Matthew Merrigan

Public corporate political stances have become a prominent — and often explicitly partisan — form of engagement, yet we know little about their political spillovers among employees. This paper studies whether stance-taking affects employee campaign contributions, distinguishing between donations to outside political actors and employer-linked channels such as corporate political action committees (PACs). Using hand-collected stance events linked to administrative records on employment and itemized giving, I first estimate a stacked difference-in-differences design with industry- and year-matched controls. Corporate stances increase employee giving on both the intensive and extensive margins, with effects concentrated among ideologically aligned employees. To sharpen inference about mechanisms and moderation — especially for institutionally mediated PAC giving — I field a survey experiment that randomizes employer stance direction across three issue domains and measure behavior using an incentive-compatible $1.00 donation-allocation task. Experimental results mirror the field evidence: aligned respondents increase PAC giving, while misaligned respondents exhibit backlash via increased co-partisan giving. Substantively, the findings reposition the firm as an intermediary institution in democratic politics, shaping not only whether employees participate but where political money flows through organizational channels, with downstream implications for polarization, representation, and firms' role in contemporary democracy.

Businesses spend far less on campaign contributions than the law allows, a puzzle Li (2018) explains through internal constraint: access-seeking PACs must raise funds through voluntary donations from their sponsor firms' employees, and employees withhold donations when PACs give to the out-party. Li identifies this mechanism on observed donors. I revisit the claim using an original panel of 205,522 individual-PAC-cycle observations linking employment records, voter registration, and DIME contribution histories across 2012–2022 — the full eligible-employee population rather than only prior donors. Li's within-donor mechanism holds on ever-donors but does not extend to the broader sample. A specification exploiting cross-individual variation identifies a precisely estimated extensive-margin response: co-partisan employees are substantially more likely to enter donation than out-partisan ones. Mechanisms identified on behaviorally-selected samples — donors, voters, activists — often fail to extend to the populations they are meant to characterize. Population-scale claims require population-scale data.