BUSI 642 Week 4 Human Resource Policy Project: Annotated Bibliography Example

Four Policies for a Changing Workforce: An Annotated Bibliography on Hybrid Work, Pay Transparency, AI in Hiring and Employee Mental Health

Purpose and Scope

This bibliography supports a human resource policy manual for a composite civil engineering and construction firm of about 1,200 employees in Virginia and North Carolina, where I work in human resources. The four topics are hybrid work arrangements for the firm's roughly 500 office employees, pay transparency, the use of artificial intelligence in recruiting and selection and support for employee mental health. I searched Business Source Complete, ABI/INFORM and PsycINFO through the Jerry Falwell Library, limited to peer-reviewed articles published from 2021 onward, and chose sources that measure outcomes or synthesize research rather than describe practice.

Annotated Sources

Baker, M., Halberstam, Y., Kroft, K., Mas, A., & Messacar, D. (2023). Pay transparency and the gender gap. American Economic Journal: Applied Economics, 15(2), 157-183. https://doi.org/10.1257/app.20210141

Baker et al. (2023) studied Canadian laws that made public the salaries of university faculty whose pay exceeded a set threshold. Because the laws took effect in different provinces at different times and affected institutions and departments differently, the authors could compare salary changes where the laws applied with those where they did not, using administrative data covering most faculty in the country. They found that the disclosure laws reduced the gender pay gap by roughly 20 to 40 percent. The study's design is strong, an event study exploiting staggered adoption, but its setting is narrow: unionized public universities with structured pay scales differ from a private construction firm. Still, it is one of the clearest pieces of causal evidence that making pay visible can narrow unexplained gaps between men and women. I will use it to support the equity rationale for the pay transparency policy and to argue that publishing salary ranges by job grade, rather than individual salaries, captures much of the benefit with less intrusion on privacy.

Bloom, N., Han, R., & Liang, J. (2024). Hybrid working from home improves retention without damaging performance. Nature, 630(8018), 920-925. https://doi.org/10.1038/s41586-024-07500-2

This randomized controlled trial assigned 1,612 graduate employees of a large Chinese technology company either to a schedule with two home days each week or to full-time office work for six months in 2021 and 2022. Bloom et al. (2024) found that the hybrid group reported higher job satisfaction and quit at a rate about one third lower, and the drop in resignations was concentrated among staff without management roles, female employees and those facing long trips to work. Performance reviews and promotions over the following two years did not differ between the groups, and managers' initially negative views of hybrid work became more positive by the end of the trial. Randomization makes this the strongest evidence among my sources. Its limitation for my purposes is the setting, a software firm whose engineers' output is easier to measure than a civil engineer's. It will serve as the main evidence for allowing two home days a week for eligible office roles and for the retention argument in the business case for the hybrid policy.

Choudhury, P., Makridis, C. A., Khanna, T., & Schirmann, K. (2026). Is hybrid work the best of both worlds? Evidence from a field experiment. The Review of Economics and Statistics, 108(4), 1134-1140. https://doi.org/10.1162/rest_a_01428

Choudhury et al. (2026) report a field experiment that varied how many days employees worked from the office rather than simply comparing remote and in-person work. Employees who spent around two days a week on site said their work and home lives were better balanced, liked their jobs more and felt less cut off from colleagues than people assigned more or fewer office days. Performance ratings did not differ across the conditions. The study complements Bloom et al. (2024) by addressing the question my firm's managers ask most often: not whether to allow hybrid work, but how many office days to require. Its strength is random assignment of the number of office days; its limitation is a short paper with a single organizational setting, so the precise optimum may differ elsewhere. I will use it to justify a two-anchor-day standard for office teams and to support the policy's provision that teams coordinate their office days so that collaboration happens when colleagues are present together.

Cullen, Z. (2024). Is pay transparency good? Journal of Economic Perspectives, 38(1), 153-180. https://doi.org/10.1257/jep.38.1.153

Cullen (2024) reviews the growing evidence on pay transparency policies, noting that 71 percent of OECD countries have adopted some form since 2000. The review distinguishes three types. Horizontal transparency, which reveals pay among coworkers doing similar work, has narrowed coworker gaps but can provoke unproductive comparisons and lead employers to bargain harder, lowering average wages. Vertical transparency, which shows pay across levels of seniority, can make employees' beliefs about their earning potential more accurate and optimistic, raising motivation. Cross-firm transparency, which reveals pay differences among employers, helps workers move to better-paying firms and negotiate higher pay. As a synthesis by a leading researcher in the field, this article is the best overview among my sources, though it summarizes others' studies rather than presenting new data. I will use it to structure the pay transparency policy around vertical and cross-firm elements, publishing salary ranges by grade and in job postings, while avoiding the drawbacks of disclosing each coworker's individual pay.

Cullen, Z. B., & Pakzad-Hurson, B. (2023). Equilibrium effects of pay transparency. Econometrica, 91(3), 765-802. https://doi.org/10.3982/ECTA19788

Cullen and Pakzad-Hurson (2023) ask how employers respond when workers can learn what coworkers earn. Their bargaining model predicts that transparency weakens each worker's individual bargaining power, because an employer that grants one worker a raise invites costly renegotiation with the others and can therefore credibly refuse. They test the prediction with an event study of U.S. state laws protecting private sector workers' right to discuss salaries with coworkers and find that average pay fell by roughly 2 percent once those laws took effect, with the smallest declines where workers already had little individual bargaining power, such as unionized settings. The study combines theory and evidence rigorously, although its outcome is average wages rather than equity within firms. For my policy manual, it is an important counterweight to Baker et al. (2023): it shows that transparency has costs as well as benefits. I will use it to argue for publishing ranges and criteria for pay decisions, which addresses fairness without inviting the bargaining dynamics the authors describe.

Fleming, W. J. (2024). Employee well-being outcomes from individual-level mental health interventions: Cross-sectional evidence from the United Kingdom. Industrial Relations Journal, 55(2), 162-182. https://doi.org/10.1111/irj.12418

Fleming (2024) used survey data from 46,336 workers in 233 UK organizations to compare employees who took part in common individual-level well-being interventions, including resilience training, mindfulness programs and well-being apps, with colleagues who did not. Across several measures of subjective well-being, participants were no better off than nonparticipants. Interpreting the results through the job demands-resources model, the author argues that such programs do not give workers resources that match the demands of their jobs, and that improving working conditions may matter more. The cross-sectional design is a real limitation, since people who seek help may already be struggling, and the author discusses this selection problem openly. Even so, the sample is very large and the result consistent across measures. I will use this source to argue that the firm's mental health policy should not rely on an app subscription and should instead address job demands, such as scheduling and workload during the summer construction season.

Folger, N., Brosi, P., Stumpf-Wollersheim, J., & Welpe, I. M. (2022). Applicant reactions to digital selection methods: A signaling perspective on innovativeness and procedural justice. Journal of Business and Psychology, 37(4), 735-757. https://doi.org/10.1007/s10869-021-09770-3

Folger et al. (2022) examined why digital selection methods sometimes make employers more attractive to applicants and sometimes less. Drawing on signaling theory, they proposed that digital methods send two signals, that the organization is innovative and that its procedures may be less fair, and tested the model with a vignette experiment among 475 potential applicants and a field study of 335 actual applicants. Both studies found that applicants saw digital methods as more innovative, which raised attractiveness. Potential applicants also judged digital interviews as less fair, but that negative effect was not significant among actual applicants. The use of two samples, one experimental and one from real hiring, strengthens the findings, though the methods studied were broader than AI screening alone. I will use this article to support the AI hiring policy's requirement that applicants be told when digital tools are used and how decisions are reviewed, since explaining the process can preserve fairness perceptions while keeping the innovativeness benefit.

Hunkenschroer, A. L., & Luetge, C. (2022). Ethics of AI-enabled recruiting and selection: A review and research agenda. Journal of Business Ethics, 178(4), 977-1007. https://doi.org/10.1007/s10551-022-05049-6

Hunkenschroer and Luetge (2022) systematically reviewed 51 articles on the ethics of using artificial intelligence in recruiting and selection, from writing job advertisements and screening résumés to analyzing video interviews. They mapped the ethical opportunities, such as reducing some forms of human bias and making screening more consistent, against the risks, including discrimination built into training data, lack of transparency, privacy concerns and the loss of human judgment, and they summarized proposed safeguards such as audits, explainability and human oversight. As a systematic review it offers breadth rather than new evidence, and the authors note that the field is young and many claims remain untested. For my policy manual, the review supplies the framework for the AI hiring policy: it identifies the specific risks the policy must address and the practical safeguards that researchers recommend. I will organize the policy's requirements, including vendor audits, applicant notice and human review of rejections, around the risk categories the review identifies.

Kelan, E. K. (2024). Algorithmic inclusion: Shaping the predictive algorithms of artificial intelligence in hiring. Human Resource Management Journal, 34(3), 694-707. https://doi.org/10.1111/1748-8583.12511

Kelan (2024) challenges the common claim that AI will remove human bias from hiring, arguing that algorithms may equally reproduce and amplify bias and embed it in technology. The article examines exclusion and inclusion in AI-supported hiring through three areas: the data used to train systems, including how organizational fit and demographic categories are encoded; the design process, including which groups dominate the teams building the tools; and decisions, including how humans and machines share them. It concludes that evidence is still lacking on how algorithms can be audited and regulated effectively for inclusion. This is a conceptual article rather than an empirical study, so its value lies in the questions it raises. For my firm, which is adopting a résumé screener from an outside vendor, the article identifies what to ask before signing: what data trained the model, how it defines a good candidate and who reviews its decisions. I will use it to support the policy's vendor due diligence section.

Wu, A., Roemer, E. C., Kent, K. B., Ballard, D. W., & Goetzel, R. Z. (2021). Organizational best practices supporting mental health in the workplace. Journal of Occupational and Environmental Medicine, 63(12), e925-e931. https://doi.org/10.1097/JOM.0000000000002407

Wu et al. (2021) conducted a narrative review of peer-reviewed research on workplace mental health to define criteria for a national award recognizing employers. They identified eight categories of best practice: culture, robust mental health benefits, mental health resources, workplace policies and practices, a healthy work environment, leadership support, outcomes measurement and innovation. The review is broad rather than deep, and because it was written to support an award, it describes promising practices without weighing the strength of evidence for each. Its value for my project is that it treats mental health as an organizational system rather than an individual program, which aligns with the finding of Fleming (2024) that individual-level programs alone show little benefit. I will use the eight categories as the outline of the mental health policy, with particular attention to leadership support, benefits access through the employee assistance program and measurement, since the firm currently tracks none of these.

Closing Note

Together, these ten sources cover all four policies with at least two each. The hybrid work policy rests on two experiments, Bloom et al. (2024) and Choudhury et al. (2026). The pay transparency policy draws on the review by Cullen (2024) and on the opposing findings of Baker et al. (2023) and Cullen and Pakzad-Hurson (2023). The AI hiring policy uses Hunkenschroer and Luetge (2022), Kelan (2024) and Folger et al. (2022). The mental health policy uses Wu et al. (2021) for structure and Fleming (2024) for caution about individual-level programs. The sources disagree in useful places, and the policy manual will be stronger for showing where it chose between them.