Does Corruption Constrain The Growth of Indian Manufacturing Firms?

This article attempts to address this gap. It focuses on institutional factors, and it aims to identify the key institutional determinants of the evolution and persistence of the missing middle in the Indian manufacturing sector. It follows a novel approach. It uses data self-reported by Indian firms on a variety of impediments they face. This data is drawn from the Enterprise Surveys2 conducted by the World Bank for formal and informal firms. The data is combined into a pooled data set that covers the continuum of firms in Indian manufacturing—from the smallest to the largest.

The data are a combination of objective and subjective measures of the constraints firms face. For example, firms are asked on how many days they take to obtain a construction permit, which is an objective measure. They are also asked how important they see a factor (say, labour regulations) in constraining their activities, which is a subjective measure. The combination of objective and subjective measures in the data guards against the weakness of one set of measures against the other.
The Enterprise Surveys are rich and allow to test for those set of institutional constraints that are more binding for the mid-size Indian manufacturing firms than for the small and large ones. This is done by the means of a descriptive analysis of institutional measures and by econometric analysis that explicitly tests which constraints are more important for mid-size vis-à-vis other firms, after controlling for other possible influences on firm size.

The ‘Missing Middle’ in Indian Manufacturing

This section sets out the missing middle problem in Indian manufacturing. Firm-level data for the entire manufacturing sector, including both formal and informal firms, are needed to describe the missing middle problem.

The data set is constructed by combining the microdata on formal firms obtained from the Annual Survey of Industries (ASI) with the data on informal firms drawn from the National Sample Survey Office (NSSO) surveys of the unorganised manufacturing sector. This data set covers three periods: 2000–01, 2005–06, and 2010–11 and represents a continuum of firms from the smallest (except household enterprises) to the largest.
To put the size distribution of Indian manufacturing firms into perspective, informal sector firms that employ 6–9 workers (employing mostly hired labor) are included with formal sector firms (Figure 1). This analysis shows a dualistic structure and that the size distribution of firms is bipolar—in line with the available evidence.
Two prominent modes are found on the left of the employment distribution (represented by the 6–9 and 10–49 categories) and on the right (represented by the 500+ category), and a striking trench is found in the share of employment in the intermediate size categories (50–499 workers).
Almost 75% of the Indian manufacturing workforce is concentrated in the smallest categories (6–9 and 10–49 workers) and the largest categories (500+), and the remaining 25% is distributed among the intermediate categories. Thus, the Indian manufacturing sector has many small firms and some large firms but few medium-sized firms.
There has been little change in the distribution of firm size between 2000–01 and 2010–11. The economic distance between small and large firms is substantial: firms in the 500+ category were about 13 times more productive (in 2010–11) than firms in the 6–9 size category in 2010–11 the productivity gap has also widened, from 1:11 in 2000–01 to 1:13 in 2010-11. Creating the institutional environment to help small firms grow to become mid-size firms can improve the manufacturing sector’s growth and productivity, and it calls for identifying the constraints to the growth of firms in terms of size and scale.

Data

To identify the key determinants, a source of data across the entire continuum of firm sizes for the Indian manufacturing sector is required. Few data sets impart information for both small (informal) and large (formal) firms in India and include indicators that allow for the objective measurement and comparison of the business environment, its binding constraints, and the quality and integrity of supportive and regulatory public services. Many public databases supply information separately for formal and informal firms.

However, combining the data sets for formal and informal sectors into one will have less utility as these data sources lack information on the availability of the types of institutions and their quality as perceived by firms.
A data set is needed that covers firms along the continuum of the entire manufacturing sector and that imparts important insights into what is needed to improve the business environment—based on what firms themselves say about the conditions they need to grow and the constraints they face. The World Bank Enterprise Surveys on Indian firms constitute such a data set.
For informal firms, data are drawn from a survey conducted among the informal enterprises in India by the World Bank in 2006. The basic survey units were establishments with 10 or fewer full-time paid employees. The data obtained from the NSSO’s 56th survey of unorganised manufacturing enterprises formed the base frame for sampling.
The survey employed a combination of the stratified, cluster, and snowballing techniques to decide on the enterprises to include in the sample. Cities were identified using stratified techniques and split into industrial clusters, and snowballing techniques were applied to identify enterprises in various industry sectors.
The informal manufacturing Investment Climate Survey (ICS) covered 1,549 enterprises in six major industrial clusters in four regions of India: Delhi and Ludhiana (north), Mumbai and Thane (west), Howrah (east), and Hyderabad (south).
Based on the proportion of shares in the aggregate sectoral output, the sample was drawn from nine manufacturers: auto components, chemicals, electrical goods, electronics, food processing, garments, leather, metal, and machine tools and textiles.
Data on formal firms are obtained from a similar enterprise survey of formal firms—“Firm Analysis and Competitiveness Survey of India” (FACS)—conducted by the World Bank in conjunction with the Confederation of Indian Industry (CII) in 2005. Manufacturing sector firms constituted the universe for this survey, and the sampling frame for manufacturing establishments was obtained from the ASI.
The survey was conducted from March 2005 to July 2005 on a random selection of 2,286 manufacturing firms sampled from 49 cities in 16 states Andhra Pradesh, Bihar, Delhi, Gujarat, Haryana, Jharkhand, Karnataka, Kerala, Maharashtra, Madhya Pradesh, Orissa (renamed “Odisha”), Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh, and West Bengal.
The survey covered industries such as auto components, drugs and pharmaceuticals, electrical goods, electronics, food products, garments, leather and plastic products, metal products, paper and paper products, rubber products, and textiles.
These surveys were conducted separately in adjacent years (2005 and 2006), but they collected comparable information that include subjective evaluations of obstacles and objective hard data numbers with direct links to growth and productivity. The ICS and FACS data were merged so that the data on the key institutional variables were comparable.
Though both surveys used the same survey instrument for data collection, the utmost care has been taken to match the information in the data sets. Following matching, the final data set held 3,835 firms (1,549 informal firms and 2,286 formal firms).
While the formal sector data set is nationally representative, the data set for the informal sector is non-random and under-sampled. The informal sector survey covers only a few cities and sectors. The results from the merged data set should be viewed with caution, due to the less representative nature of the data for the informal sector, and circumspection exercised in generalising these results to the population of firms.

Construction of Variables

This article aims to examine the role of institutional bottlenecks in firm transition. Therefore, variables pertaining to three broad categories—firm size, institutional environment, and firm characteristics—are constructed.

Firm size: To see whether these institutional constraints are more binding on certain types of firms and less binding on others, this study classifies firms into seven categories based on the number of workers they employ (firm size) as in Mazumdar and Sarkar (2009, 2013) and Ramaswamy (2013): fewer than six workers (1–5); more than five and fewer than 10 (6–9); 10–49 workers (10–49); more than 50 but fewer than 100 (50–99); more than 99 but fewer than 200 (100–199); 200–499 workers (200–499); and firms employing 500 or more workers.
Most of the sample (38%) was made up by firms in the 10–49 category, and the proportion of firms in the 1–5 and 6–9 categories is almost the same (21%). Of all the firms, 7% are in the 50–99 category, 6% in the 500+ category, and 3% each in the 100–199 and 200–499 categories.
Institutional environment: This article aims to examine the relationship between the types of institutions and the transition of firms in the Indian manufacturing sector. Variables that represent the types of institutions in the country need to be constructed.
The matched data set is used to identify questions that can be used to measure these distinct types of institutions. Based on the information available in the data set, five broad dimensions of institutional quality are identified: property rights and taxation; social and physical infrastructure; law and order; corruption; and the regulatory environment. Variables are assigned to capture the nature and quality of institutions for each of these dimensions.
Firm characteristics: The usual set of variables used in firm-level studies is employed in this study as control variables. These variables help to control for the influence of firm-specific characteristics and, thereby, capture the exact role of institutional variables in firm transition.
The firm-specific characteristics employed in this article control broadly for the firm owner’s age, ownership, experience, and education, and for the workers’ education and training. The variable “age” represents the age of the firm, and it is defined as the number of years (plus one) elapsed since the year the establishment began operations.
To avoid ages of zero, one year is added. The “partnership” variable is derived from the question on a firm’s current legal status, and it is constructed as a binary variable.
The categories of a firm’s legal status are sole proprietorship, limited partnership, partnership, shareholding company, and others. Firms that are not proprietorships take the value 1, and proprietorship firms take the value.
The influence of owner- or worker-level characteristics on firm transition is also controlled for. The owner-specific control variables are the manager or owner’s average number of years of experience (expown) and education (ednown). “Ednown” is an ordered variable.
It is coded 1 for “Illiterate” managers or owners or for those with fewer than nine years of schooling or those who did not complete secondary school; 2 for those who completed secondary school or higher secondary school; 3 for those with vocational training or some university training or those with some college education but not a graduate degree; and 4 for firm owners with a graduate, postgraduate, or doctoral degree.
The variables for worker-specific characteristics are “training” and “avgedn.” “Training” represents a dummy variable for training; it takes the value 1 for firms with trained workers (or workers given internal or external training or both) and 0 for others. The variable “avgedn” is an ordered variable; it is ordered 1 for 0–3 years of education, 2 for 4–6 years, 3 for 7–12 years, and 4 for 13 or more years of education
In this data set, the median age of firms is 12 years, and the mean is 15 years. Over 60% of the firms are owned and run by sole proprietors; the rest are partnerships. The firm owners averaged 12 years of experience, and over 50% are graduates. The level of education at the firm level averaged 7–12 years. Only 12% of firms are reported to employ workers with some training, formal or informal.

Methodology

Along with graphical illustrations, a simple multivariate regression analysis is employed to see whether firms’ perceptions on different growth constraints are evenly or unevenly distributed across firm size categories. Our basic model takes the form
ICj = α0 + ∑k>0 βk sizej + ∑i>0 γi FIRMj + μi + δc + εj … (1)
where IC stands for the specific institutional constraint, the vector size contains variables of interest that include dummy variables for the six size categories (6–9, 10–49, 50–99, 100–199, 200–499, and 500+), with the 1–5 category being the base category.
If certain institutional constraints are more relevant for mid-size firms than for small and large firms, the coefficient of firm size is expected to be significant for mid-size firms and larger in magnitude. The vector FIRM contains variables standing for firm-specific characteristics.
Six firm-specific characteristics are controlled for age, partnership, expown, ednown, training, and even. μi stand for industry dummies and δc for city dummies. Equation (1) is estimated by the ordinary least squares method when the institutional constraint is a continuous variable and by ordered probit when it is an ordinal measure.

Perceived Obstacles to Firm Operation and Growth

This section takes a first stab at the combined firm-level data compiled by comparing the average obstacle levels for firms in distinct size categories. Histograms are constructed for each size category to show the percentage of firms that reported an obstacle as constraining, and these histograms are categorised under five broad dimensions.

The histograms constructed to understand the relationship between firm size and property rights and taxation show a hump-shape—these are more serious impediments to growth for mid-size firms than for small and large firms .
More mid-size firms than small and large firms report problems with labour laws, taxation, and obtaining permits and licences. Mid-size and large firms report that the government is inefficient in delivering infrastructure-related services.
Access to power, transport, and telecommunications are their top three constraints. Small firms worry over land access more than medium and large firms. The results of this analysis show that all firms are equally constrained in obtaining finance, contrary to the idea that small and medium firms are financially more constrained than large firms. The finding on access to skilled labour suggests that it is harder for medium and large firms than small firms to find educated and skilled workers and it is a more serious constraint to growth.
Regarding constraints relating to law and order, priorities change according to size. Large firms are concerned over making payments towards security services and the judiciary, but mid-sized firms find settling overdue payments a critical concern.
On the other hand, the percentage of firms that report crime as a growth obstacle is evenly distributed across firm size categories. A hump-shaped relationship between firm size and obstacles on all indicators of corruption shows that corruption affects mid-size firms more than small and large firms.
On the regulatory environment, large firms report that visits by government officials and inspectors seriously constrain growth, while small firms are concerned mainly over visits by the police and the time taken to obtain an operating licence. Mid-size firms, on the other hand, are more concerned about the time taken to get access to public services such as power, telecommunications, and water supply.
This descriptive analysis suggests that, overall, mid-size firms face stronger institutional constraints than small and large firms, evident particularly with respect to corruption, poor law and order, and the absence of property rights, where mid-size firms seem to face the highest constraints.

Empirical Analysis

There are differences between the constraints faced by mid-size firms and those faced by small and large firms. To test whether these differences are statistically significant in this regression analysis, the estimates of Equation 1 are presented. Firm size is captured by a series of dummies: 6–9, 10–49, 50–99, 100–199, 200–499, and 500 and above.

The firms in the 1–5 size category serve as the comparator. The main aim is to see whether firm size changes a firm’s perceptions of institutional constraints. All these estimations control for other firm characteristics: the age of the firm (age); ownership (partnership); the owner’s education (ednown) and experience (expown); training given to workers (training); and the worker’s education (avgedn).
Finally, industry dummy variables are included to control for sector-specific effects and dummies for cities are introduced to control for region-specific effects. These controls allow for the incorporation of other influences on institutional constraints, such as the industry or city the firm is located in.
Further, since the institutional measures are self-reported data, firms with better-educated or more experienced managers or owners may report constraints more accurately than those less educated or experienced.
The purpose of the regression analysis is to show whether the inverted U-shaped relationship between institutional variables and the size distribution of firms remains true in a multivariate framework; the intention is not to make causal inferences on whether firm size itself is a determinant of the institutional constraint that a firm may face.
Reports the regression results for two broad dimensions: property rights and taxation, and social and physical infrastructure. Under the dimension of property rights and taxation, three institutional constraints as perceived by firms—“labregu,” “licpmt,” and “taxobst”—are considered. Labour regulations, licences, and permits constrain firms in the 10–49 and 50–99 categories more than small and large firms. This is clear from the positive and highly significant coefficients of these two size categories in regression estimations.
Labour regulations, permits, and licensing laws apply only to firms that employ 100 or more workers, and, relative to other firms, firms in this data set that employ fewer than 100 workers perceive these requirements as major growth constraints.
There are few mid-size firms in the Indian manufacturing sector and firms are not upwardly mobile. These phenomena may be explained by the necessity for obtaining permits and following licensing laws and labour regulations such as minimum wage legislation, mandatory non-wage benefits, and job security guarantees, argue a few researchers (Mazumdar and Sarkar 2013; Ramaswamy 2013). Possibly, thus, the results in this study support their stance.
Relaxing labour regulations, however, may reduce job security for workers under their purview.3
A positive and significant coefficient is observed across all firm size categories for “taxobst,” implying that high taxes and tax administration affect entrepreneurs of all sizes (relative to the residual category) severely; that is, it does not discriminate between firm size.
This is expected, as high taxes reduce an entrepreneur’s post-tax income and can lessen
entrepreneurial activity and growth for all firms, irrespective of size. Since taxes are a significant cost of doing business, it is not surprising that most firms regard them as being too high (Batra et al 2003).
The institutional constraints that represent the social and physical infrastructure dimension are important for all firm size categories. Firms in all categories stress on the importance of skilled and educated workers, sufficient power supply, and adequate telecommunications for smooth operation and growth, and they report these obstacles as major constraints to their growth as compared to firms in the 1–5 category.
The coefficients of all firm size categories are positive and significant for “ednwrk,” “power,” and “telcom”—barring a few exceptions, such as power for larger firms—and that suggests that these constraints affect firm growth and operation significantly irrespective of size.
Firms need quality power and a qualified workforce to grow (Raj and Sen 2016). There is no significant difference between firms of any size in their perception of “access to finance” as an obstacle to growth.
Under the dimension of law and order, four variables—crime, security, overdue, and court—are identified: The results show that the coefficient of security is positive and significant for all firm size categories and the coefficients of crime and overdue payments are statistically insignificant.
That implies that firms in all categories consider payments made towards security, but not crime or overdue payments, a serious constraint to growth. The legal system and conflict resolution represent a major constraint for the operation and growth of firms that employ 50 or more workers; the coefficient is positive and significant for firms in the 50–99, 100–199, 200–499, and 500+ categories and insignificant for firms in the 5–9 and 10–49 categories. This finding suggests that smaller firms report lower legal obstacles than larger firms.
Enforceable property rights and contracts are universally believed to be important for growth, and a well-functioning, the efficient judicial system is expected to have an important bearing on firm growth (Weder 2003). This study finds, however, that smaller firms report lower legal obstacles than larger firms.
This finding is in line with the argument that large firms rely much more than small firms on long-term financing and larger loans on and are, therefore, more likely than small firms to tax the resources of an underdeveloped financial or legal system (Beck et al 2005).
Inefficient financial and legal systems could increase the effects of institutional obstacles on the largest firms. It is also possible that small firms are unlikely to have much experience with the working of the judiciary and are hence unlikely to report larger problems (Schiffer and Weder 2001).
Does corruption constrain the growth of Indian manufacturing firms? Many studies show that corruption reduces private investment and growth, but few disentangle the varying effects corruption has on firms of different sizes.
This study investigates whether the perception of corruption differs between firms of different sizes by using a number of measures that proxy for firms’ perception of corruption. Two indicators of corruption are used: perception of corruption, which captures a firm’s perception of corruption as a constraint for the operation and growth of its business; and, more directly, graft incidence, which measures whether the firm was requested to pay a bribe for obtaining permits and licenses (such as a telephone, electricity, or water connection; construction-related permit; import or operating license; or in a meeting with government officials).
Regression estimations are performed on firm size categories for eight such variables while controlling for the influence of firm-specific characteristics. The results indicate that firms perceive corruption as a deterrent for their business operation and growth, as is evident from the positive and significant coefficient of all size categories but 500+ for corruption.
This implies that the effect of corruption is evident for small and medium firms but not for large firms, and it possibly lends support to the De Soto view: small firms report larger problems with bureaucratic corruption than do large firms.
By contrast, the variables representing the incidence of graft are not found to be significant growth deterrents for any firm size category and, barring very few exceptions, the coefficients of size categories are insignificant for all proxies of graft incidence. This suggests that though firms of all size categories are exposed to corruption, it affects small and mid-size firms the most.
The existing studies consider that the regulatory environment is important to firm growth and suggest that existing regulations constrain the operation and growth of manufacturing firms. To capture the regulatory environment faced by firms, this study uses proxies that intend to capture the treatment—favorable or unfavorable—meted out by regulatory agencies in terms of the time firms wait to obtain permits and services and the number of inspections by officials.
The results show that a few obstacles are critical for firms in larger firm size categories, some affect firms across the board, and for other obstacles, the effect does not differ by firm size.
Compared to small and mid-size firms, large firms report being more constrained by inspections by government officials, as is evident from the difference in the magnitude of the coefficients of “visit govt” and “visit un” between larger firms and smaller firms.
This is also true of inspections overall (“n tot visit”); large firms were reported to be experiencing more visits than small firms. On the time spent dealing with regulations (“pertimregu”), small and large firms are equally constrained, as all firms, irrespective of size, treat this as a serious issue.
The time firms wait to obtain various public services does not differ by firm size; however, the time taken to obtain a telephone connection is found to be significant for all firms, irrespective of size.
These findings imply that large firms are more exposed to inspections by officials as these firms are more likely than small firms to enjoy higher profits, more visible, and more susceptible to blackmail and more prone to offer kickbacks. These findings also imply that certain aspects of the regulatory environment affect all firms alike.5

Conclusions

This article attempted to understand the conundrum of the missing middle in the Indian manufacturing sector. Self-reported firm-level data on objective and subjective measures of institutional quality, and on other potential barriers to the entry of mid-size firms, are used.
These data are taken from two surveys of firms, one for the formal sector and the other for the informal sector, and these are combined into a pooled data set of firms across all size categories. Some of the observations from the analysis of the data provide empirical confirmation for some commonly held truths, while some provide little evidence for others.
In facing impediments to doing business, mid-size firms recognize themselves as being more constrained than small and large firms. The perception of some general constraints is not statistically different for mid-size firms from small or large firms, and labor regulations, permits, and licensing laws are significantly more likely to constrain mid-size firms severely.
Many observers of India’s growth story emphasise that laws and regulations explain the missing middle, which affects overall growth and productivity. Corruption, weak infrastructure, and poor law and order constrain firms of all sizes equally. Large firms show a statistically higher degree of a constraint than small firms on most constraints related to the regulatory environment.
Overall, the results show that the missing middle problem is an institutional problem and that it has little to do with infrastructure and finance constraints. The most important set of institutions is the ones that could be termed “predatory” institutions—the corruption that mid-sized firms face in their day-to-day interactions with the state.
The current policy approach to the manufacturing sector is to improve the ease of doing business by reforming regulatory institutions, but our findings suggest that such reforms in themselves are not enough to solve the missing middle problem.
More attention needs to be given to discipline lower-level bureaucrats who engage in petty corruption and to make government procedures more transparent and accountable so that there is less scope for corruption.

Notes

There is a rich literature on how institutions affect firm performance in the global context (Yasar et al 2011; Commander and Svejnar 2011).
The Investment Climate Surveys have been replaced by the Enterprise Surveys, which are conducted at present by the Enterprise Analysis Unit of the World Bank.
Contract workers are outside the purview of the strict labour regulations that apply to workers in permanent contracts, and formal sector firms increasingly use contract labour (NCEUS 2009). The possible under-reporting of workers by firms in surveys, partly to avoid taxes and regulations, may explain the missing middle.
The exceptions are the significant coefficients of “contract,” “conpmtgift,” and “oplicgift” for firms in the 10–49 category; “paygovtoff” in the 50–99 category; and “contract” and “tel gift” in the 100–199 category.
An anonymous referee suggested that all the empirical analyses be replicated by using only the sample of formal firms and 500+ firms as the benchmark category. The findings in this study—large firms are more visible and, therefore, more exposed to inspections, and certain aspects of the regulatory environment affect all firms uniformly—are upheld. These results are available from the authors upon request.