Working Papers

Do Performance Ranks Increase Productivity? Evidence from a Field Experiment

While performance-based ranking may induce workers to increase effort because of status concerns, such information may also demotivate them or make them wary of outperforming peers. This paper disentangles the effects of demotivation, social conformity, and status associated with ranking. I implement a randomized experiment at a Bangladeshi sweater factory that pays employees on piece rates. Treated workers receive monthly information on their relative performance either in private or in public. A simple theoretical framework shows that intrinsic status concerns induce Private Treatment workers to increase or decrease effort depending on the feedback they receive from the intervention. Workers in Public Treatment respond similarly but face two additional incentives - social status (positive effect) and social conformity (negative effect). Empirical evidence shows that Private Treatment workers increased (decreased) effort upon receiving positive (negative) feedback. Public ranking led to lower net effort relative to Private Treatment because of a strong preference not to outperform friends. The negative effects from demotivation and social conformity may explain why the existing literature finds mixed evidence of impact of ranking workers.

Layoffs and Productivity at a Bangladeshi Sweater Factory

(with R. Akerlof, R. Macchiavello, and A. Rabbani)

Conflicts between management and workers are common and can have significant impacts on productivity. We study how workers in the manual knitting section of a large Bangladeshi sweater factory responded to management’s decision to lay off about a quarter of the workers following a period of labor unrest. Our main finding is that the mass layoff resulted in a large and persistent reduction in the productivity of surviving workers. It is specifically the firing of peers with whom workers had social connections – friends – that matters. We document evidence consistent with deliberate shading of performance by workers in order to punish the factory’s management, and a corresponding deliberate attempt by the factory to win the angry workers back by selectively giving them tasks that are more rewarding. By combining ethnographic and survey data on the socialization process with the factory’s internal records, the paper provides a rare glimpse into the aftermath of an episode of labor unrest. A portrait of the firm emerges as a web of interconnected relational agreements supported by social connections.

The Effect of Political and Labour Unrest on Productivity: Evidence from Bangladeshi Garments

(with R. Macchiavello, A. Rabbani, and C. Woodruff)

This paper examines the effects of political strikes and labour unrest on production in 33 large ready-made garment factories in Bangladesh. We find that the political strikes (hartals) lasting one or two days have very little effect on productivity or worker absenteeism. We see some negative effects on factories for strikes lasting five days or more. The main channel for negative effects appears to be supply chain disruptions rather than worker absenteeism. Labour unrest has a more immediate negative effect, increasing absenteeism and quality defect rates, and decreasing output. Where protests extend to a second day, we find that production falls by around 10 percent among factories in the affected area. We also compare the magnitude of hartals and labour unrest with the effects of temperature on production.

Research in Progress

Revisiting Piece Rates: Quantity - Quality Conundrum

It has been empirically established that piece-rate pay scheme increases the productivity of workers compared to fixed salary. However, while piece-rate pay scheme may induce a worker to maximize only the quantity of her output, a firm may want to maximize both quantity and quality. To check whether piece-rate pay scheme increases quantity at the cost of quality, I intend to do a difference in difference analysis with detailed administrative data from two semi-automatic knitting sections at a Bangladeshi sweater factory. One of the two sections switched from fixed salary to piece-rate pay scheme, while the second always kept using a fixed salary regime. I study two kinds of defects. The first is a major defect that, once detected by quality inspectors, a worker needs to mend herself; such defects cost her time to mend and hence she may internalize the need to minimize such defects. The second is a minor defect that is instead passed on to a separate mending section for mending. Since this defect is cost-less for the worker, it may increase after the introduction of piece-rate regime.

Change of Plans: How Firms Adapt to Shocks in a Developing Country

(with R. Macchiavello and C. Woodruff)

Firms in developing countries are vulnerable to frequent shocks and uncertainties resulting from political and industrial instabilities. Using data from 19 Bangladeshi garment factories, in this paper we check how they change their operation plans to adapt to two different kinds of shocks - (a) political strikes, and (b) industrial unrest.