IMAGE: BOOGIE FOTO
THE ECONOMIST Dani Rodrik coined the term “premature deindustrialisation” in 2016 to describe a pattern he saw emerging across the developing world: countries whose manufacturing sectors were peaking at much lower levels of income than they had in earlier industrialising nations, and contracting before those countries had accumulated the capital as well as skills and productivity gains that manufacturing historically provides.
The classic path ran from agriculture through manufacturing to services; manufacturing was the ramp that converted cheap unskilled rural labour into higher-productivity urban employment, built export capacity, generated the foreign exchange to import technology and created the middle-income consumer class that services could then serve.
That ramp is missing in Nepal; the country has gone from agriculture (still employing 65% of the labour force) to services (62% of GDP) while manufacturing contributed only 4.37% of GDP in fiscal year 2024-25, less than half its share in 1996 and below the world average.
Understanding how this happened requires going back to choices Nepal did not make at certain historical moments.
The first moment was the Multi-Fibre Arrangement. From the 1970s onwards, Nepal’s garment industry grew under the MFA’s quota system, which allocated export volumes to developing countries and allowed even small, inefficient producers to access American and European markets at preferential terms.
By the early 2000s, garments were Nepal’s largest export sector (worth some Rs12bn). The quotas were scheduled to end in January 2005; everyone in the industry knew this a decade in advance.
The Bangladeshi garment industry used that decade to invest in productivity, consolidate into larger plants, improve quality systems and build the commercial relationships with buyers that would survive the quota’s removal.
Nepal’s garment firms, for the most part, stayed small; continued re-labelling imported fabric for quick margins; and did not invest in the competitive capabilities that the post-quota era would require.
When January 2005 arrived, Bangladesh’s industry absorbed the shock and accelerated; Nepal’s largely collapsed. Bangladesh now generates roughly $40bn in annual garment exports. Nepal generates about $90m. That was the industrial ramp Nepal most clearly had and discarded.
The second reason is geography, but not in the way it is usually invoked. Nepal is landlocked (a real constraint on export logistics); but Vietnam is not especially well-positioned geographically and built a $350bn manufacturing export sector in three decades. Bangladesh is a cyclone-prone delta with periodic flooding and built $40bn in garments. The relevant geographic constraint in Nepal’s case is not the absence of a port but the presence of an open border with India, which is simultaneously Nepal’s largest trading partner, its largest competitor and the source of imported goods that arrive faster and cheaper than domestically produced equivalents for almost every non-tradeable product.
A Nepali soap manufacturer competes with Hindustan Unilever products that cross the border with minimal friction; a Nepali food processor competes with Parle and Maggi; a Nepali plastic fabricator competes with goods from Surat and Ludhiana. The price competition from India, via an essentially unmonitored 1,700-kilometre border, has functioned as a permanent industrial policy against domestic manufacturing.
The third reason is the political economy of tariff revenue. Nepal collects some Rs1trn a year at its customs points (duties, VAT, excise); this is close to half of all government treasury receipts. A government that funds its civil service salary bill through import volumes has an incentive to keep imports flowing.
Import substitution policies (higher tariffs, domestic production mandates) would reduce this revenue stream in the short term and require the government to find other sources; over thirty years, no government has been willing to make that transition. The state has therefore maintained an implicit subsidy to trading over manufacturing without ever articulating it as policy.
The cement sector is the exception that clarifies the rule. Two decades ago, Nepal imported 90% of its cement; today it produces its own, with a sector worth some Rs150bn and some exports to India.
The conditions that allowed this are the following: limestone deposits in Nepal’s hills; a product too heavy to import cheaply from distant sources; demand large enough for scale economies; foreign equity partners who financed large plants; and state protection through export subsidies and cheaper industrial electricity tariffs. Cement succeeded because the competitive conditions lined up in Nepal’s favour, not because industrial policy in general worked. The same conditions do not exist for most manufactured goods; they cannot be manufactured into existence by announcing another policy.
What premature deindustrialisation costs Nepal is not primarily the GDP percentage point, though that matters. It costs the learning-by-doing that manufacturing provides and services do not. A factory worker who moves from agriculture to a textile plant improves their productivity through routine experience; they acquire technical skills and supply-chain knowledge that transfers to adjacent sectors. This is how East Asia built human capital without universal higher education: factory floors as schools. Nepal’s workers, moving from agriculture to construction, remittance-funded retail and the informal service sector, are accumulating different and less transferable skills. The IT sector (100,000 workers, $1bn in exports) is an alternative learning environment; but it absorbs 100,000 people in an economy that needs to find productive work for 450,000 new labour market entrants every year. The gap is what manufacturing was supposed to fill.
Rodrik’s original warning was that countries stuck in premature deindustrialisation tend to grow more slowly and with greater income volatility than countries that went through a full manufacturing phase; they have weaker export bases; more dependence on commodity cycles or remittances; and less resilience to external shocks.
Nepal’s fiscal year 2025-26 growth rate of 3.85% (against a target of 6%) in an economy carrying 28.2% of GDP in remittances is a reasonably clean illustration of what he described. The ramp was always available. The question is whether it still is; and the true answer is that it is considerably harder to build a manufacturing sector from 4.37% of GDP, against Indian competition on an open border, with interest rates at 9 to 12% and a banking system more comfortable lending to traders, than it would have been to maintain one from the 10% the sector held thirty years ago. ■
