In this week’s Demo Day, I shared my paper published at the Center for Growth and Opportunity in June. “How does targeted cash assistance affect incentives to work?” analyzed a program Mayor Sumbul Siddiqui proposed in Cambridge, Massachusetts to provide $500 per month for 18 months to all families with dependents and income below 200% of the poverty line.

Targeted programs like these are common in guaranteed income pilots, and in some enacted policies, and I find that it would cost-effectively reduce poverty: if expanded to Massachusetts, it would cost $1.2 billion per year and cut child poverty 42%.

However, that targeting comes at a cost. Using the OpenFisca US microsimulation model (supported by the Center for Growth and Opportunity and cataloged by the Policy Simulation Library), I find that the program would deepen an existing welfare cliff at 200% of the poverty line. For example, a family of four would lose over $19,000 total—$9,000 from the cash assistance and $10,000 from other benefits—once they earn a dollar above 200% of the poverty line (about $55,000). To recover those lost benefits, they would have to earn an additional $26,000, a range I call the “earnings dead zone”.

My presentation reviews these trends in both slides and the PolicyEngine US app for computing the impacts of tax and benefit policy. For example, I show how repealing the SNAP emergency allotment would smooth out welfare cliffs, while reducing resources available to low-income families, and how a universal child allowance avoids work disincentives while less cost-effectively reducing poverty.

Policymakers face trade-offs between equity and efficiency, and typically labor supply responses consider marginal tax rates. With their infinite marginal tax rates, welfare cliffs are a less explored area, even though they surface in several parts of the tax and benefit system. This paper makes a start, but more research is yet to be done.