Working Papers

Property Tax Pass-Through to Renters: A Quasi-Experimental Approach

I use a unique, quasi-experimental setting to estimate the pass-through of idiosyncratic property tax shocks to renters in Berkeley, California. Proposition 13 (1978) in California limits annual tax reassessment of property to 2% per year unless the property is sold. When sold, a large, semi-random increase in property taxes occurs, with the amount of the change largely depending on the amount of time since the property was last sold. This creates very large discrepancies in property tax liability among otherwise similar rental units. By comparing changes in market-level rents for units in similar sold versus unsold buildings, I am able to directly estimate the effect of an idiosyncratic cost shock on rent. My novel dataset of unit-level rents covers the vast majority of renters in Berkeley from 1996-2022. I estimate an elasticity of rent to taxable property value of 0.044-0.08, which equates to tax shifting of $0.50-$0.89 per $1. This finding can be theoretically justified by reference-dependent landlords and downward sloping demand for rental housing. The results are robust to differences in pricing behavior by landlord size, improvements to unit quality, and the inclusion of a property's most recent purchase price.

Event Study of Log Rent Before and After Sale

I Just Want The Best (Tax Rate) For My Child: Behavioral Responses to Inheritance Tax Changes

With Kristy Kim

This paper investigates the behavioral response to a change in taxation of inherited property in California. California has a unique property tax system in which property is taxed on its year-of-purchase value (or 'base value') plus 2% per year, instead of its current market value. As home values typically appreciate much faster than 2% per year, property taxes do not increase proportionally with home value. This system allows homeowners to benefit from lower real tax rates the longer they remain in the same home. Previously, a homeowner's child could inherit a home's base value---and thus preferential property tax rate---along with the home. Voters revoked this tax advantage by referendum in November 2020, but the change in policy did not take effect until four months later. This created a transition window during which many parents rushed to transfer their home to their child to preserve a low property tax rate for their child. We document parent-child transfers before and during this window for two populous counties in California: San Francisco and Los Angeles County. We find a large behavioral response to the policy change, with an additional 18 months of transfers occurring during the transition window in San Francisco County and an additional 11 months of transfers in Los Angeles County. The corresponding elasticity is e=614 in San Francisco County and e=106 in Los Angeles County. We disaggregate responses by Census tract income, and find that the top income decile in each county is the most elastic group (e=734, e=490 for SF, LA). These estimates are much larger than what is typically found in the inheritance tax literature (e=[0.1,0.2]), and challenge the "control of wealth" theory of bequest planning. The magnitude of the short-run behavioral response cautions against transition windows for tax policy changes, as they can 1) delay expected government revenue sources and 2) exacerbate inequality in taxation if high-income households are more responsive than other groups.

Parent-to-Child Home Transfers in San Francisco County, Pre- and Post-Reform, Principal and Non-Principal Residences

Five Facts About (Rental) Prices

With Caleb Wroblewski

We document some new facts about rental prices in a novel, near-universal, unit-level dataset from Berkeley, California. First, we document strong growth in rent for new tenants over time, with very few periods of price stagnation. Second, we document that rents are sticky, with nearly half of landlords choosing to keep rent the same when the unit turns over within two years. Third, rents are downwardly rigid, but we show that more price decreases occur in non-expansions. Fourth, we show that there is predictable seasonal demand for rental housing, with different price distributions for peak versus off-peak housing searches. Finally, we show that landlord size does not differentially impact rent-setting in a significant way.

Density of Nominal Rent Changes, Sold Buildings (Landlord Change) Versus Unsold Buildings (No Landlord Change)