There is a substantial gap between the promise and reality of artificial intelligence in human resource (HR) management. This article identifies four challenges in using data science techniques for HR tasks: complexity of HR phenomena, constraints imposed by small data sets, accountability questions associated with fairness and other ethical and legal constraints, and possible adverse employee reactions to management decisions via data-based algorithms. It then proposes practical responses to these challenges based on three overlapping principles—causal reasoning, randomization and experiments, and employee contribution—that would be both economically efficient and socially appropriate for using data science in the management of employees.

In 1973 Granovetter formulated the strength-of-weak-ties hypothesis (SWT), which became the foundation of a vast sociological literature on social networks in labor markets. Until now, SWT has never been directly tested but treated instead as a surrogate for the relationship between an actor's network and labor market outcomes such as characteristics of a job obtained. The paper restates SWT as a proposition about the probability of getting a job as a function of within-actor differences in tie strength and tests it with data on hires carried out in one Russian city in 1998. In support of SWT, the results show that a worker is more likely to get a job through one of her weak ties rather than strong ties. The advantages of weak ties lie in their abilities to provide timely access to non-redundant information and to influence employers directly. In contrast, strong ties are associated with indirect influence on employers through well-connected intermediaries. The estimates come from a within-worker fixed-effect conditional logistic regression and thereby provide rare evidence of an association between information and influence transferred through social ties and labor market outcomes, independent of workers' individual characteristics.

Although the existing theory predicts that a referral's chances of being hired increase with the job performance of the referrer, no empirical evidence is available to support this claim. To address this discrepancy, we decompose the recruitment process into objective selection, subjective selection, and self-selection and theorize that the likelihood of passing a particular recruitment stage increases with the performance of the referrer under objective selection and self-selection, but remains undetermined at a stage of subjective selection. Our analysis of unique comprehensive data on online recruitment of sales agents in a virtual call center supports these arguments. The effectiveness of personnel as a recruitment channel varies with the type of the recruitment stage and performance of the referrer. When the firm evaluates candidates by an objective criterion, the advantage of a referral increases with the performance of his or her referrer; those referred by relatively high-performing workers are significantly better than the applicants who learned about the job from Internet ads. When job candidates self-select into the next stage of the online application process, the referral of any agent is more likely to continue than a nonreferral, and this likelihood increases with the performance of the referrer. On a subjective stage, the outcome is contingent on the intricacies of the recruitment process. In our case, an applicant's chances of being hired increase with the performance of his or her referrer because the firm rejects the referrals of low-performing workers at a higher rate than it does nonreferrals, while it treats equally the referrals of high-performing workers and nonreferrals. The study's contributions to the literature on social networks in labor markets are discussed.

Price is a central analytic concept in both neoclassical and old institutional economics. Combining the social network perspective with old and new institutionalist approaches to price formation, this article examines technological, economic, institutional, and political factors that shaped the earliest pricing systems for electricity used in the United States, between 1882 and 1910. We show that certain characteristics of electricity supply led to ambiguities in how the product should be priced, which created a politics of pricing among electricity producers. In particular, we investigate why the "Wright system," arguably inferior in productive efficiency to other alternatives, was widely adopted by 1900. We argue that this outcome resulted in part from the political and organizational clout of its supporters, as well as from their particular conceptions of the boundaries and future of the industry itself. The Wright system best suited the "growth dynamic" strategy promoted by the managers of large central stations in their fierce competition with smaller and more decentralized installations. Thus, even in this apparently highly technical and mainly economic issue of how to price the product, there was ample room for social construction and political manipulation. The outcome reached was by no means inevitable and had a highly significant impact on the shape of the American industrial infrastructure.