What makes a worker productive? There is growing evidence that worker productivity may actually be quite contagious. Research by Falk and Ichino (2006), Mas and Moretti (2009), Bandiera et al. (2010), and De Grip and Sauermann (2012) all demonstrates that co-workers can exert economically significant effects on their peers via channels often not explicitly created by their firms.
The mechanisms that most researchers have in mind when discussing peer effects in worker productivity fall into two broad categories: work place norms, which includes mechanisms such as peer pressure and conformist behaviour; and worker complementarities, which includes mechanisms like knowledge spillovers, collaborative efforts, and/or task specialisation.
Our reading of the literature leads us to conclude that there is strong evidence supporting the idea that there exist meaningful peer effects in worker productivity among workers performing low-skilled tasks, and that these effects arise primarily from conformist behaviour or peer pressure. On the other hand, the evidence of peer effects among high-skilled tasks such as teaching or research is mixed (Jackson and Bruegeman 2009, Waldinger 2012). Evidence of peer effects among high-skilled workers is typically attributed to knowledge spillovers.
Peer effects in the workplace: Social networks approach
In a recent paper (Lindquist et al. 2015), we contribute to the discussion on peer effects in the workplace by applying new methods from the social networks literature. Our network approach allows us to address a broad set of important questions including several that have not been analysed before. Does co-worker productivity affect worker productivity? Do productivity increases from on-the-job training spread from trained workers to their untrained co-workers? How does the structure of co-worker networks enhance or impede the spread of productivity through the network? Is it possible to identify key workers in the firm? If so, how can such information be used to aide personnel managers when making important decisions about who to train and who to retain? Can we use information about co-worker networks to help firms organise work hours and teams in a more optimal fashion or to improve upon the design of their training programmes?
We study these questions using a high-quality data set from an in-house call centre of a multi-national mobile network operator that covers a period of two years. The data include detailed information on the performance of individual workers, their characteristics, their team affiliation, and the exact times that they punch in and out of work. These data allow us to create co-worker networks for each week, where a weighted link between two agents indicates the amount of time they have been working together on the same team during a week. That is, we have very precise knowledge of the co-workers that a worker is exposed to during the week, together with the intensity of this exposure. Unlike most network papers, we do not identify exposure to peers off of the stable part of the network, since the stable part may be prone to non-random sorting and to common shocks. We show that co-worker networks are as random after conditioning on team, week and individual fixed effects. In this setup, every worker receives a unique, exogenously varying dose of co-worker productivity each week due to worker turnover, due to changes in the scheduling needs of the company, and due to idiosyncratic changes in one's own work schedule and the work schedules of teammates.
We present a formal model of worker productivity that includes both local average network effects and local aggregate network effects. The local average effect represents the role of social work norms (e.g. conformist behaviour or peer pressure), while the local aggregate effect represents strategic complementarities (e.g. knowledge spillovers). As mentioned above, these are the two main mechanisms that are typically put forward to explain productivity spillovers among workers. We use this model to guide the empirical part of our paper, and to test which of the two mechanisms is most relevant in our particular setting. The model also allows us to identify ‘key workers’. Key workers are not necessarily the most productive workers, but workers who, once removed from the network, reduce total productivity the most. They are therefore the workers who are ‘crucial’ to the firm.
We run two different regression experiments. We first use our exogenous network exposure matrix to study the effect of co-workers’ productivity on own productivity.
- We show that there are indeed strong network effects in worker productivity.
- A 10% increase in the current productivity of a worker's co-worker network leads to a 1.7% increase in own current productivity.
- We attribute this productivity spillover to conformist behaviour.
Low tenure workers react particularly strong to the work norms of their co-worker network. The estimates are close to those found in earlier studies, which were around 1.4-1.5% (Falk and Ichino 2006, Mas and Moretti 2009; see also Herbst and Mas 2015).
In our second regression experiment, we exploit data from a field experiment with random assignment of workers to a one week on-the-job training programme to analyse how exogenous changes in worker productivity due to on-the-job training affect co-worker productivity, including non-trained workers.
- We show that being exposed to trained workers increases the productivity of non-trained workers.
- Adding one trained co-worker to a worker's network increases the worker's own productivity by 0.7%.
- We show that this effect is driven by the local aggregate effect (i.e., knowledge spillovers), and not by the local average effect (i.e., conformist behaviour).
We also demonstrate how our network model of worker productivity can be used to inform personnel policy and to increase firm productivity. Which workers should be retained by the firm? Who should be trained? We compare the optimal policy when considering network effects, with the alternative policy being that the firm does not take network effects into account.
Imagine, for example, that we ask a personnel manager to pick out 10 workers that she feels the company should work the hardest to retain. One reasonable strategy would be to pick out the 10 workers with the highest average productivity. Instead, our (model-based) strategy would be to pick out the 10 workers with the highest average network ‘inter-centrality’ measures, i.e. the key workers according to our model. Our choice of such workers depends on a worker's own productivity and on the network effect (productivity spillover effect) that this worker generates. Each worker affects her co-workers to a different extent due to her own characteristics and due to her unique position in the network of co-workers.
This result also has implications for the optimal design of shift schedules and work teams. ‘Bridge workers’, i.e. workers who work more than one shift during the day, play an important role because they are necessary to facilitate the spread of network effects across shifts.
After picking our 10 key workers, we compare our workers to the 10 most individually productive workers chosen by the firm’s personnel manager. Despite a strong correlation between our measure of inter-centrality and individual productivity (0.78), there are only 3 workers that are on both of our lists. According to our model, the total productivity loss incurred by losing our 10 key workers is 22% higher than the loss incurred by the 10 workers picked using the naive strategy based solely on a worker's own productivity. In other words, our 10 key workers have a much higher overall value to the firm than the 10 workers with the highest average individual productivity. This is because these particular workers generate large positive externalities.
We also analyse the choice of who the firm should train using our network approach. We show that the firm may have been able to improve on the returns to training by more strategically assigning workers to the training, i.e. by considering the unique position that each worker has in her own co-worker network.
We also examine the effect on worker productivity of changes to the structure of co-worker networks. Our policy conclusions concerning the optimal structure of co-worker networks can be summarised as follows:
- The firms that we study should increase team size;
- They should hire more ‘bridge’ workers; and
- They should increase the average between-ness (i.e. increase connections between workers) in each network.
Our research indicates that worker productivity is indeed contagious. The presence of such network effects impact the answer to a wide variety of policy questions faced by personnel managers on a daily basis. Our hope is that the literature on social networks will expand more vigorously into the field of personnel economics. Given the detailed data often available in firms, and the importance of peer effects in the workplace, we believe that this field is particular suited for network methods and models.
Bandiera, O, I Barankay, and I Rasul (2010), “Social incentives in the workplace,” Review of Economic Studies 77, 417-459.
De Grip, A and J Sauermann (2012), “The effects of training on own and co-worker productivity: Evidence from a field experiment,” Economic Journal 122, 376-399.
Herbst, D and A Mas (2015), “Peer effects on worker output in the labouratory generalize to the field,” Science 350, 545-549.
Jackson, C Kirabo, and E Bruegmann (2009), “Teaching students and teaching each other: The importance of peer learning for teachers,” American Economic Journal: Applied Economics 1, 85-108.
Lindquist, M, J Sauermann, and Y Zenou (2015) “Network Effects on Worker Productivity,” CEPR Discussion Paper No. 10928.
Falk, A and A Ichino (2006), “Clean evidence on peer effects,” Journal of Labor Economics 24, 39-57.
Mas, A and E Moretti (2009), “Peers at work,” American Economic Review 99, 112-145.
Waldinger, F (2012) “Peer effects in science - Evidence from the dismissal of scientists in Nazi Germany,” Review of Economic Studies 79, 838-861.