A0670
Title: An instrumental variable method for point processes: generalized Wald estimation based on deconvolution
Authors: Shizhe Chen - University of California, Davis (United States) [presenting]
Zhichao Jiang - Harvard University (United States)
Peng Ding - University of California, Berkeley (United States)
Abstract: Point processes are probabilistic tools for modelling event data. While there is fast-growing literature studying the relationships between point processes, how such relationships connect to causal effects remains unexplored. In the presence of unmeasured confounders, parameters from point process models do not necessarily have causal interpretations. An instrumental variable method for causal inference is proposed with point process treatment and the outcome. Causal quantities based on potential outcomes are defined, and nonparametric identification results with a binary instrumental variable are established. The traditional Wald estimation is extended to deal with point process treatment and outcome, showing that it should be performed after a Fourier transform of the intention-to-treat effects on the treatment and outcome. Thus, it takes the form of deconvolution. This is termed the generalized Wald estimation, and an estimation strategy based on well-established deconvolution methods is proposed.
