them down. Arthur Rock of Hayden, Stone finally brokered $1.5 million from Sherman Fairchild, and they founded Fairchild Semiconductor. Each founder put in $500 and received equity, which was rare at the time. (Traitorous eight)

Who were the early chip companies’ largest customers? In 1962 Texas Instruments won the integrated-circuit order for the Minuteman II missile guidance system; by 1965 the Minuteman missile was the largest single buyer of integrated circuits. (Computer History Museum) In the early 1960s government orders accounted for almost all integrated-circuit orders, and the unit price of a chip fell from $32 in 1961 to $1.25 in 1971. (FedTech)

The economist Mariana Mazzucato carries the story further in The Entrepreneurial State. She argues that every technology that makes the iPhone “smart” (the internet, GPS, the touchscreen, Siri) was developed with government money, and venture capital entered later: “the public sector socializes risks, while rewards are privatized.” (Mazzucato) Critics at the Cato Institute argue that she overstates the government’s role. In semiconductors, at least, the government’s position as the first large buyer is undisputed.

So the American version of “the court is willing to give you money because the court needs you to build houses” was, in the Cold War, “the Defense Department is willing to give you orders because the Defense Department needs missiles.” The difference is that the Defense Department bought products and did not directly decide who became the boss. People like Arthur Rock decided that.

This stage added two small parts. The Small Business Administration (SBA) was created in 1953. The Small Business Investment Act of 1958 let private investment companies borrow from the SBA at a 1-to-2 ratio and invest the money in small businesses (the SBIC program).

What changed the scale of venture capital was a 1979 rule interpretation by the Department of Labor.

The Employee Retirement Income Security Act of 1974 (ERISA) required pension managers to follow the “prudent man” rule, which was originally understood to forbid high-risk assets. In 1979 the Department of Labor clarified that prudence is judged across the whole portfolio, so a small allocation to venture capital does not break the rule. (NBER w6906)

The effect came quickly. In 1978 new venture funds raised $424 million in total, 15% of it from pensions; by 1986 they raised more than $4 billion, over half from pensions. (Gompers & Lerner, JEP 2001) The same group of venture lobbyists also got the capital gains tax cut from 49% to 28% in 1978. The Bayh-Dole Act of 1980 then let universities own patents from federally funded research.

I think this is the most important turning point in the whole American story, and the one closest to the core mechanism of the Chinese essay. Through one rule interpretation, ordinary workers’ pensions became fuel for venture capital. Ordinary people did not know where their money went; they only knew that a sum came out of each paycheck. The essay says “the money here is for you to use, so don’t be shy.” After 1979, Silicon Valley’s venture partners probably thought the same, though they never said it.

Stage five (1980–2007): ordinary people’s money enters in full

On August 15, 1971, Nixon closed the gold window, and foreign central banks could no longer exchange dollars for gold. (Federal Reserve History) From that day the dollar’s international value was no longer anchored to gold, and the quantity of dollars was set mainly by the Fed and by bank lending.

In 1979 Paul Volcker became Fed chairman. To beat inflation, the federal funds rate approached 20% in late 1980 and early 1981. (Federal Reserve History) High rates crippled half of American manufacturing, and they also set up the next forty years of falling rates. Rates fell from 20% all the way to zero, and over those forty years the valuations of all long-term assets were lifted. This falling curve is the premise for understanding everything that follows.

The Revenue Act of 1978 contained an obscure provision, 401(k). Around 1980 the benefits consultant Ted Benna combined employer matching with pretax deductions and invented the 401(k) plan as it exists today. At the end of 2025, total U.S. retirement assets were $49.1 trillion, including $10.1 trillion in 401(k) plans and $19.2 trillion in individual retirement accounts. (ICI)

This is the American “money house,” created by tax law: put money in and you pay less tax; once it is in, the money automatically buys stock funds. Chinese households put their money in banks; American households put theirs in the S&P 500.

Once the money was there, people found ways to lever it. The 1980s belonged to Michael Milken. He turned junk bonds into a market, letting small companies and acquirers raise money with high-yield debt. In 1987 he alone was paid $550 million. In 1988 KKR bought RJR Nabisco for $109 a share, about $31 billion including debt, financed mostly with junk bonds underwritten by Drexel Burnham Lambert. The record stood for 17 years. Drexel went bankrupt in 1990, and Milken pleaded guilty. (Britannica)

The logic of the leveraged buyout resembles the essay’s “take the money house’s money and contract out some project”: borrow against the target company’s own assets and cash flow, buy the company, cut staff, sell assets, raise dividends, and sell it or take it public a few years later. The buyer’s own money is a small share.

The savings and loan crisis came in the same period. From 1986 to 1995, 1,043 savings and loan institutions failed, with $519 billion in assets, and taxpayers paid about $124 billion. (Federal Reserve History) These institutions had been deregulated only a few years earlier, and they used the money of depositors protected by deposit insurance to make commercial real estate loans they did not understand.

The real estate line kept evolving. In 1960 Congress created the real estate investment trust (REIT) so that “small investors” could also own rental property. (Nareit) The 1981 tax law made real estate a tax shelter, and the Tax Reform Act of 1986 removed the “shelters within the shelter.”

Then the financial industry expanded itself. The Gramm-Leach-Bliley Act of 1999 repealed Glass-Steagall’s ban on combining banking businesses. (Federal Reserve History) The economists Greenwood and Scharfstein calculated that the U.S. financial sector’s share of GDP rose from 4.9% in 1980 to 7.9% in 2007. The growth came mainly from two sources: asset management fees, which rose with asset valuations (especially stocks), and household credit, mainly mortgages. (Greenwood & Scharfstein, JEP 2013)

This point matters. Financial industry revenue follows asset valuations, so the whole industry naturally wants valuations to rise. Another economist, Philippon, found that the unit cost of U.S. financial intermediation has barely changed in 130 years, at about 1.5% to 2% of intermediated assets per year; information technology did not make it cheaper. (Philippon, AER 2015)

When the dot-com bubble burst in 2000, the Nasdaq fell from 5,048 to 1,114, a 78% drop, and did not regain its peak until 2015. The Fed cut rates from 6.5% to 1%, and money moved into housing. From 2003 to 2005 home prices rose about 12.5% a year; subprime loans grew from 8% of new mortgages in 2003 to 22% in 2006; private-label mortgage securities hit a record $1.19 trillion in 2005, surpassing Fannie Mae and Freddie Mac for the first time. (FDIC)

Later research by the economists Mian and Sufi explains this thoroughly. After the 1980s, the savings of America’s richest 1% rose sharply, forming a “saving glut of the rich” comparable in size to the global saving glut. This money passed through the financial system and before 2008 was lent mainly to the middle class; after 2008, mainly to the federal government. (Mian, Straub & Sufi, NBER w26941)

Here the facts run opposite to the Chinese essay. The essay says ordinary people save and clever people borrow. In America before 2008, the rich saved and the middle class borrowed. The middle class used the borrowed money to buy homes; home prices rose for a while and then crashed. In House of Debt, Mian and Sufi calculated that from 2007 to 2010 the net worth of the poorest 20% of homeowners fell from $30,000 to almost zero. Debt makes the borrower take losses first.

Stage six (2008–2021): zero rates, when stories were worth the most

Lehman Brothers went bankrupt on September 15, 2008, with $639 billion in assets, the largest bankruptcy in U.S. history. Fannie Mae and Freddie Mac were taken into conservatorship and drew $191.4 billion from the Treasury. Congress passed the $700 billion Troubled Asset Relief Program (TARP); its final net cost was $31.1 billion, and its investments in banks even made money. (CBO)

The bigger change was at the Fed. Over three rounds of quantitative easing from 2008 to 2014, the Fed’s balance sheet grew from about $0.9 trillion to about $4.5 trillion. QE3 bought $85 billion a month with no cap. (St. Louis Fed)

This is the Cantillon effect in modern form. Cantillon, an eighteenth-century Irish banker, wrote that new money raises prices in the order in which it is spent: the first to get the money buy at old prices, and the last to get it, such as landlords living on fixed leases and workers on fixed wages, have to bear the new prices. (Cantillon, Essai, Higgs translation) Quantitative easing money entered the bond market first, pushed down long-term rates, and then pushed up stock, property and private-market valuations. Wages moved last.

Zero rates were also venture capital’s golden age. In November 2013 the investor Aileen Lee coined the term “unicorn” in TechCrunch. She had counted only 39 U.S. software companies founded after 2003 and valued at more than $1 billion, so rare that they needed a mythical animal for a name. (TechCrunch)

In May 2017 SoftBank’s Vision Fund held a first close of $93 billion, with Saudi Arabia’s Public Investment Fund committing up to $45 billion and Abu Dhabi’s Mubadala $15 billion. (PIF) This is the purest version of what the Chinese essay calls “telling a new story.” WeWork, a company that subleased office space, was valued at $47 billion in January 2019. In August it filed to go public, showing $1.8 billion in revenue and a $1.9 billion loss; in September it withdrew the filing; in November 2023 it filed for bankruptcy with $18.65 billion in liabilities. (CNBC)

Government lending in the same period produced the pair of examples most often compared. In 2010 the Department of Energy lent Tesla $465 million to produce the Model S in Fremont. In May 2013, after a stock and convertible bond offering, Tesla repaid the loan nine years early. (Tesla IR) The same program guaranteed $535 million for the solar company Solyndra. Solyndra went bankrupt in September 2011 and 1,100 people lost their jobs. The Energy Department’s inspector general later found that Solyndra had made misrepresentations in its application and that department staff had been under “tremendous pressure” to approve the loan. (POWER) Mazzucato’s comment: the government bore Solyndra’s losses without sharing in Tesla’s gains, and the problem lay in the contract design.

On the real estate side, the 2017 tax reform created “Opportunity Zones”: investors who put capital gains into designated low-income communities can defer or even eliminate capital gains tax. In 2018, 8,764 census tracts were designated. (GAO) This is almost the American tax-code version of the Chinese essay’s “cornfield turned CBD”: the state gives no money directly and tells wealthy people that they pay less tax if they put their money here.

In the 2020 pandemic, the Fed’s balance sheet grew from $4.3 trillion in mid-March to $7.2 trillion in June and peaked at $8.97 trillion in April 2022. (Federal Reserve) The CARES Act provided $2.2 trillion. Its Paycheck Protection Program (PPP) made 11.5 million loans totaling $793 billion, and by October 2022 about 96% of the amount had been forgiven. (Pandemic Oversight) The SBA inspector general estimated that PPP and the disaster loan program together may have lost more than $200 billion to fraud, about $64 billion of it in PPP.

PPP came closer than anything else in American history to “a policy money house announces a policy; if you meet the application criteria, the money is all yours.” The rules were simple, banks handed out the money, the government guaranteed it, and the loans were forgiven afterward.

Stage seven (2022–2026): rate hikes, state equity stakes and AI debt

From March 2022 to July 2023 the Fed raised rates 11 times, by a total of 525 basis points, the fastest round since 1982. (Richmond Fed) The zero-rate era ended, and stories became harder to sell.

Silicon Valley Bank fell first. On March 9, 2023, depositors withdrew $42 billion in one day, about a quarter of deposits, and requested another $100 billion the next day; the bank was closed that day. About 94% of its deposits exceeded the insurance limit. (Fed OIG) SVB’s depositors were mainly venture-backed startups; their money came from venture funds, and the venture funds’ money came from pensions and endowments. The whole chain was exposed within two days. The Fed promptly launched the Bank Term Funding Program, which accepted bonds as collateral at face value regardless of market price.

Meanwhile the United States began doing something it had hardly done in forty years: industrial policy. The CHIPS and Science Act of August 2022 appropriated $52.7 billion, $39 billion of it for manufacturing subsidies. (CRS) For the clean energy provisions of the Inflation Reduction Act, passed the same month, the Congressional Budget Office estimated about $391 billion at the time; because many tax credits had no cap, later estimates by Penn Wharton and Goldman Sachs exceeded $1 trillion. (Penn Wharton)

When the administration changed, the money changed direction at once. The One Big Beautiful Bill Act (OBBBA) of July 2025 repealed most of the Inflation Reduction Act’s clean electricity and electric vehicle credits, cutting about $500 billion over ten years. (PGPF) The same law made Opportunity Zones permanent; on November 5, 2026, the Treasury designated 6,576 new zones, effective January 2027. (CPA Practice Advisor) The Energy Department’s Loan Programs Office was renamed the Office of Energy Dominance Financing (DOE). In January 2026 it announced it would claw back more than $83 billion in loans and conditional commitments approved by the previous administration (Solar Builder) and turned to lending for the grid and nuclear power, such as $5.1 billion to the utility Alliant. (DOE)

The bigger change was direct state ownership. In July 2025 the Defense Department became the largest shareholder of the rare-earth company MP Materials with $400 million of convertible preferred stock, and it committed to a ten-year price floor of $110 per kilogram for NdPr (the Chinese market price was then under $60). (MP Materials) In August the government converted chip subsidies due to Intel into stock, taking a 9.9% stake at $20.47 a share, $8.9 billion in total. (CNBC) Lithium Americas, Trilogy Metals, a “golden share” in U.S. Steel and USA Rare Earth followed. By The National’s count, there have been 39 such deals since January 2025, totaling $27.7 billion. (The National)

How far this step goes becomes clear only against history. From 1957, when the RFC closed, to 2025, the U.S. government rarely took equity stakes in companies in peacetime outside of crisis rescues (such as forming Conrail in the 1970s, or rescuing banks and automakers in 2008). Now “the court taking a stake” has appeared in America too.

The last line is AI debt. In January 2025 SoftBank, OpenAI, Oracle and Abu Dhabi’s MGX announced “Stargate” at the White House, with $500 billion to be invested over four years. (CNN) In October 2025 Meta and the private credit firm Blue Owl formed a joint venture to build the Hyperion data center in Louisiana for about $27 billion. Blue Owl holds 80%. The financing is $27.3 billion of bonds at 6.581% maturing in 2049, kept off Meta’s balance sheet; Meta leases the campus and provides a 16-year guarantee of its value. (Meta 10-Q) Oracle issued $18 billion of bonds in September 2025 and was later sued by bondholders, who alleged it had concealed the scale of its later borrowing at the time of issue. (DCD)

On the venture side, the numbers no longer look much like venture capital. In the first three quarters of 2026, U.S. venture deal value reached $515.8 billion, 44% above the full-year record of 2021; 82.7% of it went to AI. (PitchBook-NVCA Venture Monitor Q3 2026)

Private credit began to show problems. In September 2025 the auto parts company First Brands and the subprime auto lender Tricolor failed one after the other, with combined liabilities of more than $11 billion. JPMorgan’s Jamie Dimon said, in substance, that when you see one cockroach there are usually more. (CNN) In the first quarter of 2026, redemption requests at the largest nontraded private credit funds averaged 12.1% of fund size, far above the usual 5% cap; Blue Owl limited redemptions in February, and Moody’s cut its outlook for the sector to negative in April. (CNBC)

The Fed itself is turning too. It stopped shrinking its balance sheet in December 2025; in early November 2026 the balance sheet stood at about $6.75 trillion. (FRED) In September 2026 the Fed raised rates 25 basis points to 3.75%–4.00% and held them in October, with two governors voting for further hikes. (Federal Reserve)

Summary: seven handoffs

Lining up the seven stages shows how “who is the gatekeeper” changed:

The pattern: the state stepped in directly and at scale mostly in crises or wars (the 1930s, the Cold War, 2008, 2020, and today’s U.S.-China competition). In normal times private financial institutions are the gatekeepers and the state only sets the rules. And since 1979, more and more of the money in private gatekeepers’ hands has come from ordinary people’s retirement accounts.

3. Horizontal analysis: five channels that hand out money in America in 2026

I divide the channels through which America today hands other people’s money to a few people into five, plus the public market as the final link, and then compare them with China, Japan and Germany.

Channel one: retirement funds and endowments, the biggest “purse”

The scale first. At the end of 2025, U.S. retirement assets were $49.1 trillion, 34% of household financial assets. (ICI) State and local defined-benefit pensions held about $6.4 trillion. In 2025 public pensions held $1.91 trillion in alternative assets, $831.9 billion of it in private capital; in 2009 that figure was only $424.5 billion. (Equable)

The most representative example is the California Public Employees’ Retirement System (CalPERS). Its September 2026 investment report shows total assets of $635.4 billion, with $123.6 billion in private equity, about 19.4%, 3.9 percentage points above the interim target. (CalPERS) Nearly a fifth of the retirement assets of California’s teachers, police officers and public employees is managed by private equity managers.

University endowments are another form. Yale’s endowment stood at $44.1 billion in fiscal 2025 and contributed $2.1 billion to the university’s budget that year, about a third of operating revenue. (Yale News) The core of the Yale model is a large allocation to private equity and venture capital. According to ai-CIO, Yale sold about $2.5 billion of private fund interests on the secondary market in 2025. (ai-CIO)

Insurance companies are the largest buyers of private credit. In 2024 life insurers held $849 billion of private credit, 14% of their assets, with the growth coming mainly from insurers acquired by private equity firms. (Chicago Fed) In other words, private equity firms buy insurance companies and then use policyholders’ premiums to buy private credit they originated themselves.

In August 2025 an executive order told the Department of Labor to clear the way for 401(k) investment in private assets, and in March 2026 the department published a proposed safe harbor rule. (Federal Register) What was done to pensions in 1979 is being done again to the 401(k).

Who owns this purse? In 2022, 91.1% of households with income over $150,000 had retirement accounts, against only 13.2% of households under $30,000. (CRS) So the American version of “ordinary people saving” is mostly the middle and upper-middle class saving. The truly poor do not even have a purse.

Channel two: venture capital, where stories get told

In the first three quarters of 2026, U.S. venture deal value was $515.8 billion across about 14,200 deals, and AI took 82.7%. (Venture Monitor) By one trade outlet’s count, OpenAI and Anthropic together raised more than $200 billion in the first half of 2026. (Startup Fortune)

These figures make venture capital look like a huge, open market. The fundraising side shows a very different picture. In the first three quarters of 2026, venture funds raised $108.5 billion. Funds of $500 million or more were only 6% of funds by count and took 78.1% of the money; established managers on their fourth fund or later took a record 88.2%; first-time funds raised only $4.9 billion. Andreessen Horowitz alone raised $23.8 billion in 2026, more than a fifth of all U.S. venture fundraising. In the first half, three firms took 48.1% of the money raised.

This is what the Chinese essay means by “the staff inside the money house want to give it to you too; they all have quotas to lend the money out.” Pensions and endowments must allocate to venture capital, but they cannot pick companies, only funds; and the safest choice when picking funds is the biggest, best-known few. So the money grows ever more concentrated.

Venture returns to investors are also weaker than the stories suggest. In 2012 the Kauffman Foundation reviewed the 100 venture funds it had invested in: only 20 beat the public market by more than 3 percentage points a year, and the overall net return was just 1.31 times. The report’s title was “We Have Met the Enemy… and He Is Us.” (Kauffman) According to the same PitchBook-NVCA report, cumulative net cash flow from venture funds to their investors since 2022 is negative $202 billion. The 992 active unicorns are valued at $5.7 trillion combined, and only 18 went public in the third quarter of 2026.

From the founder’s side, the typical entry point is Y Combinator: $125,000 for 7% of the company, plus $375,000 on an uncapped SAFE. The Winter 2026 batch had 196 companies, an acceptance rate of about 1%. (YC)

The counterfact that most needs stating: venture capital serves very few founders. A Kauffman Foundation report based on Census data shows that at least 83% of founders used neither bank loans nor venture capital at startup; nearly 65% relied on personal and family savings, about 10% used credit cards, and only 0.5% used venture capital. (Kauffman 2019) Using Census data from 1981 to 2005, Puri and Zarutskie found that venture-backed firms were about 0.11% of new firms but 4% to 7% of U.S. employment. (Puri & Zarutskie)

So venture capital is a narrow, high passage: very few get in, and those who do get very large sums and very high valuation multiples.

Channel three: private equity and private credit, the leverage machine

Global private equity assets under management are about $10.5 trillion (June 2025); private credit exceeds $2 trillion, or about $3.2 trillion including semi-liquid products. (PIMCO)

The chain in this channel: pensions and insurers → private equity funds → buyouts financed by borrowing against the target company → loans from private credit funds → whose money in turn comes from pensions and insurers. Matt Levine puts it simply: private credit managers lend out their clients’ money, and those clients are insurance companies, pension funds and endowments. (Bloomberg)

Risk accumulates along this chain too. In 2024 a record 110 U.S. companies backed by private equity or venture capital went bankrupt. (S&P Global) The Fed’s May 2026 Financial Stability Report says about half of surveyed market contacts listed private credit as a top risk. (Federal Reserve) According to press reports, the Financial Stability Board judges that private credit, at its current size and scope, has not yet been tested by a severe downturn. (Angel Investors Network)

This channel most resembles the Chinese essay’s “you contract out some project, you earn only one small link in the whole chain, and you can already call yourself retired early.” Private fund managers charge a 2% management fee and 20% of profits, using other people’s money and debt borrowed by the companies they buy.

Channel four: mortgages and real estate tax law, America’s “land finance”

American local governments do not sell land, but America has a whole body of tax law that makes land and buildings the best capitalization tools.

Mortgages first. Fannie Mae’s guaranty book stood at $4.14 trillion at the end of 2025. (Fannie Mae 10-K) Federally backed agencies (Fannie Mae, Freddie Mac and Ginnie Mae) support most American mortgages, and Fannie and Freddie remain in government conservatorship today. The mortgage interest deduction cost about $45.5 billion in 2025. (JCT) The 30-year fixed-rate mortgage is itself a financial product that exists at scale only in America; without government-backed Fannie and Freddie, private banks would have a hard time carrying 30 years of interest rate risk.

Then the tax machine. The Joint Committee on Taxation’s estimates of tax expenditures for fiscal 2025 to 2029 (same source):

Put these together and you get the standard playbook of the American property owner: buy a building, offset rental income with accelerated depreciation, never sell when it appreciates, trade up to a bigger building through a 1031 exchange to keep deferring, borrow against it when cash is needed (loans are not income), and finally pass it to the children, whose cost basis resets to market value at death, so a lifetime of appreciation pays no tax at all. Every step of this playbook is legal, and every step requires owning a building first.

Local governments have their own version, called tax increment financing (TIF): draw a district and use the future increase in its property tax to finance development. Chicago has 108 TIF districts, which collected $1.62 billion in tax year 2025, 17.9% of the city’s property tax. (Cook County Clerk) This is the closest American mechanism to China’s land finance that I could find: both use future land value increases to finance construction today. The difference is that TIF uses taxes, and China uses land sale revenue.

On Opportunity Zones, the Opportunity Zone funds tracked by Novogradac had raised $42.76 billion of equity by the end of 2025. Novogradac believes it tracks only about a third, so the actual total may exceed $100 billion. (Novogradac)

The economist Matthew Rognlie made an important finding: the long-run rise in the capital share of income in developed countries, net of depreciation, comes almost entirely from housing. (Brookings) Piketty said the return on capital outruns economic growth; Rognlie added that what outruns it is mainly housing. The Chinese essay says “the Earth God has nearly caught up with the God of Wealth,” and that holds in America as well.

Channel five: direct government money, growing in recent years

This channel has always existed in America, small in scale and poor in reputation. It has grown since 2022.

• Small Business Administration: in fiscal 2025 it made a record 84,400 7(a) and 504 loans totaling $44.8 billion. (SBA) The SBIC program portfolio is $53 billion.
• Energy Department loans: loan authority exceeds $289 billion, and the direction has shifted from wind and solar to nuclear, the grid and fossil energy.
• Chip subsidies: as of July 2026, the Commerce Department had awarded $31.2 billion in direct subsidies for 49 projects at 24 companies, of which $13.1 billion had been disbursed. (GAO)
• Defense contracts: Palantir has a ten-year Army enterprise agreement with a $10 billion ceiling (August 2025), and Anduril has an Army enterprise agreement with a $20 billion ceiling (March 2026). Anduril’s president himself said the agreement comes with no money attached and is only a contracting vehicle. (Washington Technology)
• State equity stakes: Intel, MP Materials, Lithium Americas and others, 39 deals totaling about $27.7 billion.

This channel is closest to the Chinese essay’s “court.” It has one American feature, though: the money follows elections. More than $83 billion in energy loans and commitments approved by one administration were announced for clawback within a year of the next administration taking office; most of the Inflation Reduction Act’s clean energy credits were repealed three years after passage. In America, the biggest risk of doing “policy business” is the change of administration every four years.

The final link: the public market, the valuation amplifier

The five channels above hand money to a few people; the public market prices the assets in their hands by their future cash flows.

At the end of 2025 total U.S. stock market capitalization was $68.9 trillion, 43.7% of the world total and 4.4 times China’s. (SIFMA) In mid-October 2026 the S&P 500’s forward P/E was about 19. (FactSet) That means a listed company earning $100 million a year gets a market valuation of about $1.9 billion.

This is the difference between running a stall and building a company. A vendor who earns 50,000 a year has 50,000. A company that earns 50,000 a year, if the market believes it will keep growing, is worth 1 million. By selling part of the company, the founder can turn twenty years of future profits into cash today.

Who owns these stocks? In the second quarter of 2026, America’s richest 1% held 50.9% of stocks and mutual funds, up from 42.8% in 1989. (FRED / Fed DFA) Add the 37.2% held by the 90th to 99th percentiles, and the richest 10% hold about 88%.

Comparison group: China, Japan, Germany

Start with the structurally cleanest set of figures, from the Bank of Japan’s March 2026 flow-of-funds comparison: (BOJ)

American companies finance themselves mainly with equity, and American households mainly hold stocks; Japanese companies rely on bank loans, and Japanese households mainly hold deposits. This determines who the gatekeeper is: in Japan, the banks; in America, the capital markets and the people who manage them.

China. The International Monetary Fund estimates local government financing vehicle debt at about 71.4 trillion yuan at the end of 2025, about 51% of GDP; including special construction funds and government guidance funds, augmented government debt is about 127% of GDP. (IMF) Land sale revenue peaked at about 8.7 trillion yuan in 2021, fell to 4.15 trillion in 2025, and fell another 28.6% year on year in the first eight months of 2026. At the end of 2024 there were 2,178 government guidance funds with a target size of 12.84 trillion yuan and 7.7 trillion actually subscribed. (ChinaVenture Research) China Development Bank had total assets of 18.62 trillion yuan at the end of 2024 and only 0.79 trillion in customer deposits; it funds itself mainly by issuing bonds. (HKMA)

Scholarly criticism of the Chinese model has two main lines. Song, Storesletten and Zilibotti, in “Growing Like China,” argue that private firms are more efficient but can invest only their own savings, and that state firms survive because they have “better access to credit.” (AER 2011) Michael Pettis argues that financial repression transfers income from households to companies and the government, holding household consumption’s share of GDP down. Nicholas Lardy adds an important point: from 2010 to 2012 private firms received 52% of new loans and state firms only 32%, and China’s growth was driven mainly by private firms; after 2012, state banks steered more credit to state firms and crowded out private investment. In 2015 the return on assets of state industrial firms was only 1.9%, against 10.6% for private firms. (Cato Journal review)

Japan. The main bank system: a company’s largest long-term lender also holds its shares, monitors management and leads the rescue when trouble comes. In April 2026 the three megabanks announced they would sell more than 1 trillion yen of cross-shareholdings within three years; the system is still slowly being unwound. (Nikkei Asia)

Germany. The policy bank KfW made €98 billion in promotional loans in 2025, but it almost never lends directly. It on-lends through each company’s own bank (the Hausbank principle): the house bank does the credit review, KfW provides the refinancing and takes at most half the risk. (KfW)

The four countries side by side:

Counterevidence: what this logic cannot explain

Looking only at the evidence above, it is easy to conclude that American wealth, too, comes from “getting the money first.” Three sets of evidence do not support that conclusion.

First, top earners in America mainly live on their own labor. Smith, Yagan, Zidar and Zwick matched 11 million firms with their owners and found that America’s top earners are mainly working-age owners of mid-sized, skill-intensive, closely held firms, such as law firm partners, dental chain owners and car dealers. When an owner retires or dies, the firm’s profit falls by three quarters. Their conclusion: “the typical top earner derives most of her income from human capital, not financial capital.” (QJE 2019)

Second, America’s top fortunes rely less and less on inheritance. Kaplan and Rauh found that today’s Forbes 400 are less likely than those of a few decades ago to have inherited wealth or come from wealthy families; their fortunes come mainly from scalable industries: technology, finance and mass retail. (JEP 2013)

Third, a company can become large without outside financing. The email marketing company Mailchimp never took venture capital and was bought by Intuit in 2021 for about $12 billion. (Intuit 8-K) Still, Mailchimp’s founders got rich through the valuation multiple in the purchase price; their accumulated profits were the smaller part. They did not get around the capitalization link either.

There is also a less comfortable set of data. Chetty’s team found that about 90% of Americans born in 1940 earned more than their parents, against only 50% of those born in the 1980s; the decline came mainly from a more unequal distribution of growth, and slower growth mattered less. (Science 2017) Another study found that children from the richest 1% of families are ten times as likely to become inventors as children from families below median income; even among children with math scores in the top 5%, rich children are more than twice as likely to become inventors. The researchers call the children who lacked the chance “lost Einsteins.” (Opportunity Insights)

So in America, “getting the money first” mostly determines how far a good business can be amplified; it determines less about who can build a good business. But “who gets access to the chance to build a good business” is itself highly correlated with family background.

4. Where the axes meet: how much of the Chinese essay holds in America

How history shaped today’s position

Laying the vertical axis over the horizontal one, I see three structures set by history.

First, America’s purse is in the stock market, and two decisions in 1978 and 1979 are the key reason. The 401(k) provision and the Department of Labor’s interpretation of the “prudent man” rule channeled ordinary people’s retirement savings into stocks, venture capital and private equity. Japan and continental Europe did not take this path (for more reasons than these alone). Today deposits make up 47% of Japanese households’ financial assets, 31% in the euro area and only 11% in America. America’s “money house” is the capital market, and its gatekeepers are fund managers. The Chinese essay’s “the staff inside the money house want to give it to you too; they all have quotas” corresponds in America to CalPERS’s private equity allocation targets and venture funds’ fundraising cycles.

Second, the American state steps in during crises every time, and it leaves slowly. The RFC ran from 1932 to 1957; Fannie Mae was born in 1938, taken over in 2008, and still had not been released by 2026; the Fed’s balance sheet went from $0.9 trillion in 2008 to $6.75 trillion today and never went back. With each crisis the scope of the state’s backstop widens by another ring. And the promise of a backstop is itself a form of capitalization: if someone covers the worst case, assets are worth a higher price.

Third, American property capitalization runs on tax law and does not depend on land sales. Chinese local governments’ land sale revenue has dropped from 8.7 trillion to 4.15 trillion yuan, and the whole model is shrinking. In America, none of the 1031 exchange, accelerated depreciation, the step-up in basis at death or Opportunity Zones has been repealed, and Opportunity Zones were just made permanent. America’s version of land finance is spread across thousands of local governments and millions of property owners and does not depend on a single stream of land sale revenue, so it carries no risk of collapsing all at once and no chance of being dismantled all at once. One more thing follows. In China’s land finance, the benefits land first with local governments, and when land prices fall, the government takes the hit first. In American property tax law, the benefits are distributed by the size of the assets held; people who already own buildings get the most, and these provisions have survived every change of party. The Chinese model shrinks with land sale revenue; the American model swells with asset prices, and the latter favors people who already own assets.

The historical roots of strengths and weaknesses

The strengths of the American machine: many gatekeepers, low cost of failure, easy to start over. When a venture fund picks wrong, the fund’s investors lose; when a company fails, bankruptcy law lets the founder start again; when a state-backed project is a bad pick, there are congressional hearings and inspector general reports. Solyndra’s failure was debated over and over for ten years, and that debate is itself a form of error correction.

The weaknesses have the same root as the strengths: once money is handed to private gatekeepers, the gatekeepers cluster toward safety. In the first half of 2026 three venture firms took nearly half of the money raised; in the first three quarters 88% went to established managers, and first-time funds got only $4.9 billion. Pensions do not dare invest in small funds, so there are few small funds, and new founders have fewer doors to knock on. This is the other side of the 1979 “prudent man” interpretation: prudence means following everyone else.

There is also a historical burden. From 1981 to 2021 interest rates fell from 20% to zero. For forty years the valuations of all long-term assets were lifted, and everyone who “got the money first” had the wind at their back. After 2022 the wind changed, and in 2026 the Fed is still raising rates. Redemption pressure in private credit, the failure of First Brands and the bondholder suit against Oracle are different forms of the same thing: assets priced in the zero-rate era must be repriced at 4% rates.

Three scenarios

The most likely scenario: two machines running in parallel. Private channels (venture capital, private equity, capital markets) keep handing out most of the money, ever more concentrated and flowing ever more to AI and data centers. State channels (equity stakes, defense contracts, energy loans) expand in industries tied to “national security,” but change direction every four years. For founders, there are two ways to get money: tell a story the market believes, or tell a story Washington believes. The first is judged on growth, the second on which side you are on. Two sets of figures support this judgment: 39 state equity deals since January 2025, after decades of almost none in peacetime; and 82.7% of venture money going to AI in the first three quarters of 2026, the highest share on record.

The most dangerous scenario: private credit and AI debt fail together. AI data centers are built largely with off-balance-sheet financing and private credit (Meta’s Hyperion is the typical case). Private credit money comes from insurers and pensions, and the insurers are in turn owned by private equity firms. If AI revenue does not keep up with the investment, data center leases default, private credit funds take losses, and the pressure travels back through insurers and pensions to ordinary people’s retirement accounts. The Financial Stability Board says private credit has not yet been tested by a severe downturn; this scenario would be that test. When it comes, the state will almost certainly backstop once more, and the scope of the backstop will widen by another ring.

The most optimistic scenario: the threshold for capitalization falls. With 401(k)s opened to private assets, if transparency and fee regulation are done well, ordinary people could share in private-market returns that only institutions could reach before. If AI really lowers the cost of starting a company (this is an assumption; I found no reliable data), more companies could grow large without outside financing, as Mailchimp did. If this scenario comes true, the advantage of “getting the money first” shrinks and the value of “making a product people pay for” grows. I have the least confidence in this scenario, because since 1979, every time “ordinary people’s money” was let into alternative assets, the first to benefit were the people who managed that money.

Back to the Chinese essay

Where the essay is right:

1. The key to business is getting other people’s money. Brandeis said the same in 1914, and American venture capitalists, private equity firms and property owners all practice it.
2. Money flows out from the center, and the closer you are to the center, the earlier you benefit. The Cantillon effect, quantitative easing and PPP are all examples.
3. Capitalizing land and assets is the core of amplifying wealth. Rognlie’s research, American property tax law and the S&P 500’s P/E ratio all support this.
4. Earn one small link in the chain and you can retire early. The private fund manager’s 2% management fee plus 20% of profits is exactly this.

Where it is wrong or imprecise:

1. In bank credit, the order is reversed. The essay says ordinary people save first and the money house then lends the money to you. A 2014 Bank of England article states it clearly: “Whenever a bank makes a loan, it simultaneously creates a matching deposit in the borrower’s bank account, thereby creating new money.” And: “Saving does not by itself increase the deposits or ‘funds available’ for banks to lend.” (Bank of England) Schumpeter held a similar view in 1911: an entrepreneur must first become a debtor before becoming an entrepreneur, and the banker is “the ephor of the exchange economy.” America’s other channel, though, where pensions and 401(k)s flow into venture capital and private equity, really is save first and invest later, matching the order the essay describes. So a more accurate statement is: bank loans depend on persuading those with the power to create credit, and equity investment depends on persuading those who manage other people’s savings.
2. The money does not have to be paid back to the court. In America, loans are repaid to banks, and equity investors exit through an IPO or acquisition. What the government gets back is mainly taxes.
3. In America, money flows both ways. The research of Mian, Straub and Sufi shows that household debt before 2008 was mainly the savings of the rich lent to the middle class. The retirement channel, as the essay says, hands wage earners’ savings to fund managers to invest. The essay saw only the second direction.
4. “The most valuable thing for the rich is controlling a large crowd of obedient poor people”: the evidence does not support this. American data show that top earners rely mainly on their own human capital and that top fortunes rely less and less on inheritance. These studies do not directly refute the “control” claim, but they show that American wealth mainly comes from sources other than other people’s obedience. I think a more accurate statement is that the two most valuable things for the rich are leverage (the ability to borrow other people’s money) and capitalization (the market’s willingness to price you at a multiple of future cash flow).

The question at the start of the essay: why do some people build a company that lasts a generation while others can only run a stall?

My answer: the stall owner sells today’s goods and earns today’s markup; the company builder sells future cash flow and earns the multiple the market gives. The door between them is, in China, “the court’s money house,” and in America, a group of fund managers holding other people’s pensions.

To get through that door in America, the hardest credential is still real customers who keep paying. Stories can be told and multiples can be negotiated, but the $12 billion Mailchimp sold for was calculated from the subscription fees it collected each year.

Sources

All sources below were accessed on November 9, 2026. English quotations in the text use the original wording where it could be confirmed; elsewhere the text paraphrases without quote marks.

Theory

• Cantillon, Essai sur la nature du commerce en général, Higgs English translation: https://oll-resources.s3.us-east-2.amazonaws.com/oll3/store/titles/285/Cantillon_0039.html
• McLeay, Radia, Thomas, “Money creation in the modern economy,” Bank of England, 2014: https://www.bankofengland.co.uk/quarterly-bulletin/2014/q1/money-creation-in-the-modern-economy
• Brandeis, Other People’s Money, 1914: https://www.gutenberg.org/cache/epub/57819/pg57819-images.html
• Minsky, “The Financial Instability Hypothesis,” 1992: https://www.levyinstitute.org/pubs/wp74.pdf
• Schumpeter’s “ephor of the exchange economy,” quoted in Storm (2018): https://onlinelibrary.wiley.com/doi/full/10.1111/dech.12385
• Mazzucato, The Entrepreneurial State: https://marianamazzucato.com/books/the-entrepreneurial-state/
• Rognlie, Brookings 2015: https://www.brookings.edu/articles/deciphering-the-fall-and-rise-in-the-net-capital-share/

U.S. empirical research

• Mian, Straub, Sufi, “The Saving Glut of the Rich,” NBER w26941: https://www.nber.org/papers/w26941
• Greenwood & Scharfstein, “The Growth of Finance,” JEP 2013: https://www.aeaweb.org/articles?id=10.1257/jep.27.2.3
• Philippon, AER 2015: https://www.aeaweb.org/articles?id=10.1257/aer.20120578
• Smith, Yagan, Zidar, Zwick, “Capitalists in the Twenty-First Century,” QJE 2019: https://www.nber.org/papers/w25442
• Kaplan & Rauh, JEP 2013: https://www.aeaweb.org/articles?id=10.1257/jep.27.3.35
• Gompers & Lerner, JEP 2001: https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.15.2.145
• Kauffman Foundation, Access to Capital 2019: https://www.kauffman.org/wp-content/uploads/2020/06/Access-To-Capital_2019.pdf
• Kauffman Foundation, “We Have Met the Enemy and He Is Us,” 2012: https://www.kauffman.org/reports/we-have-met-the-enemy-and-he-is-us/
• Puri & Zarutskie, Journal of Finance 2012: https://www.nber.org/system/files/working_papers/w14250/w14250.pdf
• Chetty et al., “The Fading American Dream,” Science 2017: https://opportunityinsights.org/paper/the-fading-american-dream/
• Bell et al., “Who Becomes an Inventor in America?” QJE 2019: https://opportunityinsights.org/paper/losteinsteins/

History

• Federal Reserve History (First Bank of the United States, Panic of 1907, the Fed’s formative years, Glass-Steagall Act, RFC, 1981–82 recession, savings and loan crisis, GLBA, Smithsonian Agreement): https://www.federalreservehistory.org/
• Liberty Street Economics, the 1792 crisis: https://libertystreeteconomics.newyorkfed.org/2014/05/crisis-chronicles-central-bank-crisis-management-during-wall-streets-first-crash-1792
• National Archives, Pacific Railway Act, Homestead Act, Interstate Highway Act: https://www.archives.gov/milestone-documents/pacific-railway-act
• Library of Congress, Crédit Mobilier, Securities Exchange Act of 1934: https://guides.loc.gov/this-month-in-business-history/september/Credit-Mobilier-Scandal
• EH.net, RFC: https://eh.net/encyclopedia/reconstruction-finance-corporation/
• FRASER, HOLC report: https://fraser.stlouisfed.org/files/docs/publications/holc/pages/61351_1950-1954.pdf
• AEI, origin of the 30-year mortgage: https://www.aei.org/housing-center/housing-finance/housing-finance-fact-or-fiction-fha-pioneered-the-30-year-fixed-rate-mortgage-during-the-great-depression/
• Mapping Inequality: https://dsl.richmond.edu/panorama/redlining/howandwhy
• MIT Museum, ARD: http://museum.mit.edu/150/78
• Computer History Museum: https://www.computerhistory.org/siliconengine/aerospace-systems-are-first-the-applications-for-ics-in-computers/
• CBO, TARP: https://www.cbo.gov/system/files/2024-04/59919-TARP.pdf
• Tesla, DOE loan repayment: https://ir.tesla.com/press-release/tesla-repays-department-energy-loan-nine-years-early
• TechCrunch, Welcome to the Unicorn Club: https://techcrunch.com/2013/11/02/welcome-to-the-unicorn-club/
• CNBC, WeWork bankruptcy: https://www.cnbc.com/2023/11/07/wework-files-for-bankruptcy.html
• Pandemic Oversight, PPP: https://www.pandemicoversight.gov/media/file/ppp-loan-forgiveness-fact-sheet-october-2022-updatepdf
• Fed OIG, Silicon Valley Bank: https://oig.federalreserve.gov/reports/board-material-loss-review-silicon-valley-bank-sep2023.pdf

Current (2025–2026)

• ICI, retirement assets: https://www.prnewswire.com/news-releases/ici-data-shows-retirement-assets-total-49-1-trillion-in-fourth-quarter-2025–302726690.html
• Equable, State of Pensions 2025: https://equable.org/state-of-pensions-2025/
• CalPERS investment report: https://calpers.ca.gov/documents/202609-invest-agenda-item06b-01-a/download?inline=
• Yale endowment: https://news.yale.edu/2025/10/24/yale-reports-investment-return-fiscal-2025
• Chicago Fed, life insurers and private credit: https://www.chicagofed.org/publications/working-papers/2025/2025-09
• PitchBook-NVCA Venture Monitor Q3 2026: https://nvca.org/wp-content/uploads/2026/10/Q3-2026-PitchBook-NVCA-Venture-Monitor.pdf
• Y Combinator: https://www.ycombinator.com/blog/ycs-500-000-standard-deal
• PIMCO, size of private credit: https://www.pimco.com/us/en/insights/how-large-is-private-credits-total-addressable-market-really
• Federal Reserve Financial Stability Report, May 2026: https://www.federalreserve.gov/publications/2026-may-financial-stability-report-leverage.htm
• JCT tax expenditures, JCX-45-25: https://www.jct.gov/getattachment/8c830c45-1680-4f7e-a649-2a0106f6b6e3/x-45-25.pdf
• Fannie Mae 10-K 2025: https://www.sec.gov/Archives/edgar/data/310522/000031052226000015/fnm-20251231.htm
• Cook County Clerk, Chicago TIF: https://www.cookcountyclerkil.gov/sites/default/files/pdfs/2025-chicago-tif-overview.pdf
• Novogradac, Opportunity Zones: https://www.novoco.com/notes-from-novogradac/more-than-200000-homes-financed-by-qofs-tracked-by-novogradac
• SBA fiscal 2025: https://www.sba.gov/article/2025/09/30/trump-sba-delivers-record-capital-small-businesses-fy25
• GAO, CHIPS: https://www.gao.gov/assets/gao-26-109121.pdf
• CNBC, Intel stake: https://www.cnbc.com/2025/08/22/intel-goverment-equity-stake.html
• MP Materials and the Department of Defense: https://mpmaterials.com/news/mp-materials-announces-transformational-public-private-partnership-with-the-department-of-defense-to-accelerate-u-s-rare-earth-magnet-independence/
• The National, government equity portfolio: https://www.thenationalnews.com/news/us/2026/09/30/whats-inside-the-trump-administrations-27bn-portfolio/
• Meta 10-Q 2025 Q3: https://www.sec.gov/Archives/edgar/data/1326801/000162828025047240/meta-20250930.htm
• PGPF, OBBBA energy provisions: https://www.pgpf.org/article/energy-tax-policy-under-the-obbba/
• FRED, Federal Reserve total assets: https://fred.stlouisfed.org/series/WALCL
• FRED, stock share held by the top 1%: https://fred.stlouisfed.org/data/WFRBST01122
• SIFMA Fact Book 2026: https://www.sifma.org/news/blog/2026-capital-markets-fact-book-key-findings
• FactSet: https://insight.factset.com/sp-500-earnings-season-update-october-16-2026
• Federal Reserve FOMC statement, October 2026: https://www.federalreserve.gov/monetarypolicy/files/monetary20261028a1.pdf

International comparison

• Bank of Japan, international comparison of flow of funds: https://www.boj.or.jp/en/statistics/sj/sjhiq.pdf
• IMF, China Article IV consultation 2026: https://www.elibrary.imf.org/view/journals/002/2026/044/article-A001-en.xml
• ChinaVenture Research, government guidance funds: https://research.pedaily.cn/202502/546594.shtml
• China Development Bank financial disclosure (HKMA): https://vpr.hkma.gov.hk/statics/assets/doc/100296/fd_fin/fd_fin_1225_eng.pdf
• Song, Storesletten, Zilibotti, “Growing Like China,” AER 2011: https://www.aeaweb.org/articles?id=10.1257/aer.101.1.196
• Lardy, The State Strikes Back, Cato Journal review: https://www.cato.org/cato-journal/fall-2019/state-strikes-back-end-economic-reform-china-nicholas-r-lardy
• KfW 2025 annual results press release: https://www.kfw.de/%C3%9Cber-die-KfW/Newsroom/Aktuelles/Pressemitteilungen-Details_886400.html
• Nikkei Asia, Japan’s three megabanks sell cross-shareholdings: https://asia.nikkei.com/business/finance/japan-s-top-banks-aim-to-unwind-6bn-in-cross-held-shares-in-3-years

Unverified claims: the share of First Bank of the United States stock that could be paid in government bonds; some 2026 figures come from secondary reports and may be revised later.

Methodology

This post uses horizontal-vertical analysis, proposed by Khazix (数字生命卡兹克): trace the full history of the subject on the vertical axis, compare it with peers at the present moment on the horizontal axis, and cross the two axes to reach conclusions.