The argument has a certain seductive logic: raise taxes on the wealthy, and they'll simply leave — taking their capital, their businesses, and their economic activity with them, leaving behind a smaller tax base and, ultimately, less revenue than before. It is an argument that has been made in virtually every debate over progressive taxation in the United States for decades.

The empirical evidence for this claim is, at best, mixed — and in important respects, actively contradicted by what research has found.

Studies of tax-motivated migration among high-income individuals have generally found that while some wealthy people do relocate in response to tax changes, the numbers are far smaller than the rhetoric suggests. A Stanford study examining millionaire migration patterns in California found that the population of millionaires actually grew following a tax increase on high earners, and that the share who left was statistically minimal.

The reasons aren't hard to understand. Wealthy people are deeply embedded in the places where they've accumulated their wealth. Their businesses, their social networks, their professional relationships, their family ties — all of these create powerful anchoring forces that a marginal change in tax rates does not easily overcome. Moving is expensive, disruptive, and in many cases strategically counterproductive. You cannot simply transplant a New York private equity firm to a no-income-tax state and expect all the deal flow to follow.

There are exceptions. Retirees with portable wealth and no business obligations are genuinely more mobile. The very wealthy with truly global operations have more flexibility than most. And at sufficiently extreme tax rates, migration effects might become significant. But the consensus range of proposed reforms — returning top marginal rates to levels seen in the 1990s, closing carried interest loopholes, reinstating higher estate taxes — does not approach those thresholds.

The "they'll all leave" argument functions primarily as a political deterrent. The data suggests the deterrent is, in most cases, more powerful than the underlying phenomenon.