Document(s) 9 de 16

The evidence presented in Chapter 8 of the generally low employment elasticity in organized manufacturing suggests that much of the growing labor force outside agriculture has been absorbed either in the tertiary sector or in unorganized manufacturing. Since the gap in labor productivity and earnings between the unorganized and the organized sectors is large, this leads to the phenomenon of 'dualism' in the manufacturing sector – which is alleged to be the source of welfare loss both from the efficiency and equity angles. There are in fact two aspects to 'dualism'. The first has its origin in the productivity gap and distribution of employment between the 'formal and the 'informal' sectors, where the demarcation line between the two is fixed by the internationally comparable definition of the use of more than five workers in the establishment. Second dualism would be accentuated if, within the formal sector, distribution of employment is skewed heavily to large firms, with relatively small representation of small and medium enterprises. In other cases the distribution of employment, even within the 'formal' sector, might be strongly bi-polar, with two peaks of employment at the low and the high end of the size spectrum, and a wide range of size groups with a relatively small number of workers. This is the phenomenon of the 'missing middle'.

Categories in the 'unorganized' manufacturing sector

Before embarking on a detailed analysis of the problem it is important to be clear about the sub-groups within informal manufacturing (or the unorganized sector, as it is called in India), and which sub-sector we are considering in our comparison with the formal (organized) sector.

The data on the sub-sectors in manufacturing outside the ASI sector (analyzed in the last chapter) as well as trade and services come from the sample surveys organized by the National Sample Survey. The surveys used the sample frame of Economic Censuses conducted by the Central Statistical Organization (CSO) of the Government of India in selected years. The Unorganized Sector Surveys define three categories of enterprises, classified by the size of enterprise and the type of labor used:

1. Own-account manufacturing enterprises (OAME). These units make use of only of family labor, and most of them are on the household premises.
   
   
2. Non-directory establishments (NDME) which have at least one hired worker, but the total size of all workers (including family labor) do not exceed five.
   
   
3. Directory manufacturing establishments¹ (DME) which have at least one worker hired more or less on a regular basis and their total complement of workers is six but less than ten. If it is more than ten then the enterprise is under the purview of the Factory Act and is part of the formal sector covered by the Annual Survey of Industries.

Within the unorganized manufacturing sector nearly 67 percent of the workers were in own account enterprises in 2001 (Table 2). The proportion of workers in own account units was even larger in rural areas 79 percent, compared to urban, 43 percent. Less than 20 percent of workers were in the larger unorganized enterprises with 6–10 workers (DME) and about 14 percent were in the NDME, less than 6 workers. Overall workers were engaged in larger enterprises in urban areas, 30 percent in DMEs and about 26 percent in NDMEs. Over the period 1983 to 2001, the share of employment in the larger units increased marginally.

(Unni 2006, p. 219)

The OAME units represent the 'pure' informal sector, based on households and in most cases in the same premises. They generally pursue the traditional or craft activities. A great deal of the NDMEs also fall into this type of activity, although they make use of 'at least one' hired worker (who indeed might be partly a house servant). For our purposes we will use the DME sector as representing a part of the formal sector – the other component of the latter being the sector covered by the ASI including units with ten or more workers. Admittedly this demarcation is to some extent arbitrary, determined by the practices of the National Sample Survey. But it stands to reason that the establishment has entered a more 'modern' economic relationship when it has graduated to a six-worker employment size. This size group is also a highly convenient one because it permits comparison with other countries in the region (see below).

The DME sector is then be distinguished, at the lower end, from the informal sector on the one hand – comprising the very large household sector – and the smaller non-household units. At the upper end it is demarcated from the larger-scale component of the 'organized' manufacturing sector covered by the ASI (which was discussed in the last chapter). Table 9A.1 gives the size distribution of the entire spectrum of enterprises, including the sub-sectors of the unorganized as well as the formal sector in manufacturing.

Distribution by size-groups in manufacturing could be considered with respect to either value added or employment. In fact, the former is the more basic of the two and is the product of two separate variables: first, the distribution of employment by size groups; and second, the differences in productivity or value added per worker as between size groups. In what follows we will work with these two variables to shed more light on the economic processes involved. Wages generally increase proportionately with labor productivity. Thus the extent of productivity differentials between small and large units would reflect differences in wage levels between them. In so far as informal sector undertakings would have wage and productivity near to the levels found in the smallest size-group in the formal sector, the large-small productivity differential in the formal sector would also be a measure of the economic distance between the informal and the formal sector firms in the economy concerned.

The size distribution and labor productivity differentials by size-group: India relative to selected Asian countries

This section seeks to present a snapshot of the Indian pattern of size distribution and productivity differential by size in manufacturing relative to selected Asian countries. It should be noted that we are confining ourselves to the size-distribution within the 'modern' sector. That is to say, our international comparison refers to the second aspect of 'dualism' mentioned above: the establishments employing five or more workers. Data could be assembled for only a few Asian countries, for various years in the late 1980s.

Table 9.1 presents data on the distribution of employment by size groups, while Table 9.2 sets out the data on relative labor productivity for the various size groups. The data reported in Tables 9.1 and 9.2 for other Asian countries also have a similar cut-off points at the lower end based on employment size they are comparable to the Indian statistics.

Basically three 'types' can be distinguished within this small sample:

1. a fairly even-size distribution in which small, medium and large firms plays more or less equally important roles and the productivity difference between the size classes is small;
   
   
2. the pattern in which the distribution of employment by size groups is distinctly skewed to the large firms. Typically in this pattern the productivity difference between large and small firms tends to be substantial; and
   
   
3. the 'dualistic' pattern in which there is a strong mode at both ends of the distribution – a relatively large proportion of employment is found both in the small and the large size groups. Within this 'type' two sub-types can be distinguished depending on the extent of the productivity differential between small and large firms.

(i) The first group is classically represented by the case of Hong Kong. As can be seen in Table 9.1 employment was fairly evenly distributed among the various size-groups, with the small enterprises playing as much a role in the island's manufacturing structure as medium and large enterprises. At the same time, the difference in labor productivity between the largest and the smallest size-group is the smallest in the sample (Table 9.2).

The pattern of distribution in Hong Kong could be usefully compared with that in the Japanese economy which has been characterized by the strong role of small establishments. It will be seen from Table 9.1 that although the modal size group for both Hong Kong and Japan is the small enterprises of 10–49 workers, the proportion of employment in large enterprises of 500+ workers is significantly larger in Japan. Further, the data in Table 9.2 show that productivity differences between small and large firms were much less in Hong Kong. The wage differential between small and large units was accordingly much smaller. Average earnings in Hong Kong in 1982 were only 55 percent higher in establishments with more than 1,000 workers than in those with 1–9 workers. In Japan the wage differential was twice as much.²

Hong Kong comes closest to a free-market model of development in Asia. Beng (1988) observes that 'within the proclaimed laissez faire environment in Hong Kong the government does not seem to have a policy towards manufacturing not to mention any policy towards the SSIs' (p. 88). An obvious hypothesis emerging from the Hong Kong case is that left to itself modern industry makes efficient use of small enterprises in a striking way. Also, in the absence of the usual set of policy biases which protect both capital and labor in large firms, labor productivity and wage differentials are kept within fairly narrow bounds.

Of the other countries represented in the sample, Taiwan comes close to the Hong Kong pattern. The size distribution is very similar. While the productivity difference in Taiwan would seem to be larger if we compare the lowest and the highest size-groups, closer examination shows that this appearance is largely due to the high relative productivity of the largest (500+) size-group in Taiwan. Value added per worker rises very gently up to the level of the large firms of 500 plus workers, and then seems to take a big jump.

Differences in wage levels, as measured by average earnings of the workers between the smallest and the largest size-groups, are almost the same for Taiwan and Hong Kong.³

(ii) The second pattern in our sample is a size distribution of employment which is skewed to the right, with the modal size-group employing 500+ workers.

The countries in our sample which show this distribution are Korea and Thailand, although, as of 1986, Korea had a larger presence of smaller firms than Thailand, particularly in the 20–50 size-group. But Korea had been consciously trying to develop its small and medium sectors since about a decade earlier. In 1976, when the proportion of employment in the largest size-group peaked at 45 percent, the Korean distribution was much more skewed – almost at par with Thailand's.

Malaysia is another country which, in 1981, showed a pattern of distribution skewed to the large size-group. But it can be seen from Table 9.2 that the productivity differential between small and large firms is much smaller than in the case of Korea. Thus we would expect different economic forces operating on the size distribution in the case of these two countries.

(iii) The 'dualistic pattern' is characterized by, first, the strong presence of both small establishments and large firms, and second, the substantial economic distance between small and large firms. The classic case of this type is Japan. The 'dualistic' pattern of Japanese industrialization has a long history. It has its roots in the initial surplus-labor conditions prevailing in Japan during its initial industrialization (which contributed to labor-market segmentation) and the simultaneous development of a complex tying large industry, the state and financial conglomerates which accentuated capital market dualism.

The other less developed countries in Asia–the Philippines, Indonesia and India–all share with Japan the dualistic pattern in their modern (formal) manufacturing sector.⁴ There is, however, a big difference with Japan, which is brought out in Table 9.2. The productivity difference between the small and the large size-groups of firms is much larger in these Asian countries than in Japan. Thus while the 'surplus labor' situation in Asian countries makes the 'dualistic pattern' emerge in a wide variety of Asian economies, Japan had, by the middle 1980s, succeeded in narrowing the gap in productivity between small and large firms which typically characterizes the dualistic development. We will return to this point later.

In South Asia, the extreme peculiarity of the Indian structure is immediately apparent. India has an exceptionally large proportion of employment in the lowest size-group of 6–9 workers and an exceptionally low relative value added per worker in this group. Furthermore, the size distribution is characterized by a large presence of the 500+ group of firms with a conspicuous 'missing middle'. This pattern resembles that of Japan in terms of a 'dualistic' development, but is wildly exaggerated in the Indian case. There can be little doubt that this outcome is basically due to the protectionist policy adopted by the government since 1950 which favored the small scale.

Figure 9.1 The missing middle manufacturing firms–India compared to other countries (source: Figures are taken from Table 9.1).

The problem of the missing middle

Figure 9.1 shows the extreme case of the formal manufacturing sector in India–how employment was concentrated in the two extreme size-groups compared with the other Asian countries.

It would seem in the Indian case there are formidable obstacles to the small units growing beyond a threshold size into middle-sized ones. This is a serious barrier in so far the middle-sized entrepreneur is often the most dynamic, and the competitiveness (and hence efficiency) of the manufacturing sector depends a good deal on the buoyancy of such units.

The productivity gap between the informal and the formal sectors in manufacturing

The second problem distinguishing the Indian case from the other Asian countries is the much larger productivity gap noticed in India. In Japan, Korea and Taiwan the labor productivity in the largest units (employing more than 500 workers) was around three times that in the smallest units. In India it was eight times as large. Even in less developed Asian country like Indonesia the larger units had labor productivity no more than five times that in the smaller units.

The policy of SME development in India

Both the two exceptional characteristics of the Indian case can be traced to have its origins in the peculiar policy of industrial controls practiced in the first 40 years or so after independence. In India a dual system of protection has been in effect since the beginning of independence. On the one hand, the policy has been to protect the small-scale from the competition of the large – the policy of 'reservation', under which a long list of items has been designated as the exclusive preserve of the small-scale (defined in terms of the value of capital assets). The capacity of production of these items by large-scale units has been frozen at the level at the time of the legislation. At the same time, the import substituting industrialization has protected all domestic units – small and large – from the competition of foreign firms. The result has been that small and large firms have developed their own niches of markets in different lines of production without too much competition between them or from foreign firms.

This method of fostering the growth of SMEs was first introduced in 1967 and the list of items "reserved" for the small-scale have been progressively increased and in the mid-nineties it comprised a total of around 830.⁵ The value of the limit in plant and machinery has been increased over time in nominal terms, but the increase in value of this limit after allowing for inflation has been small.

Initially, this approach encouraged the establishment of large numbers of new SE units which were protected from competition from the large-scale sector. But the problem with the continuation of such policies is two-fold: (i) in attempting to select labor-intensive products or industries, it misses the point that labor-intensive enterprises are found in many, nearly all industries, not just a limited set which can be easily identified; and (ii) it is not sufficiently biased towards small enterprises which show potential for growth.⁶

(i) Industry-based policies of reservation overlook the fact that small enterprises are not confined to specific product lines, and that their importance in different product groups is constantly changing. What is needed is for policies to have a pervasive effect so that all small enterprises, no matter in what product groups, could potentially take advantage of the assistance measures available.

Small enterprises are generally more labor-intensive than large ones, especially if size is defined in terms of fixed investment rather than employment. But it does not mean that they are concentrated in industries where the mean capital – labor ratio is particularly low. SMEs are found in many industries. There is no reason that in any economy the number employed (or the proportion of total output or investment) in SEs would be larger in those industries which have a less than average K/L ratio than in those in which the ratio is above the average. This is because there is a spectrum of techniques within each industry, and enterprises of different sizes and capital intensities will be found in most of them.

(ii) Turning to the second point of criticism, it is important that SE support policies do not discourage the growth of small into medium enterprises. Here the approach of the Indian package of polices has been the opposite of what is desired. Along with the reservation policy, there have been a number of fiscal subsidy programs and other forms of support which provide benefits to enterprises below a certain size. Thus there is a built-in disincentive for enterprises to go beyond this size limit. Labor laws on wages, benefits and job security are applied to units above the critical size. Enterprises graduating out of the protected small sector are thus faced with extra costs even as they are denied the benefits of fiscal subsidies and other programs.

The effect has been a polarization of the industrial structure. The small-scale and large enterprises have increasingly occupied different niches of the market in the same industry. Even when industries are defined narrowly in terms of specific product lines, there is generally a great deal of difference in the quality of the product. Small enterprises with low wages and less mechanized techniques occupy the lower end of the spectrum, catering to the demand of low-income consumers, while larger mechanized firms serve the high price segment of the market. The classic example is the textile industry. Small units with non-automatic, often reconditioned, looms ('powerlooms', as they are called in India) produce cheap cloth, while the large factories with automatic looms produce more durable cloth for the upper-class domestic and export markets. This type of polarization accentuates dualism and increases the productivity and wage gap between the small and large sectors.

Post-reform developments

The package of reforms in the last decade could be expected to have made a serious dent to the traditional policy of protection for the small scale. On the one hand, liberalization of import controls, particularly on a range of consumer goods, should have reduced the strength of the protective umbrella. At the same time the relaxation of the licensing system for large-scale industrial units could be expected to have reduced the effectiveness of the policy of 'reservation' for small enterprises. What is the evidence on the effect of these developments on the size structure of manufacturing?

The relevant data culled from the NSS Establishment Surveys and ASI for the various dates are given in Tables 9.3 and 9.4. They are graphed in Figure 9.2.

As far as the distribution of employment is concerned the only change over the period covered seems to have been a significant reduction in the number employed in very large firms (1,000 and above). The distribution is, however, still bi-polar with strong modes at the employment size groups at the two extremes (6–9 and 500+).

The productivity differentials by size-groups seem to have changed even less. If anything the extreme 'dualism' noticed in India compared with other Asian countries seems to have worsened since 1984–1985, though much of the deterioration occurred in the first half of the nineties.

Size structure of enterprises in the ASI sector

The problem of the 'missing middle' carries over into the large-scale or ASI sector of manufacturing. The inability or unwillingness of smaller non-ASI units to grow is reflected in the paucity of small-scale establishments in the latter. Little et al. (1987) pointed out the peculiarly biased structure of employment in the ASI sector for the early year of 1977. The proportion of total employment in size-groups increased progressively as one went from the smaller to the larger groups. In other words, the dualism one noticed in formal manufacturing as a whole, taking the ASI and the DME sectors together, was not apparent when one looked at the ASI sector by itself. Rather, the distribution was heavily skewed to the largest size-group. This picture contrasted starkly, not only with the experience of Japan, which had a 'cascading' employment structure with the proportion of employment falling as the size category increased, but also with the United States, Korea and Taiwan. All three of these had an employment distribution more or less resembling the 'normal' curve and had the highest proportion of employment in the 100–500 size-group (or what might typically be called the 'medium-scale' units). The dominance of the very large firms in India had been whittled down somewhat between 1956 and 1977, but even so in 1977 the establishments with more than 1,000 workers accounted for no less than 45 percent of the total employment in the ASI sector.

Figure 9.2 India–distribution of employment and productivity by size groups. Panel A: Distribution of employment (in %) in manufacturing firms by employment size groups. Panel B: Index of labor productivity by different size of firms (source: ASI data).

Note
The slabs have been adjusted to take care of the varying length of the size groups (see Little et al. p. 88).

Figure 9.3a Size structure of ASI GVA.

Figure 9.3b Size structure of ASI employment.

Note
The slabs have been adjusted to take care of the varying length of the size groups (see Little et al. 1987: p. 88).

How does the size structure look in more recent years? The package of reforms in the last decade could be expected to have made a serious dent to the traditional policy of protection for the small scale. On the one hand, liberalization of import controls, particularly on a range of consumer goods, should have reduced the strength of the protective umbrella. At the same time the relaxation of the licensing system for large-scale industrial units could be expected to have reduced the effectiveness of the policy of 'reservation' for small enterprises. What is the evidence on the effect of these developments on the size structure of manufacturing?

The evolving picture in the last 25 years of the last century is presented in Figures 9.3 which give the distributions of value added and employment respectively for different years. It is clear that the whittling down of the largest size-groups, which had been noticed earlier between 1956 and 1977, continued in the recent decades. But it is also clear that much of this change took place before the post-reform period. In fact the biggest reduction in the importance of the size-groups of 1,000 and above had taken place in the early eighties and was associated with the closure of many cotton and jute mills.

Uchikawa (2002, p. 46) found that the

average gross profit ratio between 1979–1980 and 1997–1998 was highest in units employing 50–99, second in units employing 100 to 199 and lowest in units employing 1,000 and above. Medium scale units were (therefore) dynamic sector to gain employment and investment. That is a reason why medium scale units increased employment.

He emphasized that this did not mean that medium-scale units were more profitable than large–scale ones in the same industry. In fact, many of the large units were in the depressed textile industries which were suffering from gross inefficiency and had to be closed down or retooled. These industries pulled down the gross profit ratio for large units in the manufacturing sector as a whole.

Nevertheless, there were some new industries which had a growth rate of value added in excess of 10 percent per annum (between 1979–1980 and 1997–1998) and were also home to many of the medium-scale units. These industries were chiefly NIC 26 (textile products or wearing apparel), NIC 29 (leather and leather products) and NIC 30 (chemicals and chemical products). Firms in these industries contributed to the improvement of average gross profits in the medium-scale units. There was a good deal of churning of establishments in these industries. Many medium-scale units failed, but others merged to take their place.

We can use the ASI to present a disaggregated picture of the industrial structure, and its changes in the post-reform years. It is worth emphasizing that the persistence of dualism in India appears to cut across most manufacturing sub-sectors. Data disaggregated by sectors show that in 14 of the 18 two-digit industries the largest shares of workers are found in the smallest of the ASI groups (10–99 group). Comparing 2000–2001 figures with those of 1994–1995, it is seen that the loss in the share of employment was found in the 100–199 class for as many as 13 out of 18 industries (Table 9.5). It is likely that both these developments are due to the barrier imposed by the job-security legislation which comes into play for units with 100 or more workers. There is a distinct disincentive for establishments to cross this threshold. One way industrial units could deal with the difficulty of crossing the employment barrier of 100 workers is to resort to subcontracting of some of their product lines or components. This topic is discussed in the next sub-section.

To sum up, the size structure of industry in the organized sector, as in the manufacturing sector as a whole, showed only limited change in the post-reform years. There was one clear change: the relative decline of very large units of 1,000 and above. But this was mainly due to the sickness and decline of old industries – chiefly cotton and jute. There is some indication of medium-scale units emerging. But the trend is not a particularly striking one. The ASI sector is still skewed to large-scale establishments. At the same time dualism in manufacturing continues to be striking because of the large employment in very small units outside the ASI sector (even if we leave out of consideration the non-directory establishments and household units).

The role of subcontracting in Indian manufacturing

The fall in the share of 100–199 size-group in a large number of industries, which we infer from the data in Table 9.5 has its counterpart in the rising practice of contract labor. Contract intensity (the percentage of contract labor used to the total labor force in the establishment) in export-oriented or import-competing industries peaks in the 100–199 group when we look at disaggregated groups of industries (Ramaswamy 2006, Table 9.5). This happens probably in response to firms searching for more flexible ways to respond to changing market conditions facing the firms more exposed to global competition.

Use of contract labor is of course not the only form of devolution of activity by larger firms to smaller units. Much the more important method is that of outsourcing – getting smaller firms to produce some components of the final product or producing inputs used in the manufacture of the final product. Subcontracting in this sense has been historically a very important part of the dual structure of the Japanese manufacturing economy, and it has been praised as an ideal system of ensuring the co-existence of small and large firms without sacrificing the efficiency and competitiveness of the sector. But this favorable outcome depended crucially on the subcontracting system graduating from a dependant to a technologically advanced system. It has been noted in the literature that, with advanced technology spreading rapidly, the quality of subcontractors' output became increasingly important, along with their costs. Accordingly, large primary enterprises came increasingly to monitor and upgrade the quality of subcontractors as well as select them carefully (Kaneda 1980, p. 43). This type of keiretsu or 'vertical inter-firm hierarchy' spread rapidly in the fast growing industries like machinery, automobile, metal working and electrical appliance industries. A 1981 survey conducted by the Central Bank for the Ministry of Commerce and Industry revealed that 51.5 percent of the more than 1500 subcontracting firms surveyed claimed that their technology was equal to or even superior to their parent companies (Koshiro 1990, p. 202). The traditional subordination of the subcontractors faded away over a large area of the SME sector with this increasing technological independence. In matters of pricing, negotiated rates for the products supplied by the subcontractors increasingly replaced the old system of dictated rates.

What is the evidence of subcontracting in India developing in the direction of the Japanese model? Ramasawmy (2006) has used data on this point from the ASI and classified them by size-groups of enterprises and the type of industry. His results are reproduced in Table 9.6.

It is clear that in India the largest incidence of product outsourcing takes place, not in larger firms, but in the smallest size groups. This would suggest that the motivation for outsourcing is rather different in India than in the Japanese model. It would seem that small firms are anxious not to grow beyond the employment size of nine workers as it would make them come under the purview of labor laws. This is an altogether different scenario from the Japanese model where outsourcing was perceived more as a policy of vertical disintegration by large firms.

Further evidence on this point is provided in the evidence collected by the 'Unorganized Sector Survey of the NSS. Using this source, Unni (2006) reports that 30 percent of firms in this sector undertook subcontracting work for other firms. The industries in which this percentage was particularly high were: tobacco products (89 percent), textiles (56 percent), chemical products (67 percent) and office accounting and computing. Consistent with the small size of the typical firm which outsourced work, the firms catering to the demand for such work were even smaller. The distribution of subcontracting firms by place of work showed that the overwhelming portion of them (81.2 percent) operate at home, and only 15 percent in business premises. It seems that much of this activity was non-mechanized. While 88 percent of the firms reported receiving raw materials from the contractor, and 93 percent reported working to design specified by the latter, only 7 percent were supplied with any equipment for the work.

One is left with a strong impression that the subcontractors in Indian manufacturing are yet to graduate from the dependent status vis-à-vis the parent firm with a low level of technology. It is a far cry from the Japanese model in which many of the subcontractors were dynamic small firms actively seeking out large producers anxious to collaborate in the production of finished products, and were often at the technological frontier, with a high grade of specialized workforce.

The problem of 'dualism' in Indian manufacturing

We have, in the material provided so far in this chapter, discussed the size distribution in Indian manufacturing and brought the peculiarity of the Indian size structure compared with other Asian countries. "Dualism" and the problem of the 'missing middle' were identified as a dominating characteristic of the Indian structure. Why is this a problem, and in what sense can we consider this phenomenon to be a major factor retarding manufacturing growth in India? We turn to a discussion of this group of questions in this part. Following a discussion of the impact of 'dualism' we discuss the proximate causes of its origin. This discussion leads us to an analysis of the persistence of this phenomenon even when some of the policy distortions seem to have been removed in the post-reform years.

Why is dualism a problem for manufacturing growth?

Why should we regard the phenomenon of 'dualism' in manufacturing as drag on the growth and performance of the manufacturing sector? Of the many points relevant here the more important are the following:

1. the impact on allocative efficiency and wage inequality;
   
   
2. the dynamic impact on the growth of skilled labor and entrepreneurship;
   
   
3. the stagnation in the growth of markets for manufactured goods.

Allocative efficiency and inequality

The large gap in productivity between the firms in the two extreme size-groups, as described in the data on manufacturing presented above, suggests the existence of a large gap in the marginal products of labor and capital between the two classes of firms. We know from independent evidence that large firms have access to capital supplied by the formal financial institutions, while small firms mostly have to depend on local informal sources of finance and the interest-rate differential between these sources can be huge (Little et al. 1987, Chapter 15). It is also well known that wage levels follow differences in labor productivity and large firms have a wage per worker which, even after we have controlled for measurable human capital attributes, are much higher in the large firms. Again the detailed study in a specific labor market (Bombay City) reported in Little et al. (1987, Chapter 14) revealed that, after allowing for the effect of education, training, occupation and knowledge of English, wage per person of manual workers in the largest-size-class of factories employing 1,000+ workers was almost twice the level in 'small' enterprises with less than ten workers. Two conclusions are suggested by this evidence of size-related factor price differentials: first, the larger the differential the larger is the loss in welfare in terms of static allocative efficiency theory; second, since employment in the 'dualistic' is concentrated in the smallest and the largest size-groups, inequality in the distribution of wage per person is very unequal.

Impact on dynamic efficiency

In a more dynamic sense the missing middle implies a weak process of graduation of small firms and the development of entrepreneurship. It is arguable that the dispersion of entrepreneurship as well as industrial technology over a wide spectrum of spatially and economically distributed regions is dependent on the mushrooming of medium-scale enterprises, into which the small units are able to graduate.

Similarly dualism slows down the growth of the labor force with industrial skills. This is particularly true in developing economies in which many of the skill requirements of modern industry (including discipline in the workplace) are acquired by on-the-job-training rather than education in schools. The slow growth of the skilled workforce in its turn has an impact on the choice of technology. It has been established that capital-intensive techniques have been adopted in economies or sectors more in response to a shortage of skilled rather than unskilled labor. Thus a potential shortage of skilled labor of the type needed by modern manufacturing could dampen the value of employment elasticity and slow the rate of growth of employment in the industrial sector. An important result in our research work on Indian manufacturing in the preceding chapter was the evidence that although employment elasticity varied with the economic cycles it did not exceed 0.33 in the best period of the post-reform upswing. As analyzed in the research there are several important reasons for the low employment elasticity, but a perceived shortage of labor of the requisite skill and efficiency is one of them.

Dampening the growth of markets

While the last two points emphasize problems created on the supply side, 'dualism' might also affect the growth of manufacturing through its impact on the demand side – on the expansion of markets for industrial goods. The medium-sized establishments have been lauded in the literature for having the desired amount of flexibility and enterprise to seek out new export markets in new industries. But their importance in the expansion of domestic markets also needs to be emphasized. Dualism strengthens and perpetuates product market segmentation. The market for industrial products is split into low-quality products catering to the need of low-income consumers, and supplied by small-scale local producers on the one hand, and the higher-quality segments which the large establishments supply to a limited number of high-income consumers. The lack of integration of markets could be a bottleneck in the development of mass markets for manufactured consumer goods.

Causes of the emergence and persistence of dualism

What are the major factors causing the emergence of dualism in its two aspects, the phenomenon of the 'missing middle' and the unusual productivity gap between the small and the large units? What are the reasons for its persistence over time, even when the reform process reducing some of the strength of the proximate causes of dualism has been eroded?

The more important reasons behind the origin and persistence of 'dualism' will be now be discussed under the following heads:

1. segmentation in the factor markets – of (i) labor and (ii) capital;
   
   
2. education policies affecting the relative price of skilled labor;
   
   
3. the policy of protection of the small scale;
   
   
4. hysteresis causing persistence of critical causal factors.

Segmentation of factor markets

LABOR MARKETS

An important factor which cements labor-market segmentation and discourages the upward mobility of small establishments is labor legislation. The aspect of this type of legislation which is most relevant to the problem of the missing middle in India pertains to job security laws rather than those impacting on wage levels. Wage differential between small and large firms are known to have preceded the era of wage legislation and could be traced to some extent to differences in the quality of labor. Wage cost per efficiency unit of labor was probably never as large as the difference in wage per worker in small and large firms (Little et al. 1987, Chapter 13 and the literature cited). Job-security legislation, such as those embodied in the Industrial Disputes Act of 1947, however, is critical since it places enormous costs on factories in so far as they have to obtain permission from the state designated authorities before permanent workers can be laid off. It is not so much the actual compensation that has to be paid to workers who are laid off which is important as the fact that the process of obtaining permission takes a long time and has considerable uncertainty. Even if the number of workers sought to be laid of in a particular period of time is small, the law creates a fixed overhead cost (of, for example, maintaining legal officers who can pursue the long process through the labor courts), which is only mildly variant to the size of the workforce in the factory. Thus over a considerable range of size, the overhead costs would be significantly higher for smaller firms than the larger ones. Labor legislation of this type bites for factories employing more than 100 workers. Other types of labor laws involving inspection and supervision over conditions of work are applicable to units covered by the Annual Survey of Industries (employing ten or more workers). Both sets of legislation add to the fixed cost of labor for units employing more than the respective threshold.

Employers have various means of working round these legislative requirements. For instance, a common practice is to employ part of the workforce as 'casual' labor: although they work regularly in the unit they are kept in the books as non-permanent labor as long as inspectors can be kept off the practice by bribes or other means. Similarly, in some states of India employers wanting to close down factories have deliberately refrained from paying electricity bills which have led to the effective closure of the units. But all these ways of finding ways of adjusting the labor force to avoid confrontation with the law imply the incurring of transaction costs. The possibility of bearing such costs is a clear discouragement to small employers to expand vertically in size. There is an incentive to expand horizontally with a multiplication of small units rather expanding into larger size groups.

CAPITAL MARKETS

It is well known that modern financial institutions find it much easier to lend to larger establishments than to smaller ones The high transaction costs of dealing with small units, as well as the relative scarcity of collaterals, operate against the latter, The result is the interest cost of finance – even when it is available – typically tends to be significantly higher for small units. Financial liberalization and globalization have, if anything, increased the effective bias on the part of largely urban-based financial institution to lend to large-scale establishments.

State policies in India have sought to counter this bias by creating institutions like the Small Industries Development Organization (SIDO) which have reached out specifically to the units under their purview (whether defined in terms of employment or capital size). Units which expand beyond this size tend to lose both ways – deprived of the special opportunities provided under SIDO and not large enough to avail of the low-cost finance potentially available in the modern large-scale sector. This type of segmentation in the capital market strongly discourages the growth of small firms into middle-sized ones.

Education polices

Policies that have been implemented in India over the years have been biased towards the promotion of tertiary education and have neglected basic primary and low secondary education.⁷ It has been maintained in the literature (e.g., by Adrian Wood among others) that modern manufacturing requires a minimum of basic education for a workforce to be able to perform up to minimum standards in modern manufacturing. Small- and medium-sized units – adopting comparatively labor-intensive technology – benefits from an ample supply of such labor. They are contrasted with tiny units which could use nearly unskilled labor with less than primary education for low-grade production, but would find it difficult to grow beyond a certain scale with such labor. The relatively plentiful supply of skilled labor with higher education biases production to less labor-intensive industry and modes of production. Large units have a comparative advantage in using such labor which smaller units cannot afford.

Recently researchers in the IMF (2006) have explored the question if the organized manufacturing sector in India has specialized more in industries using less unskilled labor compared with other countries. The authors use the cross-country data sets for the formal manufacturing sector prepared by the United Nations Industrial Development Organization (UNIDO) to identify industries (at the 3-digit level) which are 'labor-intensive'. For labor intensity the proxy is

the share of wages in value added for the industry in a country averaged across a broad group of developing countries – examples of industries that score highest on labor intensity are clothing, printing and publishing and non – electrical machinery while those which are lowest are beverages, tobacco and petroleum refineries.

Labor-intensive industries are identified as those below the median on the range of scores thus calculated. The authors then examine the pattern of manufacturing in each of the countries in their sample to determine the share of output/employment in labor-intensive industries. They regressed the shares of each country on per capita GDP, its square and a dummy for India. It was found that even for as early a date as 1981 the Indian dummy was negative and significant in the value-added regression, signifing that it was specializing exceptionally in less labor-intensive industries. For employment the Indian scenario was in the same direction but less significant.

In the next exercise the authors analyzed skill-intensity, using the input – output matrix for South Africa–which enables them to classify industries by the share of remuneration of highly skilled and skilled workers. As with labor intensity the industries in each country were divided into above and below median skill-intensive ones and then the ratios of value added of the two subgroups (in the cross-country sample) in 1981 were regressed against GDP per capita, its square, country size and the dummy for India. The dummy was strongly significant and positive, showing that in 1981 India was abnormally specializing in skill-intensive industries.

In order to see what has been happening since the early eighties, the authors plot the evolution in the share of output generated in labor-intensive industries for India and a select group of comparer countries – which include China, Indonesia, Korea and Malaysia. India contrasts dramatically with Indonesia which shows a rising share of labor-intensive industries. Korea and Malaysia–at much higher income levels – also show mildly increasing shares. China, on the other hand, has a declining share of such industries but from a much higher level of this share (see ibid., Figure 2, p. 48). Turning to the topic of skill intensity, the graphs for the evolution in comparator countries also show that

India's share which was already high in 1980 despite its lower level of per capita income, has been increasing and is at levels reached by Malaysia or Korea at much higher levels of per capita income. There is also a striking contrast with China. China's share of output in skill-intensive industries is lower than India's and has been virtually flat.

(Ibid., p. 22 and Figure 4, p. 52)

The tilt towards skill-intensive industries in India is also reflected in exports: the share of India's exports in skill-intensive goods has increased from 25 percent in 1970 to 65 percent in 2004.

The protection of small-scale units

The policy of protecting small-scale enterprises (SSEs) has been an important aspect of Indian industrial policy since independence. It has taken the form of reservation of large number of items for production in exclusively small units and the provision of incentives – fiscal, financial and legislative – as long as the units stayed below a certain size. The threshold size was first defined in terms of the traditional employment size of five workers. It was in later years changed to a definition based on capital size and it was also increased somewhat over the years. Nevertheless, the policies have always provided an incentive for entrepreneurs to expand horizontally with more small units, rather vertically with larger middle-sized units.⁸

Hysteresis

The policy of reservation for the small scale was largely ended in the post-1991 reform process. But we have seen that the impact on the size structure of establishments in manufacturing has been minimal. This limited impact might be due to widely recognized processes in which a socio-economic system established over a long period of time tends to persist even after the original causes have disappeared. This persistence is not just due to inertia. Economic agents and institutions acquire characteristics which sustain the system. For example, entrepreneurs develop with ambitions to think in terms of horizontal rather than vertical growth. Marketing channels, financial institutions and infrastructure are geared more to supporting small units with limited market rather than dynamic units growing into larger sizes and different markets.

It is not easy to determine how much of the inertia of the industrial structure can be ascribed to 'hysteresis'. It should be remembered that there are institutional factors, particularly labor legislation and the segmentation in the capital market, which continue to reinforce the dualism in this sector, with the attendant problems discussed in this chapter.

Conclusions

This chapter has dwelt on the peculiarities of the size structure in Indian manufacturing, relative to the experience of other comparator Asian countries. The employment size distribution is pronouncedly bi-modal. This is not just the usual phenomenon, often witnessed in developing countries when we put the household enterprises in the informal sector together with the modern enterprises in the formal sector. Rather the Indian scenario is peculiar when we take the 'organized'–or what is internationally accepted as the 'formal'–sector employing five or more workers. The size distribution in this subset of organized sector firms shows two strong modes in India, in the 5–9 and 500+ size-groups. There is a very large productivity differential between these groups, and a conspicuous feature of the size distribution is the low proportion of workers in its middle part. We have discussed the analytical and empirical reasons as to why this phenomenon of the 'missing middle' could be considered to be a significant drag on the healthy development of a dynamic manufacturing sector. While the problem of the 'missing middle' might have had its origins in past policies it has shown a remarkable persistence in the post-reform era. We refer to some possible reasons for this persistence in the previous section of the chapter.

Appendix

Document(s) 9 de 16