A middle-class family leaves New Jersey for Florida. The mortgage halves. The grocery bill drops. The school they're enrolling in shows comparable test scores to the one they're leaving. By every public metric of an American move, this is a clean trade. The Stanford Education Data Archive's 2025 release contains a number nobody put on the listing. It says the school they're moving to teaches their child about three-quarters as fast as the one they're leaving.
A middle-class family leaves New Jersey for Florida. The mortgage halves. The grocery bill drops. The school they're enrolling in shows comparable test scores to the one they're leaving. By every public metric of an American move, this is a clean trade. The Stanford Education Data Archive's 2025 release contains a number nobody put on the listing. It says the school they're moving to teaches their child about three-quarters as fast as the one they're leaving.
A family in Morris County, New Jersey. Their school district scores well above grade level — a real number from the latest SEDA release, nearly six grade-equivalent units above the national baseline. The mortgage is crushing. The childcare cost is two college tuitions a year. They look at a Florida district with comparable test scores. They list the house.
The number they don't see is the slope. SEDA reports two numbers for every district: where students score and how much they gain per grade year. The score is what shows up on Niche, on GreatSchools, on the listing brochure. The slope shows up nowhere. Morris adds 0.86 grade-equivalents per grade year. The Florida district they're moving toward adds, on average, 0.71. That difference is what they will spend $40,000 to access.
Rank every state by learning rate. Florida is dead last. Alabama and Mississippi are just above it. Texas is fifth from the bottom. South Carolina is third. These are not coincidences. Five of the seven states absorbing the most interstate wealth — Florida, Texas, the Carolinas, Tennessee — sit in the bottom third of the country on learning rate. The Sun Belt isn't an upgrade on schools. It's a discount.
Zoom into Sarasota County, Florida. The achievement number is high — 6.05 grade-equivalent units, two-thirds of a grade ahead of the national mean. On any score-only ranking this is a strong district. The learning rate is 0.76 per grade. The district imports students from somewhere — retirees' grandchildren, second-home families, the inflow that gives Florida its $21B advantage — and those students arrive ahead. They leave only slightly ahead. The school is not the source of the gap. The migration is.
SEDA names these two signatures. Coasting: high test scores, low learning rate. The students arrive ahead and stay ahead, but the school itself doesn't move them. Climbing: low test scores, high learning rate. Students arrive behind and the school catches them up. These are mirror images. They cannot be distinguished by the score alone.
The most-coasting destination districts in the SEDA data — Sarasota FL, St. Johns FL, Edison NJ, Queen Creek AZ, Poway CA, Madison City AL — are all wealthy suburbs whose families arrived with academic momentum. The most-climbing districts — Compton CA, Newark NJ, Baltimore, Detroit, Philadelphia — are urban, low-income, often heavily immigrant. The schools that genuinely teach students to grow regardless of where they started are concentrated in the places people are leaving, not the places people are going.
Plot every district on two axes — achievement on the X, learning rate on the Y. Four quadrants emerge. Coasting in the lower right. Climbing in the upper left. Falling (low scores, low slope) in the lower left. Elite Growing in the upper right — a small set of districts both far ahead and gaining fast. SEDA names them because they are rare.
A family making a real-estate decision based on the score alone is buying horizontal position. They are not buying vertical. And vertical is what compounds.
Pull back. Every state. X-axis: 2023 net interstate income inflow ($B). Y-axis: average state learning rate. The upper-right quadrant — gaining wealth AND high learning rate — should be where the money is going. It is nearly empty.
Idaho is in it. So is Wisconsin, in a quiet way. Florida is at the bottom of the chart, far to the right — the state attracting the most wealth at the slowest school growth in the country. Texas is below national average. So are South Carolina, North Carolina, Tennessee, Alabama. The wealth migration is heading into the slowest-learning half of the country. The states losing wealth — New York, New Jersey, Illinois, Massachusetts — sit in the upper half on slope. The trade families are making, at the state level, is unambiguous. They are buying cost, selling slope.
The same exception as last time. Idaho ranks first in learning rate (1.02 per grade — meaning students gain slightly more than a full grade per grade year, the only inflow state where this is true). Cost of living is $40,000 below comparable-quality coasts. Net interstate inflow is $1 billion in 2023. Idaho is the deal of the dataset on housing, on quality of life, and on schools. The problem is the same as last time. Idaho has 1.9 million people. The migration cannot fit there.
The 2020–2022 learning collapse is now four years old. SEDA's recovery numbers are in. They do not say what was expected. They say recovery followed income — cleanly. The wealthiest fifth of districts recovered nearly all of their loss. The poorest fifth recovered the least, ending 2025 a fifth of a standard deviation behind their pre-pandemic baseline. The middle three quintiles sit between them in order. There is no flat middle. There is a gradient, and the gradient is steep.
The schools the migration is leaving — middle-quintile suburban districts in New Jersey, Illinois, Massachusetts — recovered some of what they lost. The schools the migration is heading toward — bottom-quintile rural Florida, low-SES Texas, struggling districts in Alabama and South Carolina — recovered nothing. The trade is not just slope-for-score. It is also recovery-for-discount.
Where the wealth went in 2023, by state. The South and the Mountain West gained. The Northeast and the West Coast bled. Florida is the singular outlier on the high end; California and New York are the singular outliers on the low. This is the map families think they are reading.
The same forty-eight states, recolored. Now the axis is learning rate. Florida is the deepest red on the map. Texas, the Carolinas, Alabama, Mississippi sit below it on cost but at the bottom on slope. New Jersey, New York, Wisconsin, Idaho color teal — the states still moving children forward at full pace. This is the map families do not think to read.
Overlay the two. The pattern collapses to four kinds of state. Red: gaining wealth, losing slope — the double squeeze (Florida, Texas, the Carolinas, Alabama, Mississippi). Amber: gaining wealth, mediocre slope (Tennessee, Nevada, Arizona, Georgia). Teal: losing wealth, fast slope — what the migration is leaving behind. Green: gaining wealth, fast slope — the lone exception, Idaho. Most of the dollars are flowing into red.
The compounding effect runs in both directions. The destination states absorb tax revenue but inherit the slope they came with. The children of migrants graduate from the same Sun Belt high schools their parents bought into — but with the same per-year growth their parents purchased for them. Over twelve years of K-12, a learning-rate gap of 0.25 per grade compounds to three full grade-equivalents. That is more than the score gap most families thought they were avoiding.
The origin states bleed the families most likely to advocate for school improvement. The tax base shrinks. The PTAs weaken. The middle-quintile recovery, already shallow, gets shallower. The wealth gap between American zip codes is not stabilizing. It is being redistributed — and the schools the country is being asked to fund at the highest standard are losing the constituencies that paid for it.
The two numbers SEDA reports — score and slope — are not the same number. The score is what's on the brochure. The slope is what the school adds. Families optimizing for cost and score are buying horizontal position on the map. Slope is what compounds over a childhood. The single act of looking at both numbers, before the move, is the entire intervention this story argues for. The data is public. The number for the destination district is one click away. Almost nobody clicks.
District-level achievement and learning rate are from the Stanford Education Data Archive, version 2025.1 (Reardon, S. F. et al., 2026; DOI 10.25740/hm970gr1371). Each value is the empirical-Bayes estimate from the pooled 2022–2025 admindist file, all-students subgroup, grade-cohort-standardized (GCS) metric. Achievement is reported in grade-equivalent units; learning rate is the per-grade slope, with 1.0 representing the national average pace of one grade-equivalent gained per grade year. State-level learning rates are linear-regression slopes across grades 3–8 in 2022–2025 (state long_gcs file, all students, math and reading pooled).
Post-pandemic recovery deltas are computed as cohort-standardized score in the 2022–2025 pool minus the corresponding score in the 2009–2019 pool, district level. Districts are classified into national income quintiles using SEDA's SES composite (sesavgall in the covariate file), an ACS-based combination of median household income, parental education, employment, and household composition.
Wealth migration figures (2023 net interstate income flows by state) are from the Internal Revenue Service Statistics of Income migration data, the same source used in The American Cost Divide (April 2026). Cost-of-living savings figures derive from the Missouri Economic Research and Information Center cost-of-living index Q3 2025 and the BLS Consumer Expenditure Survey 2024.
A note on what the data showed vs. what the proposal predicted. The original story brief proposed a "middle-quintile squeeze" — the idea that Q2 and Q3 districts would show the flattest post-pandemic recovery, caught between federal aid that helped the poorest and family wealth that rescued the wealthiest. The SEDA 2025.1 numbers do not support this. Recovery follows income as a clean gradient: Q1 (poorest) recovered the least, Q5 (wealthiest) recovered the most, with Q2–Q4 ordered between them. This story reports what the data shows. The proposal also framed Florida as a positive exception with high learning rate; the actual 2025.1 release places Florida at the bottom of every state on this metric. The framing in this piece — Florida as the headline coaster, Idaho as the lone exception — reflects the data, not the original brief.