State top income tax rates, and what followed their changes

The rates

This is the top marginal rate in each state’s own income tax: the rate charged on each extra dollar in the highest income bracket. It covers wages and other ordinary income, for a single filer, for each tax year. Statewide extra taxes on top earners are included (California’s 1% over $1 million, Massachusetts’s 4% from 2023). Local income taxes are left out. We compiled these rates for this site, since the published series cannot be republished. Each change below cites the law that made it. It also cites at least one original document, such as the text of the law (a statute or session law) or the state’s own tax form. We fetched each one and checked the rate before accepting it. Some states changed their rate partway through a year, and their tax return used a blended rate (a mix of the old and new rates). In those cases the blended rate counts as the year’s rate. Before 2001, Rhode Island, Vermont and North Dakota taxed a percentage of federal tax. For those years, their top rate is that percentage times the 39.6% federal top rate. Each state’s page shows the rate with its other measures.

The largest changes: one percentage point or more in a year

What followed

For each outcome, we take each state and year. We line it up with the change in the top rate that year, in each of the five years before, and in each of the three years after. Each state is compared only with itself over the years, and each year only with the same year in other states. The calculation also allows for the fact that a state’s years tend to resemble each other. Figures are per one-point rise in the top rate (per one-point cut where marked). A point is one percentage point, such as 5% to 6%. The table columns add up the years: one total for the three years before the change, and one for the six years from the year of the change on. The last two rows measure rises and cuts separately, each adjusting for the other and for income growth. The three years before the change are a test of the method. A tax change cannot explain what happened before it. So if an outcome is already moving before the tax changes, it is not following the tax. Dots are estimates, and bars are likely ranges (95% intervals). A bar that crosses zero means the result could easily be chance.

Net domestic migration

per 1,000 residents, 2001–2021, 41states. The chart adjusts for each state’s own income growth. The table shows results with and without that adjustment.

Net migration of $200,000-plus filers

per 100 staying, 2012–2021, 41states. The chart adjusts for each state’s own income growth. The table shows results with and without that adjustment.

Net business formation

% of establishments, 2001–2021, 41states. The chart adjusts for each state’s own income growth. The table shows results with and without that adjustment.

Private job growth

%, 2011–2021, 41states. The chart adjusts for each state’s own income growth. The table shows results with and without that adjustment.

What this can and cannot say

• Association only.No state changed its tax at random. Rates rise when budgets are short and fall when they are flush. Adjusting for each state’s income growth takes out only part of that.
• Many estimates, so some look real by chance.There are four outcomes, nine years each, and several versions. In a set this size, a few single-year estimates will pass the usual test for “unlikely to be chance” (a p-value below 5%) by luck alone. The six-year totals are the figures to read.
• The top rate is one number. It says nothing about the income where the top rate starts, about deductions, or about local taxes. A flat tax and a graduated one (with rates that climb as income rises) can share a top rate. But it is the rate the highest earners face, which is what a question about them needs.
• High earners from 2012 only. The IRS counts of movers by income start in 2011–12, so that outcome has the fewest years. The IRS changed how it processed the data before 2018. That shifted all states alike, so comparing each year only with the same year elsewhere takes it out.

Outcomes: Census population estimates (net domestic migration), IRS migration data ($200,000 or more), Census Business Dynamics Statistics (openings minus closings), BLS Quarterly Census of Employment and Wages (private jobs). Effective rates (the tax actually collected as a share of income), and state and local income tax together, are on the tax-lag page. High earners, from every public source, are on where the rich move.

Every change, with its sources