In 30 Seconds
Economist Sanjeev Sanyal has recently advanced a provocative explanation for Bengaluru’s emergence as India’s technology capital: the city’s rise was partly enabled by the exodus of entrepreneurial and technical talent from Kolkata during Jyoti Basu’s tenure as West Bengal’s chief minister. He argues that Kolkata lost its early technological lead because its institutional and political environment under the communist regime drove away entrepreneurial and technical talent, while Bengaluru offered the conditions in which that talent could accumulate and compound. Sanyal’s Kolkata-to-Bengaluru story is persuasive as a story about “where India’s technological talent accumulated”: Kolkata lost people, institutions, and entrepreneurial energy; Bengaluru accumulated them. But that still leaves the harder question untouched. Cities do not export software; firms do. Engineers do not become export revenue merely by moving south; something has to permit their skills to cross a national border and be sold abroad at scale. Bengaluru may explain the location of the cluster, but it cannot by itself explain the peculiar economic freedom enjoyed by the activity that cluster happened to specialise in. The crucial fact, then, is not simply that Bengaluru acquired the people who could write software. It is that India had created a narrow institutional corridor through which those people could sell their work to the world. Y2K would later flood that corridor with demand. But the corridor existed first.
The existing account of Y2K and Indian software (real bug, real remediation demand, real wage arbitrage, resulting path dependency) is also broadly correct and well documented. But it treats the outcome as a story about demand: an American crisis met an Indian labor surplus. That framing quietly assumes India’s economy was a level field on which software simply happened to be the sector with the biggest external customer. It wasn’t a level field. Years before anyone in Bangalore had heard of a two-digit date field, the Indian state had already built a “regulatory enclave” that made software structurally easier to produce, export, and profit from than almost anything manufactured. Y2K did not select India’s engineers because India was generically good at cheap technical labor. It selected them because India had already constructed one specific institutional channel, and only one, through which that labor could reach the world market without first fighting the Indian state.
The “enclave” was built for a different reason, a few years earlier
In 1991, alongside the balance-of-payments crisis and the broader liberalization of that year, the Government of India created the Software Technology Parks of India (STPI) scheme. Its terms are worth stating plainly, because they are usually mentioned only in passing: units registered under STPI received up to a decade of complete corporate income-tax exemption on export profits under Sections 10A and 10B of the Income-Tax Act, 100 percent foreign equity ownership without the usual approval gauntlet, duty-free import of all capital equipment, single-window statutory clearance in place of the normal multi-agency licensing process, and, critically, dedicated, high-speed international data links that let a firm bypass India’s chronically unreliable domestic telecom and power grid entirely (extended even today).
No comparable regime existed for a firm trying to manufacture auto components, machine tools, or electronics for export in the same decade. A manufacturer still faced tariff-protected input costs, land acquisition law, restrictive labor statutes governing factory-floor hiring and layoffs, an unreliable power supply it had to solve for itself, and a customs and excise bureaucracy that had not been redesigned around export promotion in the way STPI had. The 1991 reforms loosened industrial licensing generally, but they did not build manufacturing an equivalent tax-free, infrastructure-insulated enclave. They built one for software specifically.
This matters because it inverts the causal order in the standard narrative. The usual story runs: India had abundant technical labor and weak manufacturing infrastructure, so when Y2K demand arrived, labor flowed to software because that was India’s comparative advantage. The structural point is different: India’s comparative advantage in software was not simply discovered by the market in 1999; it was substantially manufactured by policy in 1991, for reasons that had nothing to do with Y2K (chiefly the foreign-exchange crisis that made any dollar-earning, non-oil export sector strategically precious). By the time Y2K created its surge in demand, the channel through which that demand could be absorbed without triggering India’s usual industrial obstacles was already open, tested, and staffed. Y2K did not build the enclave. It filled it.
Why this reframing matters more than the “arbitrage” story
The wage-arbitrage account explains why American firms wanted Indian labor. It does not explain why Indian capital and Indian graduates disproportionately flowed toward software rather than, say, export-oriented electronics assembly, where East Asian wage levels were not obviously lower than India’s and where India had genuine natural advantages in copper, textiles-adjacent light manufacturing, and a large internal market. The STPI framework supplies that missing mechanism: for a rupee of investment, software carried close to zero effective tax rate, no capital-goods duty, no dependence on Indian roads or ports, and no exposure to Indian labor law’s exit costs, since programmers were not covered by the same industrial-dispute statutes as factory workers. Manufacturing carried the opposite profile on every one of those dimensions. A rational allocator of capital or of a child’s education did not need Y2K to make this choice legible. Y2K simply supplied the demand shock that proved, at scale and under a hard global deadline, that the enclave worked.
This is a stronger and more falsifiable claim than “India had a comparative advantage in cheap English-speaking engineers,” because it identifies a specific, dated, documented policy instrument as the switch that was thrown, rather than treating the outcome as the natural unfolding of factor endowments. It also explains a puzzle the standard account leaves hanging: why India’s software export take-off (from roughly $131 million in 1990-91 to several billion within a decade) was so much sharper than its overall liberalization-era growth in any comparable export category. Liberalization loosened constraints economy-wide; STPI removed them entirely, but only in one sector.
The Y2K panic as enclave-legitimation, not enclave-creation
There is no denying that the Y2K anxiety was partly a self-reinforcing institutional phenomenon: asymmetric incentives made over-preparation rational for Western firms regardless of the bug’s true severity. The “enclave framing” adds the missing other half: that same over-preparation needed somewhere to land, and STPI-registered Indian firms were the only large, tax-privileged, telecom-insulated, dollar-invoicing pool of programming labor structured to absorb a sudden, massive, time-boxed export order. Firms like TCS, Infosys, and Wipro were not simply “well positioned” by history; they were operating inside the one institutional wrapper the Indian state had engineered to make exactly this kind of order frictionless. Their Y2K contracts were, in a very literal sense, the enclave’s first stress test at global scale, and it passed.
What this changes about the manufacturing counterfactual
One may reasonably ask whether India’s engineers would have gone into manufacturing absent Y2K, and would conclude probably not, since manufacturing required capital and infrastructure India lacked. The “enclave framing” sharpens this: it was not merely that manufacturing required more infrastructure investment than software. It was that the Indian state had made a specific, if not fully deliberate, choice to underwrite the absence of infrastructure for software (via dedicated satellite links and duty-free imports substituting for national grids and ports) while offering manufacturing no equivalent underwriting for the infrastructure it actually needed. The two sectors were not competing on a common institutional playing field for the same pool of engineers; one sector had been given a private bypass road around India’s infrastructure problem, and the other had not. Given that asymmetry, it would have been surprising if capital and talent had not concentrated overwhelmingly in software once a large external demand shock arrived, Y2K or otherwise.
None of this requires an American conspiracy, and it does not require an Indian one either. STPI was designed to solve a 1991 foreign-exchange emergency, not to anticipate a 1999 calendar bug, and its architects were almost certainly not thinking about deindustrialization when they wrote it. But it does relocate the most consequential decision in this story from Washington and Bangalore’s corporate boardrooms to New Delhi’s tax code, years before the bug that supposedly set everything in motion. The genuine hidden mechanism is not that America needed cheap labor and found it in India. It is that India, facing its own crisis, built a tax-free island economy that only software could inhabit, and then let a foreign crisis discover the island for it.
The residual question
This does not resolve whether that choice was, on balance, good policy: an enclave that generates dollar earnings, urban middle-class mobility, and a globally competitive service export sector is not a trivial achievement, and the counterfactual manufacturing sector might well have failed regardless. But it does mean the “trap,” i.e., rapid software growth without a matching manufacturing transformation, should be understood less as an unintended consequence of Y2K and more as the foreseeable output of an institutional design that, quite apart from any American crisis, had already picked a winner half a decade before the winner was needed.
