UC Berkeley

This dissertation comprises three essays on why workers and households in low-income settings underinvest in welfare-improving decisions, despite potentially high returns. A unifying answer emerges: frictions that concentrate costs at the moment of action impose large and regressive burdens on those least equipped to absorb them. Chapter 1 shows that liquidity constraints at the point of purchase suppress investment in worker-owned capital, the portable productive assets workers finance themselves but deploy on the job. A randomized controlled trial with a large Indian home-services platform firm finds that subsidizing the scooter down-payment increases adoption by 32 percentage points, raises worker weekly earnings by 13 percent, and generates a 153 percent firm return within the first year. A discrete-choice experiment confirms that adoption is 2.5 times more sensitive to the down-payment than to total price, isolating cash at hand as the binding constraint. Chapter 2 identifies the same purchase-time liquidity mechanism in India's LPG clean cooking fuel subsidy program, where a cash-back design inadvertently depresses adoption among the poor and worsens child health outcomes. Chapter 3 shows that extreme rainfall imposes sharply unequal income losses on urban gig workers, with peripheral, low infrastructure neighborhoods bearing losses 3.5 times larger than central ones, losses that platform flexibility cannot offset. Together, the three essays show that modest reforms to the timing and delivery of costs and transfers can substantially improve welfare for the poor, and that the expansion of flexible labor market arrangements, while increasing workers' ability to smooth income, does not eliminate the structural inequalities that leave low-income workers disproportionately exposed to economic and environmental shocks.