Optimal Contract Design in the Wild: Rigidity and Control in Labor Contracts

Abstract : We document determinants of incompleteness, rigidity, and delegation in union contracts using a new corpus of 30,000 collective bargaining agreements from Canada from 1986 through 2015. Using ideas and methods from computational linguistics, we extract measures of rigidity and worker control from the text of the contract clauses. We then analyze how rigidity and authority in contracts varies according to firm-level factors and external factors. We document that contracts impose obligations equally on firms and workers but give entitlements mostly to workers. Worker entitlements have increased as a share of contract clauses over the last forty years. An increase in personal income tax rates is associated with an increase in worker entitlements, consistent with a substitution effect away from taxed compensation (income) and toward untaxed compensation (amenities). Control of province government by the labor-supporting New Democratic Party is associated with higher worker authority, consistent with higher bargaining power for workers due to political support. We further document a role for contracts as reference points as proposed by Hart and Moore (2008): negative wage shocks due to low COLA adjustments mis-predicting inflation are associated with higher strike rates and strike intensity, consistent with conflict due to frustrated worker expectations relative to a reference point. However, this wage-strike effect is attenuated by contracts with higher worker authority, consistent with a better-managed relationship.

---

Men in Pink-collar Jobs: Evidence from a Recruitment Experiment

Abstract : Despite the decrease in occupational segregation by gender since the 1970s, men’s share in historically female jobs has been fairly constant between 1970 and 2009 in developed countries. In this paper, I ask to what extent firms’ personnel recruitment strategies can impact the historical inertia in the selection into female-dominated sectors and attract qualified men to these jobs. Do recruitment messages containing cues against or aligned with the job stereotype affect adult occupational search and entry into a female-dominated job? If so, through which channels and for which types of people? I provide both a theoretical framework and experimental evidence to answer these questions. Theoretically, I study the strategic interaction between an employer and potential employees and generate predictions on the potential quality/diversity trade-offs involved in attracting minority workers through targeted advertising (i.e. men). I bring the theory to the field and perform two distinct experiments in collaboration with a UK organization, which recruits workers in social care on a national scale. Both experiments introduce exogenous variation to the content of recruitment materials. The first experiment explores whether recruitment materials which portray counter-stereotypical workers encourage information gathering by people who represent a minority in the occupation. Results indicate that men show a significant positive reaction to contents portraying other men. The second experiment explores whether the effectiveness of cues against or aligned with the job stereotype depends on people’s performance expectations. People interested in applying were randomly assigned to different registration emails that manipulated 1) salience on gender of past workers and 2) expected performance. Results show that performance expectations and gender salience interact in similar ways for both men and women, triggering 4% to 6% more applications.

---

Fostering Productive Interactions: Incentives, Processes, and Randomly Paired Employees

Abstract : In a firm-based field experiment, we randomly assigned over 650 agents to 4 different managerial treatments alongside an off-site hold-out group. The treatments include a $100 team incentive, structured face to face interactions, both the team incentive and structured interaction, and two control groups (one internal and one external). We find productivity gains of 14% with team incentives, 25% with structured interactions, and 27% with both. These estimates are similar when contrasted with the internal or external control group and in levels and differences with a pre-period. We find the performance gains are persistent for the structured interactions and fade out for the team incentives---underscoring the importance of management led processes, rather than incentives alone, to capture the benefits of peer interactions. The productivity gains for agents that received team incentives and structured interactions are submodular, suggesting these practices may have redundancies or crowd each other out. We also find evidence in support of these treatments inducing knowledge transfer and increased sales effort, albeit, in different proportions.

---