The Working Ledgers: How Seven Research Institutions In...

An earlier run of this series read the deathless ledgers of the foundations and estates, the money that gives itself away long after its owner is gone. This one turns the same method on a different kind of institution: the operating research nonprofits, the ones that do not give money away but spend it, on the nation's hardest science and engineering. Seven of them filed public tax returns that landed on the same desk this week, and read side by side they answer four questions cleanly. How does each one invest? How does it stay solvent? How does it keep the work going? And for how long can it keep going? The filings invent nothing. Every number below is read straight from a Form 990, free to anyone at ProPublica's Nonprofit Explorer, and the whole point is that the answers could not be more different from one institution to the next.

The seven are the MITRE Corporation, the Aerospace Corporation, the Charles Stark Draper Laboratory, Southwest Research Institute, the Broad Institute, the Howard Hughes Medical Institute, and the Marine Toys for Tots Foundation. Six do research. One moves toys. Together they map the entire spectrum of how an American nonprofit can pay for itself, and the map has a clear left end and a clear right end.

The spectrum, in one line each

At the left end sit the pass-throughs, the institutions that hold almost nothing and live entirely on what the government pays them this year. MITRE took in $2.48 billion, about 99 percent of it federal, and holds essentially zero investments. The Aerospace Corporation took in $1.37 billion on the same model. Both spend roughly eighty cents of every dollar on their own people and keep a cushion measured in a few months. They do not invest, because a cost-reimbursement contractor has nothing to invest. Their solvency is a contract, and their future is an appropriations bill.

In the middle are the institutions that live on contracts but chose to keep a reserve. Southwest Research Institute funds itself on research contracts from hundreds of government and industry clients, and parks a conservative $270 million cushion in cash and short-term instruments. Draper Laboratory, the lab that built the Apollo Guidance Computer, does the same kind of contract work but keeps its $270 million reserve in a diversified investment pool that looks like a small endowment, global equities and private capital and hedge funds. Same revenue model as the pure pass-throughs, a very different balance sheet, because these two banked something for a rainy day and one of them put it to work in the markets.

At the right end are the endowments, where the portfolio is not a cushion but the engine. The Broad Institute stands on three legs, federal grants and philanthropy and about $1.5 billion in invested endowment, and spends roughly $750 million a year on biomedical science. And the Howard Hughes Medical Institute is the pure case: $27.6 billion in net assets, of which nearly two billion dollars of investment income in a single year funds the whole enterprise, an institution that employs scientists directly and can do so, by design, forever. Off to one side sits the outlier, Toys for Tots, a holiday charity that moves half a billion dollars in donated toys through a machine of Marine volunteers and keeps a $136 million rainy-day fund it is willing to spend into.

They all drink from the same well

Here is the thread the person who asked for this series wanted pulled, and it runs through every one of these ledgers. The institutions that invest, invest in the same equity and security markets that almost everyone's future depends on. This is not a metaphor. When HHMI's in-house investment office buys public equities, it is buying the same stocks that sit in a retiree's 401(k). When it fights for a slot in a top venture fund, it is bidding against Yale's endowment, against the California teachers' pension, against Norway's sovereign wealth fund and Singapore's, for the same scarce seats in the same deals. When the Broad Institute allocates three quarters of its endowment to alternatives, it seats Seth Klarman and Eric Schmidt on the board that oversees it, so the line between the institution and the markets it invests in nearly disappears. When Draper holds global equities and private capital, it competes for returns with every pension fund on earth. And when Toys for Tots invests its reserve, it does so in a Pacer ETF and a PIMCO income fund and an American Funds balanced fund, the exact tickers a schoolteacher holds in a retirement account.

There is one well of market returns, and everyone lowers a bucket into it: the individual saver, the corporation, the city and state pension, the sovereign wealth fund, the federal government's own trust funds, and these research institutions. In a good year the well is generous and lifts all the buckets together, which is why HHMI could book nearly two billion dollars of investment income in the year of its filing. In a bad year the level drops for everyone at once, the endowment and the pension and the retiree feeling the same drawdown in the same quarter. The institutions on the right end of this spectrum have effectively converted themselves into investors who happen to run laboratories, and they rise and fall with the market exactly as you do. The institutions on the left end, MITRE and Aerospace, have opted out of that well almost entirely, and their fate rides not on the market but on the federal budget. Neither position is safe. They are just exposed to different weather.

Aerospace shows the subtlety at the seam. It holds no endowment, so it looks like it has no market exposure at all. But it has promised its workforce a pension, and the assets behind that promise, about $410 million of them, are invested in the same markets, while the obligation itself is the largest liability on its books. So even a pure contract shop can be dragged into the well through the one door it cannot close, the retirement plan, which is the same door that has strained corporate balance sheets and state budgets for a generation.

The other side of the ledger: what they owe and what wears out

An institution is not just what it holds. It is also what it owes and what is quietly falling apart, and the four questions look different once you turn the balance sheet over.

The liabilities tell you how each one is financed. The endowed institutions borrow on purpose: HHMI carries $935 million in tax-exempt bonds even while sitting on $27 billion, because when your portfolio compounds faster than tax-exempt debt costs, it is cheaper to borrow for your buildings than to sell investments that are busy earning more. The Broad carries $221 million in the same kind of bonds for the same reason. The contract shops borrow differently or barely at all: MITRE carries about $64 million in operating lease obligations for the space it rents and paid its other debt down during the year, while Aerospace carries a mortgage on its campus and, far heavier, that pension obligation. Toys for Tots owes almost nothing at all, $4 million in unpaid bills against a half-billion-dollar operation. Read the liability side and you can tell instantly which of these institutions is an investor using leverage and which is a channel that never accumulates enough to bother borrowing against.

Depreciation is the liability nobody sends an invoice for. Every one of these institutions runs on physical capital that is always going obsolete, and the filings measure the erosion precisely. MITRE has depreciated more than half of its $1.27 billion capital base. Aerospace has written down more than half of its $1.39 billion in property. Southwest Research Institute, a campus full of engine dynamometers and spacecraft cleanrooms, has depreciated more than half of its billion dollars of plant. HHMI has written down two thirds of its $2.2 billion in buildings and equipment. Only the Broad, still pouring $30 million a year into new construction, looks young on this measure. Science is not only salaries and portfolios. It is microscopes and test chambers and buildings wearing out on a schedule, and the reason a place like SwRI reinvests its surplus into internal research and new facilities is that its edge is depreciating even faster than its books show. The reinvestment and the depreciation are the same fight, waged on two different statements.

For how long, and the honest answer for each

Put the four questions together and the "how long" answer falls out of the other three.

The pass-throughs last as long as their sponsor keeps paying. MITRE has run FFRDCs since 1958 and Aerospace since 1960, which sounds like permanence until 2025, when MITRE's funding to run the national vulnerability database nearly lapsed and the company laid off hundreds of people after federal contract cancellations. An institution with no endowment cannot miss a funding cycle, because the funding cycle is the only thing holding it up. The contract shops that kept a reserve, SwRI and Draper, buy themselves a few years of runway and the freedom to say no, which is exactly the fragility MITRE felt and they did not.

The endowed institutions last effectively forever, because an endowment that earns more than it spends has no expiration date. HHMI could lose a bad year in the market and keep paying its scientists from principal. The Broad, standing on three legs instead of one, is more diversified in its revenue but imports the political risk of the federal budget alongside the market risk of its portfolio, a proposed cut to federal research reimbursement threatening it with a $50 million hole in the same year the markets were fine. More legs carry more weight and give you more ways to stumble.

And the charity lasts as long as the giving holds and the reserve cushions the lean years, which is why Toys for Tots can run a deliberate $17 million deficit, spending into its rainy-day fund on purpose, and still stand on $198 million in reserve. Its risk is the one arrangement none of the others quite share: a long downturn would cut donations and shrink the invested reserve at the same time, the cushion deflating exactly as the need grew.

There is a single finding under all seven ledgers, and it is the one this series keeps arriving at. The differences that matter between institutions are never how important their work sounds. They are structural, and they are legible to anyone willing to read the filing: how the money comes in, whether any of it is kept, where the kept money is invested, what is owed against it, and whether the whole arrangement can survive a bad year in the one place it is exposed. MITRE and HHMI do work of comparable national weight, and they are financial opposites, one a channel that holds nothing and one a fortune that holds everything. The tax return tells you which is which in about ten minutes, and it never flatters and never lies.

The same reading, beyond the nonprofits

This hub began with seven research institutions, but neither the method nor the well stops at the nonprofit sector. The same four questions, how the money comes in, whether any of it is kept, where the kept money is invested, and how long the arrangement can last, read just as cleanly on a corporate balance sheet, a public fund, a church, or a family. A later run of the series turned that lens on fortunes, funds, and dynasties, and almost every one of them turns out to be another bucket lowered into the same market well, with a couple built deliberately to stay out of it.

On the compounding side sit the pure engines, the buckets that do little but sit in the well and fill. Insurance float is money a company holds but does not own, and Warren Buffett grew Berkshire's to $176 billion by investing it in the gap before claims come due. The Avenir story is that same engine at family scale, a physicist's $120,000 handed to Buffett in 1956 that a foundation reported as $1.24 billion in 2024. America's state sovereign wealth funds run it on oil money, paying Alaskans a yearly dividend and Texas schoolchildren their buildings. The Thrift Savings Plan runs it for seven million ordinary federal workers at a few cents per thousand dollars. And three tax-exempt reserves run it quietly and enormously: the LDS Church's Ensign Peak fund, reported near $100 billion; the Milton Hershey School trust, $23 billion that controls a public company on a school's behalf; and the donor-advised funds at firms like Fidelity, where the biggest charity in America is really a waiting room for invested money.

On the other side sit the structures built to hold and pass wealth rather than simply grow it, where the interesting question is the plumbing, not the return. The Rothschilds did it with a closed partnership and a network, not the whole-life-insurance story the internet sells. The Waltons hold half a trillion dollars in one entity, give assets away before they appreciate, and move the rest through charitable trusts. The Crowns and the Kroenkes run diversified fortunes through quiet private holding companies. The Kennedys ran a real-estate money machine that today's billionaires still copy move for move. And the de novo bank is the scarce asset itself, a charter almost no one can create anymore, which is why the wealthy, from Sam Walton onward, buy an existing one instead.

The finding is the same one this hub started with. What separates these institutions and fortunes is never how impressive they sound. It is structural, and it is legible to anyone willing to read the filing: how the money comes in, whether any is kept, where it is invested, what is owed against it, and whether the whole arrangement can survive a bad year in the one place it is exposed. Read that way, a church reserve, a retiree's account, a candy trust, and a dynasty's holding company turn out to be the same kind of object, seen at different scales.

The dynasties, family by family

The series then went wide, profiling the specific families and structures behind American and global fortunes. Read together they sort into a small number of mechanisms this hub keeps naming.

The trust-and-family-office archetypes are the model. The Rockefellers turned an oil fortune into a family office, generation-spanning trusts, and foundations, enduring across seven generations with no single billionaire. The du Ponts ran the same playbook for two centuries, their Alfred I. du Pont trust still funding a children's hospital. The Mellons preserved a banking fortune into a museum, a bank, and billion-dollar foundations. And the Gettys split their oil stock into one trust that still pays heirs and another that became the richest museum on earth.

Others separate control from ownership through dual-class shares and foundations. The Fords keep 40 percent of the vote on under 2 percent of a public carmaker through a supervoting share. Sweden's Wallenbergs control much of a national economy through foundations that own the holding company. And the families who control the news, the Sulzbergers, Murdochs, and Hearsts, use the same trust-and-dual-class tools to keep the presses in the family.

Some of the biggest are held entirely out of view. Cargill is the largest private company in America, owned by a deliberately invisible family; Koch is the second, its privacy funding both an industrial empire and a political network; and Mars hides a $129 billion candy-and-pet fortune behind a no-photographs rule. The hidden owners, from the man who really owns Aspen's Hotel Jerome to the quietest landlords in the country, prove the rule that the biggest money is the least visible.

Not every dynasty holds. The Hunts split one oil fortune across three families into a football empire, a quiet energy giant, and a catastrophic silver gamble; the Pritzkers built a thousand-trust machine and then deliberately took it apart in a famous breakup; and the Vanderbilts are the control group, the greatest fortune of their age gone in a few generations for lack of any structure to hold it. Andrew Carnegie is the deliberate opposite, the man who gave nearly all of it away rather than found a dynasty at all.

A fortune, once made and structured, has to live somewhere. Aspen and Ketchum are the trophy-home enclaves where it buys mountains behind anonymous LLCs; Jackson Hole is the tax haven where Wyoming's zero taxes and thousand-year trusts let it be parked; and William Zeckendorf is the cautionary tale of building it all on borrowed money and losing it, twice. Further afield, two Mauritian families show the same diversification playbook run on a sugar island, and Native American tribal sovereignty is a structural financial advantage the United States itself created. And underneath all of it sits the Federal Reserve, where the series separated the documented criticisms of the central bank from the myths that bury them, the same fact-versus-legend discipline it brought to the Rothschilds.

The machinery underneath

The most recent pieces turn from the families to the machinery itself, the structures that keep fortunes intact and the rules that let them. Two hotel dynasties draw the starkest contrast: the Marriotts kept the company and the money in the family, while the Hiltons routed almost the entire fortune into a foundation and lost the company to private equity. Tata, IKEA, and Agnelli are the international version of foundation and holding-company control, three empires topped by an apex entity no heir can cash out and no rival can buy. The family office is the private apparatus underneath all of it, an investment, tax, and estate department that captures advantages, including a tax deduction ordinary investors permanently lost, that only make sense above a high threshold. The line between legal and contested offshore structures, drawn through the Paradise Papers and the Irving and Cooper cases, turns out to be substance and honesty, not the jurisdiction. The car dealership is a different kind of protected asset, a heritable family business entrenched by state franchise law in a way almost no other industry enjoys. The Social Security trust fund is the public mirror image, a fund allowed to lend only to the government that then spends it. And the firm behind a political podcast shows the default state all of this is built to preserve, private wealth that even campaign-finance law only partly pierces. At the far other end of the ledger, Jeff Bezos's Courage and Civility Award is a living billionaire handing hundreds of millions to a handful of individuals to give away, generous, unaccountable, and with much of the destination still unknown.

Four of those overviews have since been given their own single-family treatment, and two of them corrected the overview in the process. Tata reads the structure from Tata Sons' own promoter table, where the widely reprinted 66 percent turns out to be 65.30 percent spread across seven trusts rather than concentrated in two. IKEA follows the franchise fee that moves between the foundation half and the brand half of a company that owns itself. Marriott is the rarer case of family control held without a supervoting share, and Hilton is its mirror, a fortune routed into a foundation and a company that left the family altogether. Loews adds the American conglomerate version, a public parent the Tisch family runs that holds majority stakes in businesses with nothing to do with one another, and that has retired most of its own shares rather than issue more.

The most recent wave turns to the people and institutions that hold the money now, and to how little of it the public record actually shows. The biggest family offices explains why the richest Americans barely register in the quarterly filings everyone reads, because a 13F captures only the sliver of a fortune that sits in listed stock. The twelve regional Federal Reserve banks are the strange public mirror of that privacy, a self-funding hybrid that no one owns and whose regional presidents out-earn the Chair. George Soros is the same three-part machine seen whole, a hedge fund that became a family office feeding one of the largest foundations on earth, read from the filings with the conspiracy theories set aside. Taylor Swift and Travis Kelce are the celebrity version of the ownership lesson, one fortune built on owning a music catalog outright and the other on small stakes in many businesses. The United Nations runs a ninety-billion-dollar pension fund and an internal tax system on Manhattan land a single family bought and gave away, tying the series back to the Rockefellers and to William Zeckendorf, who first assembled that ground. And two studies in luxury hospitality close the loop with the hotel dynasties: Relais and Châteaux, a family-run association of independent houses strung along a French road since 1954, and Aman, the opposite model, a scarcity brand that a two-country court fight left in one man's hands.

The newest run widens the lens from the fortunes themselves to the whole system they sit inside, and to what an ordinary wage-earner can do about it. It opens with the uncomfortable evidence that luck decides more of success than either side of the argument admits, then turns practical: the ladders that raise a floor almost anyone can climb, from the military benefit stack to a public pension; the legal edges and lotteries gated by which category the law recognizes you in, from tribal contracting to century-old senior water rights to the finance pedigree wall; what happens when those floor-raising programs get looted from the inside; and whether the Justice Department is still chasing the fraud it once did. It closes on the wage-earner's own position: the W2 trap and the toll economy, where utilities, water, waste, rail, and even burial collect a guaranteed return on bills no one can escape while a paycheck is the one input promised nothing, and the schemes you are not supposed to notice, a plain playbook for owning the toll instead of only paying it.

The latest run follows the money down to the level of the paycheck and the programs that surround it. It starts with the one number the government indexes and the one it does not: the cost-of-living adjustment set against the private raise, where Social Security, disability, and federal pensions get an automatic inflation raise the un-indexed wage never does. The response splits two ways: the jobs whose raise is written into a contract, from public pensions to the pilots and longshoremen who bargained double-digit escalators, and the defensive playbook for the insurance, energy, and health bills that outrun a paycheck. Alongside them sit three more pieces of public-money plumbing read the same way as Essential Air Service and Alaska bypass mail: the flood insurance program that has never paid its own way, the black lung trust fund that taxes a shrinking industry to cover a resurging disease, and the pension backstop that stayed self-funded on one side and took a taxpayer rescue on the other. And the surprises buried in the leaks, across the Paradise Papers, WikiLeaks, and the Epstein files, close the loop by showing how much public and tax-advantaged money moves, legally, through institutions the public underwrites.

A further run walks the rural-lifeline programs, the ones judged completely differently depending on whether you measure cost per unit or the public good they buy. The Alaska Marine Highway is a ferry system that recovers only about a third of its cost from fares yet is the only road to a state capital you cannot drive to. Amtrak's long-distance trains lose the most money per passenger in the system and are the only public transit for hundreds of small towns. PILT and Secure Rural Schools pay counties for federal land they cannot tax, one on autopilot and one on a timer that keeps lapsing. And the wildfire suppression machine is the cost curve that outgrew its acres, a federal and state apparatus whose bill crossed a billion dollars in 2000 and now routinely runs several times that. Each reads the same way as Essential Air Service: efficiency and access are two different yardsticks, and which one you lead with decides the verdict before you open the spreadsheet.

A further set turns to the corners of federal money most people never see. In health care, Critical Access Hospitals are paid 101 percent of their own cost to keep an emergency room within reach of empty counties, the 340B program moves tens of billions in drug discounts that big nonprofit systems capture as revenue, and community health centers deliver primary care to one American in eleven. In the nuclear enterprise, the MOX plant consumed 7.6 billion dollars before it was cancelled, the Nuclear Waste Fund holds 49 billion for a repository that was never built while taxpayers separately pay the damages, FOGBANK is the warhead material the government forgot how to make, and the Office of Secure Transportation is the unmarked armored fleet that moves nuclear weapons across the country. Who gets paid to fight wildfires follows the suppression dollar to the private air-tanker operators and the sole-source retardant contract. And the frozen paycheck on Capitol Hill is the one salary that could index itself to inflation and votes every year not to. All of these federal programs are collected, grouped by the kind of money problem each one is, in the public-money index.

Fact-check notes and sources

All figures in this overview are drawn from each organization's most recent public IRS Form 990, and each is documented in full in the individual post linked for that institution: MITRE (EIN 04-2239742, calendar year 2024), Aerospace (EIN 95-2102389, fiscal year ending September 2024), Draper (EIN 04-2505372, fiscal year ending June 2025), Southwest Research Institute (EIN 74-1070544, fiscal year ending September 2024), the Broad Institute (EIN 26-3428781, fiscal year ending June 2024), HHMI (EIN 59-0735717, fiscal year ending August 2025), and Marine Toys for Tots (EIN 20-3021444, calendar year 2024). The returns are available free at ProPublica's Nonprofit Explorer and the IRS Tax-Exempt Organization Search.
The headline figures used above (MITRE's $2.48 billion revenue and near-zero investments; Aerospace's $1.37 billion revenue and roughly $410 million in benefit-plan assets against its pension liability; SwRI's roughly $270 million conservative reserve; Draper's roughly $270 million diversified investment pool; the Broad's roughly $1.5 billion endowment and three-legged funding; HHMI's $27.6 billion net assets and nearly $2 billion of investment income; and Toys for Tots' $136 million reserve, $17 million deficit, and $198 million in net assets): each is sourced line by line in the corresponding individual post.
The comparison to pensions and sovereign wealth funds describes the well-documented reality that large nonprofit endowments invest in the same public and private markets as pension funds, sovereign wealth funds, and individual investors; the specific allocations (HHMI's private-equity and alternatives tilt, the Broad's roughly 75 percent alternatives, Draper's multi-asset pool, and the named retail funds in the Toys for Tots reserve) are read from each organization's Schedule D.

This post is informational and historical, not financial or investment advice. All figures are reproduced from public filings. Organizations and individuals are discussed from the public record as nominative fair use, with no affiliation implied and nothing endorsed by any of them. Characterizations of investment posture describe filing schedules, not recommendations.

The Working Ledgers: How Seven Research Institutions Invest, Stay Solvent, and Keep the Work Going