Empirical research · 134 heads of school · 97 institutions
Boards set head-of-school pay by tuition. Tuition is almost noise.
The intuition is reasonable. A school that charges more should pay its head more. Across 134 identified heads of school, what a school charges predicts almost nothing about what its head is paid. What the school takes in predicts a great deal.
134heads of school
97institutions
165schools in tuition panel
2017–24filing years
Scroll
A school that charges more should pay its head more.
That is how compensation committees think, and there is nothing foolish about it. Tuition is the one number every school publishes. It sits on the admissions page. It gets quoted in the local paper. It is the figure families argue about at the kitchen table. If you want to know whether another school is your peer, it is the first thing you can actually check.
So committees check it. They pull the schedules of schools charging roughly what they charge, look up what those schools pay their heads, and land in the middle. The chain feels airtight. A school charging premium tuition is presumably a more demanding institution than one charging far less. Higher price, bigger operation, harder job, larger salary.
Every link in that chain is an inference. Not one of them is a measurement.
Tuition is not a description of a school. It is a decision a school made about the families it wants to enrol. A board can hold it down on purpose. A community can keep it low for a generation as a matter of principle. A school in an expensive city can charge a great deal and still run a small operation.
So the question is not whether tuition is convenient. It plainly is. The question is whether it tells a board anything at all about the job.
Two restaurants, one street, very different jobs.
One has a short menu and high prices. Twelve tables, forty covers on a good night, six people in the kitchen. The other is a diner. Nothing on the menu breaks fifteen dollars. It feeds four hundred people a day, runs three shifts and employs thirty.
Which is the harder place to manage?
Menu prices tell you who a restaurant is pricing for. They tell you nothing about how many people it feeds, how many staff it schedules, or what happens when the walk-in fails on a Saturday. Those are questions about the size of the operation. Price is a question about the customer.
Nobody hiring a general manager would compare menus. They would ask about covers, payroll and revenue. Yet comparing menus is close to what a compensation committee does when it opens the tuition schedules of ten schools and calls them peers.
Not what a school charges. What it collects.
Run the comparison directly. Take the 134 heads of school identified individually across eight years of federal filings, at 97 institutions, and ask which feature of a school best tracks what its head is paid.
Correlation, in one sentence
A correlation is a single number, written r, for how much knowing one figure helps you guess another — 0 means it helps not at all, 1 means it tells you everything. An r of 0.60 means a school’s revenue moves you a long way toward a sensible estimate of its head’s pay. An r of 0.15 means its tuition barely moves you at all.
Total revenue — everything the institution takes in, tuition and fundraising and the rest of it — correlates with head-of-school compensation at 0.60. Published tuition manages 0.15. Same schools, same heads, same filings. One of those numbers is doing four times the work of the other, and it is not the one on the admissions page.
Total school revenue
0.60 r
Correlation with head-of-school compensation. Scale is the dominant signal.
Published tuition rate
0.15 r
What the school charges families carries almost no independent information.
What predicts head-of-school compensationAll predictors
01 — The ranking
Size first, and by a distance.
Total revenue leads at 0.60. Enrollment tier follows at 0.44. Both are measures of the same thing: how big the institution is.
Everything else on the list sits underneath. The price, the person, the path they took to get there.
02 — The anchor that fails
Sticker price is nearly noise.
Published tuition lands at 0.15. Knowing what a school charges barely narrows the range of salaries you would expect to find.
Two schools can post the same tuition and differ enormously in what they actually take in. It is the taking in that the market is pricing.
03 — Credentials
A fact about the person beats the price of the school.
Holding an advanced academic degree correlates at 0.28 — close to twice the signal carried by tuition.
Prior career track — teaching, administration, the rabbinate — correlates at 0.06 and does not clear statistical significance. Boards price the credential, not the route. And they price it above the schedule they say they are benchmarking on.
Boards are asking a question this data answers badly.
“What do schools charging what we charge pay their head?” and “what do schools our size pay?” feel like the same question asked two ways. They are not. One is answered four times as well as the other, and the distance between them is where benchmarking goes wrong.
If tuition is not measuring the school, what is it measuring?
A price is a decision. Somebody sits in a room every spring and picks a number, and the picking is about families at least as much as it is about costs. How much can this community bear. What will the aid budget absorb. What happens to enrolment if we go up again.
Schools facing similar costs answer that question differently, and the differences hold for years. You can watch it happen in a dataset with no compensation figures in it at all: an annual panel of 165 schools.
Posted tuition growth by denominationNominal growth
04 — A decision, not a size
Schools move their prices differently, and keep doing it.
If tuition tracked the scale of an institution, schools would move it in step. They do not. Across the panel, non-Orthodox schools raised posted tuition 29.2% from first observed year to last. Orthodox schools raised it 21.1%.
Same years, same inflation, different postures toward families. That is what a price is.
05 — In real terms
Almost nobody was chasing the market.
Deflate by consumer prices and most of that growth disappears. Non-Orthodox schools roughly held their real price, up 1.1%. Orthodox schools fell 4.4% behind.
45% of all schools ended the panel cheaper in real terms than they began. These are affordability decisions, held on purpose. A tuition figure is a school’s answer to a question about its families. It is not a description of the institution.
Median posted tuition by stateThe price landscape
Lower tuitionHigher tuition
06 — And the map is in there too
Geography is baked into the number.
Median posted tuition runs from about $33,200 in California to $10,600 in Quebec. A threefold spread, driven largely by what it costs to live and to hire in each place.
So a tuition figure carries a market and a posture, blended into one number that separates neither. Neither ingredient is a fact about the size of the institution somebody has been hired to run.
The field standardised on the number that was easy to get.
Revenue is not a secret. It sits in a Form 990, a filing every one of these institutions makes. But you have to know which form, find the right year, and pull a figure out of a document nobody designed to be browsed. Tuition takes a browser and eleven seconds.
Given a hard number and an easy one, the field took the easy one. That is not a scandal. It is what fields do. The failure came at the next step: treating the resulting comparison as though it described the job.
Pull the two apart and they point at different subjects. What a school charges is a statement about its families — a decision about affordability, made in a particular market, revisited every spring. What a school collects is a fact about the institution — its size, its budget, its payroll, the number of things that can go wrong on a Tuesday. A tuition-matched peer group is not a set of comparable institutions. It is a set of schools that made a similar decision about price.
Two things this cannot show. It does not show that revenue causes pay. A correlation is a pattern in the numbers, not a mechanism: bigger schools may pay more because the job is harder, or simply because they have more money to spend. Nothing here separates those. And it says nothing about whether any head is paid the right amount. That is a judgment about value. This is a study of benchmarks.
What it does show is which benchmark carries information. Boards have been reaching for the most visible number in the room rather than the most useful one. The fix is unglamorous. Find the revenue figure. It is one more form, and a far better question.
Tuition is the number everyone can see. That is not the same as the number worth looking at.
Scale Over Sticker Price · Ḥeshbon
What this analysis cannot say
Four limits worth carrying with the finding.
No pay trends, at all
An audit established that the source compensation fields are a single current snapshot back-filled across historical rows. Every compensation finding is cross-sectional.
Twenty-five matched schools
The cross-referenced institutional analysis rests on 25 schools, only 15 to 17 with complete tuition and giving data.
Eleven Orthodox schools
Denominational classification matched 94 of 165 panel schools, leaving a thin Orthodox cell — direction, not magnitude.
Associations, not mechanisms
These are correlations. Nothing here identifies how a board actually reasons, or what causes what — only what the resulting numbers look like when they are lined up.
Ḥeshbon — Center on Data and Jewish Life
A data project on how Jewish life is paid for. The work comes in seasons. This page is part of Season 1 — Jewish Education. It states its own sources, sample sizes and limits.
Working prototype. Every figure comes from the underlying research. Where a chart simplifies a published result, the page says so. Nothing here audits an individual school.