Changing tax rates is quite hard. It's a big deal. I still remember Gordon Brown's 'twopenny budget' back in 2007, when the then-chancellor cut 2p off the basic rate of income tax and corporation tax (while scrapping the 10p introductory rate of income tax but partly muted that by beefing up tax credits). But changes in tax code technicalities and language are much less salient and widely-noticed, and may well be a more fruitful way of improving tax policy. The job market paper (pdf) from Elliott Ash, a JD/PhD candidate at Columbia university, tackles this question by looking at the text of 1.6m statutes from state legislatures from 1963 until today. He tries to discern the effective tax code "the set of legal phrases in tax law that have the largest impact on revenues, holding major tax rates constant". Essentially his approach allows him to quantify elements of the tax regime that would otherwise be only qualitative—like when people quote the length of the tax code to measure its complexity.
He then associates terms with more or less revenue. For example, the phrase 'buildings and structures' has one of the strongest positive associations with extra revenue from the personal income tax, perhaps because it clarifies what kinds of property can be claimed for deductions. Similarly, 'certain motor vehicles' has a strong negative association with sales tax, presumably because the phrase is generally used to make particular automobiles exempt from sales tax.
His main finding is that Democratic takeovers of state legislatures tend to increase the overall progressivity of the tax system, by switching the language of the state tax code towards effectively raising more from income taxes. By contrast, Republican takeovers shift tax codes towards language associated with more revenue from sales taxes—more efficient but less progressive. This approach is much finer grained than the fairly blunt tool of looking at rates, which are much stickier.
It's not all that surprising, but it's a fascinating approach, and one I expect to be mined for more interesting findings in the future.