How Factories Build Trust, and Why Diversity Pays Off
A West Bengal experiment shows that diversity, though costly at first, strengthens both workplace productivity and social cohesion over time
View as PDF
Arkadev Ghosh: Duke University
SDG 8: Decent Work and Economic Growth | SDG 9: Industry, Innovation and Infrastructure
Institutions: Ministry of Labour and Employment
India’s growth strategy rests heavily on manufacturing corridors and industrial parks meant to draw millions into stable, formal work. Yet the country’s social fabric remains divided along religious and caste lines that rarely meet inside the workplace. Whether Indian factories can create cooperation is as important as creating jobs.
This matters for policy. As governments expand production-linked incentives and vocational schemes, social frictions can turn into hidden costs – lower productivity, higher turnover, and slower learning. Understanding how diversity shapes industrial performance is therefore not a moral exercise; it is an economic imperative.
Growth and Division on the Factory Floor
A food-processing plant on the outskirts of Kolkata made the challenge visible. Its workforce – roughly four-fifths Hindu and one-fifth Muslim – mirrored the district around it. This layout offered a natural test of how production technology interacts with social identity.
Some lines required near-perfect synchrony: sealing and passing packets along a conveyor. Others allowed independent work such as mixing or packaging. Teams stayed fixed for months, cooperation could be observed as it evolved, like a gradual negotiation between prejudice and performance.
Over four months, 586 men were randomly assigned to all-Hindu or mixed teams with about 35–40 percent Muslim members. Pay, supervision and tasks stayed constant.
The results were clear. Where work demanded shared timing, identity carried an economic price; where labour was solitary, difference slipped out of view.
The Short-Term Cost of Mixing
In the first weeks, high-dependency mixed teams produced roughly 20 percent less output than comparable homogeneous teams -- a sizeable loss that gradually narrowed. Over the four-month period, the overall gap was about 5 percent. In monetary terms, given the factory’s average product price of ₹32 ($0.38), this translates to an output value difference of around $1,200 per production run between a high-dependency mixed line and a low-dependency one.
The slowdown reflected attitudes and expectations. Some Hindu workers showed taste-based bias – an instinctive reluctance to coordinate closely with Muslim peers. Others displayed belief bias – underestimating the ability or effort of those from another faith. Communication barriers played no role: all workers spoke the same language, and supervision was uniform. The slowdown appeared to reflect attitudes and expectations. The evidence points to a mix of taste-based discomfort -- an instinctive reluctance to coordinate closely across religious lines -- and belief-based bias, where workers underestimated the effort or ability of those from another faith. Both mechanisms are suggestive rather than definitive, but together they explain much of the early friction. Communication barriers played no role: all workers spoke the same language, and supervision was uniform.
How Trust Takes Shape
The decline faded quickly: by week eight, mixed teams had regained most ground, and by the fourth month they matched the productivity of all-Hindu teams.
Two rounds of attitudinal surveys conducted before and after the four-month experiment confirmed what the production data suggested. Hindu workers who collaborated closely with Muslim peers reported greater willingness to share meals, take instructions, and work together again. Workers in low-dependency sections, with little daily interaction, showed no such change.
Prejudice eroded not through exhortation but through shared dependence. When productivity depended on another’s timing, suspicion met evidence – and softened. Trust rose by roughly 0.4 standard deviations on attitude indices – comparable to the effect of a wage bonus in similar studies.
When Politics Entered the Plant
The experiment coincided with the upheaval surrounding the Citizenship Amendment Act, which sharpened identity-based divisions nationwide.
When protests peaked, mixed teams – especially those needing coordination – saw productivity drop by about 7 percent, while homogeneous lines held steady. The effect vanished within weeks, but it left a warning: political polarisation can travel through the chain of cooperation.
What Managers Already Understand
A companion survey of supervisors in five similar factories revealed how managers weigh diversity.
Most expected integration to slow productivity initially in tasks demanding teamwork. Yet almost all rejected separating workers by religion, warning that it could entrench divisions and damage morale. About one in four managers even predicted, correctly, that early losses would fade with time.
Managers described diversity as a training cost: painful early, valuable later. Some noted that external political tension occasionally rippled inside the plant, confirming that trust is a fragile but essential input in production.
Patience, not partition, proved the surer route from inefficiency to cohesion.
Technology and the Geography of Cooperation
The deeper lesson is structural. Technology – the design of who must coordinate with whom – decides whether identity becomes a barrier or a bridge. High-dependency processes force repeated, measurable cooperation; low-dependency ones permit distance.
Farm work, largely solitary, seldom tests cooperation. Manufacturing does. Each production line becomes a small institution of interdependence, where rhythm , not rhetoric, shapes relationships.
Scaled up, these micro-interactions suggest a quiet social dividend. Industrialisation, then, is not only economic infrastructure – it is civic infrastructure.
Where technology demands cooperation, trust accumulates; where it doesn’t, prejudice can rest undisturbed. Diversity, forced to work in rhythm, becomes self-correcting.
Evidence in Motion
The credibility of these results lies in their realism. Workers were full-time employees under normal conditions, paid flat monthly wages unrelated to team productivity. Supervisors measured output by standard factory metrics – units per hour and material wastage.
Random assignment ensured skill or experience could not explain early friction. Coordination – not culture – determined when identity influenced performance.
If similar frictions exist across India’s medium-scale manufacturing, diversity would lead to short-term losses. Over time, as trust gains, so would long-term productivity – a permanent gain, not just recovery.
Inclusion, in other words, behaves like infrastructure – costly to build, indispensable once in place.
From Production to Policy
India’s manufacturing push will succeed only if the rhythm of its machines is matched by the rhythm of its institutions. Diversity in factories mirrors diversity in society: both create friction, both generate potential. The policy question is not whether inclusion slows growth, but how to design workplaces that learn through it.
Coordination is capital. Every system – industrial or administrative – that forces people to rely on one another creates assets that balance sheets cannot show. Conversely, political division can erode them in a day. Stability is not ornamental to growth; it is part of the production function itself.
Factories are India’s quiet classrooms of coexistence. If governments treat social trust as seriously as infrastructure, the gains will last longer than any incentive scheme. The hum of a conveyor belt teaches something policy often forgets: cooperation, once learned, compounds.
View as PDF
Author:
The discussion in this article is based on his research published in the Journal of Political Economy (Volume 133). Views are personal.