Research
Job Market Paper
Raise Fast, Cut Slow: Asymmetric Reversal of Recession-Driven State Tax Increases
Abstract (click to expand)
Under balanced-budget rules a recession should force states to adjust their tax rates, yet the most comprehensive study of state tax changes to date finds no effect of recessions on rate levels. I document a clear short-run response followed by faster and more frequent reversal, which narrows that puzzle without dissolving it. Using a panel of 50 U.S. states from 1960 to 2022, I show that recessions raise the probability of any major state tax increase by 7.7 percentage points while recoveries produce no offsetting cuts within three years. A formal test rejects equality of the recession increase and decrease responses at the 1% level, and rejects at the 5% level the equality of the recession increase and recovery decrease coefficients. Those increases are then cut sooner and more often, facing more than twice the hazard of a subsequent cut, and 28 percent survive a decade uncut against 62 percent of increases enacted in normal times. Completeness differs at five years but not detectably at ten, and the median increase of either kind still stands in full after a decade, so substantial rate persistence remains. The asymmetry is clearest in the timing and incidence of cuts rather than in how far they go. Separately, cumulative recession exposure is not associated with higher rate levels. Together these weigh against a permanent upward ratchet, though neither establishes that the subsequent cuts account for the absence of accumulation. The contemporaneous differential is a within-state gap between recession and other years, at 0.67 percentage points on the top income rate, rather than an accumulated stock.
Working Papers
Same Language, Same Market? Substitution Between Mexican and Other Hispanic Immigrants in the United States
Draft coming soon
Abstract (click to expand)
Immigrants and natives with the same education and experience are imperfect substitutes, and the leading explanation is that they differ in language and in the tasks language allows. If so, two immigrant groups that share a language should be far closer substitutes for each other than immigrants are for natives. I test this prediction for Mexican and other Hispanic immigrants in the United States in national education-experience cells over 1990 to 2024. Because the shift-share instrument used for immigrant supply has no first stage for a single origin group, I instrument Mexican supply with a demographic projection, cohorts observed five to thirty-five years earlier aged forward with arrivals ignored. The inverse elasticity of substitution between the two groups is 0.071 for workers without a high school degree, 0.067 for high school graduates, 0.078 for workers with some college, and 0.131 for college graduates, the first two precisely estimated and the four not distinguishable from one another. The implied elasticity of substitution is about 14, against 16 to 23 for immigrants and natives in the same cells and the same design. Shared language does not make immigrant groups one labor pool. National origin segments the market at every education level about as much as nativity does, so a shared language is not sufficient for two groups to be close substitutes, and language alone cannot account for the immigrant-native wedge.
Taxation after a Tax-Base Shock: Evidence from the Boll Weevil
Draft coming soon
Abstract (click to expand)
This paper asks how local governments adjust property taxation when an income shock hits the industry on which their local economy depends. Such a shock can reduce both property values and taxpayers' earning capacity. Governments can ordinarily stabilize property-tax revenue by raising rates when assessed values fall, but doing so preserves the tax bill faced by owners whose incomes have also declined. I test whether that adjustment survives, using the boll weevil's devastation of Southern cotton agriculture between 1892 and 1922. Using decadal census data for 1890 to 1930, I compare counties of differing pre-weevil cotton intensity within compact geographic regions. For a ten-percentage-point increase in cotton intensity the levy falls by 0.119 log points. Full stabilization is the restriction that the levy response is zero, and the baseline estimates reject it. The assessed base falls as well, without an offsetting rate increase, and no specification provides statistically significant evidence of one, the estimated statutory-rate response being −0.031 log points. County-purpose estimates indicate a decline of 0.150 log points in the levy counties set for themselves and 0.069 in the rate. A fixed upper rate ceiling cannot mechanically explain such a reduction. Pre-arrival patterns, sensitivity to cotton-specific calendar paths, and a change in the levy definition across waves limit causal attribution. The findings suggest that property-tax revenue stabilization can weaken where the local economic base is concentrated.
Work in Progress
Federal Retreats and State Advances: Asymmetric Vertical Fiscal Interactions Across a Century
How state tax policy responds to federal tax changes over a century, and whether that response is symmetric.
The Price of Giving: Tax Incentives, Reporting, and Real Charitable Behavior
Whether the fall in reported charitable giving after the 2017 tax reform reflects real behavior or a change in who itemizes.
Railroads and Property Taxation
Whether local governments capture an expanding tax base when market access grows, the mirror of the boll weevil test.