Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/210754 
Year of Publication: 
2019
Series/Report no.: 
Staff Report No. 902
Publisher: 
Federal Reserve Bank of New York, New York, NY
Abstract: 
We estimate the distribution of marginal propensities to consume (MPCs) using a new approach based on the fuzzy C-means algorithm (Dunn 1973; Bezdek 1981). The algorithm generalizes the K-means methodology of Bonhomme and Manresa (2015) to allow for uncertain group assignment and to recover unobserved heterogeneous effects in cross-sectional and short panel data. We extend the fuzzy C-means approach from the cluster means case to a fully general regression setting and derive asymptotic properties of the corresponding estimators by showing that the problem admits a generalized method of moments (GMM) formulation. We apply the estimator to the 2008 tax rebate and household consumption data, exploiting the randomized timing of disbursements. We find a considerable degree of heterogeneity in MPCs, which varies by consumption good, and provide evidence on their observable determinants, without requiring ex ante assumptions about such relationships. Our aggregated heterogeneous results suggest that the partial equilibrium consumption response to the stimulus was twice as large as what is implied by homogeneous estimates.
Subjects: 
marginal propensity to consume
consumption
tax rebate
heterogeneous treatment effects
machine learning
clustering
C-means
K-means
JEL: 
D12
D91
E21
E32
E62
Document Type: 
Working Paper

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.