Puerto Rico's public money lives in three floors of budgets — and no official document adds them up. Counted once, the whole building moves about $42.5 billion a year. This is the journey of that money — from Washington and your IVU receipt, down to a $3.50 lunch tray — told entirely with the government's own documents.
Some money appears in two budgets because Floor 1 sends it downstairs: the Commonwealth pays UPR's appropriation, subsidizes the highways, tops up poor municipalities, and lately lends PREPA the cash for pension checks. That money is an expense on Floor 1 and revenue on Floors 2–3. Add the three floors naively and you count it twice. So first, subtract the internal transfers:
Kitchen-table version: for every $3 the famous budget shows, roughly $1 more moves through your power bill, your water bill, the tolls, tuition, and your town hall.
Read it plainly: 40¢ of every public dollar arrives from Washington; nearly all the rest leaves a Puerto Rican pocket directly — as taxes, bills, fees or tolls. Chapters 0–11 walk the first floor in line-item detail — the central government, where $30.4 billion actually left the accounts in FY2024 — because that is where the public data reaches all the way down to the lunch tray. Chapter 12 walks the other two floors.
Every number in this report is in millions or billions — quantities no human brain can feel. Before the data, calibrate your senses.
Same dots, no compression, drawn as you scroll. The red block at the top is the entire billion above; the orange speck in the first row is the $1 million.
— end of one fiscal year. It renews every July 1. —
Not the approved budget. Not the political promise. The money that actually left government accounts in FY2024.
The top four — Education, public health insurance, the Treasury's central custody accounts, and family welfare — absorb 58% of everything. And the third-largest "agency" isn't an agency at all: it's a central account that pays debts and pensions before a single service is delivered.
Read as one dollar: about 30¢ goes straight to people as benefits, 20¢ pays government employees, and 14¢ pays for the past — pensions and old debts — before buying any service today.
The Departamento de Educación is the single biggest spender: $5.37 billion. Here is the same money sliced three ways.
In FY2024 the Legislature approved $131M for school maintenance; only 52% was executed. The government now proposes $465M for FY2026 — a tacit admission of the backlog. The bottleneck isn't only how much is budgeted. It's the capacity to actually spend it.
Two honest numbers: the system costs taxpayers ~$22,900 per student all-in; educating a child today costs ~$17,900. The $5,000 gap is each student's share of the past.
One of every three students is in private school — about triple the U.S. rate — in a territory where median income is a fraction of the mainland's. A third of families pays twice: taxes for a system they don't use, plus tuition. Cross-checked against Census data, these counts hold within a ~15% margin, with a genuine blind spot: roughly 55–60,000 school-age children appear in no enrollment count (graduates, dropouts, homeschoolers, migration timing — nobody knows precisely).
Puerto Rico now spends near the U.S. average per student. What does the world get for similar money?
The states cluster in a band between 259 and 284. Puerto Rico isn't in the band: at near-average spending, it scores 57 points below the lowest state — roughly five school years of learning. Essentially no PR public-school 8th grader (~1%) reaches "Proficient." Context that's fair to add: child poverty is double the poorest state's, a third of students (skewing affluent) are in untested private schools, and today's 8th graders lived through hurricanes, earthquakes and closures. None of it closes a 57-point chasm.
Three stories in one image: Estonia vs. Utah — same money, one extra year of learning (system design moves you vertically). New York — 181% of U.S. spending buys average results (money has diminishing returns). Puerto Rico — alone in the bottom half of the entire map. No other point is close to that corner.
Two corners of the budget that explain more than their size suggests.
The Departamento de la Vivienda runs almost entirely on Washington's money — mostly hurricane-recovery funds. Its approved budget jumps from $1.4B to $3.37B in FY2025 as the long-delayed reconstruction billions finally schedule to move. After what we saw with school maintenance, "approved vs. actually spent" becomes the accountability question of the decade.
$13.6 billion in local revenues (FY2024-25 actual, Hacienda). The biggest taxpayer isn't who you think.
Does income match spending? Yes — by law. PROMESA only allows balanced budgets, and revenues just beat projections by $410M. Puerto Rico even "saves": a 5% spending retention, a $683M electricity rate reserve, growing pension reserves. But the discipline is a cage, not a habit; part of every good year flows to bondholders through contingent value instruments; and the reserves exist precisely because both main income pillars — federal transfers and multinational taxes — are outside the island's control.
Before any service is delivered, the past collects.
27¢ of every locally-raised dollar pays for the past — and 94% of debt payments come from local taxes, not federal funds. Not shown: a slice of the IVU pays COFINA's restructured debt before ever reaching Hacienda's register.
The pattern: old, payroll-heavy agencies with shrunken workforces — schools, roads, environment — carry the heaviest past. Transfer agencies (health insurance, family welfare) carry none. Pensions are not waste: they're earned, legally protected pay for work already done. The honest charge is against decades of promising without funding — leaving today's budget to pay twice, hardest in exactly the services citizens touch daily.
Under PROMESA, no budget has legal effect without certification by the oversight board. Its reach: 100% of the $32.7B — and beyond.
This year the Junta certified the revised budget alone, after the governor and Legislature failed to agree. Everything in this report — the surpluses, the reserves, the balanced books — happens inside a cage Puerto Rico didn't design and pays to maintain. Whether that's protection or tutelage is for the reader to decide.
At bankruptcy, Puerto Rico owed ~$70B in bonds plus ~$55B in unfunded pensions — ~$125B, about $35,000 per resident. Bondholders get paid through two doors.
The Capital Appreciation Bonds — mostly COFINA — were Puerto Rico's payday loans: interest compounds silently for decades and balloons at the end. $4.3 billion borrowed; $33.5 billion owed — 785% interest, most of it money nobody ever actually lent. Fairness note: the restructured stack's 1.75x ratio is normal-mortgage territory; the scandal was never that debt costs interest, but that the old structure was predatory — and the new COFINA deal still carries CABs with balloons to 2058.
Bondholders didn't lose everything — recoveries ran 54¢ to 93¢ per dollar. Who gained and who lost depended less on the haircut than on when you bought and whether you held through the panic.
The plan paid everyone holding the same bond the same amount, regardless of what they paid. Profit was determined by entry price — and entry price was determined by someone else's panic. The abuela who panic-sold at 30¢ on the dollar and the fund that bought her bond collected from the same instrument.
The aggregate gain — roughly $16 billion under the documented blend — looks like a normal return spread over 30 years. It isn't: original par holders roughly break even in nominal terms (a loss after inflation), which means essentially all of the real gain concentrated in the minority that bought the panic.
The right question is never "is it good or bad" — it's does it create more long-term public value than it costs? Applied to Act 60's individual investors, the answer turns out to be: nobody can currently prove it either way.
The asymmetry is robust: by GDP, by taxes actually paid, or by jobs, manufacturing outweighs Act 60 individuals by 10–50×. But note the anchors: Act 60's estimated contribution ($1–5B) and its estimated forgone taxes ($0.3–1.5B) are both meaningful fractions of the entire $13.6B Fondo General — too big to ignore, too unmeasured to settle.
Same benefit, three damage assumptions: +$2.8B, +$1.0B, or −$1.7B. When a policy's verdict depends entirely on unmeasured variables, the finding isn't "positive" or "negative" — it's that the data to decide is not published. The same transparency gap as the charter schools and the bondholders' purchase prices: this project's recurring villain isn't a party or a sector; it's unpublished data.
The intro showed the building — three floors of budgets, ≈$42.5B counted once. This chapter walks the floors the Sábana never shows: the property-tax pipe that pays old debt before any town sees a cent, the 78 municipal budgets nobody certifies, and the pension bill that follows you to every floor.
One entity in the stack isn't a spender at all. CRIM collects the island's property taxes into a trust — $1.26B in FY2025 — and almost half never reaches a town hall: it's intercepted at the source to pay municipal debt.
With few exceptions, it buys the town's administration itself, road and building maintenance, trash collection, municipal police, and recreation. Essential services — schools, hospitals, police at scale — remain the central government's job.
In 2022 this fund was over 30% of the budget in 17 of the 78 towns; for the large urban centers it rounds to zero. It is being phased out — and towns like Las Marías, Utuado and Vieques now run more than half their budget on non-recurring money: reconstruction funds, one-time windfalls, construction excises. Some towns run half their government on money that isn't promised next year.
Chapter 7's finding doesn't stop at the Commonwealth. Every floor of the building pays yesterday before today: