I am a PhD candidate at UCLA; I am graduating Spring 2026 and starting as an Assistant Professor at Pomona College this fall. My current work primarily involves solving and estimating heterogeneous
agent models of the macroeconomy to analyze fiscal and monetary policy. Before starting my degree at UCLA, I was a research analyst at the Federal Reserve
Bank of New York, where I worked on topics related to empirical trade economics, applied microeconomics, and household
finance.
When fiscal policy is active and monetary policy is passive in a heterogeneous agent New Keynesian (HANK) model, deficit-financed transfers to low-asset households lead to similar cumulative inflation but greater increases in real output than transfers to wealthier households. I use the inverse of the “Phillips multiplier”, the price level sacrifice ratio, to quantify this dynamic. Household heterogeneity and targeted policy change the timing of output gaps, making this consistent with the Phillips Curve and rendering conventional sacrifice ratio intuition misleading for assessing the inflation/output trade-off between policies.
During the inflationary surge of 2021-2023, inflation expectations were slow to rise -- but when they did, expected real interest rates fell as nominal rate expectations remained roughly constant. I estimate a medium-scale sticky-expectations heterogeneous agent New Keynesian (HANK) model and study the effect of fiscal transfers under a policy configuration that matches this expectations pattern. In particular, I simulate the deficit-financed transfers disbursed by the CARES Act of 2020 and the Consolidated Appropriations Act (CAA) and the American Rescue Plan Act (ARPA) of 2021 in a scenario wherein the central bank does not systematically respond to inflation with higher interest rates and the government does not commit to retire past deficits with future taxation. The simulated payments boost real annual GDP relative to trend by an average of 0.6\% per quarter over 2020Q1-2023Q4 with an overall transfer multiplier of 2.0 and reduces unemployment by an average of 1.5 percentage points in each quarter. However, they also induce a 4.2\% surge in the price level over the same period, with a peak annual inflation rate in 2022 of 2.8\% above trend (slightly less than half of the post-pandemic peak observed in the data). This contrasts with the highly muted response of unemployment and inflation to similar fiscal shocks in an environment wherein all agents believe the central bank will raise interest rates in response to inflation. Unlike the first setting, this latter environment yields an expectations pattern that differs strongly from the post-2020 expectations data.
Linearized full information rational expectations heterogeneous agent models can be easily converted into a sticky expectations environment, even when solved in state-space form. The technique recycles the Jacobians of the full information model with only a few modifications. The process is greatly simplified by working in continuous time, which facilitates the use of natural boundary conditions to ensure agents do not violate idiosyncratic borrowing constraints, and by a block recursive structure: one can solve for the behavior of agents with average expectations, and then solve for the behavior of agents who have rational expectations. Using this technique to solve a canonical HANK model, I show that if both households and firms have sticky expectations, then stimulus checks generate more output and much less inflation relative to a full information benchmark.
I hypothesize that cash transfers to poor households improve the mental health of recipient children. Specifically, I posit that the 1993 Omnibus Budget Reconciliation Act’s expansion of the Earned Income Tax Credit (EITC) could have worked through a number of mechanisms to reduce the incidence of depression, anxiety, and antisocial behavior among children in eligible households, as reported by broad survey indexes. To test this claim, I estimate the intent-to-treat (ITT) effect of the EITC using a difference-in-differences (DID) identification strategy, with linear controls and household and region fixed effects. I find evidence that the federal tax credit expansion reduced externalizing behavior and tendencies among low-income children.
