WorthDegree Investigation · October 24, 2026

The most expensive college in California charges $99,682 a year. The cheapest hands students back $2,296. Ten years later, their graduates are 28% apart.

We ranked all 500 California institutions in the federal data by what they actually cost and what their students actually earn. The two numbers barely know each other exist.

Two schools, one state, one taxpayer system.

ICOHS College, San Diego. Annual net price after grants: $99,682. Median earnings ten years after enrolling: $62,714. Cohort: 171 students.

Moorpark College, Ventura County. Annual net price after grants: minus $2,296. Median earnings ten years after enrolling: $49,044. Cohort: 12,082 students.

That price gap is $101,978 per year. The earnings gap is $13,670 per year, or 28%. One school charges six figures. The other charges nothing — it pays — and lands within 28% of it.

If price were a measure of what a school adds, that sentence would be impossible. It is not impossible. It is the median case in the largest state in the country.

The measurement: rank correlation, not anecdote

One pair of schools proves nothing. Small cohorts, unusual programs, one bad year — any two institutions can be made to tell whatever story you want. So we measured the whole state.

For every California institution in the College Scorecard with both a reported net price and reported post-enrollment earnings — 500 of them — we ranked the schools by price and ranked them again by ten-year median earnings, then computed the Spearman rank correlation.

The result: 0.004. Not 0.4. Zero point zero zero four. In plain terms, knowing what a California college costs tells you essentially nothing about what its median student earns a decade later. The relationship is indistinguishable from noise.

Nationwide the same test is slightly less brutal but still weak: 0.193 across 5,031 institutions, which means price explains about 3.7% of the variation in earnings. And the California number is not an artifact of the state's size — it is what happens when a state contains both a $0 community college system serving two million students and a cluster of six-figure boutique art and health institutes serving a few hundred. The price axis is measuring institution type, not educational value.

What price does buy, on average

A correlation of zero does not mean price is irrelevant. It means the signal is drowned by the spread. Fitting a straight line through all 5,031 institutions: every additional $10,000 of annual net price is associated with $4,317 more annual earnings ten years later.

Put differently, the cheapest tenth of American colleges carries a median net price of $4,463 and a median ten-year income of $37,097. The most expensive tenth costs $35,785 — eight times as much — and produces $46,543. Eight times the price for 25% more pay.

Sort every institution into price bands and the averages do climb, which is exactly why the single anecdote is so misleading:

Read the middle of that table carefully, because it is where the honest version of this story lives. Going from $10,000 to $45,000 in annual price triples the share of schools that beat a high school diploma. Then the curve stops. The $45,000–$60,000 band is worse than the band below it on median earnings and on the share clearing the baseline. The top of the price distribution is not the top of the earnings distribution. It is a different distribution entirely.

Why the two numbers come apart

Look at what sits in the $55,000-plus tier: a helicopter flight school, two advertising portfolio schools, two art and design colleges, a naval architecture institute with 106 students, a college built for students with learning differences, and four institutions teaching massage therapy, East Asian medicine, audio engineering and dramatic arts.

These are not scams. Several are the only route into a licensed profession. But the price is set by the cost of delivering the training — aircraft hours, studio space, clinical supervisors, 12-student cohorts — while the earnings are set by the labor market that absorbs the graduates. Nothing forces those two numbers to meet. When a program costs $99,682 a year to run and its graduates earn $62,714, the student is not buying a return. The student is financing a cost structure.

The community college side is the mirror image. Moorpark's negative net price is not a marketing trick: it is the average annual net price for full-time first-time students after grant aid, and when Pell, Cal Grant and institutional aid exceed tuition plus living costs for the average recipient, the arithmetic goes below zero. The state subsidizes the price; the labor market sets the ceiling. Median earnings land at $49,044 — modest in absolute terms, extraordinary per dollar of private cost.

Which is the actual finding. Not that expensive colleges are worthless. That the price tag has stopped encoding the thing families think it encodes.

Three honest objections

1. The negative price is an average, not a bill. Moorpark's minus $2,296 reflects students who received enough grant aid to cover tuition and living costs. A student who qualifies for nothing pays tuition. The school is not writing checks. The figure does show, correctly, that for the median aided student the private cost of attendance is zero or below — which is the number that matters when comparing to a $99,682 invoice.

2. The two headline schools are not measured the same way. Moorpark's earnings are reported values over 12,082 students. ICOHS College's are estimated values over 171. One is a verified median from a large cohort, the other a small-sample estimate. That asymmetry is a property of the federal data, and it is precisely why the 0.004 correlation — computed across all 500 institutions, not two — is the load-bearing number here, not the anecdote.

3. Institution-level medians hide the major. Miami Ad School charges $64,951 and reports $106,192; Pacific College of Health and Science charges $57,004 and reports $26,209. Both sit in the same price tier. The difference is what they teach, not what they charge. A single median per school averages a computer science cohort with a ceramics cohort. Read the correlation as "price does not predict the median," never as "price causes nothing."

All three are true. None of them restores the price signal. If a family is choosing between two California colleges $100,000 apart and the federal data cannot tell them apart on earnings, the price is not information. It is a bill.

The complete list: every US college above $55,000

All 15 institutions nationwide with an annual net price of $55,000 or more. Red earnings are below the $46,748 high school graduate baseline. An em dash means the federal data does not report a completion rate for that institution. This is the whole population, not a sample.

Check us in four minutes

Do not take the 0.004 on trust. Open the College ROI dataset, filter to state = CA, keep only rows with both a net price and an earnings figure, rank both columns, and compute the correlation yourself. Then re-run it for your own state. Of the 41 states with at least 30 institutions, 31 come in under 0.25 and 11 are negative outright — Idaho is the lowest at −0.17, Iowa the highest at 0.53.

If you would rather test one decision than a distribution: take the two schools you are actually considering, pull each one's ten-year median earnings from the Scorecard, and compare that to the annual net price you personally would pay. That ratio — earnings over price, in your case, with your aid — is the only version of this number that belongs to you.

Related: the completion rate does not predict earnings either — across 4,895 colleges, median earnings peak in the 50-60% graduation band at $42,198 and then fall 11.7% by the 80%+ band, and thirteen schools report $70,000+ where fewer than half of students finish.