This is one of the entries in the NY Times Magazine's annual Year in Ideas:
The Rising-Tide Tax System, by Stephen Mihm, NY Times: In the last 20-plus years, overall economic growth in the United States has come at a cost: rising income inequality, which in the past few years has hit levels not seen since the late 1920s. A more progressive income tax, introduced during the New Deal, helped mitigate the problem, and it remains a likely prescription today. Yet a team of economists that includes Robert Shiller of Yale University and Leonard Burman of the Tax Policy Center recently released a paper in which they propose another way of “spreading the wealth”...
Under the proposal, the tax code would automatically be rewritten at the end of each year to reflect any changes in the relative share of national income earned by each income bracket. For example, if one year the nation's top earners saw their share of national income rise while people at the bottom saw their share grow at a slower rate (or decline), the following year's tax rates would be automatically rewritten to compensate for the new inequality. This would keep everyone's share of after-tax income at the earlier level.
The ... proposal would work both ways: if the rich saw their share of the nation's income grow more slowly relative to people lower down the economic ladder, the tax system would become less progressive...
The name of the proposal — the Rising-Tide Tax System — is an allusion to John F. Kennedy's claim in 1962 that “a rising tide lifts all boats,” a promise that general economic growth would benefit all members of society. Shiller argues that it's still possible to turn Kennedy's vision into a reality. “It's something we can engineer,” he says.
This has attractive properties, and it may help us to avoid the concentration of bubble-inducing excess liquidity in the hands of relatively few people, but how do we know which baseline level of inequality to target?
Also, this approach leads to the idea that the only reason for progressive taxation is redistribution, but that's not the case. Concepts such as equal marginal sacrifice can be used to support a progressive structure, so even if there is no redistribution at all we may still want to have progressive taxes. Thus, before we can use this approach we'd have to decide (at least) two things - what is the level of inequality we are targeting, and what is the base level of progressiveness that we use. For example, if there is no change in inequality from previous years, does that mean taxes should be flat?