Yes—high earners fled big cities in 2020-2022, but most did not return; some cities are refilling with new high earners, not old ones

Summary

Yes, and mostly no. IRS tax-return migration data confirms a real, large, income-selective outflow from major city cores between 2020 and 2022: across 10 major urban core counties (Manhattan, San Francisco, Los Angeles, Cook/Chicago, Suffolk/Boston, King/Seattle, DC, Santa Clara/San Jose, Philadelphia, Fulton/Atlanta), net adjusted gross income (AGI) lost to migration totaled roughly -$50 billion in tax year 2021 alone, versus about -$14 billion in 2019 — and in every one of these counties, every year, the people leaving earned more on average than the people arriving, with that income gap widening through 2022. As for whether they came back: largely, no. Household outflow volumes from NYC and San Francisco did normalize back toward pre-pandemic levels by 2022, but the average income of who was still leaving kept climbing (Manhattan leavers: $156k avg AGI in 2019 to $221k in 2022; SF leavers: $166k to $281k), and independent reporting through 2024-2025 shows continued net outflow of top earners from New York (~125,000 New Yorkers and ~$14B in income to Florida alone since 2018) and Chicago, even as overall city populations ticked up again. The exception is San Francisco, where population is rebounding in 2024-2025 — but reporting attributes this substantially to a new wave of AI-industry high earners (OpenAI, Anthropic, and related hiring) refilling the city, not the return of the households who left in 2020-2021.

What the IRS migration data shows, 2019-2022

AskAmerica's fiscal.county_migration_flows table (IRS Statistics of Income, county-to-county outflow files) covers tax years 2012-2022, with 2022 the latest year currently loaded (IRS publishes on roughly a 4-year lag, so 2023-2026 do not yet exist in this or most public datasets). This table does not carry an income-bracket dimension, so "high earners" is proxied two ways: (1) net AGI flow, which is disproportionately driven by higher earners since they carry more income per return, and (2) average AGI per return among leavers vs. arrivers, which directly measures the income composition of movers.

Querying the 10 largest, highest-cost urban core counties for 2019-2022:

• Net AGI outflow accelerated sharply in 2020-2021 in every core county queried, then partially narrowed by 2022 in some (Manhattan, San Francisco) but not others (Los Angeles, Cook County/Chicago, whose 2022 net AGI loss was essentially as large as 2021's).
• Manhattan lost $16.1B in net AGI via migration in TY2021 (matching the Economic Innovation Group's independently reported $16B figure for that year) before narrowing to -$4.1B in 2022 — still triple the 2019 loss.
• San Francisco lost $7.0B in TY2021, then -$4.4B in 2022 — narrower in dollar terms, but the average AGI of leavers kept rising every single year (2019: $166k -> 2022: $281k), meaning the income-selectivity of the outflow, if anything, intensified even as the dollar total eased with fewer total movers.
• Household-count outflow (net returns) did fall back close to 2019 levels for Manhattan (-8,509 in 2019 vs. -890 in 2022) and San Francisco (-4,437 vs. -4,443, essentially unchanged), but got worse for Los Angeles (-33,877 to -51,909), Chicago (-19,376 to -29,245), Seattle/King County (+2,515, i.e. net inflow, in 2019, to -4,875 in 2022) and DC.

Did they stay away? Evidence past 2022

Because the IRS county migration table ends at tax year 2022, the "did they stay" half of the question rests on independent reporting rather than this connector's own data for 2023-2025:

• The Citizens Budget Commission (via reporting in the New York Post and summarized by AIER's The Daily Economy) finds more than 125,000 New Yorkers moved to Florida between roughly 2018 and 2022-2023, taking about $14 billion in income with them — with New York City's top 1% of filers (~41,000 people) still paying more than 40% of city income tax, making continued high-earner departures fiscally significant.
• Migration-data aggregators (moving-company and IRS-derived trackers cited via Empower/Fortune/PropertyShark reporting) find that in 2024-2025, high-income ($200k+) movers still made up a large share of outbound New York City moves, and that outmigration of $200k+ earners in mid-2025 was only modestly lower than in mid-2024 — a slowdown in the rate of loss, not a reversal into net inflow.
• Overall city population figures for New York and San Francisco did turn positive again in 2023-2024, but that recovery appears driven mainly by international immigration and a broader (non-wealthy) population, not by the return of the specific high-earning households who left in 2020-2021.
• San Francisco is the clearest partial exception: 2025 reporting (Washington Post, CNN, SF Standard, Fortune) documents an AI-industry hiring boom (OpenAI, Anthropic and others each leasing roughly 1 million square feet of office space) pulling a new cohort of high earners into the city, visible in a housing-market frenzy and rents up 23-26% year over year. This is repopulation by a new wave of tech wealth, not the same households returning — a meaningfully different phenomenon than "they came back."
• Illinois/Chicago and Massachusetts reporting (Crain's, Bloomberg) describe continuing, not reversing, high-earner outmigration through 2024, consistent with Cook County's IRS data showing no 2022 improvement.

Bottom line

High earners did leave major cities in large, disproportionate, IRS-documented numbers in 2020-2022, concentrated in the biggest, highest-cost coastal metros (Manhattan, San Francisco, Los Angeles) plus Chicago and DC. The raw volume of people moving out of NYC and SF specifically has normalized back toward pre-pandemic levels, but the income skew of who's still leaving has not reversed — if anything it intensified through 2022 — and post-2022 reporting shows continued, not reversed, high-earner net outflow from New York and Chicago through 2024-2025. San Francisco's 2024-2025 population and housing rebound is real but appears to reflect a new AI-wealth cohort moving in, rather than the pandemic-era leavers moving back.

Every query behind this report

1 warehouse call ran in this session, in order. Each is reproducible against the same snapshot.

WITH counties AS (
  SELECT * FROM (VALUES
    ('36061','Manhattan (New York County), NY'),
    ('06075','San Francisco County, CA'),
    ('06037','Los Angeles County, CA'),
    ('17031','Cook County (Chicago), IL'),
    ('25025','Suffolk County (Boston), MA'),
    ('53033','King County (Seattle), WA'),
    ('11001','District of Columbia'),
    ('06085','Santa Clara County (San Jose), CA'),
    ('42101','Philadelphia County, PA'),
    ('13121','Fulton County (Atlanta), GA')
  ) AS t(fips, name)
),
outflow AS (
  SELECT origin_county_fips AS fips, year, SUM(num_returns) AS ret_out, SUM(agi) AS agi_out
  FROM fiscal.county_migration_flows
  WHERE origin_county_fips IN (SELECT fips FROM counties)
  GROUP BY origin_county_fips, year
),
inflow AS (
  SELECT dest_county_fips AS fips, year, SUM(num_returns) AS ret_in, SUM(agi) AS agi_in
  FROM fiscal.county_migration_flows
  WHERE dest_county_fips IN (SELECT fips FROM counties)
  GROUP BY dest_county_fips, year
)
SELECT c.name, o.year,
  o.ret_out, i.ret_in, (i.ret_in - o.ret_out) AS net_returns,
  ROUND(o.agi_out) AS agi_out_k, ROUND(i.agi_in) AS agi_in_k,
  ROUND(i.agi_in - o.agi_out) AS net_agi_k,
  ROUND(o.agi_out / NULLIF(o.ret_out,0),1) AS avg_agi_per_out_return_k,
  ROUND(i.agi_in / NULLIF(i.ret_in,0),1) AS avg_agi_per_in_return_k
FROM counties c
JOIN outflow o ON o.fips = c.fips
JOIN inflow i ON i.fips = c.fips AND i.year = o.year
WHERE o.year IN ('2019','2020','2021','2022')
ORDER BY c.name, o.year