Firms increasingly collect data on their employees’ performance, a management practice that they can use to promote comparisons among staff and potentially boost productivity. At the same time, there is growing corporate emphasis on employee empowerment, teamwork, and happiness – again with the goal of improving overall performance.
Are these two objectives compatible? A new study investigates how the success of a management practice depends on the underlying values articulated by senior managers. The researchers report evidence on what happened when a large US transport company tried to combine performance management with changing the workplace culture.
The company was in the process of fitting its trucks with an ‘electronic on-board recorder’ to provide drivers with information on their driving performance. Senior managers had also introduced a multi-year ‘lean management’ program to remake the company’s internal operations, the first phase of which focused exclusively on changing values towards a greater emphasis on teamwork and empowerment.
In this setting, a natural question to ask is whether the optimal management practice consists of only letting each driver know their individual performance, or alternatively, providing drivers with information about their performance with respect to other drivers. Using the driver performance data, the researchers randomized these practices across the company’s worksites.
The main result of their experiment is that the second option leads to better performance than the first option in a particular worksite if and only if it has not yet received the values intervention. Driver performance is worse in a worksite where the values intervention has taken place. This finding is consistent with the presence of a conflict between competition-based management practices and a shift to a cooperation-based value system.
More broadly, the findings highlight the role of intangible factors in determining the optimal set of management practices. What’s more, the success or failure of a management practice depends on underlying conditions at the firm. These conditions include not just the environment in which the firm operates and the presence of other management practices, but also the type of long-term relationship that the firm chooses to establish with its employees.
The key lesson for corporate managers who are considering whether to adopt a new management practice is to ask themselves not just whether the practice worked in similar firms, but more specifically, whether it worked in firms that have a similar value system for their employees.
Firms increasingly collect data on their employees’ performance, a management practice that they can use to promote quantitative comparisons among staff and potentially boost productivity. At the same time, there is growing corporate emphasis on employee empowerment, teamwork, and happiness – again with the goal of improving overall performance. This column reports what happened when an American transport company tried to combine performance management – in the form of electronic tracking of truck drivers – with changing the workplace culture – via a multi-year ‘lean management’ program. The research evidence reveals a conflict between competition-based management practices and a shift to a cooperation-based value system. The relative performance evaluation increased productivity where there was an individualistic workplace culture – but decreased productivity when more cooperative values were embedded.
There is a growing consensus among economists that management practices are an important explanation for the large observed variation in productivity among firms. Different firms often adopt different practices, even in a narrowly defined industry.
In turn, some researchers have speculated that this variety is related to differences in the less tangible attributes of firms. Such attributes include ‘relational contracts’ – the non-legally binding understandings between firms and their employees that typically describe how employees should behave and how firms will reward the expected behaviour.
In a recent study, we explore the idea that the effectiveness of a specific management practice has to be evaluated in the broader context of organizational culture. The same practice may be beneficial in one culture and detrimental in a different one.
We begin with Edgar Schein’s definition of organizational culture as existing at three levels:
- artefacts – tangible organizational structures and processes;
- espoused values – ‘the articulated, publicly announced principles and values that [the organization] tries to achieve’;
- and basic underlying assumptions – deep individual values, which usually come from a broader cultural context, such as the country, generation, or family to which the individual belongs.
The effect of adopting a particular artefact in a particular firm – in our case, a management practice – depends on espoused values and basic assumptions.
In our study, we focus on the effect of espoused values and report the results of a double treatment: similar units of a large corporation were subject to both a change in management practices, but the underlying espoused values had been changed in an earlier period.
We randomized the posting of employee performance across a company in the US transport industry that was midway through a costly, multi-year process of altering its value system (the lean management intervention). Employees working in company locations with the original individualistic relational contract responded positively to the performance postings, while employees in the worksites with the newer relational contract responded negatively.
A field experiment in the US transport industry
The specifics of our study are as follows. We ran a field experiment in a company with a large number of worksites that all perform a similar function throughout the United States. The company had recently introduced ‘electronic on-board recorder’ technology that measures the performance of drivers against a route-specific benchmark.
The introduction of this technology raised a question about the optimal management practice for sharing performance information. In particular, should drivers be made aware only of their individual performance; or should performance statistics of all drivers at a given worksite be posted and made publicly viewable? Both options provide performance feedback to drivers, while the latter practice is also likely to spur comparisons and potentially competition between drivers.
At the time that the performance monitoring system was introduced, the company was in the middle of a major reorganization. Specifically, senior managers were engaged in a multi-year program to overhaul their business practices according to a ‘lean management’ management philosophy inspired by the Toyota Production System. Accordingly, the company committed substantial resources and set a ten-year time horizon for the transition.
Crucially, at the time of our experiment, the company had only commenced the first phase of the transition (‘Phase 1’), which focused on re-centering the workplace culture towards teamwork and the empowerment of front-line workers, away from the firm’s prevailing individualistic orientation. Truckers are sometimes known as “America’s last cowboys”, due to the traditional culture of rugged individualism. This meant that some worksites had started Phase 1 while others had not.
Accordingly, we ran a ‘3-by-2’ experiment, in which we randomized the implementation of three performance information conditions – two visible performance posting treatments and one control condition – stratified across worksites by their Phase 1 status.
Contrasting effects of posting drivers’ performance
The main empirical finding of our study is that the effect of posting drivers’ performance depends on whether the worksites had received the Phase 1 intervention. Drivers assigned to pre-Phase 1 sites (more individualistic cultures) respond on average positively to public performance postings, while drivers at Phase 1 sites respond negatively.
The effects are substantial, corresponding to a differential treatment response by drivers at the Phase 1 and pre-Phase 1 worksites of between 15% and 27% of a standard deviation in driving efficiency.
These findings must be interpreted in light of the fact that Phase 1 did not change any existing incentives (or compensation of any kind) or formal processes. In Schein’s language, it affected espoused values.
A researcher who had complete site-by-site ‘hard’ information about the current management practices – but no knowledge of the fact that certain worksites had been exposed to a values intervention – would have missed a key source of site-level adoption success. As such, our experimental results highlight the importance of accounting for intangible factors by researchers seeking to understand employee behavior.
Once we establish the main result of our experiment, we then probe in two directions: first, the randomness of the Phase 1 assignment; and second, the psychological mechanism driving the experimental results.
Regarding Phase 1 assignment, while we directly randomized the performance posting treatment, we relied on the company’s pre-existing rollout of Phase 1. We made this decision because of timing differences between the performance posting and Phase 1 treatments. The full randomization of the Phase 1 intervention would have required at least 18 months to obtain statistical power.
But the company’s management required that the experiment be conducted during their four-month rollout of the electronic monitoring technology. Given this timing mismatch, full randomization was infeasible, and we instead stratified our performance postings by whether a worksite had received the Phase 1 intervention at least three months prior to the commencement of the study.
Psychological mechanisms at work
To probe the psychological mechanism underlying our experimental results, we analyze employee attitudes that the company collected through an annual engagement survey. As one would expect, Phase 1 worksites score higher on the survey questions that assess workers’ collectivistic orientation.
When we replace the Phase 1 intervention dummy with a survey-based index of collectivist orientation in our primary analysis, we replicate the pattern of results. No significant pattern is observed if instead we use a different index of employee attitudes, one that focuses on individual satisfaction with compensation and benefits. These results suggest that the differential effects are most accurately attributed to collectivistic orientation and not to individual satisfaction.
Moreover, two additional tests shed further light on the nature of the backlash we find against public performance postings and provide additional support for our reasoning.
First, we compare the outcomes of two different posting treatments. In both cases, all drivers’ scores are publicly posted, but in one case, names are revealed, and in the other case, they are replaced with employee IDs, which effectively anonymizes the results.
These two conditions enable us to isolate more precisely the effects of explicit competition among employees, since they hold constant relative performance feedback and vary only in the identifiability of peers’ performance.
We find that only the named postings treatment matters. This finding is consistent with social psychological research showing that the competitive behavior arising from the postings should be greatly reduced when one does not know the identities of one’s adversaries.
Second, when we divide our sample into quartiles of pre-treatment performance, we find that the top quartile of drivers in Phase 1 worksites significant reduce their effort in response to the performance postings.
This result is in line with our theoretical predictions: when performance postings are introduced in Phase 1 worksites, top performers reduce their effort out of deference to their lower-performing teammates’ satisfaction, which is harmed when performance differences become widely known.
So it appears that the different responses of drivers in different worksites are driven by the individualistic orientation of the initial values and the collectivist orientation of the new values. These new values are based on the Toyota Production System, which emphasizes the value of teamwork and cooperation, as well as the important role of management in helping empower and enable the front-line workers.
Research in social psychology and organizational behaviour has found that employees respond poorly to perceived inconsistencies in leaders’ and organizations’ messages. Our findings support this result, with the posting of individualistic performance postings representing a violation of the collectivist values espoused by the company.
In fact, when they were presented with the results of our experiment, the company management chose to discontinue the performance postings immediately, concerned about the risk to the new culture they were seeking to instill.
Trends in workplace practices
Our study is related to two important trends in workforce practices over the past three decades. The first is the adoption of innovative human resource management practices, particularly a trend towards team-based management and group incentives, perhaps reflecting increased diffusion of Japanese management practices.
The second parallel trend is the increased use of data-driven management, in which firms implement technologies that enable much closer monitoring of some key output factors. We show that these two trends, while potentially complementary, have complex interactions that can affect the returns to firms attempting to adopt both.
Guidance for firms adopting a new management practice
The main contribution of our research is to show that the success or failure of a management practice depends on underlying conditions at the firm. These conditions include not just the environment in which the firm operates and the presence of other management practices, but also the type of long-term relationship that the firm chooses to establish with its employees.
Corporate managers who are considering adopting a new management practice should ask themselves not just whether the practice worked in similar firms, but more specifically whether it worked in firms that have a similar value system for their employees. This result highlights the importance of measuring not just management practices but also how workers perceive their relationship with their employer.
This article summarizes ‘The Contingent Effect of Management Practices’ by Steven Blader, Claudine Gartenberg, and Andrea Prat, published in the Review of Economic Studies in 2019.
Steven Blader is at New York University. Claudine Gartenberg is at The Wharton School. Andrea Prat is at Columbia University.