English only · Odia translation in progress

Chapter 3: The Cathedral in the Village (1957-1980)

On February 3, 1959, Jawaharlal Nehru arrived in Sundargarh district to inaugurate the Rourkela Steel Plant — India’s first integrated steel plant built with foreign collaboration in the public sector. The setting was extraordinary in its contradictions. A German-designed steel mill, engineered by Krupp and Demag of Duisburg, erected by thousands of German technical personnel who constituted the largest German colony outside Germany at the time, stood in a tribal district where the Oraon, Munda, Kharia, and Gond communities had lived for centuries on subsistence agriculture and forest produce. The Indo-German Club had been established for the social life of the expatriate engineers. Restaurants serving food from across India had appeared. The township had modern infrastructure — roads, hospitals, schools, housing — built to standards that would not have been out of place in a mid-sized European industrial city. Twenty kilometres in any direction, the world reverted to what it had been for generations: subsistence agriculture, tribal villages, forest economy, colonial-era roads.

Nehru called his steel plants, dams, and power stations the “temples of modern India.” The metaphor was more revealing than he intended. A temple is a sacred structure, built by specialists, funded by a patron, operated by a priestly class. The ordinary people come to worship. They do not build. They do not own. They receive what the temple offers and return to their lives. The relationship is dependency, not reciprocity.

Rourkela, Hirakud, NALCO — the three great Nehruvian insertions into Odisha’s landscape — followed this logic precisely. They were cathedrals: massive, top-down, centrally planned, designed by distant authorities, built by imported expertise, operated according to national strategies decided in New Delhi. They were impressive. They produced output. They did not generate organic industrial culture. And the gap between the cathedral and the village that surrounded it is the structural story of Odisha’s industrialization for three decades.

The Dam That Fed the Factory

Two years before the steel plant opened, on January 13, 1957, Nehru had inaugurated another cathedral in Odisha: the Hirakud Dam across the Mahanadi River, near Sambalpur. At 25.8 kilometres including its dykes, Hirakud was India’s longest dam. The project had been conceived in 1937 by engineer M. Visvesvaraya and formally proposed in 1945. Construction began in 1948, barely a year after independence. American engineering consultancy oversaw the work. The Central Water Commission supervised it from Delhi. The project cost Rs 100.02 crore — an enormous sum in 1950s terms, equivalent to roughly Rs 10,000 crore in 2023 purchasing power.

The dam’s stated purposes were noble and legitimate: flood control on the Mahanadi (the “Sorrow of Odisha”), irrigation for the delta farmland, and hydroelectric power to fuel industrialization. The reservoir covered approximately 743 square kilometres. The designed irrigation capacity was 436,000 acres. The two power houses — Burla and Chiplima — eventually reached a combined installed capacity of 347.5 megawatts, generating 800 to 1,000 gigawatt-hours annually.

The costs were borne by people who had no say in the decision. Three hundred and twenty-five villages were submerged — 291 in Odisha, 34 in undivided Madhya Pradesh. Approximately 183,000 acres of land went underwater, including 123,000 acres of cultivated land — some of the most fertile in western Odisha, the rich silt deposits of the Mahanadi floodplain. Around 22,000 families, perhaps 110,000 to 150,000 people, were displaced. The budgeted compensation was Rs 9.5 crore. The amount actually disbursed was Rs 3.32 crore — barely 35 percent of what had been allocated. Not 35 percent of what was fair or adequate, but 35 percent of what the government itself had decided was the minimum. Seven decades later, descendants of the original oustees are still fighting for the remainder. The Hirakud Budi Anchal Sangram Samiti, the organization of dam-displaced people, has outlived most of its founding members. The grievance was inherited along with the poverty.

Nehru told the villagers of Sambalpur: “If you are to suffer, you should suffer in the interest of the country.” That sentence was not an off-the-cuff remark. It was a philosophy. It contained the entire logic of Nehruvian development: the nation comes first; local costs are regrettable but necessary; the benefits, when they arrive, will justify the sacrifice. The people who bore the costs would simply have to trust that the people who designed the project had their interests somewhere in the calculus.

But who actually captured the benefits? The hydroelectric power was not designed for the displaced farmers or the villages of Sambalpur. Its primary consumers were the Rourkela Steel Plant, NALCO’s Angul smelter (from 1987), Hindalco’s aluminium operations, and the constellation of thermal power plants and industrial facilities in the Jharsuguda-Sambalpur corridor. By 2007, industrial water allocation from the dam had increased sixfold compared to pre-1997 levels — while the farmers for whom the irrigation canals had supposedly been built watched their water diverted to factories. Industries were paying half the price for reservoir water that farmers were charged.

In November 2007, approximately 30,000 farmers from eight districts stormed the Hirakud reservoir area, demanding that water be used for its original purpose — agriculture. Over 300,000 farmers across the command area were affected by the industrial diversion. The protest received brief national coverage and changed nothing structurally. The factories needed the water. The factories had the political weight. The farmers had history on their side and nothing else.

The Hirakud template would be repeated across Odisha for seven decades, with variations in scale but not in structure: national need identified, local population displaced with systematically inadequate compensation, benefits captured by external interests, costs left with locals, and no mechanism for accountability. It is the template of what development economists call the “sacrifice zone” — a geography that pays the costs of modernity so that modernity can happen somewhere else.

The German Factory in the Tribal District

The Rourkela Steel Plant was one of three public-sector steel plants conceived under the Second Five Year Plan to expand India’s steelmaking capacity. Bhilai in Madhya Pradesh would be built with Soviet assistance. Durgapur in West Bengal would use British expertise. Rourkela, in what was then one of the most remote tribal districts in India, was assigned to West Germany.

The choice of location was driven by proximity to iron ore in the Keonjhar-Sundargarh belt, coal from the Jharsuguda-Ib Valley, water from the Brahmani River, and the Odisha government’s determination to attract industrial investment. The logic was impeccable on paper. In practice, placing a German-engineered steel mill in a tribal district created a dislocation so complete that the two worlds never integrated.

Krupp and Demag formed a joint venture called Indien Gemeinschaft Krupp Demag (IGKD) to provide consultancy, design, and commissioning. Thousands of German technical personnel came to Rourkela. The initial capacity was one million tonnes per annum of crude steel. Subsequent expansions raised capacity to 1.8 million, then 2 million, and eventually 4.5 million tonnes per annum after modernization completed in 2015-16.

The land acquisition tells its own story. Approximately 19,772 acres were acquired for the plant and township from over 32 villages. An additional 11,923 acres were taken for the Mandira Dam that would supply water to the plant, submerging 31 more tribal villages. The displaced population was overwhelmingly tribal — Scheduled Tribes constituted 68 to 90 percent of those affected. Between 13,000 and 16,000 Adivasi inhabitants were resettled from the plant area alone. Another 8,785 people were displaced for the Mandira Dam.

What does “resettled” mean in practice? The displaced families were placed in peripheral settlements — Jalda, Jhirpani, Bisra, Bondamunda — bearing no resemblance to the forested villages they had lost. A study found that 78.7 percent of displaced families lost their land entirely. Many experienced disintegration of household cohesion. They became seasonal labourers, dependent on the industrial economy that displaced them but unable to access its better-paying permanent jobs. The forest that had sustained their economy, their diet, their social structures, and their cosmology was replaced by a settlement pattern designed for industrial convenience, not human continuity.

The steel plant, meanwhile, became a node of genuine cosmopolitanism. Hindi, Bengali, Odia, Telugu, and Punjabi speakers lived side by side. Hindu temples stood alongside churches. A sports culture around hockey and football flourished. The township had hospitals, schools, and recreational facilities that outstripped Bhubaneswar. Rourkela is, in many ways, more cosmopolitan than the state capital.

But this cosmopolitanism was an island. The plant workers had salaried employment, provident funds, and medical benefits. The displaced tribals had seasonal labour and intergenerational poverty. Drive twenty kilometres in any direction from the steel city, and you entered a different century.

This is the enclave in its purest form: a node of modern industrial civilization transplanted into, but not connected to, the local economy. The plant could have been airlifted to any location in India and its internal functioning would have been identical. Its relationship to Sundargarh district was that of a foreign embassy to its host country — present in the territory but operating under a different sovereignty.

The Aluminium Machine in the Poorest District

Twenty-two years after Rourkela was commissioned, the pattern was replicated with a different metal in a different district, but with the same structural logic.

The National Aluminium Company Limited — NALCO — was established on January 7, 1981, with its registered office in Bhubaneswar. Prime Minister Indira Gandhi laid the foundation stone in March 1981. The project was conceived to exploit India’s bauxite reserves — specifically, the massive Panchpatmali deposit in Koraput district, one of the largest bauxite reserves in India. French technical assistance from Pechiney (now part of Rio Tinto Alcan) provided the technology for the alumina refinery and smelter.

The scale was enormous. Bauxite mining at Panchpatmali Hills was designed for a capacity of 6.825 million tonnes per annum — a fully mechanized opencast operation that became operational in November 1985. The alumina refinery at Damanjodi, at the foot of the Panchpatmali Hills, had a normative capacity of 2.1 million tonnes per annum, operational from September 1986. The aluminium smelter was placed not near the bauxite but at Angul, approximately 600 kilometres north of Damanjodi, where a 1,200 megawatt captive power plant fuelled by coal could supply the electricity that aluminium smelting demands. The smelter reached a capacity of 460,000 tonnes per annum.

The original project cost was approximately Rs 2,400 crore — a massive investment in early-1980s terms. This was national capital, deployed in Odisha’s soil, processing Odisha’s minerals. The Government of India owns 51.28 percent of NALCO’s equity. When NALCO pays dividends, more than half flows to the central exchequer.

Follow the money. In FY 2023-24, NALCO recorded a net profit of Rs 2,060 crore. It paid Rs 918 crore in dividends. In FY 2024-25, the dividend to the Government of India alone was a record Rs 988.88 crore. Total dividends that year reached Rs 1,928 crore. Nearly Rs 1,000 crore per year flows from Odisha’s bauxite to Delhi’s treasury.

Where does the bauxite come from? Koraput district. Where rural poverty among tribal populations has historically been measured at 74.2 percent. The aluminium is sold at London Metal Exchange prices. The profit is distributed to the Government of India, which uses it in the general budget allocated across all states based on population-weighted formulas that do not privilege the state whose ground was dug up.

The displacement at Damanjodi was more modest than at Hirakud or Rourkela — 597 families across 26 villages, with 42.55 percent being tribal families. NALCO’s rehabilitation record was better than most: 599 out of 600 land-displaced persons were given employment by 2022. But employment does not replace a web of social, cultural, and ecological relationships that formed over generations. You can give a Kondh family a job at a refinery. You cannot give them back the forest, the relationship between their village and the hill, or the cosmology that made the landscape legible as home rather than as a mineral deposit with a market price.

And the 2024 Pottangi expansion tells you how much the pattern has changed in four decades. In June 2024, NALCO executed a new mining lease for the Pottangi bauxite deposit near Serubandha Hills — 697.979 hectares, 111 million tonnes of estimated reserves, 3.5 million tonnes annual production capacity, 32-year mine life. Tribal communities in Semiliguda and Pottangi blocks responded with “NALCO Go Back” slogans, stalling the bhumi pujan ceremony in August 2024, demanding that land settlement issues be addressed before digging began. The specific complaint — that the mining lease was signed without adequate Gram Sabha approval, as required by the Forest Rights Act and the Niyamgiri precedent — is a complaint about process, not about progress. It is a demand for consent, not a rejection of modernity. And it is a demand that the cathedral model, by its nature, cannot accommodate.

The Cathedral and the Bazaar

In 1997, an American programmer named Eric S. Raymond published an essay called “The Cathedral and the Bazaar.” It became one of the most influential texts in the history of software development. Raymond was writing about two models of building software, but his framework illuminates far more than code.

The cathedral model, Raymond argued, is how institutions naturally want to build things. A small group of architects designs the system. The design is revealed to the world only when it is finished. The users receive the product; they do not participate in its creation. The code is proprietary. The architecture is controlled from the center. The cathedral is impressive, internally consistent, and beautifully engineered. It is also fragile, because its complexity is managed by a small number of people, and it does not learn from its users.

The bazaar model is the opposite. It is messy, decentralized, iterative. Many participants contribute small pieces. The system evolves through constant feedback. Bugs are found quickly because many eyes are looking. The architecture is not designed top-down but emerges from the accumulated decisions of hundreds of independent contributors. The bazaar looks chaotic. It produces systems that are adaptive and self-sustaining. Linux, the operating system that runs most of the world’s servers and all of its Android phones, was built as a bazaar. No single architect designed it. Thousands of contributors, each scratching their own itch, collectively built something no cathedral process could have produced.

The distinction maps onto industrial economics with uncomfortable precision. Hirakud, Rourkela, and NALCO were cathedrals. They were designed by distant architects (the Planning Commission, SAIL headquarters, French engineers), built by imported expertise (American consultants, German technicians, centrally recruited skilled workers), financed by a national patron (the Government of India), and operated according to a centralized plan. The local population received the product — some irrigation water, some employment, some trickle-down economic activity — but did not participate in the creation. They were users, not contributors. They could worship at the cathedral; they could not build their own.

What Odisha needed was a bazaar. An organic, bottom-up, locally owned economic system that would grow through entrepreneurial initiative, local supply chains, skill accumulation, and iterative improvement. What it got was three cathedrals surrounded by villages that were no more industrially capable after the cathedrals were built than they had been before.

The distinction is not merely aesthetic. It is about reproduction. A cathedral does not reproduce. You cannot build a second Rourkela Steel Plant by learning from the first, because the knowledge, the capital, the management, and the supply chains all came from outside and stayed outside. A bazaar reproduces organically. When a supplier to an anchor factory develops a capability, that capability can be deployed to serve other customers, which creates demand for more suppliers, which deepens the local skill base, which attracts more anchor factories. The bazaar generates its own momentum. The cathedral depends on continued patronage from the center.

This is why the most consequential fact about Odisha’s Nehruvian industrialization is not the quantity of steel produced or the megawatts generated or the tonnes of aluminium smelted. It is the quantity of industrial ecosystem that did not grow around these projects. After seven decades, Rourkela Steel Plant has no equivalent of the Adityapur ancillary cluster that grew around Tata Steel at Jamshedpur. NALCO has no downstream fabrication ecosystem in Koraput or Angul. Hirakud’s power feeds factories, but the villages that were submerged to create the reservoir have not been integrated into the industrial economy. The cathedrals produce. They do not reproduce.

Why the Ecosystem Did Not Grow

Five structural factors explain why three massive industrial projects, sustained over seven decades, failed to generate the organic industrial culture that emerged in other parts of India around similar anchor investments.

The workforce came from outside. At Rourkela, skilled workers were recruited predominantly from Bengal, Bihar, and Andhra Pradesh — states with existing industrial workforces and technical education infrastructure. At NALCO, French technical expertise built the initial operations. The local tribal population had no industrial skills and, critically, no mechanism to acquire them rapidly. A steel plant requires metallurgical engineers, mechanical fitters, electrical technicians, quality control specialists — none of which could be sourced from a subsistence agricultural community within a single generation without a massive, deliberate investment in technical education. That investment was not made at the scale required. NIT Rourkela was established in 1961 — a fine institution — but its graduates migrated to industrial clusters in other states where the jobs were. The knowledge that flowed into these projects flowed through external channels and stayed within the project boundaries.

Management was centrally appointed. SAIL’s management was and is appointed in New Delhi. Officers serve a few years and rotate. They have no local roots, no long-term relationship with the region, no economic incentive to develop local capacity. Their career advancement depends on performance metrics set in Delhi, not on the welfare of Sundargarh district. NALCO’s management follows the same pattern — central government appointments, rotating postings, national accountability structures. Compare this with a privately owned anchor company where the founding family has invested not just capital but identity in a specific place. The incentive structure is fundamentally different.

Procurement was national. SAIL’s supply chain is integrated nationally. Raw materials, equipment, spare parts, and services are procured through centralized processes governed by public-sector procurement rules. There was no mandate and no incentive to develop local suppliers. A small engineering firm in Sundargarh could not compete with an established supplier in Kolkata or Mumbai for SAIL contracts, and SAIL had no reason to help it learn. The procurement system was optimized for compliance and cost at the national level, not for ecosystem development at the local level.

The local economy was too undeveloped to supply inputs. In the 1950s, Sundargarh district was overwhelmingly tribal and agricultural. There was no industrial base from which suppliers could emerge. The gap between what the steel plant needed — precision engineering, chemical inputs, industrial-grade components — and what the local economy could provide — agricultural labour, forest produce — was unbridgeable in the short term. Bridging it would have required deliberate, sustained investment in technical education and supplier development over a period of decades. Nobody was mandated to make that investment. The cathedral was not designed to build the village’s capacity. It was designed to produce steel.

There were no forced linkages. This is perhaps the most consequential factor, because it distinguishes the Indian approach from the one approach that demonstrably worked — South Korea’s. India’s central planning did not include mandatory local procurement requirements. There was no mechanism forcing Rourkela Steel Plant to source a percentage of its inputs locally. There was no requirement for NALCO to develop downstream processing in Koraput. There was no policy linking the cathedral projects to organic local industrial development. The projects existed in the geography but were not of the geography. They were designed as nodes in a national production system, not as seeds of a local industrial ecosystem.

These five factors are not independent. They interact and reinforce each other. External workforce means no local skill development. Central management means no local accountability. National procurement means no local supplier development. An undeveloped local economy means no capacity to participate even if opportunities existed. And the absence of forced linkages means no mechanism to break the cycle. Each factor makes the others worse. The result is a stable equilibrium — stable in the sense that it persists without external intervention, not in the sense that it is desirable. The cathedral sits in the village. The village remains a village.

The Road Not Taken: Jamshedpur

One hundred and seventy kilometres southeast of Rourkela, in the Chota Nagpur Plateau of what is now Jharkhand, sits the most instructive counterfactual in Indian industrial history: Jamshedpur.

Tata Steel was founded in 1907, fifty-two years before Rourkela was commissioned. Like Rourkela, it was a steel plant placed in a remote, tribal-dominated district. Like Rourkela, it required massive land acquisition. Like Rourkela, it displaced local communities. The initial conditions were, in many ways, similar. What diverged was not the geography, not the raw materials, not the scale of investment. What diverged was the ownership structure — and from that single difference, everything else followed.

Jamshedpur was built by the Tata family. Jamsetji Tata conceived the project. His sons and their successors executed it. The family had a permanent stake in the city’s success — not a rotating posting, not a three-year assignment, but a multigenerational commitment rooted in the understanding that the company’s fortunes and the city’s fortunes were indivisible. When the company needed a reliable workforce, it invested in local education. XLRI Jamshedpur was established in 1949, NIT Jamshedpur followed, and multiple technical schools were built. When the company needed reliable city services, it built them: Tata Steel Utilities and Infrastructure Services Limited (formerly JUSCO) manages water, electricity, sanitation, and road maintenance for the city. The private company became the city’s government because the company understood that a well-functioning city was not charity — it was infrastructure.

Over decades, Tata was compelled by economic logic to develop local procurement. A private company optimizing costs recognizes that proximity of suppliers reduces logistics expenses, improves supply reliability, and enables just-in-time delivery. If you can develop a local supplier who delivers the same quality at lower total cost than a distant one, the economics demand that you do so. This incentive does not exist for a public-sector unit with centralized procurement, because the PSU is not optimizing for local cost efficiency — it is executing a national supply chain designed in Delhi.

The result, compounded over 119 years, is Adityapur — one of Eastern India’s largest industrial clusters. Approximately 1,500 industrial units, of which 600 to 700 are ancillary units dependent on Tata Motors and Tata Steel. Not ancillary in the loose sense of being nearby, but ancillary in the structural sense of being economically integrated into the anchor company’s supply chain. These firms make components, provide engineering services, maintain equipment, and supply the hundred minor inputs that a major industrial operation consumes daily. They employ tens of thousands of workers who developed their skills in the ecosystem and transmit those skills to the next generation.

Jamshedpur is a bazaar. Not in the sense of being unplanned — Tata’s investment in civic infrastructure is legendary — but in the Raymond sense: an ecosystem that grew organically over time, where many independent participants contribute to a system that is larger than any of them, where knowledge circulates through the local economy rather than being confined within a single proprietary operation, and where the system’s complexity makes it resilient. You cannot shut down Jamshedpur by reassigning a management team. The knowledge is distributed. The supply chains are local. The ecosystem sustains itself.

Now compare the numbers:

Factor	Jamshedpur (Tata Steel)	Rourkela (SAIL)
Founded	1907	1959
Ownership	Private (Tata family)	Government of India (SAIL)
Management	Permanent corporate stake	Rotating central appointments
Procurement	Increasingly local over decades	National/centralized
Ancillary industries	~1,500 units in Adityapur	No equivalent cluster
City governance	Tata-managed (JUSCO)	Standard municipality
Education investment	XLRI (1949), NIT, technical schools	NIT Rourkela (1961), graduates migrated out
Time to ecosystem	119 years of compounding	67 years without compounding

The fifty-two-year head start matters, but it does not explain the divergence. Sixty-seven years is more than enough time to build an ancillary ecosystem if the incentive structure supports it. Tamil Nadu built its auto component cluster in roughly forty years from the Ashok Leyland anchor in 1948. Gujarat built its chemical corridor from the GIDC estates established in 1962. Karnataka’s IT ecosystem is forty-one years old, counting from Texas Instruments’ arrival in 1985. The issue at Rourkela is not time. It is the absence of the mechanism that converts time into compounding.

In compound interest terms — and this is a framework worth internalizing because it applies to ecosystems, not just savings accounts — Jamshedpur compounded at a positive rate for 119 years. Each year, the ecosystem grew, the supplier base deepened, the workforce skills expanded, and the institutional knowledge accumulated. Rourkela had a zero percent compounding rate on ecosystem development. The steel plant grew. Its capacity expanded from one million tonnes to 4.5 million tonnes. Its profits increased. But the ecosystem around it did not compound. The plant got bigger; the village stayed a village. Compounding requires reinvestment of returns into the same system. Rourkela’s returns were extracted to Delhi through SAIL’s dividend payments and reinvested according to national priorities, not local ecosystem logic.

The Road Not Taken: Pohang

If Jamshedpur is the counterfactual within India, POSCO’s Pohang steelworks is the counterfactual from across the world — and it is even more damning, because Pohang started later than Rourkela, with fewer advantages, and the divergence is total.

The Pohang Iron and Steel Company was established in 1968, nine years after Rourkela was commissioned. Park Tae-joon, a retired army general and confidant of President Park Chung Hee, was appointed president. Pohang was a small fishing port. South Korea had no steelmaking tradition, no iron ore, and limited coal. On paper, the starting conditions were worse than Rourkela’s in almost every dimension. What Korea had that Odisha did not was a government that understood the difference between building a factory and building an ecosystem.

Construction at Pohang began in April 1970. The first integrated mill was dedicated in July 1973 with an initial capacity of 1.03 million tonnes per annum — roughly matching Rourkela’s initial capacity. But what happened next is where the stories diverge permanently.

Korea’s Steel Industry Promotion Law of 1970 did not just subsidize POSCO. It subsidized the entire chain — from raw materials to finished products. Low-cost foreign capital, electricity discounts, rail transport discounts, limits on steel imports. This was not merely protectionism. It was ecosystem architecture. The law was a blueprint for how POSCO’s steel would be consumed domestically, not just produced.

During construction phases, Korean engineers progressively replaced foreign experts. In Phase 2 (1974), Koreans took over specification inspection. In Phase 3 (1976), they handled material balance, facilities specification, and drawing inspection. By Phase 4 (1979), Korean engineers had supplanted foreign consultants entirely from general engineering planning. Within fifteen years of POSCO’s founding, Korea had fully autonomous steelmaking capability. This was not organic evolution. It was deliberate policy — a forced capability transfer executed over a defined timeline.

Simultaneously, the Korean government’s Heavy and Chemical Industries (HCI) drive of the 1970s explicitly linked steel production to downstream industries. Shipyards (Hyundai Heavy Industries, established 1972 at Ulsan) were built to consume POSCO’s steel. Automobile manufacturers (Hyundai Motor, Kia) were nurtured to consume steel and create downstream demand. Ulsan and Pohang were designated as linked industrial centers in late 1974, connecting steel production to automobile manufacturing and shipbuilding in a single policy architecture.

By the early 1980s, POSCO’s capacity exceeded 9 million tonnes per annum at Pohang alone. A second integrated plant opened at Kwangyang in 1985. By the 2000s, POSCO was one of the world’s largest steel producers, and Pohang had transformed from a fishing village into an industrial city of approximately 520,000 people.

The contrast with Rourkela:

Factor	POSCO (Pohang)	Rourkela (SAIL)
Government intent	Build national industrial ecosystem	Produce steel for national consumption
Local capability transfer	Mandatory, phased, complete within 15 years	Minimal; key positions still rotate from Delhi
Downstream linkages	Deliberate: shipbuilding, auto, construction	None mandated
Procurement policy	Increasingly local as capabilities developed	National/centralized
Result after 20 years	World-class autonomous steel producer	Dependent PSU enclave
City transformation	Fishing village to industrial city of 520,000	Tribal district to industrial town surrounded by subsistence agriculture

The difference was not resources. Korea had fewer natural resources than Odisha. The difference was not the scale of investment. India invested massively in Rourkela. The difference was not human capital — Korean workers in 1968 were not inherently more skilled than Indian workers. The difference was policy architecture: Korea deliberately built an ecosystem; India built a factory.

In Raymond’s terms, Korea built a bazaar using the initial cathedral as a seed. The government provided the cathedral — POSCO itself — but then deliberately ensured that the cathedral would spawn a bazaar around it. The steel was not an end product. It was an input to an industrial ecosystem. The factory was not the goal. The ecosystem was the goal. The factory was just the first node.

India built the cathedral and stopped. The steel left Rourkela on trains. It was consumed in factories across India. The value chain continued in Mumbai, in Chennai, in Kolkata. The profits returned to Delhi as dividends. The ecosystem grew elsewhere. The cathedral stood in the village, impressive and alone.

What the Cathedrals Actually Produced

It would be dishonest to treat the Nehruvian insertions as pure failures. They were not. They were partial successes that failed to become complete ones, and the distinction matters because it reveals where the system broke down.

Hirakud Dam provides flood control that has protected the Mahanadi delta for seven decades. The devastating floods that earned the Mahanadi its nickname as the “Sorrow of Odisha” have been substantially mitigated. The irrigation canals, despite the industrial diversion that followed, brought water to farmers who had been entirely dependent on the monsoon. The hydroelectric power, whatever its allocation, created a capacity that enabled further industrial development. These are genuine achievements. They represent real welfare improvements for millions of people.

Rourkela Steel Plant, by FY 2021-22, generated revenue of Rs 26,830 crore and profit before tax of Rs 6,347 crore. It employs thousands directly and supports an economic ecosystem — however limited compared to what it could have been. It introduced an entire district to industrial modernity, created a cosmopolitan town in a region that had none, and demonstrated that large-scale manufacturing was possible in Odisha. NIT Rourkela, established two years after the plant was commissioned, has produced generations of engineers — even if most of them left to build careers elsewhere.

NALCO is India’s largest integrated aluminium producer. Its operations at Damanjodi and Angul provide direct employment, tax revenue, and economic activity in two of Odisha’s districts. Its rehabilitation record at Damanjodi, while imperfect, is substantially better than the norm for Indian industrial displacement. The company is profitable, well-managed within its mandate, and contributes to India’s self-sufficiency in aluminium production.

The point is not that these projects produced nothing. The point is that they produced less than they should have — not less steel or aluminium or electricity, but less industrial culture, less local capability, less economic self-sufficiency. They produced output without producing capacity. They generated wealth that flowed upward to the center without generating the knowledge, the supply chains, the entrepreneurial energy, and the institutional infrastructure that would have allowed the local economy to capture and reinvest that wealth.

The cathedrals were mines, not gardens. A mine produces enormous output, but each extraction depletes the resource. A garden produces less per harvest, but each season strengthens the soil. The cathedrals extracted value from Odisha’s geography. They did not plant gardens.

The Invisible Tax: Freight Equalization

While the cathedrals were being built, the central government was simultaneously ensuring that no organic industrialization could grow around them. The instrument was the Freight Equalization Policy (FEP), enacted in 1952 and maintained for forty-one years until 1993.

The policy’s mechanism was straightforward: the government equalized the transport cost of key raw materials — iron ore, coal, cement, steel — across the entire country. A factory in Mumbai paid the same delivered price for iron ore as a factory in Rourkela, even though Rourkela sat on top of the ore deposit and Mumbai was over a thousand kilometres away. The natural cost advantage of proximity — the most basic economic incentive for industrial clustering near raw material sources — was eliminated by administrative fiat.

The consequences were devastating and are explored in depth elsewhere in this project. But they are essential context for understanding the cathedral problem. Even if Rourkela Steel Plant had generated local supplier interest, even if a local entrepreneur had wanted to build a small steel products factory near the ore deposits, the FEP ensured that his raw material costs would be identical to a competitor in Gujarat or Tamil Nadu who had better infrastructure, a more skilled workforce, better market access, and decades of accumulated industrial culture. The FEP did not just fail to help Odisha. It actively penalized Odisha’s natural advantages while subsidizing the advantages of already-industrialized states.

For forty-one years, from 1952 to 1993, the cathedral projects produced steel, aluminium, and electricity in Odisha, while the Freight Equalization Policy ensured that no organic industrial ecosystem could use those outputs to grow. By the time the policy was abolished, the industrial geography of India was fixed. The factories were in the west and south. The supply chains were established. The skilled workforces were trained. Odisha was left with minerals and migration.

The combination of cathedral industrialization and freight equalization was, in effect, a double lock. The cathedrals ensured that industrial knowledge stayed within the project boundaries. The FEP ensured that even knowledge that leaked out could not be commercially exploited. The village could see the cathedral. It could not learn from it. And even if it had learned, the economic system ensured that the knowledge would be worthless locally.

The Algebra of the Enclave

Step back and look at the structure. What the Nehruvian period created in Odisha was not an industrial economy. It was a set of industrial enclaves — nodes of modern production embedded in a pre-modern economic landscape, connected not to the local economy but to the national production system.

The mathematics of the enclave are important because they persist.

SAIL is a Government of India enterprise. Its profits flow to the central exchequer as dividends. Its management is appointed by the central government. Its procurement is controlled centrally. The steel produced in Rourkela is sold across India and exported internationally. Pricing, allocation, and distribution are determined by SAIL’s corporate strategy, which is determined by central government industrial policy. What remains in Odisha: wages for a declining number of workers as automation increases, local taxes and royalties set by central government formulas, and whatever trickle-down economic activity the presence of a large industrial facility generates.

NALCO follows the same logic. The Government of India owns 51.28 percent. Dividends flow to Delhi. Corporate tax is paid to the central government. Export earnings from aluminium sold at LME prices accrue to the national balance sheet. The full value chain beyond aluminium ingots — downstream fabrication, auto parts, aerospace components, packaging, electrical conductors — happens elsewhere, where the aluminium is consumed and fabricated, primarily in western and southern India.

Coal India’s subsidiary Mahanadi Coalfields Limited (MCL) follows the same logic in the coal sector. The coal is mined in the Talcher Coalfield. The profits flow to Coal India’s accounts. The national balance sheet grows. The villages of Angul and Dhenkanal host the mines and bear the environmental costs.

In each case, the algebraic structure is the same:

Local inputs (land, minerals, water, displacement) + External inputs (capital, technology, management, skilled labor) = Output (steel, aluminium, electricity, coal) —> Distribution (profits to center, product to national market, costs to local community)

This is not exploitation in the crude sense of colonial plunder. It is a structural outcome of centralized ownership applied to geographically fixed resources. The system was designed with national welfare in mind, not local extraction as an end goal. But the structural consequence for the locality is the same regardless of intent: the value is created here and captured elsewhere. The costs remain. The benefits leave.

The enclave economy is not an accident. It is not a policy failure in the sense that someone made a mistake and the outcome could have been different without changing the underlying structure. It is the predictable, structural consequence of central ownership of local resources. Change the ownership structure, and the incentives change. Change the incentives, and the outcomes change. Leave the ownership structure intact, and no amount of policy adjustment will produce an ecosystem, because the fundamental incentive — to develop the local economy — does not exist for an entity whose accountability, management, and profit distribution are all directed toward Delhi.

What Was Actually Needed

Consider what would have had to be true for the Nehruvian insertions to produce industrial ecosystems rather than industrial enclaves.

First, forced local procurement. If Rourkela Steel Plant had been required to source an increasing percentage of its inputs locally — starting at 5 percent and rising to 40 percent over two decades — the demand would have created suppliers. The suppliers would have needed skills, which would have required training, which would have required institutions, which would have anchored NIT Rourkela’s graduates in the local economy instead of exporting them. This is precisely what South Korea did with POSCO. It is not speculative. It is demonstrated history.

Second, mandatory local capability transfer. If the German engineers at Rourkela had been contractually required to train local engineers to replace them — with a defined timeline for complete handover, as Korea achieved in fifteen years — the technical knowledge would have entered the local economy instead of departing with the expatriate engineers.

Third, downstream linkage requirements. If the central government had required that a percentage of Rourkela’s steel be processed locally into higher-value products — auto components, structural steel, industrial equipment — the demand for downstream processing would have created the very factories that Odisha lacked. Korea linked POSCO to Hyundai’s shipyards and automobile plants. India linked Rourkela to nothing.

Fourth, local ownership stakes. If the Odisha government or local cooperatives had been granted a meaningful equity stake in SAIL’s Rourkela operations and NALCO’s Angul-Damanjodi operations, the dividend flows would have stayed in the state. The accountability would have been local. The incentive to develop the surrounding economy would have been structural, not philanthropic.

None of these were technically impossible. Korea, starting later with fewer resources, implemented all four. What was lacking was not capability but intent — specifically, the intent to build local ecosystems rather than national production units. The Planning Commission designed these projects as nodes in a national system. The question “what happens to the village around the factory” was never part of the design specification. It was an externality, in the precise economic sense: a cost not borne by the decision-maker.

This is where the cathedral metaphor reveals its deepest truth. A medieval cathedral was not built for the village. It was built for God. The villagers could worship in it, could marvel at its stained glass, could take shelter in its shadow. But the cathedral was designed by an architect who answered to a bishop who answered to the Pope. The local community was the site, not the purpose. The Nehruvian cathedrals were designed by planners who answered to the Planning Commission which answered to the Prime Minister. Odisha was the site. The nation was the purpose.

The Legacy in 1980

By 1980, the Nehruvian period had left Odisha with a specific and consequential inheritance.

On the asset side: one of India’s longest dams, a functional steel plant, and the foundations of what would become India’s largest integrated aluminium producer. These were real. They provided employment, generated economic activity, and connected Odisha to the national industrial system. The irrigation from Hirakud had improved agricultural productivity in the delta. The power from Hirakud and later from NALCO’s captive plant enabled further industrial development. The state was no longer entirely agricultural.

On the liability side: 325 villages submerged for Hirakud. Over 30,000 families displaced across the three projects. Tribal communities dispossessed of their land, their forests, and their way of life. Compensation promises systematically dishonoured. A cosmopolitan steel town surrounded by subsistence agriculture. An aluminium smelter that would send nearly Rs 1,000 crore per year to Delhi from one of India’s poorest districts. And — the deepest liability of all — no industrial ecosystem. No supplier networks. No local entrepreneurial class. No technical workforce rooted in the local economy. No organic industrial culture.

The Freight Equalization Policy, running simultaneously from 1952, had ensured that the natural cost advantages of sitting on mineral deposits could not be exploited for local manufacturing. The knowledge generated by the cathedrals stayed within their walls. The profits flowed to the center. The costs stayed with the people.

Manufacturing employment in Odisha at the end of this period was approximately 5 to 6 percent of the total workforce. Agriculture still employed over 60 percent. The GSDP share of industry had risen from roughly 10 percent in 1950 to perhaps 18 to 22 percent by the early 1990s — but “industry” meant mining and metals, not the diversified manufacturing that was transforming Tamil Nadu, Gujarat, and Karnataka.

The cathedrals stood in the landscape. The village remained a village. The bazaar had not been born.

What happened next — the liberalization of 1991, the mining explosion, the POSCO debacle, the announcement economy — is a different chapter. But it is a chapter that can only be understood against the structural inheritance of 1957 to 1980: three decades in which Odisha received modernity in cathedral form, without the institutional architecture that would have allowed that modernity to reproduce.

I believe with roughly 80 percent confidence that the cathedral-without-ecosystem problem was not inevitable. Korea’s experience demonstrates that a different policy architecture could have produced a different outcome from similar initial conditions. But I hold with lower confidence — perhaps 50 percent — that the Indian state of the 1950s was institutionally capable of implementing Korean-style forced linkages even if the concept had been understood. The Planning Commission was designed to allocate national resources for national production, not to build local ecosystems. The intellectual framework for ecosystem thinking did not exist in Indian economic planning until decades later. The cathedral was not a mistake within the logic of the time. It was the logic of the time. The mistake was not revising that logic when its consequences became visible.

What those consequences look like at scale — when the cathedrals are joined by the mining explosion and the announcement economy — is where the story turns next.

Sources

“Hirakud Dam.” Wikipedia. https://en.wikipedia.org/wiki/Hirakud_Dam
“Mahanadi Valley Development Hirakud Dam Project.” Central Water Commission. https://cwc.gov.in/sites/default/files/hirakud.pdf
“30,000 farmers demand Hirakud dam water.” Down to Earth, 2007. https://www.downtoearth.org.in/news/30-000-farmers-demand-hirakud-dam-water-6893
“Hirakud Dam and Plight of Its Oustees.” EPW Engage. https://www.epw.in/engage/article/hirakud-dam-and-plight-its-oustees
“Development, Displacement and Justice in India: Study of Hirakud Dam.” ResearchGate. https://www.researchgate.net/publication/311736290
World Bank Dam Rehabilitation and Improvement Project: Hirakud RAP. https://documents1.worldbank.org/curated/en/979661531725488310/pdf/Hirakud-RAP.pdf
“Rourkela Steel Plant.” Wikipedia. https://en.wikipedia.org/wiki/Rourkela_Steel_Plant
“About Rourkela Steel Plant.” SAIL. https://www.sail.co.in/en/plants/about-rourkela-steel-plant
SAIL Financial Results FY22. PIB. https://pib.gov.in/PressReleaseIframePage.aspx?PRID=1827904
B.K. Roy Burman. “Social Processes in the Industrialization of Rourkela.” Academia. https://www.academia.edu/52884791
“Impact of industrialization on tribal livelihood: case study of Rourkela.” Academia. https://www.academia.edu/85749489
NALCO About Us. https://nalcoindia.com/company/about-us/
“National Aluminium Company.” Wikipedia. https://en.wikipedia.org/wiki/National_Aluminium_Company
NALCO FY24 results. Alcircle. https://www.alcircle.com/news/nalco-s-fy-2023-24-net-profit-hits-inr-2-060-crore-amid-strong-sales-and-export-growth-112096
NALCO dividend to Government of India FY25. https://nalcoindia.com/pre-rel/nalco-pays-rs-988-88-crore-as-dividend-to-government-of-india-for-fy-2024-25/
NALCO Pottangi opposition. Alcircle, 2024. https://www.alcircle.com/news/nalco-faces-growing-tribal-opposition-over-pottangi-bauxite-mines-in-koraput-odisha-116858
Eric S. Raymond. “The Cathedral and the Bazaar.” 1997. http://www.catb.org/~esr/writings/cathedral-bazaar/
“Economy of Jamshedpur.” Wikipedia. https://en.wikipedia.org/wiki/Economy_of_Jamshedpur
“Eastern India’s largest auto cluster seeks to emerge from Tata shadow.” Business Standard, 2024.
“POSCO.” Wikipedia. https://en.wikipedia.org/wiki/POSCO
“Establishment of POSCO.” K-Developedia, KDI School. https://www.kdevelopedia.org/Development-Overview/all/establishment-posco—201412070000355.do
“History of POSCO.” SteelOnTheNet. https://www.steelonthenet.com/history/posco.html
“Pohang.” Industrial Urbanism. https://www.industrialurbanism.com/pohang
“History of POSCO.” Funding Universe. https://www.fundinguniverse.com/company-histories/posco-history/
NBER Working Paper: “The Plant-Level View of an Industrial Policy.” https://www.nber.org/system/files/working_papers/w29252/w29252.pdf
John Firth and Ernest Liu. “Manufacturing Underdevelopment: India’s Freight Equalization Scheme.” Cornell University. https://barrett.dyson.cornell.edu/NEUDC/paper_316.pdf
Felix Padel and Samarendra Das. Out of This Earth: East India Adivasis and the Aluminium Cartel. Orient BlackSwan, 2010.
Pulin B. Nayak, Santosh C. Panda, Prasanta K. Pattanaik (eds). The Economy of Odisha: A Profile. OUP, 2016.

Source Research

The raw research that informs this series.