Research

This paper introduces VRscores, a workplace-level measure of employee partisanship constructed by linking U.S. voter registrations to electronically available profiles of workers covering 2012 to 2024. The organizational-level dataset measures the partisanship of 24.5 million workers across more than 534,000 employers with at least five employees in our data. We release this employer-level dataset as well as additional datasets that report VRscores at the firm, occupation, industry, and MSA levels. We show that our data cover substantially more employees and organizations than donation-based approaches to measuring political ideology. We also show that VRscores are more representative of employees in terms of partisanship, seniority, occupation, and industry. VRscores and donation-based measures are only moderately correlated (r = 0.51) and differ in their dichotomous classification (Democratic vs. Republican) of a given workplace for one in five companies.

We estimate the extent of workplace political segregation in the United States by merging data covering over 45 million workers. We present four main findings. First, partisans are segregated by workplace. The average Democrat's coworkers are 11.7 percentage points (95% CI: [10.5, 12.9]) more Democratic than the average Republican's. After controlling for geography, industry, and occupation, segregation is 2.9 p.p. [2.5, 3.3], comparable to analogously estimated gender segregation (2.8 p.p. [2.4, 3.2]). Second, segregation is largest among the politically active (donors: 14.8 p.p. [13.2, 16.4] vs. non-donors: 11.6 p.p. [10.0, 12.0]) and those with more market power (senior executives: 14.7 p.p. [13.3, 16.1]). Third, Republicans experience higher exposure to Democrats than vice versa: the median Republican's coworkers are 50% Democratic versus 32% Republican for the median Democrat. Fourth, political segregation has changed little over time (2012: 11.1 p.p. [9.9, 12.3] vs. 2024: 11.0 p.p. [10.0, 12.0]).

In response to the Trump trade war, China, the EU, and other countries enacted politically-targeted trade retaliation (PTTR) against swing states and Republican strongholds in the United States. We argue that PTTR increases public concerns about foreign election interference and assess the effects of such retaliation across partisan affiliations. We test our predictions using a national survey experiment in the United States fielded before the 2020 election. In contrast to findings about sanctions and foreign endorsements, we find strong evidence that PTTR increases fears of election interference among both Republicans and Democrats. Partisan double standards in reaction to PTTR were strongest for retaliation targeting swing states and smaller for retaliation targeting the President's base. Overall, the evidence shows that economic policies which are not primarily intended to influence elections may nevertheless come to be viewed by the public as foreign election interference.

Organizational cultures shape not only workplace outcomes but also civic life. While prior work has largely conceptualized civic culture within geographic units (e.g., country, state, or city), we introduce and evaluate the concept of organizational civic culture—the norms and values within organizations of participation in democratic institutions beyond the workplace. Leveraging a novel dataset containing employment histories and voter turnout records for 28 million Americans, along with a quasi-experimental mover design, we find plausibly causal evidence that coworker voter turnout increases a focal worker's turnout. The influence of organizational civic culture is especially pronounced in low-turnout elections, among workers who are less civically engaged, among college-educated workers, and in instances where more coworkers share a focal worker's partisanship. In supplementary analysis, we show that these effects spill into workers' households, suggesting that civic habits acquired at work diffuse to other corners of social life. Overall, our findings identify organizations and employment as previously overlooked but important scaffolds to democratic institutions.

We examine how corporate sociopolitical stance-taking affects employee turnover. While prior studies suggest asymmetrically negative reactions from stakeholders who disagree with corporate stances, they have tended to examine low-cost behaviors. We argue that because leaving one's job is exceptionally costly, positive retention effects will instead outweigh negative attrition effects. We test this following North Carolina's 2016 "bathroom bill," which required transgender individuals to use bathrooms corresponding to the sex on their birth certificate. We link millions of employment histories to voter files, draw on survey data to impute attitudes on transgender issues, and estimate turnover effects using matched difference-in-differences. Companies publicly opposing the bill experienced increased retention in North Carolina. Consistent with our mechanism, retention effects were larger for employees who likely agreed with the stance and among young workers, for whom politically-motivated career decisions are less costly. While stance-taking may improve retention, it may also contribute to political segregation across workplaces with potential downstream consequences for organizations and society.

Where you work affects whether you vote. Using linked employment and voter records for 30 million U.S. workers across 377K employers, we estimate "Employer Turnout Effects" (ETEs)—organization-specific effects on voter turnout derived from a movers design that accounts for worker self-selection. ETEs have a bias-corrected interquartile range of 5.6 percentage points, comparable to the largest single-contact interventions in the voter mobilization literature. When workers change jobs, their turnout shifts in the direction predicted by the new employer's ETE, with asymmetric effects suggesting that turnout gains may persist after subsequent job changes. ETEs vary across industries (accommodation and food service show the lowest; white-collar sectors the highest), though within-industry variation far exceeds between-industry differences. Companies that publicly commit to providing time to vote have significantly higher ETEs. These findings identify the workplace as an overlooked but powerful lever for democratic participation.

Can firms shape the political behavior of their employees in support of their strategic goals? The workplace is a political sphere where employees are routinely exposed to political messages, including guidance on their employers' preferred political outcomes. This has led to claims that firms are “political machines” that can mobilize employees to turn out to vote in support of their political goals. Linking millions of LinkedIn worker profiles with administrative voter records and their company's campaign finance disclosures, I find that the potential of firms to mobilize employees to turn out to vote in strategically important elections is real but contextually limited. In state and local elections, employees at firms engaged in election-related corporate political activity turn out to vote at 1.86 percentage points higher rates. However, mobilization effects are dependent upon the timing of elections, the extent of shared material stakes between firms and employees, and whether elections are partisan. Turnout effects are smallest in high-turnout presidential elections and larger in lower-salience off-year and off-cycle elections, and they are larger for ballot questions with high material stakes for employees. While I find effects in non-partisan ballot question races, I do not find evidence that firms' support for specific partisan candidates can affect turnout in races for the U.S. House of Representatives. These results suggest that the political mobilization of employees constitutes a real, but contextually bounded strategic resource for firms' pursuit of strategic goals.

A growing body of research shows that corporate sociopolitical stance-taking can lead to polarized reactions from stakeholders. Thus far, research has focused on issues characterized by ideological (or partisan) divides. While these studies have emphasized the role of ideology in driving polarized stakeholder reactions, there is little evidence of whether similar patterns of polarization can emerge on issues which feature less clear ideological (or partisan) divides. We address this gap by studying consumer reactions to real-world corporate responses to a set of issues which include the Israel-Palestine conflict. While Israel-Palestine is a highly charged issue that divides Americans, divisions do not fall along clearly partisan or ideological lines. We use an incentive-compatible survey experiment which uses a naturalistic format to inform respondents about pro-Israel or pro-Palestinian stances taken by different parts of the McDonald's system (i.e., local franchises) before using an incentive-compatible mechanism to measure respondents' willingness-to-pay for a McDonald's gift card. We find that consumers consistently react negatively to stances they oppose regardless of whether the issues are aligned with ideological divisions (e.g., stances on Black Lives Matter, LGBT rights, and climate change) or issues are less directly aligned with an ideology or partisan camp (e.g., stances on Israel-Palestine). Consistent with prior work, we show that consumers penalize companies for taking misaligned stances, but do not reward them for taking aligned ones. We also demonstrate that these changes are almost entirely manifested as decreases along the extensive margin (i.e., boycotting) rather than along the intensive margin (i.e., continued consumption but reduced willingness-to-pay). Our findings suggest that polarized stakeholder reactions can occur across a range of issue areas and need not be solely attributable to ideological divides.