• (PhD job market paper) “Intrahousehold Resource Allocation in Collective Households with Large and Sparse Demand Systems," 2025, Revise and Resubmit at Journal of Business and Economic Statistics. Also available at SSRN 5261131, codes for empirical estimation and simulation (Python).

Abstract: Resource shares, defined as the share of total household expenditure attributed to each household member, are parameters of collective household models that speak to the within-household distribution of consumption. They are typically identified and estimated in the context of a small continuous demand system (often just 2 goods). However, modeling collective households using large and sparse demand systems remains underexplored. In this paper, I extend the collective household model to a random utility framework, following Lanier et al. (2023) to estimate sparse demand systems. I provide a model wherein resource shares may be identified using extremely high-dimensional data where most goods are not purchased in any given trip, such as the barcode-level scanner panels, and I provide a consistent and asymptotically normal estimator. The proposed estimation procedure leverages techniques from machine learning to improve computational feasibility and efficiency. Using data from the United States from 2004 to 2020 provided by NielsenIQ, I estimate resource shares for people in a variety of household compositions. The results indicate no significant gender inequality in resource shares between adults, and suggest that adult men's and women's resource shares are slightly positively related to the household budget.

• “Economies of Scale to Consumption in Collective Households," 2025, with Arthur Lewbel and Krishna Pendakur. Revise and Resubmit at Journal of Applied Econometrics. Download (revised Apr 2026), codes for empirical estimation (Python).

Abstract: Browning, Chiappori, and Lewbel (2013) model collective-household economies of scale in goods consumption by having each good be partly shared, instead of each good being public or private. We modify their model to achieve simple point identification and estimation of the economies of scale of consumption of each good, and we provide a new index of household level economies of scale. Our model has a linear form, permits households of varying composition, including children, and accommodates unobserved heterogeneity in both preferences and in the economies of scale of each good. We provide estimates using Canadian Survey of Household Spending data.

• “Robust Moment Restriction Based Inferences in Finite Samples under Heteroskedasticity of Unknown Form,” 2022, with Ta-Sheng Chou and Eric S. Lin. Download.

Abstract: For a long time, heteroskedasticity and nonlinear models have been essential topics in econometric studies. Recently, Lin and Chou (2012, 2018) proposed the quasi-hat matrix to extend conventional HCCME-type corrections pioneered by Eicker (1963, 1967) and White (1980) from OLS models to nonlinear GMM models. In this paper, we aim to provide an extended version of Lin and Chou’s (2012, 2018) model, which fits a framework of vector regression or conditional moment restrictions. We begin with the extension of the conventional HCCME-type corrections from a single-equation setting toward a multiple-equation setting. Next, we derive a generalized-hat matrix to extend the applicability of the HCCME-type corrections for a more general model with conditional moment restrictions. Through several Monte Carlo experiments, we find that the proposed method is able to improve hypothesis testing, estimates, and prediction, especially under heteroskedastic data with high leverage points. Finally, an empirical application of Heckman’s (1976, 1979) sample selection model is introduced to illustrate the empirical applicability of the proposed method.

• “Dimension Reduction in Consumer Demand Analysis with Scanner Data," with Ian Crawford, Krishna Pendakur, and Matthew Polisson.

• “What is the Best Way to Impute Shelf Prices in Barcode Level Consumer Scanner Data?" with Krishna Pendakur.