Every public pension in America, the funds that pay retired teachers, police, and firefighters, rests on a single guess. It is the assumed rate of return, the plan's estimate of how much it will earn on its investments each year over the long run. Change that one number, and a pension can flip from looking healthy to looking badly broke.

The reason is subtle. That same guess is used to measure how large the plan's future obligations are today. A high assumption makes the obligations look small and the plan look well funded. A lower, more cautious one makes the same obligations balloon and the shortfall appear.

The machine we are drawing is that number, and the accounting trick hidden inside it. It quietly decides whether a state reports a manageable gap or a crisis, how much taxpayers must chip in now, and how much of the bill gets quietly pushed onto the next generation.

Public pensions have been grudgingly lowering these assumptions for years. Before the 2008 crisis, the typical plan assumed it would earn around 8% a year. Today the national median has fallen to about 7%, and the average sits just below that, as plans admit the old optimism was unrealistic.

Even small cuts land hard. When the largest U.S. pension fund trimmed its assumption by a tenth of a percentage point, its reported obligations jumped by roughly $30 billion overnight. The money did not change; only the number used to measure it did.

Why this reaches the reader: these plans owe some $10 trillion in promised benefits, and his state and local taxes are what fill any gap. If the assumption is set too high, the reported shortfall is understated, contributions are kept artificially low today, and the true cost is deferred onto him and his children.

The board of a federal bank voted unanimously to lend nearly $3 billion to build a gold mine on American soil. Congress got 25 days notice. Nobody objected.

Final papers are expected in the second half of this year. The day that ink dries, three things happen at once: funding risk goes to zero, the U.S. government becomes financially fused to the project, and Wall Street re-rates the stock from speculative developer to federally backed strategic asset.

One more detail. The company's own filings cite "substantial support and partnership from the Department of War," a phrase we've never seen on a gold project. The reason: alongside its gold, the deposit holds a metal China formally banned from export to the United States. The only domestic reserve in the country.

The company is about one fiftieth the size of Newmont.

See the gold stock at the center of it >>

Start with what a pension owes. It has promised to pay retirees for decades into the future. To know whether it has enough money, it must translate all those future payments into a single value today. That translation requires picking a rate to discount the future back to the present, and the higher the rate, the smaller today's value looks.

Here is the quirk that makes public pensions unusual. For a typical plan, the rate used to discount those future promises is set equal to the rate it assumes it will earn on its investments. So the same optimistic number does two jobs at once: it predicts investment gains, and it shrinks the measured size of the debt.

The load-bearing part is therefore that single assumption. Assume 7% and the obligations look modest and the plan looks respectably funded. Assume a cautious 5.5%, closer to what safe bonds actually yield, and the very same promises swell by hundreds of billions of dollars nationally, and funded ratios that looked fine suddenly look alarming. Nothing real changed. Only the dial moved.

And the dial has a thumb on it. A higher assumption lowers the reported shortfall and lowers the contribution the government must make this year, which is politically convenient. Officials facing tight budgets have every incentive to keep the number high, because a high number makes today's problem smaller and hands the difference to whoever is in charge later.

That is why the machine can quietly endanger a giant. To justify a 7% assumption when safe bonds yield only about 4.5%, the plan must reach for that extra return by piling into riskier investments, stocks, private equity, private credit. The accounting assumption drives the investment strategy: the fund takes on more risk not because it is wise, but because it must, to make its own optimistic number come true.

The first joint is the hidden shortfall. Because a high assumption shrinks the reported debt, a plan can look much healthier than it is. Independent analysts find that valuing these promises with safe, bond-like rates would reveal far larger unfunded liabilities than the official figures admit. The gap between the comfortable headline number and the cautious one is the risk, sitting in plain sight but out of the report.

The second joint is the deferral onto taxpayers. When the assumption is too high and returns fall short, the missing money does not disappear; it becomes a bill that grows with time. Contributions that should have been made years ago must be made later, larger, by future taxpayers who received none of the benefit of the delay. The optimistic number is, in effect, a loan taken out against the next generation.

The third joint is the forced reach for risk. Holding a high assumed return pushes funds into ever-riskier assets to hit the target. That works in booming markets and backfires in bad ones, when the risky portfolio falls hard precisely as the plan can least afford it. The assumption meant to describe the future ends up dictating a riskier present, tying retirees' security to the market's mood.

What persists: the logic of long-horizon investing, which is sound. Pension funds really do invest for decades and can reasonably expect to earn more than cash over time, so some assumed return above the risk-free rate is defensible. And the long-running move to trim these assumptions toward reality is genuine progress, shrinking the gap between the comfortable story and the honest one.

What can break: the honesty of the funded ratio, whenever the assumption drifts above what the plan can truly earn. The number that is supposed to measure a pension's health instead flatters it, and the flattery is strongest right before trouble, when a plan clings to a high assumption to avoid admitting a shortfall. The danger is not the assumption itself but the temptation to set it by what is convenient rather than what is likely.

This machine connects to two others the reader's money touches. It is the public-sector cousin of the private pension transfers we have traced, the same promise to retirees measured a different way, and its hunger for that extra return is a major reason pensions pour money into the private credit and leveraged strategies we have followed. The assumption, the funded ratio, and the risky portfolio chasing it are one chain.

The next reading arrives with each plan's annual valuation and any change to its assumed rate. Those will show whether public pensions keep inching their assumptions toward honesty, or whether the single most convenient number in government finance stays set a little too high, quietly billing tomorrow's taxpayers for today's comfort.