Managing shocks to worker productivity
Emerging evidence leaves little dispute that the quality of management matters for firms’ productivity (Van Reenen 2018, Bloom et al. 2013, Adhvaryu 2018). In a previous study in Indian garment factories, we identified several key features of line supervisors that predict the productivity of their teams (Adhvaryu et al. 2019a). Still, we know little about the day-to-day strategies these effective team leaders might use to achieve high productivity. In a new study, we ask how production supervisors respond to idiosyncratic shocks to team members’ productivities (Adhvaryu et al. 2019b). Variation in exposure to air pollution allows us to identify exogenous shocks to productivity. Importantly, the productivities of workers performing various tasks respond to pollution exposure differently. The data show that many supervisors observe these productivity fluctuations across workers throughout the day and reallocate workers across tasks accordingly. More attentive managers are more likely to reallocate workers across tasks in response to productivity fluctuations and, as a result, are more able to avoid productivity losses due to pollution.
Impact of air pollution on productivity
The impact of air pollution on health is well documented (Pope and Dockery 2006). We first investigate whether air pollution could reduce productivity in this context. Figure 1 below shows the relationship between fine particulate matter (FPM) pollution and efficiency – a standard measure of productivity in the garment industry. Clearly, FPM pollution decreases productivity. The magnitude of this impact is small but robust – increasing FPM by one standard deviation decreases productivity by approximately 1% of the mean (or 0.5%).
Figure 1 Relationship between residuals of FPM exposure and efficiency
Note: Residuals are from regressions of each variable on year, month, day of week, and hour of day fixed effects as well as coarse PM. Fine PM residuals are winsorized at the 5th and 95th percentile. Scatter depicts mean residual of efficiency within integer fine PM residual bins. Solid line depicts local polynomial smooth fit and dotted lines depict 95% confidence intervals.
The opportunity for production supervisors to realize gains from reallocating workers across tasks arises from heterogeneity in the impact of pollution on productivity across workers and tasks. That is, it only makes sense to reallocate workers if some workers are hit harder than others by pollution shocks. To investigate whether this type of heterogeneity in productivity impacts of pollution is exhibited in our context, we calculate the average efficiency residual of each worker for each operation during times of low (1st quartile) FPM pollution exposure. Then, for each worker, we rank the machine operations we observe them doing from 1 to 9 in ascending order of the average efficiency residual achieved. Figure 2 shows that, compared to times of low pollution, workers become less productive during times of high (4th quartile) FPM pollution exposure. Moreover, workers experience the largest drop in productivity in high ranking operations – i.e. those operations that they would perform with highest efficiency during times of low pollution.
Figure 2 Mean efficiency residual for low and high PM levels
Note: Figure depicts the mean efficiency residual at both first and fourth quartile PM levels for the different operations we observe a worker doing in the data, ranked by how efficient the worker is at each operation during low PM times. The solid line shows, mechanically, that during first quartile PM times mean efficiency is rising in the rank of the operation for the worker. The dashed line shows that at fourth quartile PM times these same workers are significantly less productive at their high ranked operations, but not for their lowest ranked operations.
In fact, as more direct evidence of this heterogeneity, we show that the efficiency rank of each operation within a worker changes as pollution increases. Figure 3 shows that operations that rank the lowest for a given worker during times of low pollution increase in rank by close to 1 during times of high pollution, and operations that rank the highest decrease in rank by more than 2. This illustrates an opportunity for production supervisors to switch workers who are currently performing operations with high susceptibility to productivity losses during high pollution times with workers for whom the same operation is less susceptible to pollution-induced productivity losses. That is, as the same operation will appear at different points along the curve in Figure 3 for different workers, gains from reallocation of workers across tasks can be realized.
Figure 3 Within worker change in the efficiency rank of operations across first and fourth quartile PM levels
Note: plotted by operation efficiency rank at first quartile PM levels.
How managers respond to productivity shocks
Do team leaders indeed perceive and respond to these productivity shocks and opportunities to mitigate losses? We find that that if a team experiences a one standard deviation rise in FPM in an hour, the probability that someone in that team will be reallocated to a different task increases by 3%. Consistent with Figure 3, Figure 4 shows that workers tasked with operations that rank highly during low pollution are the most likely to be reassigned. On average, however, task reallocation does not seem to be utilized sufficiently to fully mitigate productivity loss. The same one standard deviation rise in FPM in an hour still leads to a 0.3% decrease in efficiency net of any realized reallocation. Is this because some managers are not responding as much as they should?
Figure 4 Mean probability of worker reallocation
Note: Figure depicts the mean probability that a worker is reallocated when exposed to above median FPM levels by the efficiency rank of the operation, restricting to lines and days on which at least one such task reallocation occurred.
Data from a survey of team leaders allows us to study if the mitigative reallocation response to productivity shocks varies. We find that a 1 standard deviation increase in the team leader’s attention (frequency at which production is being monitored) increases the likelihood of task reallocation from 3% to 10%. Interestingly, these more attentive managers also seem to re-allocate tasks both when the FPM levels rise to above average and when levels decrease to below average. Accordingly, the most attentive supervisors are able to avoid nearly all productivity losses due to pollution.
Worker productivity is prone to idiosyncratic fluctuations, often from environmental conditions. For example, our study demonstrates that ambient air pollution negatively affects productivity in Indian garment factories, and that impacts vary across workers and tasks. However, given the nature of team production and this heterogeneity in productivity impacts, team leaders can manage such shocks by reallocating workers across tasks. Importantly, we find that more attentive managers are more responsive in their reallocation efforts and are, accordingly, more able to forego productivity losses due to pollution. Our findings suggest that firms might benefit from inculcating vigilance among team leaders, and ensuring that managers in team production settings actively consider task reallocation as a strategy to mitigate the impacts of fluctuations in individual worker productivities on team output.
Adhvaryu, A (2018), “Management Quality and Worker Productivity in Developing Countries”, IZA World of Labor.
Adhvaryu, A, A Nyshadham, and J Tamayo (2019a), “Managerial Quality and Productivity Dynamics”, NBER Working Paper no. 25852.
Adhvaryu, A, N Kala, and A Nyshadham (2019b), “Management and Shocks to Worker Productivity” NBER Working Paper no. 25865.
Bloom, N, B Eifert, A Mahajan, D McKenzie, and J Roberts (2013), “Does management matter? Evidence from India”, Quarterly Journal of Economics, 128 (1), 1-51.
Van Reenen, J (2018), “Management and the wealth of nations”, VoxDev.org, 18 January.
Pope, C A, and D W Dockery (2006), “Health effects of fine particulate air pollution: lines that connect”, Journal of the Air & Waste Management Association, 56 (6), 709–742.