Inequality Project · Case Study No. 1
Forty years ago a teacher could buy the brownstone. Twenty-five years ago, an architect. Now it takes tech, finance, or a trust. What happened?
First findings · August 2026 · data through tax year 2022 / income year 2024
The question decomposes into four measurable pieces. Two now have data answers; two are queued. Everything below comes from public data — IRS county tax files, BEA county income accounts, and the Frank state top-income panel — rebuilt from scratch in this project's pipeline.
Answer: yes — on two separate margins at once
The honest version requires holding the bar fixed in real dollars, because the raw IRS bracket ($200k, nominal) drifts with inflation. Done properly: inflation explains about 60% of the raw growth in "$200k+ returns" — but the real shift underneath is still large. Households above a constant-dollar bar grew from 3.2% to 4.9% of all returns in eleven years (1.7× in count, against 11% growth in total returns), and real-dollar millionaires grew 1.8×, from 301k to roughly 547k. And on the longer horizon the bar itself moved: qualifying for New York's top 1% takes nearly triple the real income it took in 1950. More rich households, and a much richer top.
The U-curve: top 1% share of US income, 1917–2022
Percent of national income received by the top 1% of tax units
The timing matters for the thesis: the trough is 1970, and the climb begins right around 1980 — the era of banking deregulation and the shareholder-value turn. A teacher could buy the brownstone precisely when the top 1% claimed the smallest slice of national income in a century.
Answer: they were in the suburbs — then they came to Brooklyn fastest of all
Share of tax returns over $200k, New York area, 2011–2022
Percent of each county's returns in the IRS top bracket
In 2011 the old-money geography still held: Westchester and Nassau had two to four times Brooklyn's density of high earners. Rich people mostly were "being rich in the suburbs." What changed is the direction of flow — Brooklyn's high-earner count grew 3.4× in eleven years, against 1.6× in Manhattan and roughly 2× in the classic suburbs. Brownstone Brooklyn is where the new rich chose the city over the suburb, which is why the change feels most violent there.
Zooming out from New York: for the first postwar decades, poor and rich places in America were growing together — the per-capita income gap between counties shrank steadily. That ended around 1980 and reversed. The concentration you see on your block is the street-level expression of a national regime change.
Inequality between places, 1969–2024
Population-weighted Gini of county per-capita income (higher = places further apart)
The data. BEA's county personal income accounts (table CAINC1) report, for every U.S. county in every year since 1969, two numbers: total personal income earned by residents, and population. Personal income is BEA's broad pre-tax measure — wages and salaries, employer benefits, self-employment income, dividends, interest, rent, and government transfers. Dividing the two gives each county's income per person. State and national rows are dropped, leaving ~3,110 counties per year.
The calculation. For each year, imagine assigning every American their own county's average income — a Brooklyn resident gets Brooklyn's per-capita income, a McDowell County, WV resident gets McDowell's — and then measuring inequality across all ~330 million people in that flattened distribution. That is what the plotted number is: a standard Gini coefficient, computed by sorting counties from poorest to richest per capita, building the Lorenz curve of cumulative population share against cumulative income share (each county weighted by its population), and taking one minus twice the area under the curve. Zero would mean every county has the same average income; higher means places are further apart.
What it does and doesn't measure. This is inequality between places only. Inequality inside a county is invisible by construction, because everyone in a county is assigned the same average. That is why the level (0.11–0.16) is far below the familiar person-level U.S. income Gini (roughly 0.48): averaging within counties erases most inequality. The signal here is the movement, not the level — falling through 1976, flat, then rising from about 1980 to above where it started.
Caveats. No inflation adjustment is needed: the Gini is a unitless ratio computed within each single year, so the price level cancels out. It is not adjusted for local cost of living, so some of the measured gap between counties is offset by price differences (a Manhattan dollar buys less). Income is counted where people live, not where they work. Weighting by population means large counties matter in proportion to their size — without it, 100-person rural counties would count the same as Los Angeles.
Reproducible from analysis/phase1/concentration.py (column gini_county_pci_popw in data/processed/concentration_timeseries.csv).
Teacher, architect, and veterinarian median pay versus brownstone-Brooklyn housing costs, decade by decade. The plan: BLS occupational wages against Census/Zillow housing series, to date exactly when each profession lost the neighborhood.
The IRS publishes county-to-county migration flows, by income class, back to 1990. We can literally trace where households leaving Kings County went, and how the answer differs for those above and below the income bar.
Sources & method. IRS Statistics of Income county files (2011–2022) and national Table 1.1; BEA Regional Economic Accounts CAINC1 (1969–2024); Frank, M.W. (SHSU), state top income shares 1917–2022, CPI-U-RS deflated, incl. realized capital gains; Census CBSA delineations (2023, held constant). "Returns" are tax units, not people — a married couple filing jointly is one return. Constant-dollar counts use CPI-U (2011→2022 factor 1.30) with Pareto interpolation inside the IRS $200k–$500k bracket. All series rebuilt from raw sources in the project pipeline; every figure on this page is reproducible from analysis/phase1/.