Essays in Public Finance and Environmental Policy

Abstract

This thesis consists of three chapters in environmental economics and public finance. The first chapter tests for adverse selection on private information in the market for flood insurance. Using detailed flood insurance policy microdata and newly available estimates of flood risk, I develop a measure of excess flood risk not used by the insurer to price insurance policies. I then use this measure as the basis for an unused observables test described in Finkelstein and Poterba (2014) and find that the market in certain areas that FEMA categorizes as low-risk, which together make up 40% of all flood insurance policies, is adversely selected. My results hold even after conditioning on a rich set of demographic and housing characteristics and measures of flood risk information and salience, suggesting that at least part of the selection is directly risk-based.

The second chapter studies the distributional properties of federal disaster aid and how barriers to take-up vary across the income distribution. Using applicant-level microdata from FEMA’s primary emergency relief program from 2002-2019, I find that lower-income households receive more aid in expectation, both conditional on applying and conditional on receiving aid. This is consistent with the idea that lower-income households experience relatively greater uninsured necessary expenses following disasters. Despite this, I also find suggestive evidence that the non-monetary costs of applying for aid are highest among lower-income households, and estimate that program take-up sharply declines among the poorest households. Access to disaster recovery field offices, which provide assistance with filing disaster aid applications, significantly increases take-up among lower-income households but not higher-income households, implying that they increase the targeting efficiency of aid along the income distribution.

The third chapter, written with Faraz Hayat, investigates whether firms’ input decisions are sensitive to the value-added tax (VAT) they pay on purchases. The study uses rich VAT return data from Pakistan combined with variation in VAT rates on electricity from a quasi-experiment. In contrast to the standard theory that intermediate VATs have no impact on production decisions, we find that electricity demand is responsive to VAT rates. Our results are not driven by changes in tax evasion in response to changing VAT rates. Rather, we provide evidence that frictions in the VAT refund system explain our results, as many of the firms we study collect no VAT on their output and therefore rely on the government to reimburse their VAT credits.