College Costs Compound for Eighteen Years Before Anyone Enrols

The multiplier is not the rate, it is the rate compounded

The mistake that produces most of the surprise is treating cost inflation as if it applied once. It applies to every year between now and the first bill, and then again to every year of the programme. A four-year public in-state budget of 30,990 dollars, eighteen years away, at 4 percent, is not 30,990 plus a bit. Year one is 30,990 times 1.04 to the eighteenth, which is 2.0258, so 62,780 dollars. Year four is 30,990 times 1.04 to the twenty-first, which is 2.2788, so 70,619. The four years together are 266,593 against 123,960 at today's prices — a ratio of 2.15, higher than the first-year multiplier because the later years compound longer.

That structure is why the assumed rate matters so much and why it should be chosen deliberately. Published prices rose 2.9 percent for public four-year in-state students and 4.0 percent for private non-profits between 2024-25 and 2025-26. Institutions' own input costs are tracked by a different index, the Higher Education Price Index, which is not the consumer price index and does not have to move with it. Over the thirty years from 1995-96 to 2025-26, after adjusting for inflation, average published tuition and fees rose from 5,940 to 11,950 dollars at public four-year institutions — a doubling in real terms, a real compound rate of about 2.4 percent a year. Pick a rate you can defend, then look at what one point either way does to the answer: on this example, 45,872 dollars.

The sticker price is not the price

The published price and the transaction price are two different numbers, and in the United States the gap is enormous. For first-time full-time in-state students at public four-year institutions in 2025-26, average grant aid is 9,650 dollars, which takes published tuition and fees of 11,950 down to an average net of 2,300 and the published cost of attendance of 30,990 down to a net of 21,340. At private non-profits, average grant aid is 28,090, taking published tuition and fees of 45,000 down to 16,910 and the published budget of 65,470 down to 37,380. Note what that last pair implies: the average net cost of a private non-profit, 37,380, is only 1.21 times the published sticker of a public four-year in state.

So project the base you actually expect to pay. Running the same four years at the net cost of attendance of 21,340 dollars gives 183,579 instead of 266,593 — 83,015 less, from changing one input. Two cautions, though. Averages hide enormous variance: grant aid is concentrated, so a family that receives none pays close to the sticker, and the average is not a forecast for any particular household. And net price today is not net price in eighteen years, because aid formulas and appropriations change too. Projecting the sticker is a conservative choice; projecting the net is a realistic one; doing both and looking at the range is the honest one.

Tuition is a share, and the share tells you where to act

Compare the shares and the strategy writes itself. In the five European systems in the table, the institution's own charge is between roughly 3 and 11 percent of the annual cost of studying away from home; the other 89 to 97 percent is rent, food and transport. In the United States the share is genuinely large — 11,950 dollars of tuition against a 30,990 budget is 39 percent at a public four-year in state — but even there, housing and food alone are 13,900. Wherever you are, the biggest single lever on the total is whether the student lives at home, and the second biggest is where they live if they do not.

This also explains a comparison that looks paradoxical. A student who wins a full tuition waiver but moves to an expensive city may pay more in total than one who pays full fees and stays at home. In the French example, the entire three years of droits d'inscription and CVEC come to 849 euros; a hundred euros a month less in rent saves about 3,700 euros over the same three years, more than four times as much. Put the effort where the money is.

Filling the calculator in without importing someone else's assumptions

The calculator asks for three cost lines, an inflation rate, the years until enrolment and the length of the programme, and every one of those is a decision. For the cost lines, use the institution's own published cost of attendance rather than a national average if you have a shortlist — the variance between institutions dwarfs the variance between years. For years until enrolment, count from now to the September the student actually starts, including any gap year: an extra year at 4 percent adds about 4 percent to every figure in the table. For the length, four years is the American default but six-year completion is common, and a fifth year costs a fifth year.

One last honesty check. The projection assumes the family pays the whole bill from savings, which almost nobody does: part comes from current income during the study years, part from student earnings, part from borrowing. The total is the size of the problem, not the size of the pot you need on day one. Split it explicitly — how much from savings, how much from income, how much from borrowing — before you decide the monthly saving figure, because a target set against the whole total will look impossible and a target set against your actual share will not.

What the model does not do

It applies one inflation rate to all three cost lines, which is a simplification and a visible one: tuition, housing and books have not moved together historically and there is no reason to expect them to. It also assumes a continuous programme with no repeated year, no change of institution and no change of residency status — and in the United States, moving from out-of-state to in-state status, where it is possible, is worth 19,930 dollars a year of published tuition on the 2025-26 averages, more than any inflation assumption.

And it says nothing about how the money is held, which is a separate decision with its own rules and its own tax treatment. Work out the size of the target here, then take it to the savings side and ask what monthly contribution and what expected return get you there — and how much of the answer is contributions rather than growth, which is usually more than people expect.

Frequently asked questions

What cost inflation rate should I use?

Pick one you can justify, then test one point either side of it. For the United States, published prices rose 2.9 percent for public four-year in-state and 4.0 percent for private non-profits between 2024-25 and 2025-26, and over the thirty years to 2025-26 average published tuition and fees at public four-year institutions doubled in real terms — from 5,940 to 11,950 dollars in constant money — which is a real compound rate of about 2.4 percent a year on top of general inflation. Institutions' input costs are tracked separately by the Higher Education Price Index, which is not the consumer price index. The important discipline is not picking the right number, which nobody can do eighteen years out, but seeing the sensitivity: on a 30,990-dollar budget eighteen years away, 4 percent gives 266,593 over four years and 3 percent gives 220,721, a difference of 45,872 from one percentage point.

Why is the projected total so much more than four times today's price?

Because the compounding starts today, not on the first day of the programme. Each year of the programme is a different number of years away: with a newborn and a four-year degree, the four years are 18, 19, 20 and 21 years out. At 4 percent the factors are 2.0258, 2.1068, 2.1911 and 2.2788, so the four annual figures are 62,780, 65,291, 67,903 and 70,619 dollars against a base of 30,990 each. The total, 266,593, is 2.15 times the four-year sticker of 123,960 — more than the first-year multiplier of 2.03, because the later years compound for longer. Waiting a year before starting to save does not change the target; it changes how long the money has to grow, which is a separate and larger problem.

Should I project the sticker price or the net price?

Both, and then look at the gap. On the 2025-26 averages, projecting the published cost of attendance of 30,990 dollars for four years starting in eighteen years at 4 percent gives 266,593; projecting the average net cost of attendance of 21,340 gives 183,579. The gap, 83,015, is exactly the amount of grant aid you are assuming. Two warnings. Averages hide huge variance because grant aid is concentrated, so a family that qualifies for none pays close to the sticker. And today's net price is not a forecast of the net price in eighteen years, because aid formulas and appropriations change. Treat the sticker projection as the conservative bound, the net projection as the central case, and the gap between them as the risk the savings plan is carrying.

Does the calculator handle a five-year or six-year path?

Yes — set the programme length and it extends the projection year by year, which matters more than most people assume. Each extra year is both an extra annual cost and an extra year of compounding on top of the previous one, so the fifth year of a programme starting eighteen years from now is at 1.04 to the twenty-second power, or 2.37 times today's price. On the public four-year example, adding a fifth year takes the total from 266,593 to 340,037 dollars, an increase of 73,444 — more than the whole four-year sticker at today's prices divided by two. If the realistic plan involves a master's, a professional qualification or a likely repeated year, model the real length rather than the nominal one.

Is a public university always cheaper than a private one?

Not necessarily, and the American figures show why. In 2025-26 the average published budget at a private non-profit four-year institution is 65,470 dollars against 30,990 at a public four-year in state — more than twice. But average grant aid is 28,090 at private non-profits against 9,650 at public four-year institutions, so the average net cost of attendance is 37,380 against 21,340. The private average net is 1.21 times the public published sticker, not 2.11 times it. Two caveats keep this from being an argument for any particular choice. Averages are not offers: aid depends on the household's finances and on the institution's own policy, and half of all students receive less than the average. And the public in-state figure assumes residency, which is what makes it 11,950 rather than 31,880 in published tuition. Compare actual award letters, not sector averages.