Copyright © The Author(s) under exclusive licence to Editorial Office, Indian Economic Review 2023, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic.
All data used in the article are from official government sources, as mentioned above each Figure.
This paper provides a comprehensive analytical account of the major industrial policies in India since independence. It identifies three phases: 1948–1980, a period characterised by increasing state intervention; 1980–1991: a transitional period of gradual reforms; and 1991–2020: a period of extensive market-oriented reforms. Within each period, it reviews the major policy changes, and discusses possible reasons for their adoption. It also provides a brief account of industrial performance during each phase, and a more detailed review of how those policies have been evaluated by scholars from different perspectives. The discussion is supplemented by simple explanations for some economic theories and empirical methods that have been used in the literature. The review concludes with an eclectic perspective on the record of industrial policy, and some suggestions for the future.
Keywords: India, Industrial policy, Manufacturing, Trade policy, Licensing, Economic reformsThis policy review provides a more or less comprehensive chronological account of the major industrial policies adopted by the Indian government since independence, and how scholars have evaluated them from different perspectives that are often seen by their protagonists as mutually incompatible. While acknowledging their strengths and weaknesses, an attempt will be made to reconcile these perspectives wherever possible. Some economic theories and empirical methods that have been used in this literature will be explained in simple terms for the benefit of readers with only a basic knowledge of economics. The possible reasons why different policies were adopted at different times are also touched upon. The review is structured in four sections. Section 2 gives a very brief description of the state of Indian industry at independence. Section 3 covers the period of detailed state control over industry between 1948 and 1980. Section 4 briefly covers the transitional phase of gradual economic reforms during the 1980s. Section 5 deals at length with the period of far-reaching liberalization since 1991. This being a review article, it offers no original research contribution, but I shall give my own views towards the end of each section, and some suggestions for policy in the concluding Sect. 6.
Although I use the term industrial policy, the focus throughout will be on policies relating to the manufacturing sector. This excludes construction, electricity, gas and water supply, which are usually included in the category of industry. Within manufacturing, the focus will be on what is referred to by most researchers as the organized, registered, or formal sector. This is because development of this sector has been the main objective of industrial policy in both the pre- and post-reform eras, and because data on the unorganized/unregistered/informal sector is much more sparse. 1 Consequently, organized manufacturing has received disproportionate attention from researchers, although as late as 2017, the unorganized sector contributed a third of gross value added in manufacturing and employed two-thirds of the manufacturing workforce. These proportions had not changed significantly since the beginning of this century (Krishna et al., 2022a, ch.8). However, at various points, I shall briefly discuss the relationship between the organized and unorganized sectors. Covering even a limited range of issues concerning the organized sector, and the associated literature, has extended this review to an excessive length, so it has not been possible to deal with more specialized topics such as the costs and benefits of foreign direct investment, regional imbalances in industrial development, technology policy, or the growing concern with the environmental effects of industrialization. I also do not go deeply into two subjects on which I have written at length elsewhere: labor market regulation (Bhattacharjea, 2021) and competition (antitrust) policy (Bhattacharjea, 2010; Bhattacharjea & De, 2021).
India suffered from ‘deindustrialization’ in the nineteenth century, when its traditional handicraft industries were unable to withstand the competition from machine-made imports, under the colonial policy of free trade. Some modern cotton and jute mills were set up in the later part of the century, and the Tata Iron and Steel Company commenced production in 1911. More industries came up in the years between the two world wars, especially after 1923 when the colonial government began to impose modest import tariffs, principally for generating revenue. Yet, by 1951, the manufacturing sector of the newly independent country contributed only about 14% of national income and employed about 10% of the labor force, of which only 2% worked in modern (mechanized) industry. The modern manufacturing sector was based mainly on the processing of agricultural raw materials (cotton and jute textiles, paper, sugar, vegetable oils), or minerals (iron and steel, cement). All these industries were reliant on imported machinery and equipment. Domestic manufacture of capital goods was virtually non-existent, except for limited production of textile machinery and railway engines. Policies adopted by the government of independent India have to be understood in this context.
The most important industrial policies of this era can be summed up as: Planning, Protection, Public sector, Industrial licensing and Price controls.
Planning took the form of successive Five-Year Plans, beginning in 1951. The Plan documents could be viewed as vision statements for the economy, with priorities changing over time. But they also embodied different conceptual approaches. The focus of the first three plans was on increasing the rates of savings and investment, and the sectoral allocation of investment. The First Plan (1951–56) was largely a collection of several investment projects, mainly in the public sector. It also provided a growth projection based on an exercise resembling the then-dominant Harrod–Domar model, but this did not much influence actual Plan formulation. It was the Second Plan (1956–61), based on what came to be known as the Mahalanobis model, that launched the pattern of industrial development that was followed for several decades thereafter.
The Harrod–Domar model had shown that an increase in an economy’s rate of growth required an increase in its domestic savings rate, supplemented by foreign capital (which at that time took the form of financial assistance from foreign governments). It was presumed that higher savings would automatically be transformed into higher investments. But how could physical investment be increased in an economy that lacked a domestic capital goods producing sector? The Mahalanobis model addressed this ‘transformation problem’ in a particular way. It showed that if the economy could devote a higher proportion of whatever capital goods it produced to increase the capacity of the capital goods producing sector itself, its long-run growth rate would be higher, because a progressively larger capital goods sector would enable it to sustain a higher rate of physical investment. 3 Given that the productivity of capital was higher in the consumption goods sector as compared to the capital goods sector, this pattern of investment allocation in favor of greater production of capital goods would result in the production of consumer goods growing more slowly in the short run, but attaining higher levels in the future. In the meanwhile, traditional and small-scale industries would cater to consumption demand, while generating employment. With long-run growth being seen as the means for reducing widespread poverty, the model provided an intellectual justification for increasing investments in the capital goods sector of a labor-abundant country. This was the so-called ‘heavy industry bias’, which showed up in growth rates of investment and output in the machinery, metals and chemical industries exceeding those of consumer goods industries.
The Third Plan (1961–66) followed the same basic philosophy, with greater attention to inter-sectoral consistency (i.e., outputs of industries were to be planned in a coordinated manner, so as to prevent shortages in the supply of inputs required by the industries that used them). But the adverse events of the 1960s and early 1970s 4 made plan priorities take a back seat to crisis management; in fact there was a ‘plan holiday’ from 1966 to 1969. The next two plans paid greater attention to agriculture, poverty, and employment generation, but lacked the kind of overarching conceptual framework that had been provided by the Mahalanobis model—although the latter was never formally abandoned, and the Plan documents continued to specify detailed sector-wise targets for public and private sector investments and outputs.
The Mahalanobis model did not lay down specific industrial policies, but its heavy-industry bias entailed several policies that came to characterize the pre-reform period. The most obvious was the vastly increased role of the public sector. It was believed that private investment would not be forthcoming for such capital-intensive projects, characterized by long gestation lags before they could yield profits. The 1948 Industrial Policy Resolution (IPR) of the government of newly independent India had already reserved production of arms and ammunition for the central government. New establishments for the manufacture of iron and steel, aircraft, ships, telephone, telegraph and wireless equipment, were reserved for undertakings owned by the central or state governments. Existing private producers in these industries could continue, but nationalization could be considered after 10 years. The list of industries reserved for the public sector was expanded in the 1956 IPR to cover 14 industries in which new establishments were to be set up only by the state, which would also take investment initiatives to acquire a leading role in 12 more. The public sector could also enter any other industry. Now, after the 1955 Avadi session of the ruling Congress party had committed itself to building a ‘socialistic pattern of society’, the state was to control the ‘commanding heights’ of the economy.
A second implication of planning followed from the need to allocate scarce resources to the priority industries. This could not be achieved through the market mechanism, because the public sector could not bid resources away from the more profitable industries that were left for the private sector. Therefore, investment licensing was needed so as to limit private firms’ production of inessential items, and to assign them production targets as part of the Plans. It was also supposed to control the concentration of economic power, protect small-scale industries, and promote regional balance in industrialization. Under the Industries (Development and Regulation) Act of 1951, private firms had to apply for licenses to set up a new factory, significantly expand the capacity or change the location of an existing factory, or to produce new products. They also required permission to import foreign technology. Limits were imposed on the duration of technical collaboration agreements, as well as fees and royalties that were payable to foreign technology licensors.
A third implication was that new industries would need protection from import competition, because they would be operating with unfamiliar technologies and at uneconomically small scales of production. Protection took the form of both high tariffs (up to 300% for some items), and a regime of strict import licensing for almost all goods, described in more detail below. Finally, with both imports and domestic production being subject to licensing restrictions (on top of other constraints on the availability of infrastructure, capital, and skilled labor), supply could not respond readily to growing demand, so shortages of many items emerged and their prices rose. Since many of these were regarded as ‘essential’, the government imposed price controls. And, as elementary supply and demand theory would predict, this resulted in chronic shortages, so distribution controls also had to be imposed on many of these items, in the form of rationing rules to determine who could buy them, and how much. 5 In some cases, e.g., cement, cloth, and sugar, a dual pricing scheme was enforced whereby the industry had to sell a share of its output at a controlled price to the government or other authorized buyers, and the rest at market-determined prices.
The foregoing discussion suggests a certain logical coherence among the policies that were adopted, but it does not mean to imply that there was a detailed plan that integrated them. In fact, much of the regulatory apparatus was set up in 1951, before the Mahalanobis model was conceived. Significant trade restrictions were imposed only after a balance of payments crisis in 1957. Plan targets were set for only broad industrial categories; the allocation of licenses for individual products, and to individual firms producing those products, were left to administrative discretion. Policies such as import licensing or price controls were imposed on an ad hoc basis to deal with shortages as they arose.
The influences that led the government to adopt these policies also require some clarification. The need for rapid industrialization, and these policy imperatives, were of course consistent with the beliefs of Prime Minister Nehru and the ruling Congress party. But state-directed industrialization was the centerpiece of the newly emerging field of development economics, and had even been articulated in the so-called ‘Bombay Plan’, devised by India’s leading industrialists as far back as 1944. They knew that even modest tariff protection had encouraged some industrialization in recent decades, and that they lacked the capacity to undertake investments in infrastructure and capital goods industries on the scale required. The Indian planners seem to have believed that diversification of the industrial structure would enable India to export manufactured products which, unlike primary products, did not face inelastic demand and deteriorating terms of trade. This later came to be described as ‘export pessimism’. But the policies that were actually implemented went much further than what these seemingly reasonable arguments could justify. Expansion of the public sector went far beyond the vision of the Bombay Plan, which had proposed public investment only in ‘strategically important’ industries, with their eventual privatization. On the other hand, policy towards foreign direct investment remained fairly liberal until the late 1960s, despite the industrialists’ hostility. The way in which protection was granted also went far beyond what they had proposed.
Tariffs keep domestic prices somewhat in line with international prices, maintaining competitive pressure on domestic producers. Tariffs also bring revenue to the government. However, it was the system of import licensing that came to dominate the import regime. This kind of quantitative restriction or ‘non-tariff barrier’ effectively delinks domestic prices from those in the rest of the world. It allows domestic producers who have monopoly/oligopoly power to increase their prices much more than what an equally restrictive tariff would allow. Instead of the government, those who get the licenses reap the scarcity premium which is implicit in the difference between the domestic and import prices.
These negative aspects of the regime were reinforced by the manner in which it was implemented. Commercial import of most consumer goods was prohibited. For importing raw materials, capital and intermediate goods, industrialists had to apply for licenses, for which they first had to get certificates of ‘essentiality’ and ‘indigenous non-availability’ from different government agencies. The latter requirement meant that any item produced in India could not be imported, thus automatically guaranteeing domestic producers of those items complete protection from foreign competition. This also reduced the competitiveness of the industries that used those items as intermediate inputs, because they had to buy their requirements from domestic suppliers regardless of cost or quality. Moreover, the licenses were non-transferable, so that even given the foreign exchange constraint, more efficient industries and firms (or new entrants) could not legally buy their import entitlements from the less efficient ones. (A black market existed, which allowed some researchers to infer the magnitude of the scarcity premiums on import licenses.) All this especially harmed export competitiveness.
The actual process of license allocation was not based on economic criteria or sectoral priorities. The relevant authorities did not have the expertise to determine the ‘essentiality’ of inputs required for the hundreds of industries they were supposed to supervise. They lacked the information to judge whether domestic production, which would automatically rule out imports, was adequate to supply the user industries. Most licenses were allotted so as to ensure fair sharing among applicants. This could be on the basis of their past requirements, which kept out potential entrants who might have used the imported inputs more efficiently; or based on the applicants’ capacity, which encouraged firms to install excess capacity in order to secure imported inputs for their existing capacity. Or the relevant authorities simply made arbitrary allocations. Estimates of the effective rate of protection (ERP), taking into account tariffs as well as the import premiums generated by licensing restrictions on both inputs and outputs, exceeded 1000% in some cases, and varied wildly in ways that could not possibly have been intended by the policymakers. 6
Similar problems plagued the investment licensing regime. Prior to getting a license, an entrepreneur had to get approvals from different authorities for importing capital equipment and components (as above), foreign collaboration, and raising funds from the capital market. Even after obtaining all these, the absence of clear criteria for license allotment paved the way for arbitrariness, delays (often over a year), inefficient allocation, and corruption. The licensing system was supposed to be used to allocate Plan targets to the private sector. But this was an unachievable objective, because firms were seldom penalized for producing more than their licensed capacity, and could not be penalized for producing less, or for not installing it at all. Licenses were often allocated sequentially until the Plan targets were reached, without evaluating the relative merits of the proposals. As official inquiry committees showed in the mid-1960s, big business groups that could maintain offices in Delhi and had close connections with bureaucrats and politicians were disproportionately successful in getting licenses. In several cases, these licenses were not used, so it was apparent that they had been obtained only to keep out potential competitors.
Restrictions on the private sector expanded greatly after 1969 when Indira Gandhi’s government, with its renewed commitment to ‘socialism’, nationalized several chronically loss-making private manufacturing firms which were on the verge of shutting down. (This was done to save the employment of their workers, and therefore, different from the strategic nationalization of 14 major banks, coal mines, and the Indian operations of the multinational oil companies, which took place in the same years.) The Monopolies and Restrictive Trade Practices (MRTP) Act (passed in 1969 and brought into force the following year) reinforced the licensing system by requiring firms or business groups with more than Rs 20 crores in assets to get additional permissions for substantial expansion, setting up a new unit, or mergers and acquisitions. The Foreign Exchange Regulation Act (FERA) of 1973 imposed a 40% cap on foreign equity participation, except in export-oriented or high-tech activities. Companies covered by the MRTP Act and FERA were restricted to investing in a specified list of ‘core’ industries. There was an increase from around 30 to nearly 200 in the number of items reserved for small-scale industries. The number jumped to over 800 under the Janata Party regime (1977–80). In 1976, a new chapter V-B was added to the 1947 Industrial Disputes Act, making it mandatory for factories with more than 300 workers to get government permission to fire them or to close down. On the other hand, the 1970 Patents Act provided a boost to Indian pharmaceutical firms by allowing patenting of processes but not products. This allowed domestic firms to copy patented foreign drugs using a slightly different manufacturing process. Even process patents for this sector were limited to 7 years, as against 14 for others. But concurrently, price control was extended to the pharma sector, and the list of covered drugs was steadily increased over the next 2 decades.
It would be hard to find a serious scholar, even among critics of post-1980 ‘neoliberal’ reforms, who is nostalgic about the policies of the earlier period. None of them has called for a return to the ‘licence-permit raj’, as it came to be known. Nor have any of the political parties, across the ideological spectrum, which have been part of the Indian government in the last 3 decades. But that does not mean that the pre-1980 period should be regarded as an era of complete darkness. As Balakrishnan (2022) reminds us, the growth rates attained during the Nehru era were much higher not only than in the preceding half-century of colonial rule in India, but also the contemporaneous growth rate of China, and the growth rates of the United States, United Kingdom and Japan over the period 1820–1992. He refutes two common misconceptions about the Nehruvian era by showing that public sector enterprises actually generated budgetary surpluses, and private investment accelerated.
However, performance over the subsequent 15 years up to 1980 was much worse. The average rate of manufacturing growth halved from nearly 7% to 3.5% per annum in the decade upto 1975, with the growth rate of registered manufacturing falling from 8.3 to 2.7% per annum. A slight recovery in the next 5 years pulled up the overall manufacturing growth rate for 1965–80, but only to 4.6% per annum (Sivasubramonian, 2000, p. 588). Even if India’s performance had earlier been respectable by international standards, it was now falling far behind other developing countries in East and Southeast Asia. They too had begun with small manufacturing sectors based on processing of primary products, but they rapidly industrialized and became major exporters of manufactured goods. So what went wrong? While acknowledging the undeniable role of external shocks (see n. 4 above), economists have taken two views on this: a microeconomic one based on the deleterious effects of intensifying controls over the private sector, and a macroeconomic one based on intersectoral linkages and the significant decline in public investment, which had been the driver of growth in the preceding 15 years.
Even if one disregards the imaginary world of allocative efficiency based on free trade, it cannot be denied that the way in which industrialization was promoted in India entailed a range of productive inefficiencies. Licenses, as we saw above, were allocated without economic criteria. They were fragmented among firms without regard to minimum efficient scale. The practice of preserving past shares among applicants, and their almost complete protection from foreign competition, prevented more efficient suppliers from expanding at the cost of the less efficient. The expansion of MRTP and FERA firms was curbed, preventing them from competing effectively not only with smaller domestic firms, but also against each other. Limited competition meant that availability, productivity, product diversity, and quality suffered. Entrepreneurs and managers had to expend resources in chasing licenses rather than investing in better products and techniques. Large firms were constrained from producing at scale by the licensing and MRTP restrictions; small firms producing items reserved for small-scale industries (SSI) could not expand or upgrade their technology, because they were not allowed to exceed the thresholds (in terms of value of plant and machinery) that defined SSI. In order to protect the handloom industry, restrictions were imposed on the expansion of weaving capacity in textile mills. Price ceilings inhibited investment to expand capacities of items in short supply, perpetuating their shortage. Ceilings were set so as to cover the costs of the least efficient units, either uniformly across all plants (e.g., for steel), or plant-wise (e.g., fertilizers), thereby rewarding inefficiency. These policies also contributed to the growing number of ‘sick’ firms which were chronically loss making and unable to repay their loans, burdening the banks with non-productive assets. Restrictions on mergers made it difficult for other firms to acquire the land and machinery of the sick companies and put them to alternative uses.
The finances of many public sector manufacturing undertakings turned from surpluses to chronic losses. The reasons included fragmentation of public sector investments to satisfy regional balance, without regard to minimum efficient scale or locational efficiency; design flaws; delays in project execution; administered pricing to protect consumers; social obligations as model employers (well-equipped industrial townships and better wages and benefits than comparable private sector firms); and political interference in location, employment, and procurement decisions. Since public sector enterprises’ losses were borne by the public exchequer, and their managers were hard to fire, there was no incentive to keep costs under control. Moreover, the nationalization of loss-making private enterprises meant that from being a nursery for new industries, the public sector came to be relied on as a hospice for terminally sick private firms. No doubt many of these policies were geared to other important objectives like protecting workers and consumers, as well as promoting SSI and regional balance, but they had an inevitable cost in terms of productivity and quality.
Turning now to trade policy, the main line of criticism from the perspective of neoclassical trade theory would be that import protection diverts fully employed factors of production away from exporting sectors to those producing import substitutes, a reallocation that defies a country’s comparative advantage and reduces social welfare as defined in that framework. But even if we disregard the neoclassical approach, one can readily appreciate the productive inefficiencies (including misallocation of resources between firms with different levels of efficiency within the same industry) and poor quality resulting from the policy regime, as described above. 7 This showed up in poor international competitiveness. The substantial import requirements for capital goods and components which were not produced in India, on top of irreducible imports of food, oil, and defense equipment, put a chronic strain on the balance of payments. Addressing this via import controls instead of devaluation meant that the rupee remained overvalued, which further undermined export competitiveness. 8 Production for the domestic market, automatically protected by the ‘indigenous non-availability’ requirement for import licenses, was obviously far more attractive. The biggest exporter of manufactures among developing countries in the 1950s, India slipped to seventh place in 1978; her share of world exports decreased from 2.4% in 1948 to 0.4% in 1979 (Wolf, 1982, pp. 18, 37).
As for the second, macro-oriented perspective, the correlation between public investment and industrial growth is quite evident. A use-based classification of the index of industrial production shows that capital goods industries, which had the highest rate of growth during the first 15 years, recorded the sharpest deceleration in the next 15. Correspondingly, public investment grew rapidly in the first 15 years and slumped in the next decade, along with private investment (Sivasubramonian 2000, pp. 589–98; Balakrishnan, 2010, pp. 129–131). The theoretical connection between the two was articulated in a series of articles by Prabhat Patnaik, and independently by Pulapre Balakrishnan. The following is a synthesis of this perspective. Instead of being ‘crowded out’, as orthodox macroeconomics would have predicted, private investment in the first phase was ‘crowded in’ by public investment which generated exogenous demand, directly for the capital and intermediate goods produced by the private sector, and indirectly via wages and salaries of public employees being spent on consumer goods. It also facilitated private investment on the supply side by creating infrastructure and providing capital and intermediate goods, and augmented agricultural output (the Green Revolution), thereby keeping the prices of wage goods of industrial workers in check. However, Patnaik (1984) sees inherent limitations on this process. Tax and other concessions to capitalists erode the government’s share of economic surplus, so its expenditures threaten to create inflationary pressures. Unless the government can afford to allow real wages to fall, it must cut back its capital spending. To the extent that this also affects investment in agriculture, it further tightens the wage goods constraint and threatens to worsen the inflation in food prices. Public investment must then be cut down further, causing a reduction in capacity utilization and, therefore, investment in the private sector. After sufficient demand deflation has brought inflation under control, public investment grows again, initiating another cycle.
Balakrishnan (2010) independently restates some of these arguments, and fleshes them out with an econometric demonstration of how cycles in public investment preceded cycles in private investment. From this ‘macro’ perspective, the cutbacks in public investment from the mid-1960s took their toll on private investment. Balakrishnan et al (2017) extend this to ‘producer services’ such as power, transport, communications, banking and insurance which are used as inputs into manufacturing. The authors develop a formal model of endogenous growth, and empirically demonstrate the positive feedback between such services and manufacturing after 1965. They suggest that public investment enabled these services to attain the threshold scale required to cover setup costs prior to 1965, after which the internal dynamics resulting from the interaction between services and manufacturing gave rise to successive accelerations. (To these mechanisms, I can add the role of knowledge spillovers and mobility of trained personnel from the leading firms.) All these mechanisms further enhance the productivity of industry, so ‘growth begets growth’ in a process of cumulative causation. Changes in trade and industrial policy, which are crucial determinants of growth in the ‘micro’ framework, play only a secondary role here.
A summary judgment on the first 30 years of industrial policy would be that public investment in heavy industries and infrastructure, protection of the home market, and licensing restrictions on consumer goods production, all encouraged the creation of a diversified manufacturing sector, which could produce a range of products that would otherwise have been inconceivable in a capital-scarce, technologically backward economy. However, the arbitrary way in which controls were implemented gave rise to accumulating inefficiencies. Investment incentives for the private sector, as illustrated by the huge inter-industry variation in ERPs, were surely the antithesis of planned development. Even the ‘socialistic’ goals of the governments of the pre-reform era were ill-served by the way in which controls were implemented. Price ceilings were imposed on steel, produced mainly in the public sector, but not on the private firms that used it to produce consumer goods to sell in the protected domestic market. Loans from public sector financial institutions, given at below-market rates, were often not repaid. This effectively meant that the public sector was subsidizing the private sector. Public sector losses were financed by a tax system that depended largely on regressive indirect taxes, or by deficit financing that fuelled inflation. Both of these had adverse distributive effects. The freeze on textile mills’ weaving capacity led to the growth of small powerloom factories that used the same technology with lower wages and worse working conditions, thus displacing employment in handlooms. Moreover, although controlling the concentration of economic power was one of the stated intentions of industrial policy, it was manipulated by oligopolistic business groups to protect themselves from both domestic and foreign competition. The MRTP Act was powerless to prevent them from forming cartels and dominating many markets (Bhattacharjea, 2010). For all these reasons, industrial policy in the pre-reform era can hardly be regarded as heralding a ‘planned’ economy, much less a socialist one.
Most earlier narratives of India’s post-independence economic policies treated 1991 as the crucial dividing line between the pre- and post-reform eras. But as shown by several scholars (Kohli 2006a; Nagaraj 2003), there was no significant acceleration in the economy’s growth rate for at least a decade after 1991. Balakrishnan (2010, ch.3 and Appendix) shows that for registered manufacturing, the sector most directly affected by the 1991 reforms, the acceleration actually began in 1982–83. 9
What were the reasons for this turnaround? Some earlier scholars who identified an upturn in growth from around 1980 have attributed it to Indira Gandhi’s return to power in that year, but they do not provide evidence of any specific policies other than the opening of some industries to private investment (Rodrik and Subramanian 2005; Kohli 2006a; and earlier authors cited by them). Their preferred explanation is that a shift occurred in the government’s attitude. Although its rhetoric changed from redistribution to growth and productivity, it adopted a ‘pro-business’ approach that strengthened incumbent private firms, rather than ‘pro-market’ policies that promoted foreign and domestic competition. But there were specific policy changes as well. Balakrishnan (2010) takes note of the Industrial Policy Statement of 1980, which allowed up to 25% automatic expansion of licensed capacities, ‘regularized’ capacities installed over the licensed limit in several industries, and doubled the ceiling on value of plant and equipment that defined SSI. ‘Broadbanding’ was allowed in some industries, whereby licenses could now be used to produce other items within the same broad industrial category. (Some of these measures had been introduced in 1975, but were ineffectively implemented.) I can add significant relaxation of price controls on cement and steel.
However, Balakrishnan questions the evidence of a productivity surge in the 1980s, which is an important part of the Rodrik–Subramanian argument. I too find their explanation inadequate. At least in two areas that I have studied intensively, legislation took an anti-business turn in the early 1980s. Chapter V-B of the Industrial Disputes Act was amended in 1982 so as to get around a Supreme Court judgment that had struck down part of the 1976 amendment, and to reduce from 300 to 100 the employment threshold at which establishments required government permission for layoff, retrenchment or closure. Then, the MRTP Act was amended in 1984 to reduce the threshold market share at which a firm or business group could be regarded as dominant in that market from 33 to 25%, and to reverse a landmark Supreme Court judgment that had allowed economic arguments to be raised in defense of trade practices that were regarded as ‘restrictive’ under the Act. 10 Of course, there has always been a huge gap between legislation and implementation, and the government’s pro-business attitude could have taken the form of slackening enforcement (the MRTP Act was already quite ineffective: see Bhattacharjea, 2010). But these amendments did increase the scope for harassment and sent out a different signal to the business community.
So, apart from any attitudinal shift towards big business, what lay behind the turnaround in the 1980s? Balakrishnan (2010) identifies two developments in the late 1970s as the main propellants: first, buoyant rural demand due to a succession of good harvests, and secondly, the revival of public investment in the mid-1970s, which was followed with a lag by private corporate investment. Rodrik and Subramanian (2005, pp. 212–214) show that productivity in manufacturing responded to public investment with a lag. However, applying the Bai-Perron methodology (see n.9) to a longer time series, Balakrishnan et al., (2017, Table 1) now find an acceleration in manufacturing growth from around 1985-86. This suggests that we should look more closely at developments in the second half of the decade.
What seems to have been overlooked by most authors are the significant changes introduced by the government of Rajiv Gandhi, who came to power in 1984 after the assassination of his mother. A longstanding corporate insider has recalled that Rajiv had been very open to listening to industrialists’ views even while he was General Secretary of the Congress party during his mother’s prime ministership. With his succession to that office, there was a decisive improvement in government–business relations (Das, 2017). Three major reforms were implemented in 1985. Industries that contributed nearly half of registered manufacturing output were removed from the scope of investment licensing (Aghion et al., 2008). Second, the MRTP Act was amended to quintuple the value of assets above which a firm or corporate group was subject to approvals for expansion and merger. A quarter of the firms that had been covered by this restriction were now freed from it. The list of industries open to companies remaining under MRTP/FERA was also expanded. Third, a New Textile Policy removed the freeze on mills’ weaving capacity, and allowed them full flexibility in choosing the mix of cotton and synthetics. Apart from these policies specific to industry, the Rajiv government began the task of modernizing India’s information technology and telecommunications infrastructure, which ultimately facilitated the software boom and also improved manufacturing performance.
Reforms on the trade policy front were somewhat mixed. Many capital and intermediate goods were transferred to the Open General Licence (OGL) category of imports which did not require government approval—a process that had begun earlier after the 1977 report of the Alexander Committee. Most of them were not produced in India, so this did not signify an increase in import competition. However, it did permit some modernization. The government also strengthened some measures to allow duty-free import of intermediate goods for export production, or to refund import duties on such imports. Apart from these formal policy changes, implementation of some of the remaining controls became more business-friendly, accelerating a trend that had begun a few years earlier. Investment licenses and MRTP clearances were granted more quickly, and foreign technological collaboration agreements were granted more liberally (World Bank, 1989, paras 30, 42, 52, 93, 3.14–3.16, 3.19, 3.27).
The effect of these supply-side relaxations was reinforced towards the end of the decade by a demand stimulus from two other developments which do not fit either the deregulation or public investment perspectives. One was real depreciation of the rupee, which was now linked to a basket of currencies. This stimulated exports and curbed imports, which were further restrained by increased tariffs. The other development was that the government’s consumption expenditure increased rapidly, even as its capital spending eased. But the consequential fiscal deficits sowed the seeds of what happened next.
The deteriorating fiscal situation was exacerbated by the political instability that followed the electoral defeat of the Rajiv Gandhi government in 1989. Growing current account deficits deteriorated into a full-blown crisis when the first Gulf War of 1990 led to a spike in oil prices and a fall in foreign exchange remittances from migrant workers who had to be repatriated from West Asia. This depleted India’s foreign exchange reserves, to the point where they could barely cover two weeks’ worth of imports. The political economy had also undergone a transformation. Kohli (2006b) points to the growing influence of the Confederation of Indian Industry (CII), which was more pro-reform than the older Associated Chambers of Commerce (ASSOCHAM) and Federation of Chambers of Commerce and Industry (FICCI), which were dominated by industrialists who feared competition. This was the context for the decisive break in economic policies that was pushed through by the government of Prime Minister Narasimha Rao, with Manmohan Singh as his Finance Minister. The then Director-General of the CII recalls their tenure as “The Glory Years” when “the changes in the [business-government] relationship during Rajiv Gandhi’s prime ministership were carried forward and institutionalized” (Das, 2017, p. 226).
There can be no doubt that 1991 did usher in a new era of more consistently ‘pro-market’ reforms. Remarkably, despite some partial reversals that I note below, the direction of reform has remained unaffected by the many changes of government since then. The specific reforms are best summarized in bullet points:
Almost all remaining industries were delicensed in 1991, leaving less than ten percent of 3-digit industries under licensing. By 2020, the only industries that remained under the licensing regime were electronic aerospace and defense equipment, industrial explosives, a few hazardous chemicals, and tobacco products.
The MRTP Act was amended in 1991, to remove completely the restrictions on expansion and mergers by large firms. Its remaining clauses on control of monopolistic and restrictive practices were ineffectively enforced until it was repealed altogether in 2009, making way for the new Competition Act, whose merger-related clauses were brought into effect only in 2011. Whereas a significant proportion of mergers had been blocked under the MRTP Act, there was no restriction on mergers on competition grounds between 1991 and 2011, and not a single merger has been blocked under the Competition Act.
Ten out of the 18 industries earlier reserved for the public sector were now opened to private investment. Out of the remaining eight, only one (arms, ammunition and defense equipment) was a manufacturing activity.
The FERA restrictions on FDI in manufacturing were progressively relaxed, beginning with automatic approval for upto 51% foreign equity in almost all manufacturing industries. This went up to 100% by the end of the decade.
Firms could now access the equity market without the range of approvals that had been required earlier.
Import licensing was abolished in 1991 for capital and intermediate goods. It remained in place for consumer goods for another decade, when India lost a dispute at the World Trade Organization (WTO) and was forced to abolish import licensing altogether, except for a small number of items that remain restricted on grounds of health, safety, morality, or compliance with international agreements, e.g., on narcotics or endangered species.
In 1991, the highest tariff rate on imports was reduced from 355 to 150%. The ‘peak’ Most Favoured Nation (MFN) rates were decreased thereafter to 50% in 1995–96, followed by a slight reversal, and then again more gradually to 10% by 2007–08. This still remains the peak rate. 11 From 1995, the upper limits on MFN rates for most products were ‘bound’ under the Uruguay Round WTO Agreements, but India’s actual rates have fallen far below these ceilings. Average MFN tariff rates on manufactures declined from 126% in 1990–91 to about 12% in 2014–15. Tariffs below these rates proliferated in the decade after 2004, as India entered into multiple preferential trade agreements, principally with countries in South and South-East Asia, and also gave duty-free access to imports from least-developed countries. Thanks to various concessions, by 2008–09 the weighted average of actual collection rates (total import duties collected divided by total imports) declined into the single digits. 12 One concern that has emerged is that tariffs on many consumer goods have fallen more rapidly than on raw materials and intermediate goods, resulting in inverted duty structures for many industries (Pathania & Bhattacharjea, 2020).
Trade liberalization has witnessed two tendencies in the reverse direction. First, after 1997 India became the world’s biggest user of anti-dumping duties. (These are additional duties imposed on specific exporters from specific countries, on the basis of complaints from domestic industries that can demonstrate that they have been injured by ‘dumped’ imports.) Imports of chemicals (including pharmaceutical ingredients), especially from China, were the main targets. There has been an increasing reliance on such duties in recent years. 13 However, because such duties were targeted rather than general, and were in many cases lifted after five years, this cushioned rather than reversed the impact of trade liberalization. Second, in 2018 the basic MFN rates on thousands of products were increased, followed by increases in some more products in the succeeding years. This has raised the average tariff rate from 13% to almost 18% (as calculated by Chatterjee & Subramanian, 2020), which is still far short of the levels prevailing before 1991.
In order to deal with the immediate balance of payments crisis in 1991, the rupee was devalued by about 19% in two stages. This somewhat offset the reduction in import duties by raising the c.i.f. price of imports in rupee terms. In the next few years India moved to an entirely market-determined rate, with the Reserve Bank of India occasionally intervening to maintain stability. The real effective exchange rate fell sharply after the initial devaluation, but then remained fairly stable with recurring episodes of real appreciation, caused by high domestic inflation relative to India’s trading partners, and/or inflows of foreign portfolio capital.
Beginning in 1997, products that had hitherto been reserved for SSI were progressively dereserved. From over a thousand products in 1996, only 22 remained reserved by 2008, and none by 2015 (Martin et al., 2017).
Most of the remaining price ceilings were abolished or relaxed, with the exception of pharmaceuticals, for which a new regime was instituted in 2013.
Another major change followed from India’s accession to the WTO in 1995, which required compliance with the Agreement on Trade Related Aspects of Intellectual Property Rights (TRIPS). After utilizing the transition period allowed to developing countries, the Patents Act was amended to provide a uniform 20-year duration for all patents, and to allow product patents for pharmaceutical products. Indian pharmaceutical firms would now have to obtain licenses from foreign patent holders to manufacture products that they had earlier copied with impunity.
The Special Economic Zones (SEZ) Act was passed in 2005. This gave tax benefits to businesses for developing and operating SEZs, and also to units located in those zones. The latter were also allowed, for their export production, to import intermediate inputs duty free and to buy domestic inputs exempted from indirect taxes. (In 2019, some SEZ benefits and other export-promotion measures were held to violate WTO rules, so the scheme will have to be revamped.)
Some state governments facilitated the creation of SEZs as well as other investment projects using their power of eminent domain under the 1894 Land Acquisition Act to acquire land for private entrepreneurs instead of public projects as hitherto. Many episodes of dispossession of landowners created a political backlash which forced the government to pass a new law, the Land Acquisition, Rehabilitation and Resettlement Act, 2013, which required the consent of 80% of the landowners for land acquired for private projects, a social impact assessment, and compensation, rehabilitation, and resettlement of those whose lands were acquired.
Non-recovery of corporate debt has been a chronic problem in India. It has loaded banks’ balance sheets with non-productive assets (NPAs) and made them unable or unwilling to lend to new borrowers. With workers, suppliers and tax authorities also making claims, the recovery proceedings stretched on for years, keeping thousands of ‘sick’ companies holding on to land and capital that could have been put to productive use by other firms. In 2016, the Insolvency and Bankruptcy Code (IBC) was passed by Parliament. It put in place a time-bound procedure whereby, with the consent of a committee of creditors, such a company can be taken over by a resolution applicant (who could be an individual, a trust or a corporate bidder) who submits an acceptable plan to revive it and repay a proportion of its debts. If no resolution plan is acceptable to the creditors, the company is liquidated so that its assets can be sold piecemeal. Since the assets would have been severely depreciated, the proceeds from the sale would usually fall short of the total value of claims, so the Code also specified the order of priority for settling various types of claims. This would reduce the banks’ burden of NPAs, and release the idle assets for alternative use. In addition, successive Union budgets provided a massive infusion of funds to recapitalize the public sector banks so that they could increase their lending operations.
All the reforms listed above brought about significant changes in the product, capital, land and technology markets. Reform of the market for industrial labor came to be known as the ‘unfinished agenda’ of reform. But this too changed after 2014, when the National Democratic Alliance came to power. To begin with, NDA governments in several states amended the Industrial Disputes Act to increase from 100 to 300 the employment threshold at which government permission is required for layoff and retrenchment of workers, and closure of plants. Some states also doubled (from 10 to 20) the employment threshold for registration under the Factories Act (which regulates working hours, health and safety provisions for workers, and is enforced through multiple compliance forms and inspections). Then, in 2020, the central government passed four new Labour Codes to replace 29 central labor laws. Among many other changes, two of the Codes will extend these two increased thresholds to the entire country. However, as of December 2022, they had not yet been brought into force because many state governments had not passed the regulations required to implement them.
Apart from these legislative changes, many procedural changes were periodically announced by NDA governments at the center and the states in order to promote the ‘Make in India’ and Ease of Doing Business agenda. These measures simplified the number of forms required to comply with various laws, allowed for voluntary compliance and self-certification in many cases, and systematized the regime of inspections.
Special credit schemes and preferential procurement by public sector undertakings have been introduced for micro, small and medium enterprises. Since 2020, production-linked incentives (PLI), in the form of subsidies based on incremental sales from capacity expansion over a specified period, are being announced for large firms in a widening range of industries.
Before we proceed to an assessment of the post-1991 policy regime, we need to question the tendency to view it in terms of a redrawing of the boundary between the state and the market. Almost all the measures described above no doubt allowed freer play of market forces in determining the pattern of production and resource allocation. Note, however, that this did not always imply less state intervention. Implementation of the TRIPS Agreement, the land acquisition laws, and the IBC all involved the use of the state’s authority to assign or reallocate property rights, to create or reinforce markets, and to lay down and enforce rules on how they would work. Moreover, the post-1991 period saw the setting up of market-regulating institutions such as the Securities and Exchange Board of India (SEBI) as a statutory body in 1992 and the Competition Commission of India (CCI) in 2003.
We also need to question the tendency to attribute the policy changes to the conditions imposed by the international lending agencies, the World Bank and the International Monetary Fund, who took advantage of India’s balance of payments crisis. There is no doubt that economic policy changed in the direction of what was known as the Washington Consensus. 14 Indeed, several researchers have argued that precisely because the reforms were externally imposed, they were exogenous to the performance of specific industries, and so econometric estimates of their impact do not suffer from endogeneity bias. I return to this while discussing the literature below. On the other hand, Indian economists like Bhagwati (1993) quickly claimed that the reforms followed their longstanding recommendations. More recently, some of the government advisers of that time have given their own accounts. Mohan (2017) describes how successive versions of the new Industrial Policy Statement had been prepared by him and his colleagues well before the crisis, and cosmetically repackaged to get it through Parliament in 1991. Ahluwalia (2020, pp. 109–115) acknowledges authorship of what was known at the time as the ‘M document’ of 1990, which provided a blueprint for the reforms that were rolled out the next year. Although both of them had earlier worked at the World Bank, they give ample credit to the home-grown civil servants who helped them to guide the political executive. Balakrishnan (2010, p. 121) quotes two very senior advisers who had been closely involved in administering the control regime, and had recognized its distortions and inefficiencies much earlier.
Given these claims and attributions of responsibility in elite academic and government circles, it is important not to lose sight of the developments discussed in the previous section: the committee reports and significant reforms of the 1980s, and the changing domestic political economy. The crumbling of ‘actually existing socialism’ in Eastern Europe, the evident success of the East Asian ‘miracle’ economies, and the market-oriented reforms in post-Mao China no doubt had an influence, even though the latter two examples were certainly not advertisements for the Washington Consensus. Somewhat more speculative is Kohli’s (2006b) argument that by 1991, the negotiations that ultimately led to the formation of the WTO in 1994 made it evident that significant trade liberalization was inevitable, so Indian industry supported the removal of regulatory barriers to mergers and expansion by domestic firms, in order to be able to compete with imports when the time came. Be that as it may, attributing the reforms to the Washington Consensus is as simplistic as attributing the earlier policy regime to Nehruvian socialism.
The impact of India’s reforms has been intensively studied in hundreds of papers, including several in the leading professional journals. It is not possible to survey this vast literature, but I shall selectively draw attention to some notable contributions. I begin with a broad overview based on aggregates for the entire manufacturing sector. Balakrishnan et al. (2017) show that when the appropriate econometric techniques are applied, there was an acceleration in the mid-1980s and again after 1999, but no discernible break around 1991 in the growth rate of manufacturing output. The investment rate (ratio of investment to value added) in manufacturing displayed sharp cyclical fluctuations, first rising to over 50% in 1997, falling back to 35% in 2001, surging above 55% in 2007, and declining thereafter—a concern that continues to this day (Krishna et al., 2022b, Fig. 3 ). Employment in the organized manufacturing sector, the supposed beneficiary of the reforms, remained stagnant at around 6 million through the 1980s, rose to 7.6 million by 1995, but then fell back to 6 million in 2001 (despite unchanged employment protection laws). It grew rapidly during 2004–11, and more slowly thereafter (Fig. 1 ). However, the growth in organized sector employment was offset by a fall in employment in the unorganized sector, especially in labor-intensive, export-oriented industries, after the 2008 Global Financial Crisis. These data are based on enterprise surveys; data from household surveys show a substantial fall in manufacturing employment between 2011 and 2017 (Nagaraj, 2020; Thomas, 2022; Krishna et al., 2022a, ch. 8). The disappointing employment record since the 1980s is associated with high and growing capital intensity in most manufacturing sectors, which in turn is attributed to pro-worker labor regulation (Hasan et al., 2013); lower prices of capital goods due to trade liberalization (Sen & Das, 2015); and real (product) wages rising relative to the real price of capital goods (Krishna et al., 2022a, ch. 9.4.3). 15
Workers employed in organized manufacturing, 1981–2020.
Source: Annual Survey of Industries 2019–20, Table 1: Annual series for principal characteristics
Growth rate of gross value added in manufacturing at constant (2011–2012) prices.
Source: Author’s calculations from Ministry of Statistics and Programme Implementation (MOSPI), National Accounts Statistics 2022, Statement 8.18.1
Since the proximate objective of the liberalizing reforms was to improve allocative and productive efficiency, their impact should be judged on the basis of trends in productivity rather than growth rates of outputs and capital and labor inputs separately. Several studies over the last few decades have used growth accounting to estimate Total Factor Productivity Growth (TFPG) as the residual after accounting for increases in capital and labor inputs. Many of the earlier studies, which found increases in TFPG after the reforms of the 1980s and 1990s, suffered from methodological deficiencies. Balakrishnan (2004) summarized several papers that used an improved methodology and found no increase—rather a possible decrease—in the rate of TFPG in the 1980s, and a definite decline in the 1990s. Krishna et al., (2022a, p. 7) mention other studies, including some that cover the unorganized sector, that come to similar conclusions. On the other hand, Bollard et al (2013) found a large increase for 1993–2007 compared to 1980–1992. The apparent contradiction between these studies can be resolved if we note that the latter paper clubs the 1990s with 2003–2007, a period that witnessed a dramatic spurt in manufacturing growth which could not be sustained, as indicated by the macro-aggregates above and is further discussed below. A recent paper by Das et al (2021), which covers a much longer period with much richer data, including an improved capital series, finds that TFPG in manufacturing was negative for both the 1950–64 and 1965–79 sub-periods, offsetting the growth of factor inputs. During 1980–92, TFPG turned slightly positive, but during 1993–2001, it declined even more rapidly than in the pre-reform era. Only after 2002 did it pick up to unprecedentedly high levels. However, capital accumulation remained the major contributor to manufacturing growth throughout, while employment growth decelerated.
As for industrial workers, apart from the worrying trends in employment, real wages in organized manufacturing fell between 1996 and 2006, and rose again thereafter, but much more slowly after 2012 (Basole & Narayan, 2020). 16 In the last 2 decades, an increasing proportion of workers has been hired through labor contractors, depriving them of the protection of the Industrial Disputes Act and other social security benefits that should be available to workers directly hired by industrial establishments. But even for directly hired workers, a small and declining share actually receive such benefits (see sections V and VI of Bhattacharjea, 2021, for details). There was a sharp decrease in the share of wages in value added since the late 1980s, as wages lagged behind increases in labor productivity. The wage share too has improved in recent years, although it remains far below the pre-reform level (Fig. 2 ). 17
Share of wages in net value added in organized manufacturing, 1981–2020.
Source: Author’s calculations from Annual Survey of Industries 2019–2020, Table 1: Annual series for principal characteristics
The combined picture we get from these trends for manufacturing as a whole, or for broad groups of industries, is that the benefits of the reforms initiated in the 1980s remained quite modest for over 2 decades, with periods of actual retrogression in employment and TFPG. Aggregate trends, of course, can be affected by macroeconomic policies, political instability, fluctuations in world demand, energy prices and the like, which can occlude the impact of microeconomic reforms. Moreover, most of the reforms (apart from delicensing and MRTP amendments) were unfurled in a staggered manner for different industries, making it hard to identify their impact at the aggregate level. A large and growing literature has tried to establish this connection with data at the disaggregated industry or firm level. Researchers have had to confront at least three issues in this context. The first is quantification of each type of reform. Analyses are typically at the 2- or 3-digit level of the National Industrial Classification (NIC). These are broad categories of industries. One-off changes which affect a particular industry from a particular year, like permission for FDI, abolition of investment licensing, or dereservation for SSI or the public sector, can be cleanly captured by a dummy variable that switches from 0 to 1 in that year for that industry, if data at that level of disaggregation are used. But trade policy indicators are harder to compile. Each 3-digit level classification includes a large number of diverse industries which were affected by reforms at different times, so calculation of a weighted measure of reform becomes essential. Weighting of tariff rates on each item by their respective shares of total imports in that 2- or 3-digit category perversely gives a low weight to high tariffs precisely because they reduce imports. Tariff rates also exclude anti-dumping duties, which have been of growing importance after 1997. Import licensing is typically measured by a ‘coverage ratio’: the proportion or import share of items covered by licensing in a particular year. This captures the complexity of the licensing system very crudely. 18
A second problem in establishing a relationship between policies and performance at the industry or firm level is that if different reforms that affected an industry were unfurled concurrently, then unless all the relevant policy variables are included in the regression equation, the coefficients on the included variables will be biased if they are correlated with the omitted variables. This is unlikely to be a problem for investment delicensing and MRTP amendments, which were almost entirely confined to the years 1985 and 1991, or to SSI dereservation, which was initiated much later and gathered steam only in the 2000s, by which time the impact of the other reforms would have played out. The problem arises in the assessment of trade and FDI liberalization, which were rolled out concurrently in the 1990s.
A third issue is the possibility of reverse causality: if a policy change that could affect a particular industry is itself motivated by the performance of that industry, the estimated coefficient for that policy variable will suffer from endogeneity bias. For example, if a poorly performing industry succeeds in lobbying for higher protection, or even a slower pace of import liberalization, its poor performance may wrongly be attributed to relatively higher tariffs, instead of vice versa.
I now provide a highly compressed summary of the findings of some of the more notable papers in the literature. Industrial delicensing and SSI dereservation were probably the policies least vulnerable to the problems discussed above. Aghion et al (2008), who covered the period 1980–97, showed that delicensing was associated at the national level with an increase in the number of factories, but not in output, which actually fell in some states. What they called their “key result” was that “when delicensing occurred, industries in states with pro-employer regulation experienced larger increases in output relative to those located in pro-worker states” (Aghion et al., 2008, p. 1403). As I have argued in Bhattacharjea (2021), the classification of states in terms of their labor regimes is problematic, but there can be no quarrel with the finding that delicensing had very different effects in different states. Aghion et al. also showed that the expansion of output after delicensing was greater for industries in the top one-third of the distribution of labor productivity, and those in states with more development expenditure and bank branches. More recently, Alfaro and Chari (2014), covering 1989–2005, show that the industries that were delicensed in 1991 exhibited greater entry at the lower end of the firm-size distribution, expansion of incumbent firms at the top end, and shrinking of those in the middle. Market concentration, average firm size, and market shares all declined, especially in the delicensed industries. However, delicensing had only a weak negative effect on price–cost markups, and was associated with even lower TFP in already less productive firms and higher TFP in more productive firms. Kandilov et al (2017) find a significant positive effect of delicensing on firms’ investment in fixed assets during 1989–2000, in states with above-average private bank credit. The increase was largest for firms in the lowest quartile of the size distribution, followed by those in the highest quartile. Bas and Paunov (2018) find a significant post-delicensing increase in firms’ probability of investing in research and development, a tendency that was strongest for the larger and more productive firms.
Most of these studies control in different ways for the concurrent trade and FDI reforms. The coefficients on the trade variables are almost always insignificant; in some papers, FDI liberalization is found to be negatively associated with firms’ assets and investment. Since these policies were not of central concern in these studies, they had little discussion of measurement or endogeneity problems. They assumed the 1991 reforms were exogenous because they were externally imposed. Other studies that focus on the impact of trade liberalization investigate the endogeneity issue more carefully. They find no statistical correlation upto 1996 between tariff changes and prior performance of industries, but a significant correlation thereafter. This suggests reverse causality might cause problems in the later period, so the estimation is confined to the earlier one (Goldberg et al., 2010; Topalova & Khandelwal, 2011). Nataraj (2011) covers the period upto 2000 after similarly finding no correlation. These studies also feature two other methodological advances. Econometrically, they employ a method of estimating TFP that avoids the problem of simultaneity between a firm’s inputs and unobserved productivity shocks. Conceptually, the authors incorporate insights from ‘new’ trade theory and endogenous growth models. These emphasize the role of trade liberalization in allowing greater access to a wider range of cheaper and better imported intermediate inputs, which enables user industries to increase their productivity. Newer trade theory emphasizes the heterogeneity of firms within the same industry. Between them, the studies that apply these insights to India show that the input tariff reductions greatly expanded the range of imported inputs, and were associated with improved TFP among the firms that used them. This effect was reinforced by the disciplining effect of lower output tariffs (i.e., those on competing imports), especially on the larger and more productive firms, and by concurrent industrial delicensing and FDI reforms. A greater range of imported inputs was associated with a greater range of output varieties. Bas and Berthou (2017) show that reduction of input tariffs during the same period increased the probability of firms importing capital goods.
Another strand of trade theory with heterogeneous firms emphasizes the role of liberalization in enhancing productivity through reallocation of market shares from less to more efficient firms. Using data on listed companies, Alfaro and Chari (2014) find that reallocation contributed to higher labor productivity after 1991, especially in delicensed industries. On the other hand, Harrison et al (2013), for a much more comprehensive panel of plants spanning 1985–2004, and industry-specific tariffs, find that this mechanism was relatively unimportant as compared to increases in within-plant TFP. The latter was attributed to reduction of input tariffs and FDI liberalization, and to a lesser extent the reduction of output tariffs. Investment delicensing seemed to increase TFP only for larger plants with over 100 workers. Harrison et al. also rule out reverse causality by showing that there was no correlation between reforms during 1990–2004 and trends in industry characteristics in the five preceding years. Similarly, Bollard et al (2013) find substantial productivity growth over 1993–2007 within large plants (in their case, those with over 200 workers), with little evidence of a reallocation effect. But they are unable to relate the productivity growth to industrial delicensing, tariff or FDI liberalization, or SSI dereservation.
I now turn briefly to some studies on the effect of reforms on the price–cost margins (markups) of manufacturing firms. Goldar and Aggarwal (2005), using industry-specific data on tariff and non-tariff barriers for the period 1980–98, found that margins were depressed by trade liberalization, especially in highly concentrated industries. But they confirmed the findings of some earlier studies that margins had actually risen in the 1990s, and related it econometrically to a decline in the share of labor in value added, which offset the pro-competitive effect of the reforms. Workers seem to have borne the burden of increased import competition. Using more recent econometric techniques, de Loecker et al (2016) also show that output tariff reductions had the expected pro-competitive effect, but firms did not fully pass on the benefits of reduced input tariffs to consumers; they increased their price–cost margins. Although this result suggests that liberalization benefited producers more than consumers, the authors argue that it is consistent with consumers benefiting from improved quality and more variety. Their paper assumed exogenously given wages, but a recent paper by Brooks et al (2021) uses similar econometric techniques to estimate not only firms’ markups in their product markets, but also ‘markdowns’ (of wages relative to the value of labor’s marginal product) arising from the exercise of monopsony power in their labor markets. The two combined led to a substantial reduction in wages as well as the labor share of value added, relative to a counterfactual competitive benchmark. The gap was highest around 2001, but fell thereafter. Their study does not explicitly include any policy variables, but it covers the 1999–2011 period over which the supposedly pro-competition reforms discussed above had been completed.
However, dereservation of products for SSI did occur in this period. Martin et al (2017) investigate the consequences, with data limitations forcing the study period to be confined to organized sector establishments during 2000–2008. They incorporate a product-level time trend in their regressions, to control for endogenous selection of products to be dereserved on the basis of past performance of the respective industries. They show that dereservation was followed by significant increases in output, labor productivity, and real wages per worker. However, these average effects masked important differences between establishments of different kinds. Out of the establishments that were earlier producing reserved products, the smaller ones actually shrank after their products were dereserved, while the larger ones expanded in terms of output, investment and employment. Establishments that had not previously been producing reserved products entered into their production after dereservation. The authors also compute the proportion of district-level employment attributable to establishments producing reserved products, and find that districts that were more affected by dereservation actually expanded their manufacturing output and employment.
Most of the micro-studies focus on organized manufacturing, and many of them use data only on large companies. Therefore, they miss out on any adverse effects on the informal sector, which constitutes the largest number of establishments and provides the largest share of employment in manufacturing. Only Nataraj (2011) undertook a comparable analysis of the effect of tariff reforms on the unorganized sector. Data on this sector are collected only at 5- or 6-year intervals, so she had to use three separated years: 1989–90, 1994–95, and 2000–2001. She found that reduction of output tariffs increased average productivity of informal firms, but this was due to exit by the less productive firms. Other studies on the informal sector provide mixed results, which are hard to summarize in this already lengthy survey. One issue that has been studied intensively is whether increased competitive pressure on formal sector firms has induced them to subcontract parts of the production process to unorganized sector units, and whether this has helped or harmed the latter. Gupta (forthcoming) provides a survey of the literature and presents some econometric findings (based on the latest available survey of 2015–16) that subcontracting actually benefits unorganized sector units in terms of the surplus they can earn over costs, although female-owned units benefit much less.
Readers might be puzzled by the contrast between the conclusions of the two bodies of literature surveyed above: the aggregative studies show only modest gains for manufacturing during the post-reform period, while the studies based on industry and firm level data are much more upbeat. The key takeaway seems to be the heterogeneous effects of reforms, with many gains accruing only to the largest and most productive firms, and many adverse effects on the smaller and less productive. This would have attenuated the impact of reforms in the aggregate, especially on output and employment. What is undeniable is that the reforms had many losers as well as gainers. The losers were smaller firms and many workers in larger firms. Corporate India also witnessed a massive churn. As shown by Goswami (2017) and Piramal (2017), respectively, many famous old companies and business families that had ranked in the top 50 by market capitalization in 1991 fell far down in the rankings by 2016. Of course, their promoters would not have suffered the kind of deprivation that the other losers were forced to undergo.
Readers might also be wondering why most of the studies surveyed above cover the post-reform period only upto around 2008. One reason is that the new industrial classification that was initiated in that year is hard to match with earlier classifications, so it is difficult to construct industry-wise panel data beyond that year. Another reason is that for almost a decade thereafter, there were no significant changes in industrial policy that could be subjected to the same kind of economic analysis. Assessment of the specific policy initiatives of more recent years (‘Make in India’, increases in tariffs, PLI schemes, reduction in corporate profits taxes, and labor law reforms in some states) will take time. Researchers need data for several years after each such change to evaluate its consequences, and there are long lags in collection and dissemination of microdata by official agencies, followed by the time taken for careful analysis and discussion of research findings before they get published.
In any case, industrial performance since about 2003 has been dominated by macroeconomic developments, arising from both external and internal sources. The relevant data being more accessible, macroeconomic analyses have been plentiful. 19 Industry enjoyed its heyday in what Nagaraj (2013) called ‘India’s Dream Run, 2003–08’, fuelled by a boom in world trade, foreign capital inflows, and easy domestic credit, as well as allotment at low prices of state-owned or state-acquired lands for industry and mining. Despite the so-called ‘policy paralysis’ of the then government, manufacturing output and exports grew rapidly. Then came the Global Financial Crisis of 2008. The economic downturn in India was actually mild, compared to other countries, because of stricter financial sector regulation. A generous boost of fiscal spending propped up the Indian economy, but once again it resulted in high inflation and a deteriorating balance of payments situation, exacerbated by high oil prices from 2010 to 2014. The reckless lending during the boom, the collapse of export markets, and the delays in land acquisition, allocation of telecommunications spectrum, and environmental clearances, especially after the backlash against the scams around these issues, resulted in stranded projects and massive NPAs on banks’ balance sheets, which deterred them from further lending. The counterpart of this was massive liabilities on borrowers’ balance sheets, which deterred them from borrowing. This ‘twin balance sheet’ or ‘debt overhang’ problem resulted in a continuing fall in the ratio of investment to value added in manufacturing. The speedy resolution of NPAs that the IBC seemed to promise got bogged down in legal delays, and banks have instead written off huge amounts of outstanding loans in order to clean up their balance sheets.
After 2012, the government reversed gears and tried to reduce the fiscal deficit. The Reserve Bank of India (RBI) resorted to monetary tightening, and also forced commercial banks to recognize the NPAs on their balance sheets, which inhibited their lending. These measures imparted a deflationary bias to the economy. Then, there were the three shocks of demonetization in November 2016, imposition of the GST from mid-2017, and finally the COVID-19 pandemic in 2020–21. The impact of demonetization is clearly visible in Fig. 3 , which is truncated at 2019–20 because the huge fall in output due to COVID, followed by the recovery, severely distorts the vertical scale of the chart and flattens all earlier fluctuations. The effects on manufacturing supply chains of the war in Ukraine, the sanctions on Russia, and the draconian Chinese lockdowns, are yet to be fully reflected in the data. The high interest rates recently imposed by the RBI to curb inflation and short-term capital outflows will certainly impinge on investment, as will the looming threat of recession and protectionism in major developed country markets.
It makes little sense to evaluate the lasting effects of recent policies in such an era of uncertainty and crisis management. However, Fig. 3 also shows that the growth rate of manufacturing output began slowing down almost 2 years before the COVID shock. 20 The investment rate (ratio of investment to value added) in manufacturing decreased almost every year from 2008 to 2017, and the average over this period was lower than either the 1994–2002 or 2003–2007 periods, which were marked by the previous slump and boom, respectively (Krishna et al., 2022b, Figs. Figs.3 3 and and4). 4 ). The declining investment rate would have slowed down capacity creation and technological improvement in the form of technologies embodied in new capital equipment. These trends indicate that any recovery from the current slump is likely to be modest, prolonging the lackluster record of industrialization that is revealed if we zoom out to get a longer-term perspective. Krishna et al., (2022a, pp. 52–53) show that the share of manufacturing in GDP remained in the range of 16–18.5% from 1980 to 2017, with the highest share being attained in the mid-1990s. Over the same period, the employment share of manufacturing in the workforce inched up from 10.5% to 11.5%, remaining stagnant since 2003 (ibid, p. 78). The share of manufacturing in GDP has been decreasing further in recent years (Fig. 4 ), and is now not much higher than at independence, although the data represented in this figure are not strictly comparable with those of earlier periods due to changes in the estimation methodology. All this suggests that even if we get through the current ‘polycrisis’, we have to confront much deeper and more longstanding problems that have resulted in India’s failure to industrialize despite all the twists and turns in industrial policy that have been chronicled in this paper.
