1. Introduction

In 1990–1, the Indian economy underwent a severe balance-of-payments crisis. By the summer of 1991, India's foreign exchange reserves covered less than two weeks of imports. The immediate cause of the crisis was the increase in world oil prices and the drop in the remittances of migrant workers from the Gulf following the annexation of Kuwait in September 1990. There was a realisation among Indian policy-makers, however, that 'the roots of the crisis were more structural in nature and lay in the import-substituting industrialisation (ISI) strategy followed by successive Indian governments since independence' (Agrawal et al. 1995: 161). While the ISI regime had enabled India to develop a large and diversified manufacturing sector, the net result of the protectionist policies was 'the growth of a high-cost, capital-intensive domestic industry that was by and large incapable of withstanding international competition' (p. 175). Not only did these policies severely inhibit India's export performance, they also served to limit the possibility of growth based on domestic demand.² In spite of four decades of import-substitution policies, production in the Indian manufacturing sector remained greatly import intensive. As a consequence, with India's trade regime providing little incentive to export, growth based on domestic demand would lead to balance-of-payments problems sooner or later.³

In June 1991, the new government that assumed office (led by P. V. Narasimha Rao) embarked on an economic reform programme along with several macro-stabilisation measures. One of the major long-term objectives of the reforms was to increase India's international competitiveness, both in relation to its past and to the fast-growing economies of East Asia. While the 1991 reforms could be seen as a continuation of the deregulation measures that were initiated in the mid-1980s by the Rajiv Gandhi government, they were far more comprehensive in scope and radical in substance. The macroeconomic stabilisation programme initiated in 1991 yielded immediate benefits, with foreign exchange reserves recovering from just over 1 billion USD at the time of the crisis to over 6.4 billion USD at the end of 1992–3. The inflation rate, which had peaked at 17 per cent in 1991, came down steadily to 7 per cent in 1992–3. Real output growth, which had dipped to 1.2 per cent in 1991–2, recovered to 4 per cent in 1992–3. It is far from clear, however, whether the long-term goal of the 1991 reforms with regard to international competitiveness has been achieved.

International competitiveness refers to the ability of a country to expand its share in world markets. The standard view on competitiveness is that it is essentially determined by factor endowments and comparative advantage. Thus, to exploit comparative advantage, it is necessary to minimize various distortions in the economy and 'to get the prices right'. In recent years, two more views have emerged: One argues that technological differences across firms, industries and countries are an important determinant of competitiveness. The other view locates competitiveness in the firm's investment decisions and hence in its ability to obtain investible funds. Both these views stress the importance of non-price factors and provide an important role for the government to build technological capabilities and to ensure a financial environment that is able to identify and allocate resources to the best investment projects.

In this study, we examine the international competitiveness of India's manufacturing sector. We take the view that competitiveness is a multifaceted issue and that no single theory (and the associated measures) adequately captures its complexity. We thus use several measures of competitiveness to examine the relative importance of various factors that influence it. Section 2 deals with competitiveness at the aggregate and sectoral level. Here we assess the relative importance of the real exchange rate and trade specialisation patterns in explaining India's trade flows. We also examine the link between labour costs and competitiveness. Further, we explore the technological intensity of India's exports. In Section 3, we analyse one important determinant of competitiveness, i.e. the financial environment. Specifically, we ask the question: to what extent has the Indian financial sector provided an enabling environment for successful export performance by manufacturing firms in the post-1991 period? We attempt to answer this question in two steps. First, we examine firms' sources and uses of funds to discern whether there is any systematic relationship between export performance and financing patterns. Next, we estimate investment functions for a sample of firms in the Indian manufacturing sector to see whether finance constraints are less severe for the exporters as compared to firms whose sales are primarily to the domestic market.

We begin with a more explicit consideration of trade, industrial and financial sector policies in India and the periodisation of the policy regime that we have used in our study.

Policy regimes and periodization

The Indian policy regime can be categorised into three distinct phases. The first phase was the era of planning from 1951 to 1984 when the state had strict control over resource allocation. The second period was a period of partial deregulation from 1985 to 1991 when the state retained a major role in resource allocation even as private agents were given greater freedom in investment decisions. Finally, in the post-1991 period, resource allocation was primarily market driven.⁴ In what follows, we provide a brief overview of the economic policies followed in each period.⁵

1951–84

During this period, India had a highly restrictive trade and industrial policy regime. Nearly all imports were subject to discretionary import licensing or were 'canalised' by government monopoly trading organisations. The only exceptions were commodities listed in the Open General License (OGL) category. Capital goods were divided into a restricted category and the OGL category. While import licenses were required for restricted capital goods, those in the OGL could be imported without a license subject to several conditions. Intermediate goods were also classified into the banned, restricted and limited permissible categories plus an OGL category. As these names suggest, the first three lists were in order of import licensing stringency. The import of consumer goods was, however, banned. Like imports, exports were also subject to an elaborate licensing regime. To counteract the anti-export bias of the trade regime however, there were a large number of export incentives for manufactured goods.

The principal instrument of industrial policy was an elaborate industrial licensing framework under the Industries Development and Regulation Act of 1951. The Act stipulated that no new units (above a certain size) could be set up nor substantial expansion be made to existing units without a license from the government. The Monopolies and Restrictive Trade Practices Act (MRTP) became effective in 1970 to ensure against concentration of economic power and check restrictive trade practices. Foreign investment in India was regulated by the Foreign Exchange Regulation Act (FERA) of 1974.

With respect to financial sector policy, there was a period of increasing financial repression from the early 1970s. In 1969, fourteen of the largest commercial banks were nationalised followed by six more in 1980. Moreover, commercial banks were increasingly pressured to lend to the 'priority sector', comprising agriculture, small-scale industry, retail trade, transport operators, professionals and craftsmen. While the commercial banks essentially provided short-term credit to the manufacturing sector, long-term loans were provided by All India Development Banks like Industrial Development Bank of India and Industrial Credit and Investment Corporation of India. These term-lending institutions depended a lot on the government for resources (usually subsidised heavily), and their allocation of long-term loans to firms was strictly monitored by the government according to plan priorities. Interest rates both of commercial banks and term-lending institutions were controlled by the government. The stock markets too were controlled by the government with respect to pricing, quantum and timing of new issues.

Finally, with respect to exchange rate policy, the rupee was pegged to the pound sterling till 1975 (except for a brief period when the rupee was pegged to the US dollar). In September 1975, the peg was altered to a basket of currencies with undisclosed weights. For much of the period, the peg was 'passive', with the sole intention of keeping the real exchange rate constant.

1985–91

With the advent of the Rajiv Gandhi government in 1985, piecemeal reforms were initiated in trade and industrial policy. Several initiatives were taken to limit the role of licensing, expanding the scope for contribution by large business houses to growth, encouraging modernisation and allowing existing firms in certain industries to achieve minimum economic level of operations. The shift from quantitative import controls to a protective system based on tariffs initiated in the mid-1970s was considerably quickened from 1985 onwards. Also, beginning in the mid-1980s, there was a renewed emphasis by the new administration on export promotion. The number and value of incentives offered to exporters were increased and their administration streamlined. The allotment of REP licenses – tradable import entitlements awarded to exporters on a product-specific basis – became increasingly generous. There was also a steady devaluation of the Indian rupee during this period. Effectively, India operated an 'active' crawling peg from 1986 onwards to produce a sharp real depreciation of the rupee in the period 1986–90.

Post-1991

As noted earlier, the year 1991 marked a watershed in Indian economic policy. As a part of the structural adjustment programme, quotas on the imports of most machinery and equipment and manufactured intermediate goods were removed. REP licenses were abolished and a large part of the import licensing system was replaced by tradable import entitlements linked to export earnings. There was also a significant cut in tariff rates, with the peak tariff rate reduced from 300 per cent to 150 per cent and the peak duty on capital goods cut to 80 per cent. There was, however, little change in trade policy with respect to consumer goods which remained banned. With respect to industrial policy, industrial licensing was abolished altogether except for a select list of environmentally sensitive industries. MRTP was substantially revised so that regulations restricting the growth and merger of large business houses were eliminated. FERA was altered in 1993 so that the earlier policy of restricting foreign investment became one of actively promoting it.

From 1991 to 1993, India moved gradually to full current account convertibility of the exchange rate, first in March 1992, with the replacement of the tradable import entitlements, with a dual-exchange rate system, and then in March 1993, moving to a unified 'market-determined' exchange rate system (i.e. a managed float). Nonetheless, strict controls over the capital account, especially capital out-flows, remain.

In the financial sector, from the point of view of the financing decisions of firms, the two most important changes were the deregulation of interest rates (both of commercial banks and term-lending institutions) and the freeing of pricing restrictions on new issues of shares through the stock markets.

Our study is mostly confined to the 1985–91 and post-1991 policy regimes. In the next section, we attempt to trace the effects of these policy changes on export competitiveness of the Indian manufacturing sector. Specifically, we look for breaks in the trend in competitiveness across these two periods. In Section 3, where we analyse financial factors at the firm level, we confine ourselves to the post-1991 period for obvious reasons.

2. Productivity and the price determinants of competitiveness

In this section, we assess the relative importance of the real exchange rate and labour productivity (and domestic costs) in explaining India's trade performance in the recent past. We begin with overviews of India's trade performance and the evolution of the current account. We then attempt to determine the importance of the real exchange rate in explaining India's competitiveness in both total and manufacturing exports. Next we examine in detail India's trade specialisation patterns. We compute export shares and indices of revealed comparative advantage to assess competitiveness at a sectoral level. 'Winner' and 'loser' industries are then identified and the links between competitiveness, labour productivity and domestic costs are explored. We end by examining alternate measures of trade specialisation such as intra-industry trade and the technological complexity of exports.

Overview of trade flows

India had a persistent deficit in the trade account during the period 1971–96 (Figure 4.1). The trade deficit as a percentage of GDP was smaller in magnitude in the 1990s as compared to the 1980s. This, in spite of a rapid increase in imports as a ratio of GDP, was due to a strong performance by the export sector. It is clear that due to the sharp increase in both the ratios of exports to GDP and imports to GDP since the mid-1980s, the economy has been increasingly 'open' during this period (Figure 4.1).⁶ There has also been a steady increase in manufacturing exports as a proportion of India's total exports since the 1980s, from less than 60 per cent in 1979–80 to about 75 per cent in 1995–6. Nonetheless, market shares of India's total and manufacturing exports in world exports have not improved substantially and continue to remain at less than 1 per cent. There does not seem to be any perceptible increase in the annual growth rates of both total and manufacturing exports in the post-1991 period. For the period 1981–90, the average annual growth rates for total and manufacturing exports were 9.4 and 11.8 per cent, respectively, while for the period 1991–6 the average annual growth rates for total and manufacturing exports were 8.9 and 9.9 per cent, respectively. Therefore, the 1991 reforms do not seem to have had any perceptible positive effect on India's export performance.

Figure 4.1 India's exports, imports, trade balance, current account balance and openness measure.

Evolution of the current account and the real exchange rate

It is evident from Figure 4.1 that it is only in the early 1980s that India had large deficits in the current account. In the 1990s, while India still had a deficit in its current account, the current account deficit to GDP ratio was considerably lower than in the 1980s. We have already observed earlier that India had a rapidly falling deficit in its trade balance from the early 1980s as exports grew rapidly during this period (see Figure 4.1). Moreover, the real effective exchange rate (REER) had been steadily depreciating since the mid-1980s (Figure 4.2). During this period, India followed a policy of steadily devaluing the rupee in combination with other export promotion measures to boost exports. Clearly then, the worsening current account deficit in the 1980s cannot be attributed to a weakly performing export sector.

Figure 4.2 Real effective exchange rate of the rupee (1979 = 100).

Joshi and Little (1994) argue that the increase in the current account deficit to GDP ratio in the 1980s could be linked to an increase in the investment–savings gap. Underlying this was the widening fiscal deficits of the central government, with the public investment–savings gap increasing from 7.1 per cent of GDP in 1982–4 to 8.4 per cent of GDP in 1985–9. With the fiscal retrenchment initiated in 1991, there was a narrowing of the investment–savings gap in the 1990s and a consequent decrease in the current account deficit to GDP ratio. Thus, the large current account deficits of the 1980s could be attributed to a macroeconomic imbalance (related to a widening fiscal deficit) rather than a stagnant export sector or an inappropriate real exchange rate. The structural adjustment programme of 1991 led to some correction in this imbalance and, hence, a more sustainable current account deficit. It should be noted, however, that in contrast to its behaviour in the mid-to-late 1980s, the real exchange rate (RER) has shown a slight appreciation in the very recent past.

We have observed earlier that India followed a discretionary crawling peg in the 1970s and 1980s to maintain an 'appropriate' level of the RER. Yet there were periods, particularly in the early 1980s, when the nominal exchange rate was kept fixed in spite of a high inflation rate prevailing in the domestic economy. It is commonly agreed that sustained RER misalignment may contribute to severe macroeconomic disequilibria and a balance-of-payments crisis. Moreover, there is evidence to suggest that more 'successful' countries owe much of their success to having been able to maintain the RER at its 'appropriate' level (Edwards 1994). To what extent can it be argued that India had 'misaligned' RERs during the period under consideration? Elbadawi (1994) estimates the degree of misalignment in India's real exchange rate for the period 1965–88. The degree of mis-alignment is defined as the deviation of the actual RER from the equilibrium RER. The latter is itself the level of the RER which allows the economy to simultaneously attain internal equilibrium (i.e. the non-tradable market clears, the budget is balanced and portfolio equilibrium holds) and external equilibrium (the current account is in balance). Elbadawi has developed a model of the equilibrium RER where the equilibrium RER is determined by domestic absorption and government expenditures (both as ratios of GDP), terms of trade and a measure of the degree of 'openness' of the economy. Elbadawi finds that except for 1965 and 1986, which witnessed episodes of overvaluation of 16.3 per cent and 10.6 per cent, respectively, the period is characterised by single-digit RER misalignments, most of which are actually quite small. This, according to the author, supports the view that 'India, while maintaining an elaborate ensemble of economic controls, has nonetheless adopted a rather conservative macroeconomic policy' (Elbadawi 1994: 126).

Aggregate competitiveness

To measure competitiveness at the aggregate level, we use the constant market share (CMS) analysis. According to the CMS method, the proportionate increase in exports over time comprises a number of effects: (a) standard growth effect, (b) commodity composition effect, (c) market distribution effect, and (d) a residual effect which may be termed 'competitiveness'. In other words, the increase in exports can be 'explained' in terms of four factors: the general growth of world exports to the focus destination; the commodity mix of exports and differential growth in import demands; the extent to which the particular market represents growing centres of demand; and finally, a residual term which captures the net gain or loss in the market shares presumably due to changes in the relative price and/or quality of the product, not to mention the marketing effort and skill of the exporters.⁷

Figure 4.3 India's competitiveness and export growth – all commodities (SITC two-digit level).

The estimates of each of the above-mentioned effects depend on the 'standard' against which the focus country's exports to the focus destination is to be compared. This study has used the world standard, assuming that the commodity composition of world exports bears a reasonably good relationship to that of the focus exporter.

The CMS methodology is used to decompose the annual change in India's total exports, all commodities and manufacturing commodities separately, over the period 1970–92.⁸ The data set used is the World Trade Database from Statistics Canada made available through the NBER (Feenstra et al. 1997). Based on the trade data from the United Nations Statistical Office, this database provides on a consistent basis the annual bilateral trade values for all countries of the world over 1970–92.⁹

In Figure 4.3, we plot the competitiveness measure for all commodities as obtained from the CMS methodology along with the annual growth of total exports. Similarly, in Figure 4.4, we plot the competitiveness measure only for manufacturing commodities along with the annual growth of manufacturing exports. In both cases, the change in competitiveness is correlated with export growth. However, there is a closer correlation between the growth rate and the change in competitiveness of manufacturing commodities than there is between the growth rate and the change in competitiveness for all commodities (the correlation in the former case is 0.861 as compared to 0.796 for the latter case). This indicates that competitiveness may play a greater role in determining the export performance of the manufacturing sector than it does for all other sectors.

Figure 4.4 India's competitiveness and export growth – manufacturing commodities (SITC two-digit level).

The CMS methodology decomposes the change in a country's exports into four components – the world trade effect, the commodity composition effect, the market effect and the competitiveness effect. In Table 4.1, we decompose exports into these four components for all commodities and for manufacturing commodities. We find that the relative importance and the direction of change of the four components for all commodities is quite different from the relative importance of these components for manufacturing commodities for most subperiods. For example, in 1971–5 and in 1986–90, the competitiveness effect is large in magnitude (and opposite in direction, for the period 1971–5) for manufacturing exports as compared to all exports. This may indicate that the factors explaining competitiveness for manufacturing exports may be different from those explaining competitiveness for all exports. The periods 1971–5 and 1986–90 are striking in that we find that for manufacturing exports, the competitiveness effect is negative and large in magnitude in the first period and positive and, again, large in magnitude for the second period. What explains these large variations in the aggregate competitiveness of both total exports and total manufacturing exports? We examine this below.

The real exchange rate and aggregate competitiveness

The real exchange rate is often viewed as the most important determinant of the overall competitiveness of an economy. We examine this relationship for aggregate competitiveness measured over all commodities (CMSA) and over manufacturing commodities (CMSM) as estimated earlier using the CMS methodology (Table 4.2). Towards this, we regress CMSA on the change in the real exchange rate (RER) (Model 1a), and on the change in the nominal exchange rate (NER) and the inflation differential between India and the US (INF) (Model 1b). Similarly, we regress CMSM on the change in the real exchange rate (RER) (Model 2a), on the change in the nominal exchange rate (NER) and the inflation differential between India and the US (INF) (Model 2b), on the change in the sector-specific real exchange rate (RERM) (Model 2c), and, finally, on NER and on the sector-specific inflation differential between India and the US (INFM) (Model 2d). A linear functional form was specified and estimated using ordinary least squares (OLS) over the period 1971–92.

For CMSA, the change in RER is positive and significant at the 5 per cent level, albeit with a lag (Model 1a). The current change in RER was found to be insignificant. Decomposing the RER into its components, we find that it is the change in NER with a lag that explains the variations in CMSA (Model 1b). Similarly, for CMSM, the change in RER is positive and significant with a lag (Model 2a), with the decomposition again indicating that it is the change in NER that matters (Model 2b). Further, sector-specific RER does not have as much explanatory power as the economy-wide RER (Models 2c and 2d).

As noted earlier, India has followed an active exchange rate policy to boost exports since the mid-1980s. The evidence above shows that such a policy has indeed been effective. With a shift towards a more market-determined exchange rate since 1991 however, such a policy option may no longer be available.

Trade specialisation patterns

Data

The database used is obtained from the International Economic Data Bank (IEDB) at the Australian National University and provides trade and industry data at the ISIC four-digit level. The source of the industry data is UNIDO's Industrial Statistics databank, which in turn is compiled from the Annual Survey of Industries published by the Central Statistical Organisation, India. The export data is obtained from the United Nations Trade Database and uses a commodity concordance developed by the United Nations and further refined by the IEDB. The commodity concordance involves a mapping from the SITC classification system used by the Trade Database of the United Nations in reporting export data to the ISIC classification system used by the UNIDO in reporting industry data. While all the commodities that are usually included in the SITC definition of manufacturing exports (SITC 5 to 8 less 68) have been reclassified according to their industry of origin at the ISIC four-digit level, the ISIC classification contains some additional commodities not included in the SITC classification. As is well known, one limitation of the SITC classification of manufacturing exports is that it excludes processed food items and tobacco products (which are included in SITC 0 and 1). In contrast, the ISIC (i.e. industry-based) classification of manufacturing includes all such commodities in ISIC 311 (food products), 313 (beverages) and 314 (tobacco products). Furthermore, the ISIC classification of manufacturing also includes non-ferrous metals (ISIC 372), which are usually excluded from the SITC-based classification of manufacturing. Therefore, the coverage of manufacturing exports using the ISIC-based definition (i.e. the definition used in this chapter) may be considered to be more comprehensive than the more commonly used SITC-based definition.

Export shares and indices of revealed comparative advantage

In Table 4.3, we present the top two dozen commodities (at the ISIC four-digit level) in terms of export shares in India's total manufacturing exports over the period 1971–96. It is evident from the table that the shares of ISIC 3211 (spinning, weaving and finishing of textiles) and ISIC 3231 (tanneries and leather finishing) have declined significantly in the period under consideration from a total of around 34 per cent in 1971–5 to less than 14 per cent in 1991–6. On the other hand, the shares of ISIC 3220 (wearing apparel excluding footwear) and ISIC 3901 (jewellery and related articles) have increased in this period from a total of less than 8 per cent in 1971–5 to around 32 per cent in 1991–6. Basic industrial chemicals (excluding fertilisers, ISIC 3511) also seem to be increasingly important in India's manufactured export basket over time. The shares of most other commodities do not show any significant change in trend over the period 1971–96. It is also evident from Table 4.3 that these twenty-four commodities have consistently accounted for more than 85 per cent of the manufacturing exports during this period. This seems to suggest that India's manufacturing exports have not diversified over the past twenty-five years.

We computed the revealed comparative advantage (RCA)¹⁰ of India's manufacturing exports for each year over the period 1971–96.¹¹ The RCA computations showed that for a vast majority of industries, India is just not competitive in export markets as indicated by RCAs that are less than 1 over the entire period. In the post-1991 period, India was most competitive in ISIC 3901 (jewellery), followed by ISIC 3214 (carpets). In the case of jewellery, in particular, the increase in RCA has been dramatic, from 2.4 in 1971–5 to 12.8 in 1991–6. Other commodities whose export competitiveness has been increasing over the period 1971–96 are ISIC 3116 (grain mill products), ISIC 3220 (manufacture of wearing apparel excluding footwear), ISIC 3233 (manufacture of leather excluding footwear, apparel) and ISIC 3551 (tire and tube industries). Commodities with declining competitiveness are ISIC 3212 (manufacture of made-up textile goods excluding wearing apparel) and ISIC 3231 (tanneries and leather finishing).

Winner and loser sectors

In order to determine which industries 'gained' and which industries 'lost' in competitiveness, we adopt a non-parametric approach involving essentially a t-test (and an associated F-test) on the sample mean of RCAs across different subperiods of interest. The theme of the t-test is to split the whole time series of RCAs into two subsamples (say, Period I and Period II), compute the means of the series over the subsamples and test for equality or inequality of these two subsample means. At a given level of significance, a significant positive (negative) t-statistic would indicate a significant increase (decrease) in the mean level of the RCAs in Period II compared to Period I. An insignificant t-statistic would indicate equality of the mean level of the RCAs between the two subperiods, i.e. the RCAs are more or less constant over the full sample. The mathematical expression for the test statistic can be found in Brockett and Levine (1984) and Kanji (1993).

As we have noted in Section 1, the Indian economy has undergone two sets of reforms in the recent past, once in 1985, and the second in 1991. To see whether these two rounds of reforms have had any discernible effect on external competitiveness of the Indian manufacturing sector, we conduct the t-test on the sample means of RCAs once between the periods 1970–84 and 1985–91 and a second time between the periods 1985–91 and 1992–6. An industry whose RCA showed a significant increase (decrease) is considered to be a 'winner' ('loser') industry over the relevant period. Industries whose RCA did not show a significant change are considered to be 'stagnant'. We confine the t-tests to those industries which had RCAs greater than one for at least one of the subperiods. The results are tabulated in Table 4.4. A summary of these results is reported in Table 4.5.

From these results it is clear that some industries have gained in competitiveness while others have lost following the two rounds of reforms. Furthermore, there have been more winners than losers after the 1991 reforms as compared to the earlier reforms of 1985. Only one industry, namely leather products (excluding footwear and apparel), has been winning over both rounds of reforms. In contrast, three industries have lost in both rounds of reforms. These are food products (NEC), textile goods (excluding wearing apparel) and product of tanneries and leather finishing. There have been some industries which gained in one round of reforms but lost in another round, such as footwear (excluding rubber and plastics) and sugar factories. One possible explanation for this could be that these industries may have gained/lost (as the case may be) due to inter-industry effects of the reform measures that dominated the direct effects of reforms.

Evolution of labour productivity and unit labour costs

It is well recognised in the literature that a key determinant of external competitiveness is unit labour costs (see Fagerberg 1988). To what extent this hypothesis is relevant in the Indian context is of great significance given that India is perceived to be a labour-surplus economy. There has been a significant increase in labour productivity in the manufacturing sector since the early 1980s, with a levelling off in the 1990s (Figure 4.5). Real wages followed labour productivity for much of the 1970s and 1980s, leading to no perceptible change in unit labour costs during this period. In the late 1980s however, there was a slight decline in unit labour costs in the manufacturing sector, as labour productivity growth overtook growth in real wage per worker. In the early 1990s, with stagnation in labour productivity, unit labour costs began to increase. We observe that the movements in unit labour costs during the 1980s and early 1990s seem to have a fairly strong negative correlation with India's market share in world manufacturing exports.

Figure 4.5 Labour productivity, real wages and unit labour costs (ULC).

During the early 1980s, with little change in unit labour costs, there was no significant change in India's market share. With the decline in unit labour costs in the late 1980s, India's market share improved. Finally, in the early 1990s, with a slight increase in unit labour costs, there was a fall in India's market share. There is preliminary evidence, then, that at the aggregate level, the behaviour of unit labour costs may have played an important role in determining India's international competitiveness in the period under consideration.

Data on changes in unit labour costs by industry show that there is no consistent pattern on unit labour cost growth across industries (Table 4.6). In keeping with the trend in unit labour costs at the aggregate level however, a larger proportion of industries witnessed declining unit labour costs in the period 1986–90 as compared to the periods 1982–5 and 1991–2. The correlation coefficients between growth in unit labour costs and the change in RCAs across industries indicate that for the period 1982–5, growth in unit labour costs in a particular industry may be negatively correlated with the change in the international competitiveness of that industry (the correlation coefficient between the two is –0.25). On the other hand, there is little correlation between growth in unit labour costs and the change in RCAs for the other two subperiods. Moreover, when we attempted to relate changes in unit labour costs with the classification of industries into winners and losers, no discernible pattern emerged at the sectoral level on the linkage between domestic costs and export competitiveness (see Table 4.6). It should be noted nonetheless that such an analysis is incomplete until we can compare the evolution of unit labour costs at the sectoral level in India with a world norm. Clearly, what is of relevance for export competitiveness of a particular sector is the relative movement of its domestic costs with respect to the domestic costs of the destination country and that of other competitors in the same sector.

Alternate measures of trade patterns

Hitherto, our analysis has been based on an implicit assumption that trade specialisation is based on comparative advantage emanating from perfectly competitive domestic and international markets. In reality however, markets, both in India and abroad, would generally be characterised by product differentiation and economies of scale. We look at two measures of competitiveness that incorporate such assumptions.

Measures of intra-industry trade¹²

If a significant proportion of the industrial sector is characterised by imperfect competition, measures of intra-industry trade may indicate the extent of product differentiation and the presence of economies of scale in a particular industry. Furthermore, with trade reforms, one would expect an increase in the share of intra-industry trade in total industry trade as firms specialise in the production of certain products and not in others within an industry group (Helpman and Krugman 1989). As is clear from Figure 4.6, there has been a significant increase in aggregate intra-industry trade in the Indian manufacturing sector since the mid-1980s. Interestingly, one notes a slight downturn in total intra-industry trade in the mid-1990s. Measures of intra-industry trade by industry¹³ show that the industries with the highest share of intra-industry trade in total trade (0.8 and above) are ISIC 3215 (cordage, rope and twine industries), 3311 (sawmills, plying mills), 3312 (wooden and cane containers), 3521 (paints, varnishes, lacquers), 3620 (glass and glass products), 3691 (structural clay products), 3710 (iron and steel basic industries), 3819 (fabricated metal products), 3831 (electrical industrial machinery), 3833 (electrical apparatus and supplies) and 3843 (motor vehicles).¹⁴ It is an open issue, however, to what extent the high volumes of intra-industry trade evident in these industries are due to the existence of scale economies and differentiated products or due to industry classifications that are not comprehensive enough (see Loertscher and Wolter 1980).

Figure 4.6 Aggregate intra-industry trade.

Technological complexity of exports

A classification of India's manufactured exports by technological complexity indicates that India's manufactured exports are very much at the low end of the 'technology spectrum' (Table 4.7).¹⁵ Labour-intensive and resource-based products are the two dominant categories in India's manufacturing export basket. There has been some increase in the total share of scale-intensive, differentiated and science-based products in India's manufacturing exports from 18.1 per cent in 1980 to 23.2 in 1995. Nonetheless, it is far below that of China (38.4 per cent), Malaysia (79 per cent) and Thailand (53.6 per cent).¹⁶ A closer look at the 'winners' in either of the two subperiods, 1985–91 and 1992–6, shows that these are either labour-intensive, resource-intensive or scale-intensive products.

The comparison with China is particularly revealing. As of 1995, 9.7 per cent and 16.3 per cent of China's manufactured exports were in science-based goods and differentiated products, respectively, as compared to 5 per cent and 4.1 per cent for India. Differentiated products are technology-intensive engineering products while science-based products use leading-edge technologies (Lall 1998). Both these types of goods could be classified as 'high technology'. While China and India are both large labour-surplus economies with comparative advantage in labour-intensive manufactures, China is also diversifying into the low-medium technology end of export-oriented activity, with India doing poorly in this area. Clearly, the relatively slow progress in 'climbing up the technology ladder' with respect to exports may act as a constraint on India's long-term export performance and growth potential.

The evidence presented in this section does not allow for an unequivocal interpretation of the role of price factors in determining India's external competitiveness.

While at the aggregate level, the real exchange rate and unit labour costs seem to have a definite link with external competitiveness in the Indian context, the picture is far less clear at the sectoral level. This may indicate the importance of firm-level and industry-level non-price factors that may impinge on export performance. We explore in the next section one important determinant of competitiveness at the firm level, namely the availability of external finance. This factor acquires greater significance in the context of the wide-ranging reforms in the Indian financial sector since 1991.

3. Competitiveness and finance

There is widespread agreement in the literature that price competitiveness is a necessary but not a sufficient condition for export success. Among the non-price factors, the ones most commonly identified in Indian policy discussions are technology upgradation, product quality and infrastructural bottlenecks. One non-price factor that has received less attention, however, is the financial environment, i.e. the extent to which the financial sector provides an enabling environment for successful export performance. In the context of this study, an important question that arises is whether there has been any relationship between export performance of firms and financial factors in the Indian context. This question can be framed in two parts. First, is there a systematic relationship between export performance and the financing patterns? Here, we classify firms in certain selected industries into three categories, namely 'domestic firms', 'winning exporters' and 'losing exporters'. For each of these categories, we study the Sources and Uses of Fund Statements as well as a few other financial performance indicators to look for differences in their financing patterns and financial performance.

Second, do successful exporters face less information-based capital market imperfections than the not-so-successful exporters? Modern theories of finance which attempt to explain differences in financing patterns across firms emphasise differences in costs associated with different providers of funds. It stresses the lack of substitutability between internal sources (retained profits and depreciation) and external sources (different types of debt and new equity) of funds. This imperfect substitutability arises primarily due to asymmetric information between the suppliers and users of funds and incentive problems between managers and owners of the firm. It has generally been argued that these information asymmetries and incentive problems make external funds more costly than internal funds. In the new equity markets, this manifests as a 'lemons premia' (as pointed out by Myers and Majluf 1984) and in credit markets as credit rationing or loan mis-pricing (as pointed out by Stiglitz and Weiss 1981, and others). Further, this view contends that the cost differential between internal and external funds would vary across firms depending upon the extent of the information asymmetry. Besides, this view also suggests that simple transaction costs might also vary across firms. The implication of a higher cost of external funds is that internal funds would be more important than external funds in financing investments. Clearly, to the extent that a firm is forced to depend on internal sources for investment, its growth is said to be finance constrained.

If exporting firms are finance constrained then this would be a major impediment to sustaining their competitiveness in international markets.

In what follows, we explore this hypothesis by estimating investment functions which explicitly allow for the presence of finance constraints, i.e. models that allow the costs of internal and external sources of finance to be different (see Hubbard (1997) and, in the Indian context, see Athey and Laumas (1994)).

Classification of firms

Classification of firms into the above three categories proceeds as follows: firms are first categorised as 'domestic firms' and 'exporting firms' based on the share of exports in their total sales. If the exports to sales ratio of a firm exceeds 5 per cent over more than half the number of years in the sample period, then the firm is considered to be an 'exporting firm' whereas it is a 'domestic firm' otherwise. The reasoning behind this first level of categorisation is that there exist a large number of firms even in the tradable sector (be they winner or loser industries) who primarily sell only in the domestic market.¹⁷ Issues such as export competitiveness obviously are of little relevance to these 'domestic firms'. 'Exporting firms' are then further classified into 'winning exporters' and 'losing exporters' based on a comparison of the annual growth rates of their exports vis-à-vis the annual growth rate of exports for the industry to which they belong. If the growth rate of exports of a firm exceeds the industry export growth rate for more than half the number of years in the sample period, then the firm is classified as a 'winning exporter'; otherwise it is considered a 'losing exporter'. It may be noted here that this way of classifying exporting firms into winners and losers is largely consistent with the procedure adopted earlier for classifying industries.¹⁸

Furthermore, this procedure allows for the possible existence of winning exporters within a losing industry and vice versa. Consider, for example, a losing industry (the analogy runs similarly for winning industries also), i.e. an Indian industry whose exports are growing but whose RCA is falling over time.¹⁹ The growth rate of exports of some firms in this industry may be higher than the industry average. We consider such a firm to be a winner firm as it has outperformed the industry. This indicates that there could be some firm-specific characteristics (unobserved as yet) that enable such firms to outperform the industry. Similarly, a losing firm is one which has not been able to match the industry performance in terms of export growth, again perhaps due to certain firm-specific characteristics. We feel that it may be important to distinguish these two types of firms.

This analysis is done for firms belonging to five industries. Earlier, we have identified 'winner' and 'loser' industries based on whether their RCAs have been increasing or falling over time, respectively. Three winner industries from that classification, namely ISIC 3220 (manufactured wearing apparels excluding footwear), ISIC 3511 (basic industrial chemicals excluding fertiliser) and ISIC 3901 (jewellery and related articles), and two loser industries, ISIC 3211 (spinning, weaving, finished textiles) and ISIC 3839 (electrical apparatus and supplies NEC), have been selected for this analysis. The average share of these industries in total exports over the period 1991–6 was 19.7 per cent, 5.3 per cent, 12.1 per cent, 11.9 per cent and 0.5 per cent, respectively (see Table 4.3).

The database used is PROWESS provided by CMIE, Mumbai. The PROWESS names for the above industries are readymade garments, industrial chemicals, gems and jewellery, cotton textiles and electrical machinery, respectively. It must be noted that the mapping from ISIC to PROWESS may not be perfect. Only those firms for which data are available for all the years of the sample period are considered for the analysis here. Table 4.8 reports the number of firms in the balanced panel for the three categories for the sample period, 1993–7.

Descriptive statistics on financial variables

The indicators of the financial performance of firms used here are assets, export–sales ratio and profitability ratio (profit before interest, depreciation and taxes to sales). Movements in these indicators over the time period 1993–7 are studied by pooling firms across industries within each category. We present in Table 4.9 the average values of these variables. It may be noted that these summary statistics for the 'exporters' reported in these tables are estimated over winning and losing exporters combined. The following broad conclusions emerge:

1. The average size and the rate of growth in assets are found to be the lowest for domestic firms. Among exporters, winning exporters outperform the losing exporters in both average asset size and growth.
2. The export to sales ratio, which was similar for both winning exporters and losing exporters at the beginning of the period, grew for the former while it fell for the latter.
3. Winners and domestic firms, on an average, have been more profitable than the losers. Moreover, this ratio has been more or less stable over the years for all three groups.
4. We also found that the correlations between assets (i.e. firm size), profitability and the export to sales ratio were all insignificant.²⁰

The above patterns must, however, be interpreted with caution as all these three variables show substantial variation across firms within the groups for each year.

With the exception of profitability, the other two variables have coefficients of variation greater than 1.0.

We now examine the sources and uses of funds to see if there are differences in the financing pattern of firms in these three categories. The sources and uses of funds statements for the domestic firms, and all exporters (winners and losers) are reported in Table 4.10. These statements for winning exporters and losing exporters are reported in Table 4.11. The following broad patterns emerge:

1 In 1993, the average amount of funds raised and used was more or less identical across the three groups. Over time, however, winning exporters on an average have been able to raise more funds from various sources than losing exporters and domestic firms.

2 Across all the three groups and over the entire period, external sources are the most important, accounting for more than 60 per cent of the funds raised.²¹ For domestic firms, the importance of internal sources has risen by over 10 percentage points. For winning exporters, the importance of internal sources has fallen over time by around 5 percentage points. For losing exporters, no clear pattern is found in the share of internal sources.

3 Within external sources, funds raised through capital markets have been most significant for the winning exporters.

4 On an average, borrowings have been more important for domestic firms and losing exporters than for winning exporters.

5 The share of gross fixed assets in the uses of funds has risen substantially for the winning exporters. There seems to be no discernible trend for losing exporters and for domestic firms.

The broad conclusions above seem to suggest that exporters as a group are likely to be less financially constrained than domestic firms. Further, it is likely that winning exporters are less financially constrained than losing exporters. In the next section we attempt to estimate investment functions separately for these categories to test the extent of financial constraints that firms in various categories face.

Investment function

Specification²²

In the empirical literature on firms' investment behaviour, two sets of hypotheses relating to finance constraints are usually tested. First, the presence of a finance constraint is explored using the specification for a panel of firms in eqn (4.1):

where I is investment, K is capital stock, Q is an estimate of Tobin's q, IS is internal sources of funds, ε is the error term, i is the firm subscript and t is the time subscript. If in the above specification the estimated coefficient c turned out to be positive and significant, it is taken as evidence in favour of the finance constraint hypothesis. Sometimes, besides IS, a variable measuring leverage is also added to the above specification.

The second hypothesis explored in this literature is the so-called 'excess sensitivity hypothesis', which states that the degree of finance constraints varies across firms of different characteristics representing inter-firm differences in information costs. In order to test the excess sensitivity hypothesis, firms are grouped into 'high information cost' and 'low information cost' categories based on some a priori criteria (such as firm size). A higher value for the estimated coefficient c for the 'high information cost' group points to the excess sensitivity of this group to financing constraints.

Empirical work within this framework in the developing country context is sparse. In the few studies available, Tobin's q is replaced by a traditional sales-accelerator model of investment. In this approach, fluctuations in output/sales motivate capital spending. To such a model, cash flow and leverage ratios are added to capture finance constraints. A typical specification is as follows in eqn (4.2) (see Harris et al. 1994):

I is investment, K is capital stock, S is sales, IS is internal sources of funds, D is debt, ν is error term, λ is the time-invariant firm-specific effect, η is a common time effect, ε is the idiosyncratic component of the error term, i is firm subscript and t is time subscript. Positive and significant estimates of α2 indicate the presence of finance constraints. Tests of the excess sensitivity hypothesis can be done as described earlier. The coefficient α3 reflects the premium above the safe rate that must be paid as the debt to capital ratio increases and it may vary across groups of firms.

In this study, we intend to estimate an investment function such as eqn (4.2) to test for the presence of finance constraints. Earlier, we had seen that the share of external finance in the total sources of funds is larger for exporters than for domestic firms. Moreover, external funds raised through capital markets as a percentage of total external funds are higher for exporting firms. The relative success of exporters in raising funds through capital markets possibly suggests that these firms might belong to the low information cost category while the domestic firms might belong to the high information cost category. From the perspective of the suppliers of funds, the quality of investment projects is likely to be superior for exporters who have a proven record in international markets – given that international markets are perceived to be highly competitive. The flip side to this is that success (or mere continued presence) in domestic markets, which were largely protected in the pre-1991 regime, is not sufficient assurance that the firm will remain successful in the increasingly competitive environment that is evolving in the domestic product markets since 1991. This suggests that finance constraints are likely to be more severe for domestic firms than for exporters (excess sensitivity hypothesis). In an exactly analogous way, even among exporters, losers are likely to face more severe finance constraints than winners (winners and losers as we defined above). In what follows, we attempt to empirically test the above propositions.

It may be noted here that the criteria we have used to classify firms into high and low information cost categories are, to our knowledge, unlike any used hitherto in the literature. Traditionally, a common criterion used to distinguish firms into high and low information cost has been firm size (usually measured in terms of net fixed assets). We also attempt to evaluate if the above-mentioned excess sensitivity hypothesis (exporters versus domestic firms) holds after controlling for firm size.

Prior to 1991, the Indian Government strictly controlled the creation of new firms and the expansion of existing firms through a rigid licensing regime in accordance with plan priorities. The plans had both industry-specific real capacity targets and a financial plan to ensure the realisation of these targets. Control was exercised on the financial side by public ownership of financial institutions providing long-term loans to the private corporate sector. The government provided subsidised credit to these financial institutions, which were in turn directed to the private corporate sector at a fixed rate of interest implying that these institutions had a limited screening role to perform. Private corporate firms faced severe restrictions on the pricing, quantum and timing of new issues and the government also forced certain industry-specific debt/equity ratio norms on firms, leaving little leeway for firms to choose their capital structure.

In such a scenario, de facto, finance did not matter for investment and the traditional finance literature that focuses on informational asymmetries and agency costs faced by suppliers of funds can be argued to be of little relevance for the pre-1991 period. Furthermore, during the period 1991–3, rapid changes were taking place in the financial sector so we chose to exclude these years from our analysis. We, therefore, estimate the investment function for the period 1993–7.

Empirical results

The investment functions are estimated using the pooled data, namely for five years (1993–7) across 204 firms (total 1,020 observations). For the construction of the dependent variable (I/K) and explanatory variables (ΔS/K, IS/K and D/K), we need a measure of the real capital stock. We estimate the beginning of the period capital stock from book value using a method similar to that of Athey and Laumas (1994). The following assumptions are made:

1 All the firm's capital has an identical useful life Li.

2 The firm's initial end of period capital stock equals the book value of net fixed assets in current rupees.

3 Firms use the straight line method of depreciation and actual depreciation is exponential with depreciation 1/Li.

4 All investments are made at the beginning of the year and all depreciation is subtracted at the end of the year.

We estimate the beginning of the period's capital stock by eqn (4.3):

where P is the wholesale price index of capital goods.

Besides the above three explanatory variables, we also construct various dummy variables to represent firms according to different categories such as domestics, winners, losers, etc. Table 4.12 lists the variable notations used and also their definitions.

Finance constraints – Overall sample

The investment function – eqn (4.2) – for the entire sample (i.e. no distinction is made between domestics, exporters, etc.) is estimated using panel data techniques (a) in levels allowing for both firm and time effects, and (b) in first differences allowing only for time effects. Time effects were found to be insignificant in both cases whereas firm-specific effects were found to be significant in the levels. Table 4.13 reports the GLS estimates for the levels regression and OLS estimates for estimation in first differences. The positive and significant coefficient for IS/K shows that for the entire sample financial constraints are important in explaining investment behaviour.

Exporting firms versus domestic firms

To test the excess sensitivity hypothesis between exporting and domestic firms, dummy variables are introduced into the regression for both the intercept and the slope coefficients attached to IS/K and D/K. The dummy variable takes the value one for exporting firms and zero for domestic firms. The estimation is carried out using OLS without allowing for any firm-specific effect or time effect.²³

The regression results in both levels and in first differences are reported in Table 4.14. It is seen that the slope dummy attached to IS/K is negative and significant in both the levels and first-difference regressions. This suggests that finance the constraint is less binding for the exporting firms than for domestic firms upholding the excess sensitivity hypothesis. We, therefore, re-estimate the investment function separately for the domestic firms and for exporting firms (Table 4.14). These show that the finance constraint is unambiguously binding for domestic firms. For exporting firms the coefficient of IS/K is clearly lower than that for the domestic firms in both the levels and first-difference regressions. There is, however, some ambiguity as to the significance of the coefficient between the levels and first-difference regressions.

Small versus large firms

As indicated earlier, firm size has often been used as a criterion to classify firms into high information cost and low information cost categories. From our point of view, it is important to ensure that the excess sensitivity hypothesis between domestic firms and exporters continues to hold after controlling for firm size. Towards this end, we first define a firm to be a small firm if its net fixed assets are less than Rs 25 crore (i.e. Rs 250 million).²⁴ Dummy variables for small firms and large firms are accordingly defined. Investment functions incorporating size effects are estimated first without distinguishing exporters and domestic firms (Table 4.15) and next by making this distinction (Table 4.16).

From Table 4.15 we find that the slope dummy for small firms attached to IS/K is negative and significant (in both levels and first differences), indicating that finance constraints are less important for small firms than for large firms – a somewhat counter-intuitive result. A similar conclusion was arrived at by Athey and Laumas (1994), who attribute it to the government's policies that favoured small firms.

We now turn to the question of whether the excess sensitivity hypothesis between exporters and domestic firms holds after controlling for size. We concentrate on the slope dummies attached to IS/K which correspond to small domestic firms, small exporting firms and large exporting firms. From the results reported in Table 4.16, we see that these slope dummies are negative and significant (in both levels and in first differences). Moreover, the results suggest that the finance constraint is less severe for (a) small exporters than for small domestic firms, and (b) large exporters than for large domestic firms.²⁵ Thus, the excess sensitivity of domestic firms to finance constraints over exporting firms holds well across firms of similar size.

Winning and losing exporters

A similar approach as above is adopted to test the excess sensitivity hypothesis between winning exporters and losing exporters. Two sets of dummy variables are used to distinguish the winners and losers vis-à-vis the domestic firms. As above, the dummy variables are used in both the intercept and slope terms. Table 4.17 reports the estimation results. The slope dummies attached to IS/K for both winners and losers turn out to be negative and significant as expected. Nonetheless, when the investment function is estimated separately for winners and losers, the coefficient of IS/K turns out to be substantially lower for both these types of firms than for the domestic firms (Table 4.14). In fact, in first differences the coefficient is insignificant for both these categories.

These results should, however, be interpreted with caution. While the excess sensitivity of domestic firms versus the exporters is clearly established, the same cannot be said between winners and losers. Possibly this is due to the fact that sample size is rather small for the winners and losers. It may be pointed out here that while the criteria used to categorise firms into domestics and exporters are rather straightforward, the same cannot be said about the criteria used to categorise exporters into winners and losers. This is because we have essentially compared the annual growth rate of a firm's export with a benchmark growth rate of the relevant industry's exports. One may expect that the growth rate in exports is more volatile for individual firms than for the industry as a whole. Considering that we have only five years of data this could lead to misclassification of some firms as either winners or losers, thus affecting our results.

Another possible reason for not obtaining clear results at the level of winner and loser firms could be that we have not controlled for the fact that a winner/loser firm belongs to a winner/loser industry. We attempt to control for this factor below.

Winner/loser–industry/firms

A priori we would expect that the information cost would be the least for winner firms in winner industries, followed by loser firms in winner industries, winner firms in loser industries, loser firms in loser industries, and finally domestic firms in that order. Accordingly, the severity of finance constraints would increase in the above order.

We define dummy variables that distinguish firms into four categories, namely winner firms in winner industries, winner firms in loser industries, loser firms in winner industries and loser firms in loser industries. Distinction of domestic firms by winner/loser industries is not made here since the focus is on the severity of finance constraints within different categories of exporting firms. Table 4.18 reports the results of this dummy variable regression.

These results indicate that:

1 The severity of finance constraints is highest for domestic firms followed by exporting firms in loser industries and is the least for exporting firms in winner industries (compare the coefficients of WWIS/K and LWIS/K on the one hand with those of WLIS/K and LLIS/K on the other).

2 The above expected progression in the severity of finance constraints is found to hold true for exporting firms within loser industries but not for exporting firms within winner industries (compare the coefficients of WWIS/K with LWIS/K and that of WLIS/K with LLIS/K).

Coefficients of ΔS/K and D/K

The coefficient of the accelerator (ΔS/K) is in most of the cases positive and significant in both levels and first-difference regressions as expected. The exceptions are the regressions in first differences for exporters, winners and losers, and in levels for winners and losers. With respect to the coefficient of the leverage term (D/K), however, no clear pattern emerges either on the sign of the coefficient or its significance. This could perhaps be due to the small size of our sample. We may note here that some other studies have also reported similar results (see Hall 1992; Harris et al. 1994).

Examining the financing patterns of firms, we have found that winning exporters have been able to raise more funds from various sources than losing exporters and domestic firms. Furthermore, there is evidence to suggest that while financial constraints are important in explaining investment behaviour for all firms in our sample, they are less binding for exporting firms, particularly those in the winner industries.

4. Stylised facts on micro–macro and trade–finance interactions

Trade specialisation and sustainability of current account

From the policy-maker's perspective, issues such as patterns of trade specialisation and competitiveness are of interest primarily because of their implications for the link between economic growth and current account sustainability. It is well known that the current account balance is influenced both by micro-factors (such as trade specialisation and competitiveness) and macro-factors (mainly the savings–investment gap). In India, in the 1980s and 1990s, macro-factors played a dominant role in determining the current account balance (as has been noted in Section 2). Our findings suggest that with greater amounts of resources flowing into the exporting sectors since the 1991 reforms, poor export performance due to resource constraints is unlikely to be a source of concern for current account sustainability. Other factors, however, such as the lack of adequate infrastructural facilities (roads, ports, power, etc.) may prove to be a generalised constraint on the supply side, which could affect export performance and thus ultimately the trade balance.

Labour costs and trade competitiveness

The non-dynamism of the export sector has long been an issue of policy concern in India. Given India is a labour-surplus country, it has been suggested that trade patterns should follow comparative advantage and that India should specialise in the export of labour-intensive commodities. The 1991 reforms were an attempt to provide an impetus to India's manufacturing exports, especially of the labour-intensive type. There is little evidence, however, of a significant increase in manufacturing exports (labour intensive or otherwise) in the post-1991 period. Moreover, we do not observe any correlation at the sectoral level between unit labour costs and export competitiveness. The lack of importance of price factors in determining competitiveness at the sectoral level could be taken to provide support for the view that non-price factors (such as finance constraints) may be important determinants of competitiveness at the firm and industry levels. Equally, it could also be due to the possibility that, notwithstanding the economic reforms of 1991, there remain severe distortions in the Indian economy, which are far too deep and in extent to enable the country to exploit its labour resources.²⁶

Financial environment and finance constraints

In the restrictive policy environment prior to 1991, financial intermediaries were passive conduits of funds from the government and the banking sector to manufacturing firms. The 1991 reforms have empowered financial intermediaries to play an active role in resource allocation. We have argued that in a regime where resources are allocated according to government directives, the very concept of finance constraints is of little consequence (as all real plans are backed by a financial plan). Finance constraints (caused by informational and agency costs) are relevant when financial intermediaries screen projects. In our empirical work, we have demonstrated that in the new environment financial intermediaries seem to take export performance as an indicator of a firm's competitive strength. Thus, investments by exporters in general and, in particular, among exporting firms in winning industries are not restricted by the availability of internal funds. On the other hand, financial intermediaries seem to view firms that operate primarily in domestic markets as lacking in competitive strength and, consequently, investments by these firms are constrained by the availability of internal funds.

Notes

¹ We gratefully acknowledge comments by Mustapha Nabli, José María Fanelli, Ari Kuncoro, Paolo Guerrieri and other participants in the Interim Workshop held at the Philippines Institute for Development Studies, Manila, during 21–23 April 1998, and in the Final Workshop held at the Trade and Industrial Policy Secretariat, Johannesburg, during 30 November–2 December 1998. Usual disclaimers apply.

² As we shall see in Section 2, there was, in fact, an improvement in India's export performance since the mid-1980s, due in great part to the export-promotion policies followed during this period.

³ There was another mechanism by which the foreign exchange constraint would prove to be binding on domestic demand-driven growth in the Indian context. An increase in aggregate demand would lead to higher food prices and, hence, inflation. Given the aversion of Indian policy-makers to high inflation, this would invariably trigger off deflationary fiscal and monetary policies (as the lack of adequate foreign exchange reserves precluded the possibility of large-scale imports of food).

 ⁴ It should be noted that such a periodisation is widely accepted in the literature. For example, see Ahluwalia (1991) and Joshi and Little (1994).

 ⁵ Further details of these policies can be found in Ganesh-Kumar et al. (1998).

 ⁶ Openness as conventionally defined is the sum of exports and imports as a ratio of GDP.

 ⁷ A full discussion of the CMS methodology is available in Kumar, Sen and Vaidya (1999).

 ⁸ The definition of manufacturing used here is the SITC-based one and includes all commodities in the SITC categories 5 to 8 excluding 68. It should be pointed out that this definition differs from the definition of manufacturing exports used in Section 2 beginning with 'The real exchange rate and aggregate competitiveness'.

 ⁹ It should be noted that the CMS analysis ends in 1992 while the rest of the empirical results in this section are until 1996. The CMS analysis requires data on bilateral trade flows for all commodities and all countries. Such detailed data for the post-1992 period were not readily available at the time of the study.

 ¹⁰ The RCA measure expresses the share of country i's export of product j in total world exports of product j, as a ratio to the share of country i's total exports of manufactures in world total exports of manufactures. An RCA of unity would imply 'normal' export performance of product j relative to the size of country i, as an exporter, while a ratio of 2 would suggest that the product j's share in country i's exports is twice the corresponding world share, and so on. An RCA of more than unity is usually taken as an indicator of competitiveness, while an increase in the RCA supposedly suggests a strengthening of the competitiveness so revealed (see Balassa (1965) for further details).

 ¹¹ Commodity-wise time series of RCAs are not presented here but will be made available upon request.

 ¹² We use the Grubel and Lloyd (1975) measure of intra-industry trade (IIT), defined as

where Xit is India's exports of commodity i at time t, and Mit is India's imports of commodity i at time t. The variable IITit can be between 0 and 1, with higher values indicating greater intra-industry trade.

 ¹³ Details available from the authors upon request.

 ¹⁴ We have excluded ISIC 3909 from the above list as it includes manufacturing industries not elsewhere classified. Therefore, by definition, ISIC 3909 would have a high volume of intra-industry trade.

 ¹⁵ We have used Lall's (1998) classification of the technological complexity of various exports, which is a refinement of OECD (1998).

 ¹⁶ See Lall (1998) for further details.

 ¹⁷ The presence of such domestic firms can be explained primarily in terms of the extremely restrictive trade and industrial policy regime that prevailed prior to 1985.

 ¹⁸ Recall that an Indian industry was classified as winner/loser by comparing its export performance with a world norm for that industry: winner if it outperformed the world norm; loser otherwise.

 ¹⁹ Note that RCA of an industry can rise/fall over time even when the world trade in that industry is falling over time. None of the industries (both winners and losers) chosen below in our analysis fall in such a category. Hence this case is not discussed further.

 ²⁰ We have not reported the correlations matrix for brevity.

 ²¹ The figures for 1997 for both winners and losers seem to be influenced by some extreme values and thus need to be interpreted with caution.

 ²² We draw upon Hubbard (1997), Athey and Laumas (1994) and Harris et al. (1994) in this discussion.

 ²³ Note that introducing a dummy variable for exporting firms and simultaneously allowing for firm-specific or time effects through relevant dummy variables would result in the equation being collinear. Hence, we ignore firm and time effects.

 ²⁴ Note that the Government of India defines a small firm as one whose gross fixed asset is less than Rs 1 crore for its priority lending policies. Historically, this limit was much lower.

 ²⁵ The small size of our sample in each of these four categories prevented us from estimating the investment function separately.

 ²⁶ Two such distortions that may be relevant here are the lack of an exit policy in labour market and the widespread reservation of products for the small-scale sector.

References

Agrawal, P., Gokarn, S., Mishra, V., Parikh, K. S. and Sen, K. (1995) Economic Restructuring in East Asia and India: Perspectives on Policy Reform. Basingstoke: Macmillan.

Ahluwalia, I. J. (1991) Productivity Growth in Indian Manufacturing. New Delhi: Oxford University Press.

Athey, M. J. and Laumas, P. S. (1994) 'Internal Funds and Corporate Investment in India', Journal of Development Economics 45: 287–303.

Balassa, B. (1965) 'Trade Liberalisation and Revealed Comparative Advantage', Manchester School 33(2): 99–123.

Brockett, P. and Levine, A. (1984) Statistics and Probability and Their Applications. New York: CBS College Publishing/Saunders College Publishing.

Edwards, S. (1994) 'Real and Monetary Determinants of Real Exchange Rate Behaviour: Theory and Evidence from Developing Countries', in J. Williamson (ed.), Estimating Equilibrium Exchange Rates. Washington, DC: Institute for International Economics.

Elbadawi, I. A. (1994) 'Estimating Long-Run Equilibrium Real Exchange Rates', in J. Williamson (ed.), Estimating Equilibrium Exchange Rates. Washington, DC: Institute for International Economics.

Fagerberg, J. (1988) 'International Competitiveness', The Economic Journal 98: 355–74.

Feenstra, R. C., Lipsey, R. E. and Bowen, H. P. (1997) 'World Trade Flows, 1970–1992 with Production and Tariff Data', National Bureau of Economic Research Working Paper, No. 5910.

Ganesh-Kumar, A., Sen, A. K. and Vaidya, R. R. (1998) 'Finance and Changing Trade Patterns in Developing Countries: A Case-Study of India', Indira Gandhi Institute of Development Research, Mumbai PP Series, No. 38.

Grubel, H. and Lloyd, P. J. (1975) Intra-Industry Trade. London: Macmillan.

Hall, B. H. (1992) 'Investment and Research and Development at the Firm Level: Does the Source of Finance Matter?', National Bureau of Economic Research Working Paper, No. 4096.

Harris, J. R., Schiantarelli, F. and Siregar, M. G. (1994) 'The Effect of Financial Liberalization on the Capital Structure and Investment Decisions of Indonesian Manufacturing Establishments', The World Bank Economic Review 8(1): 17–47.

Helpman, E. and Krugman, P. (1989) Trade Policy and Market Structure. Cambridge, MA: MIT Press.

Hubbard, R. G. (1997) 'Capital Market Imperfections and Investment', National Bureau of Economic Research Working Paper, No. 5996.

Joshi, V. and Little, I. M. D. (1994) India: Macroeconomics and Political Economy, 1964–1991. New Delhi: Oxford University Press.

Kanji, G. K. (1993) 100 Statistical Tests. New Delhi: Sage Publications (India) Private Ltd.

Lall, S. (1998) 'Technological Capabilities in Emerging Asia', Oxford Development Studies 26(2): 213–43.

Loertscher, R. and Wolter, F. (1980) 'Determinants of Intra-industry Trade: Among Countries and across Countries', Weltwirtschaftliches Archiv 116: 280–93.

Mayers, C. (1990) 'Financial Systems, Corporate Finance, and Economic Development', in R. G. Hubbard (ed.), Asymmetric Information, Corporate Finance, and Investment, National Bur