Small Bus Econ (2012) 39:383–400

DOI 10.1007/s11187-011-9318-7

Firm growth and barriers to growth among small firms

in India
Alex Coad • Jaganaddha Pawan Tamvada

Accepted: 14 January 2011 / Published online: 15 March 2011

 Springer Science+Business Media, LLC. 2011

Abstract Empirical work on micro and small firms Keywords Entrepreneurship  Developing
focuses on developed countries, while existing work countries  Micro and small businesses 
on developing countries is all too often based on Firm growth  Firm age  Barriers to growth 
small samples taken from ad hoc questionnaires. The Declining firms  Female entrepreneurs 
census data we analyze here are fairly representative Robust regression
of small business structure in India. Consistent with
findings from prior research on developed countries, JEL Classifications L26  L25
size and age have a negative impact on firm growth in
the majority of specifications. Enterprises managed
by women have lower expected growth rates.
Proprietary firms face lower growth on the whole, 1 Introduction
especially if they are young firms. Exporting has a
positive effect on firm growth, especially for young Small firms are major drivers of industrial develop-
firms and for female-owned firms. Although some ment and economical advancement. In developed
small firms are able to convert know-how into countries, entrepreneurial small firms play critical
commercial success, we find that many others are roles in introducing new products and new techniques
unable to translate it into superior growth. into the market through technological innovations
(Pavitt et al. 1987; Acs and Audretsch 1990). The
industrial organization literature has also suggested
that the entry of new entrepreneurial ventures is a
A. Coad (&) source of competition that keeps markets functioning
Science and Technology Policy Research (SPRU),
University of Sussex Freeman Centre, well.1 Another advantage of small firms is that they
Falmer, Brighton BN1 9QE, UK are associated with more flexible production routines
e-mail: [email protected] (You 1995).
However, in developing countries the roles of
J. P. Tamvada
Aston Business School, Aston University, Aston Triangle, small firms differ from those in developed ones. One
Birmingham B4 7ET, UK major reason why micro and small firms exist in
e-mail: [email protected]
1
A. Coad  J. P. Tamvada In reality, however, the competitive threat of new entrants
Max Planck Institute of Economics, appears to be quite limited because they enter at a small scale
Kahlaische Strasse 10, 07745 Jena, Germany and many die shortly afterwards (Geroski 1995).

123
384 A. Coad, J. P. Tamvada

developing countries is that they offer individuals a It is estimated that, in 2001, SSI businesses in India5
livelihood and a source of independent revenue.2 In accounted for around 95% of the industrial units in the
many cases, new small businesses are founded as a country, contributed 40% of the manufacturing sector
last resort rather than as a first choice (Beck et al. output and close to one-third of India’s exports (SIDBI
2005). The work is typically labor-intensive, but 2001). By 2009, the share of SSI’s contribution to
levels of labor productivity (and also capital produc- manufacturing output increased to 45%, and the share
tivity) are typically low (Little 1987). Formal finan- of exports rose to 40% (MSME 2009).
cial markets are of limited use for these small Government support for micro and small firms,
businesses, and so they remain small and do not however, is not uncontroversial.6 While some view
become capital intensive. For such firms, ‘‘the entrepreneurship as the ‘prime mover’ which can
entrepreneur must be his own banker’’ (Leff 1979, initiate the development process (Leff 1979), or as a
p. 48) and must rely on informal capital markets means to attain a more efficient allocation of resources
(such as personal savings or small loans from friends (You 1995), others focus on the low productivity of
and family) for financing (Kozan et al. 2006; Allen these enterprises and are pessimistic about the role
et al. 2006). Micro and small firms in developing they can play in economic development.7
countries generally do not correspond to the most Given the compelling interest in the growth of
efficient scale of production (Little 1987), yet they small enterprises in developing countries, the aim of
allow people to be independent and make a living.3 this paper is to provide novel empirical evidence on
Given both the special role of small businesses in this subject. We contribute to the existing literature
providing economic opportunities for a large number by providing new insights into the nature and
of people and the obstacles to growth that they dynamics of small businesses in India. We exploit
encounter, developing countries have often taken a data from a final sample of more than 600,000 firms
favorable position towards micro and small firms. compiled in a nationwide census survey conducted by
India has, historically, been especially supportive of the Indian government. Within this database we are
its small enterprises (Little 1987). For example, the able to explore the growth performance of types of
Indian government has granted exclusive production small firms that are relatively unheard of in the
rights of certain goods and services to Small Scale
Industry (SSI) businesses and even though the Indian 5
The term Small Scale Industry encompasses small firms with
economy has undergone considerable deregulation in an initial investment of less than 10 million Indian rupees, an
recent years, the production of many goods continues amount approximately equivalent to US$ 200,000 (http://
to be reserved for small businesses.4 www.dcmsme.gov.in/publications/circulars/circularmay1994.
html#icoty, accessed on 25 October 2010).
6
A recent contribution to this debate by Beck et al. (2005)
examines the role of the small and medium-sized enterprise
2
(SME) sector in 45 countries (not including India). They begin
It has even been argued that, in the case of India, better by reviewing the prior literature and conclude that ‘‘firm-level
educated individuals are quick to leave self-employment and studies do not provide an empirical foundation for subsidizing
pursue alternative career paths, whereas less educated individ- SMEs.’’ (Beck et al. 2005, p. 201). In their empirical analysis,
uals have fewer opportunities to leave self-employment they observe a robust positive relationship between the relative
(Nafziger and Terrell 1996). size of the SME sector and economic growth, although the
3
Even in developed countries, however, many entrepreneurs SME–growth relationship is not robust to the use of instru-
are not ‘true’ entrepreneurs, in the sense that they do not bring mental variables to control for endogeneity. Put differently,
innovations or bring about reform in stagnant markets (San- they write that ‘‘although a prosperous SME sector is a
tarelli and Vivarelli 2007). Many enter for less ‘noble’ reasons, characteristic of flourishing economies, we cannot reject the
such as overoptimism on the part of the founder, the pursuit of view that SMEs do not cause growth.’’ (p. 224). They also fail
a relaxed lifestyle, or the flight from unemployment. In fact, to observe any significant relationship between the size of the
many entrants are far less productive than incumbents (even SME sector and poverty alleviation.
taking into account their liability of small scale). The concept 7
As such, there is some skepticism about the economic
of new firm entry covers a particularly heterogeneous group of foundations of government support of small enterprises—
enterprises. ‘‘India is …exceptional in the extent and range of its policies
4
Most of these items are foodstuffs and consumer products. that directly support SSEs. …They have been romantic, rather
For a complete list of protected items, see http://www.small than economic.’’ (Little 1987, p. 232). SSE refers to small-
industryindia.com/publications/reserveditems/resvex.htm. scale enterprises.

123
Firm growth and barriers to growth among small firms in India 385

existing literature, including traditional sector firms small business—is described in Sect. 3, and our
(such as firms whose principal source of energy is methodology is discussed in Sect. 4. The empirical
firewood) and older small firms (who reach ages of results are presented in Sect. 5, where we investigate
100 years or more). It seems to us that little work has the determinants of growth rates, and in Sect. 6,
been done on mature small firms to date (even in the where we focus on a subsample of struggling firms
literature concerning developed countries), for sev- and the difficulties caused by specific growth barriers.
eral reasons. First, it is not always easy to obtain data Section 7 presents a summary of the main results and
on small firms. Second, it is not always easy to obtain the conclusion.
data on firm age. Third, researchers have often tended
to assume that all firms want to grow larger with age
and have tended to overlook this category of firms. 2 Theoretical background
The main themes discussed as we present the results
are the growth of firms of different size and age, the The emerging body of literature on opportunity and
gender of the owner, and export and access to necessity entrepreneurship in developing countries
external knowledge sources. We also consider ques- suggests that although the magnitude of entrepre-
tionnaire responses to specific growth barriers faced neurial activity is higher in developing countries, a
by firms (such as the lack of demand or the lack of significant proportion of entrepreneurs are driven by
working capital) and investigate how different firms necessity (Wennekers et al. 2005; Acs 2006; Tam-
are vulnerable to different types of growth obstacles. vada 2007). Recent work has found it useful to
Consistent with prior research on developed coun- distinguish between firms on the basis of their growth
tries, the results in this paper suggest that size and age performance. Roughly speaking, the intuition is that
have a negative impact on firm growth in the majority high-productivity entrepreneurial firms will grow
of specifications. Enterprises managed by women fast, while those who turn to self-employment simply
have lower expected growth rates. Proprietary firms because they have no other option are more interested
face lower growth on the whole, especially when they in survival and remain at a small scale. For example,
are young. The results suggest that exporting has a Rogerson (1996) distinguishes between ‘survivalist’
positive effect on firm growth, especially for young and other small firms, while Autio (2008) separates
firms and for female-owned firms. An interesting high-aspiration entrepreneurs from low-aspiration
finding is that although some small firms are able to entrepreneurs. It should be noted that these classifi-
convert technical knowhow into commercial success, cation schemes rest mainly on different growth
many do not have any access to technical knowledge performances of small firms.9
and many others are unable to translate it into superior The negative dependence of growth on size is well
growth.8 Finally, the results pertaining to the barriers documented in the literature, especially for small
in firm growth models suggest that the main problems firms, and is taken by many authors as a ‘stylized
faced by female firms seem to concern raw materials fact’ (see Coad 2009, Chap. 4, for a survey). We
and market problems. Rural firms are vulnerable to therefore hypothesize that smaller firms will grow
problems concerning raw materials, equipment, man- faster in our database.
agement, and power shortages, while they are rela- A negative dependence of growth on age has been
tively less exposed to problems concerning labor and found by a long list of authors.10 In the case of small
lack of demand.
The paper is organized as follows. In the following 9
This body of literature is also related to the debate on
section, we provide the theoretical background on formal–informal labor markets in less developed countries and
small firm growth in developing countries. The the nature of self-employment in such contexts (Harris and
dataset we use—an Indian government census on Todaro 1970; Blau 1985; Maloney 2004; Fields 2005; Günther
and Launov 2006; Tamvada 2010).
10
A negative relationship between age and growth has been
8
The database has information on the source of technical found by Fizaine (1968) for French establishments, Dunne
knowledge of the firm. Each firm in the database is asked if it et al. (1989) for U.S. establishments, Evans (1987a, b) for U.S.
has acquired technical know-how from foreign firms or manufacturing firms, Variyam and Kraybill (1992) for U.S.
domestic firms and scientific institutions. manufacturing and services firms, Liu et al. (1999) for

123
386 A. Coad, J. P. Tamvada

firms in India, however, two recent studies (Das into growth and in converting external know-how into
1995; Shanmugam and Bhaduri 2002) report that superior growth performance.
while larger firms have lower growth rates, age is Further, there is ample evidence that female enter-
significantly positively related to firm growth. Das prises in developing countries experience slower
(1995) examines the growth of firms in the young, growth than male enterprises (McPherson 1996; Mead
fast-growing computer hardware industry in India, and Liedholm 1998), presumably because of conser-
while Shanmugam and Bhaduri (2002) focus on a vative social values and attitudes concerning women in
sample of 392 Indian manufacturing firms. We are industry (El-Namaki 1988). Banerjee and Duflo (2007,
therefore curious to see how age influences growth in p. 162) explain that ‘‘If you have few skills and little
our sample of Indian small businesses. capital, and especially if you are a woman, being an
Although young firms have higher expected growth entrepreneur is often easier than finding an employer
rates, they do not have an easy existence. They may with a job to offer.’’ Female entrepreneurs in develop-
well suffer from the ‘liability of newness’ (Stinch- ing countries are less ambitious about growth and
combe 1965) and face difficulties because they lack financial performance than their male counterparts
experience and knowledge, and they have not yet been (Hisrich and Ozturk 1999; Singh et al. 2001). They are
legitimated or recognized in the marketplace. In less concerned about achieving their high-growth
contrast to the liability of newness, however, it is business aspirations than they are about providing for
useful to consider the concept of a ‘liability of their families. Female-led enterprises are typically
senescence’ (Barron et al. 1994). This adversely smaller and pay less to their employees (Singh et al.
affects old firms who, after years of repeating the 2001). We therefore anticipate that female firms will,
same production routines and accumulated rules and on average, have lower growth rates.
structures, become more inert. Furthermore, there is Many recent studies show that exporting has a
the concept of ‘liability of obsolescence’ (Barron et al. positive impact on performance. As Grossman and
1994) which holds that older firms are no longer a Helpman (1991) suggest, foreign buyers may provide
good fit to the new market environment. Younger technical assistance to exporters leading to improved
firms, on the other hand, are more flexible and can production efficiency. Clerides et al. (1998) show that
internalize outside knowledge in their routines. international trade could improve firms’ access to
Although at first glance these perspectives appear to advanced technical knowledge. Clerides et al. (1998)
be conflicting, it is important to recognize the subtle find that firms that export are more productive than
differences between the liability of newness and the non-exporting firms in many developing countries. For
liabilities of senescence and obsolescence. Young these reasons, we expect a positive effect of exporting
firms are fragile and vulnerable, they are unable to and external knowledge on firm growth. However, we
weather adverse shocks, and they suffer from a lack of anticipate that these benefits of exporting and external
legitimation and recognition. For instance, in our know-how will be stronger for younger firms than for
‘barriers to growth’ analysis, we would expect young older firms which may well be too inert to benefit from
firms to suffer from access to working capital (and we these challenging opportunities.
do in fact observe this). Old firms, in contrast, are Firms that enter the market small often remain
more inert, and we would expect them to be less able small because they face formidable barriers to
to face new challenges (such as exporting) and to growth. Indeed, several authors have commented on
integrate new external knowledge into their produc- the ‘missing middle’ in the firm size distribution that
tion routines. We would therefore expect that old arises when large firms grow larger because small
firms are less successful in converting export activity firms rarely grow into the next size category (Tybout
2000; Sleuwaegen and Goedhuys 2002). These
enterprises remain small for a whole host of reasons,
Footnote 10 continued such as continuing financial constraints, transport
Taiwanese electronics plants, Sleuwaegen and Goedhuys costs, limited infrastructure, and a lack of suitable
(2002) for Ivorian manufacturing firms, Reichstein and Dahl
management resources. In particular, difficulties in
(2004) for Danish limited liability companies, Geroski and
Gugler (2004) for large European companies, and Yasuda adopting new technologies and crossing over to the
(2005) for Japanese manufacturing firms. ‘modern’ sector have been singled out as impeding

123
Firm growth and barriers to growth among small firms in India 387

the growth of small firms in developing countries of Small Scale Industries in India. We use firm level
(Aftab and Rahim 1989). In our analysis of ques- data from the third census of registered small-scale
tionnaire responses on perceived barriers to growth, firms. This census was conducted for the financial
we investigate which types of firms are more year 2002–2003 (the reader interested in the macro-
vulnerable to certain specific growth barriers. For economic context of our analysis is referred to other
example, we might expect that small and young firms sources).12 Our database is essentially a cross-section
are likely to have problems obtaining access to at a given point in time. This rich dataset has data on
capital, while exporting firms might have problems in 1.5 million small firms. The Ministry of Small Scale
finding (skilled) employees. Industries defines firms having an initial investment
of less than 10 million Indian rupees as Small Scale
Industries. In an attempt to have some degree of
3 Database and summary statistics homogeneity between the firms we analyze, we
restrict the sample to firms in the manufacturing
In this section, we begin by describing the database sector. Each firm was asked a series of questions,
used in this paper (Sect. 3.1), before presenting the such as its year of initial production, the sector of its
distributions of some key variables (Sect. 3.2) and operation, the source of its technical knowledge and
finally providing some summary statistics (Sect. 3.3). energy, the gross output for three consecutive finan-
cial years, ownership type, and whether it experi-
3.1 Database description enced ‘significant decline’ over the last 3 years. If the
firm experienced a decline, it was asked the causes
Previous work on entrepreneurship in developing for its decline. The sample contains new small firms
countries has tended to use small samples obtained (including those that are growing rapidly) as well as
from correspondence with entrepreneurs through old firms that have not grown beyond the size
questionnaires. Undeniably, this approach has pro- threshold despite their age.
duced valuable new results in those areas where data In theory, one might be concerned about the
were sparse. In our paper, however, we use a statistical effects of having an upper threshold on
government census, and there is a very high response the sample. At the top of the distribution, near the
rate.11 Our work thus complements previous work by threshold, relatively large firms may grow or decline.
providing estimates based on a large database. We do Growing firms might be excluded from the sample
not focus on the details and specificities of particular (because they exceed the threshold), whereas declin-
regions or categories of businesses, as other work has ing firms would stay in the sample. This might lead to
done, but instead provide an aggregate view of the a downward bias on the growth rates of large firms. In
characteristics of all Indian firms. Previous work practice, however, this does not seem to be an
focusing on small samples can thus be compared to important issue. First, only a very small proportion
our estimates. However, the estimates reported here of firms are close to the upper size threshold. Second,
have an advantage over much previous work because firms are sorted according to their initial size (i.e., size
our large sample size ensures the representativeness at the beginning of production), rather than their final
of our results.
The main source of data for examining the 12
The rapid transformation of the Indian economy from a
determinants of firm growth comes from the Ministry state-controlled license regime to a liberalized regime is widely
documented in the literature (we refer the reader to Aghion
et al. 2005; Kochhar et al. 2006; Rodrik and Subramanian
11
The survey was conducted on all registered small firms and 2005). These economic transitions have had implications for
was conducted by the Office of the Development Commis- industry-wide productivity and competition (see Krishna and
sioner, Government of India. According to the Development Mitra 1998; Balakrishnan et al. 2006). India is the second-most
Commissioner, India, ‘‘All the SSI units permanently regis- populated country in the world, with a population exceeding
tered up to 31-3-2001 numbering 2,262,401 were surveyed on one billion. For further indicators of economic development in
complete enumeration basis, of which 1,374,974 units (61%) India, such as gross domestic product (GDP), life expectancy,
were found to be working and 887,427 units (39%) were found literacy rates, inequality, and the structure of Indian industry
to be closed.’’ (http://www.dcmsme.gov.in/ssiindia/census/ (shares of agriculture, industry and services), the reader is
highlights.htm, accessed on 25 October 2010). referred to Allen et al. (2006).

123
388 A. Coad, J. P. Tamvada

Fig. 1 Kernel densities of the size distribution (log output in computed for equispaced points using an Epanenchnikov
the financial year 1999–2000; left) and age distribution (right), kernel (using the gbutils ver. 5.1 software package)
both plotted with a logarithmic y-axis. Kernel densities are

size, such that firms that are large enough to be near 3.2 Univariate distributions
the threshold will not be censored even if their final
size is above the threshold. Third, we repeated our In this section, we present some univariate distribu-
econometric analysis using a different (lower) size tions of some of the main variables.
threshold and we obtained similar results.13 Figure 1 shows the size and age distributions for
The main variables in the empirical analysis the small businesses in our sample. The size distri-
include age of firm, size (gross output), ownership bution (Fig. 1; left) shows a smooth unimodal shape,
type, gender of the entrepreneur, exporting status, which is in line with the previous literature on firm
source of technical know-how and, of course, firm size distributions. There is a concave decay on the
growth rates (calculated as log-differences of size). upper tail, which indicates that the firm size distri-
The control variables include industry dummies, bution has a faster decay than the Pareto bench-
sources of energy, and regional dummy variables. mark.15 Although there is an upper size threshold
In keeping with previous studies, our measure of which determines whether or not firms are required to
growth rates is calculated by taking the differences of answer the questionnaire, this threshold does not
the logarithms of size: produce any dramatic cut-off point at the upper end
of the size distribution.16
GROWTHit ¼ logðXit Þ  logðXi;t2 Þ ð1Þ
The age distribution is shown in Fig. 1 (right).
where X represents firm size and is measured in terms Empirical age distributions are hard to find in the
of gross output.14 literature, presumably because data on age are not
always easy to obtain [but for an early contribution,
see Huergo and Jaumandreu (2004); see also Coad
(2010) for a survey]. Our plot of the firm age
13
By upper threshold, we refer to the upper limit of the distribution is therefore a useful addition to this
investment of 10 million Indian rupees that is used to define nascent strand of literature. Most of the businesses
small-scale firms in India. In order to check for the robustness
of the results, we artificially lower this threshold to exclude
15
firms that have initial investment greater than 9 million Indian The Pareto distribution (also known as a power law
rupees and find that the results are consistent. distribution) can be represented as a straight line of negative
14
Growth is measured in terms of changes in gross output, slope on log–log axes.
16
rather than in terms of an alternative measure of firm size, such By definition, all firms in the survey have an investment in
as number of employees, because of data constraints. For fixed assets in plant and machinery that is less than 10 million
example, we only have data on firm employment levels for the Indian rupees. This is the upper threshold for firms to be
single financial year 2001–2002, which means it is not possible classified in the SSI sector in India at the time when the survey
to calculate employment growth rates. was conducted.

123
Firm growth and barriers to growth among small firms in India 389

Fig. 2 Annual growth rate distributions [i.e., (log) growth of in the figure on the right at ±loge10 = ±2.303, which
output from financial year (FY) 2000–2001 to FY 2001–2002] correspond to tenfold growth rates. Kernel densities are
for the raw dataset (i.e., before trimming), plotted on a linear y- computed for equispaced points using an Epanenchnikov
axis (left) or a logarithmic y-axis (right). Note the local peaks kernel (using the gbutils ver. 5.1 software package)

are young, with the median age being 12 years and Looking at the growth rate distribution, we are in
75% of the businesses being 18 years or younger. The fact able to detect an unnaturally high occurrence of
decay in the distribution is roughly linear over most tenfold growth rates when the data are plotted on
of the support, which allows us to speculate that an logarithmic axes (see Fig. 2, right), although these
exponential law would be a suitable representation.17 local peaks are barely visible on standard axes (see
There is a local peak at around 100 years, however, Fig. 2, left). We conclude that measurement error and
which is relatively insignificant in magnitude, data entry mistakes are present in our dataset (this is
although it is clearly discernable when logarithmic discussed in more depth in Sect. 4).
axes are used. This local peak could be due to
misreporting, as firms approximate their age to a 3.3 Summary statistics
century.18
Summary statistics for our database are presented in
Table 1.19 The summary statistics in Table 1 show
that the average age of a firm in the sample is
17
Incidentally, this lends some support to the assumption of an 13.9 years and that the average annual growth rate is
exponential distribution of firm age in the model of the firm between 6 and 7%. Single ownership is the most
size distribution in Coad (2009). In this model, the lognormal
distribution of firm size obtained from a Gibrat process is
prevalent form of firm ownership structure. Of the
mixed with an exponential firm age distribution to obtain the total number of firms, 86.5% are owned by
Pareto firm size distribution suggested by a number of proprietary owners and 9.2% are partnerships. Only
empirical studies. 8% of the firms are managed by women entrepre-
18
Self-reported data on the age of small businesses have been neurs. While 70% of the firms in the sample have
described by Phillips and Kirchhoff (1989) as being less than
perfectly reliable, even in developed countries such as the
electricity as the main source of power, 19.7% firms
USA. In the case of India, it should be remembered that many do not use any source of power, 6% of firms depend
people do not know exactly how old they are! As such, there on coal and oil, and 3.3% of firms use non-
may well be a small degree of measurement error in the age conventional and traditional sources of power, such
distribution and, as a result, we do not seek to explain every
feature of this distribution. Rather, at this preliminary stage, we
as firewood.
merely wish to emphasize the broad shape of the distribution.
For instance, the age distribution as represented in Fig. 1 is
19
powerful enough to quickly dispel the myth that all small firms In our baseline sample (e.g., in Table 2 column 2) we have
are young. around 700,000 observations.

123
390 A. Coad, J. P. Tamvada

Table 1 Descriptive statistics (all firms)

Summary statistics Mean Standard deviation Minimum Maximum

Growth 2002/2001–2000/1999 0.1441 0.2305 -0.9609 0.9651

Ln (Output 2002/2001) 12.1283 1.7553 0.6931 22.7830
Ln (Output 2001/2000) 12.0614 1.7571 0.6931 22.8357
Ln (Output 2000/1999) 11.9842 1.7610 0.6931 22.7388
Age 13.9261 10.2137 1 102
Ln (age) 2.4031 0.6989 0 4.6250
Proprietary ownership 0.8646 0.3422 0 1
Partnership 0.0918 0.2887 0 1
Private company 0.0299 0.1702 0 1
Co-operative 0.0036 0.0601 0 1
Other ownership 0.0102 0.1004 0 1
Woman 0.0809 0.2728 0 1
Know-how from abroad 0.0103 0.1011 0 1
Know-how from domestic sources 0.1418 0.3488 0 1
No power 0.1966 0.3975 0 1
Coal 0.0242 0.1536 0 1
Oil 0.0362 0.1867 0 1
Liquefied petroleum gas (LPG) 0.0060 0.0775 0 1
Electricity 0.7041 0.4565 0 1
Non-conventional energy 0.0045 0.0670 0 1
Firewood 0.0284 0.1661 0 1
Rural 0.4189 0.4934 0 1
Export 0.0070 0.0836 0 1
n 671,159

4 Methodology plotted with a logarithmic y-axis, displays local peaks

at ±loge10 = ±2.303, which correspond to tenfold
It is likely that there are several sources of measure- growth rates. This unusually high frequency of
ment error. Although this survey is considered to be tenfold growth events is at odds with the existing
an authoritative source of information that has been literature on firm growth rate distributions,20 and we
used to guide industrial policy in India, it cannot be suggest that it is due to misreporting and data entry
expected that the records are 100% accurate. First, mistakes [for example, tenfold growth would occur if
there may well be data entry problems. For example, the figure for gross output in any year contains an
if the data are mistakenly entered with an extra extra digit (e.g., 500,000 instead of 50,000) or a digit
‘zero’, or if a digit is accidentally omitted, then the is missing (e.g., 5000 instead of 50,000)]. Therefore,
firm will experience a tenfold growth rate (positive or we have shown that by examining the growth rate
negative growth). Second, although all of the firms in
the database are registered as small firms, some of
them do not keep detailed accounts of their activities,
and so their responses may not be entirely accurate.
20
The growth rate distribution presented in Fig. 2 Previous work on growth rate distributions has suggested
suggests that our data contain a non-negligible that the empirical distribution is well approximated by a
unimodal ‘tent-shape’ distribution, in particular, the Laplace
amount of measurement error in the form of data (or symmetric exponential) distribution. For a survey of growth
entry mistakes. The growth rate distribution, when rate distributions, see Coad (2009, Chap. 3).

123
Firm growth and barriers to growth among small firms in India 391

distribution one can make deductions about the likely points (Western 1995), although some drawbacks are
presence of outliers. that LMS is computationally intensive (Rousseeuw
To take these caveats into account, we trim our and Leroy 1987), and standard errors cannot be
dataset for outliers. In our main specification, we computed for the coefficient estimates (Blasnik 1998;
omit firms experiencing growth rates that are greater Thorson 1994). In order to detect extreme observa-
than 400% in magnitude.21 Furthermore, we use tions on the explanatory variables, we perform LMS
median regression [also known as the least absolute regressions and then inspect the LMS residuals. We
deviation (LAD) regression, or the L1 estimator] then remove those observations that have LMS
which performs better than ordinary least squares residuals with magnitudes [2.5 standard deviations
(OLS) regression in the case of non-Gaussian distri- (i.e., the cut-off point suggested in Rousseeuw and
butions of the dependent variable. While least- Leroy 1987, p. 17). This removes about 3.5% of our
squares estimators minimize the sum of squared observations.23 Once these outliers have been
residuals, median regressions minimize the sum of removed, we perform LAD regressions on the
absolute residuals. Median regressions have indeed remaining observations. Our methodology therefore
been popular in growth rate regressions because of bears similarities to the reweighted least squares
the heavy-tailed and non-Gaussian nature of growth estimator which gives weights of zero to leverage
rate distributions [see, for example, Bottazzi et al. points (cf. Rousseeuw and Leroy 1987) and also to
(2011) and Coad (2007)]. However, in a robustness the weighted LAD estimator in Giloni et al. (2006).
analysis we also investigate OLS estimates (not
reported). Generally, these tend to be similar to
median regression results. To reduce the influence of 5 Growth rate regressions
unreliable data, we also smooth our growth rate
variable by measuring growth over a 2-year period In this section, we estimate the following firm growth
(from financial year 1999–2000 to financial year equation for our cross-sectional dataset:
2001–2002)—not just for 1 year—so that the impact
Firm Growth
of short-term growth shocks can be smoothed out.22
Extreme observations on the dependent variable ¼ a þ b1  Size þ b2  Age
(i.e., ‘outliers’) are therefore dealt with by removing þ b3  ProprietaryOwnership
extreme growth events and by using an estimator that þ b4  Woman þ b5  PowerSource
is robust to outliers. A further complication, however, þ b6  TechnicalKnowledge
is that there may well be extreme observations in the
þ b7  Industrial Sector
predictor space (i.e., ‘leverage points’). Indeed, we
have no reason to suspect that mismeasurement and þ b8  Rural þ b9  Export
data entry mistakes are exclusive concerns of the þ b10  RegionalDummies þ e
dependent variable; consequently, they may also
The main explanatory variables of interest are firm
affect the explanatory variables. We are also aware
size, firm age, a dummy variable for female-owned
that standard regression techniques, such as least-
firms, the power source of the firm, the use of
squares estimators and median regressions (i.e.,
external technical knowledge, and dummies for rural
LAD), are highly sensitive to extreme observations
firms and for exporting firms. We include a full set of
in the form of leverage points (Rousseeuw and Leroy
industry and regional dummies as control variables,
1987). In contrast, the least median of squares (LMS)
although the coefficients are not reported in the
regression is very robust to outliers and leverage
results tables. More information on these variables

21
In a further analysis (not reported here) we also explored the 23
For example, without removing outliers we would have
robustness of our results using a 200% growth cutoff point, and 695,757 observations in column 2 of Table 3, but when outliers
obtained similar results. are removed, we have 671,159 observations. This corresponds
22
A similar approach can be found in Liu et al. (1999). to a loss of 3.535% of observations.

123
392 A. Coad, J. P. Tamvada

can be found in the variables definition table in the The finding of a negative dependence of growth on
Appendix (Table 5). age is thus able to reconcile the dynamics of Indian
We present the estimation results for the whole small firms to the broad regularities observed in the
sample in Table 2 and the estimation results for the rest of the world (as mentioned in Sect. 2). Further, as
subsample of female firms in Table 3. In both tables, the statistically significant and negative coefficients
the basic specification estimating the effect of age of variable ‘rural’ in columns 3, 4, and 5 of Table 2
and size on growth is presented in column (1). In suggest, young rural firms experience slower growth
column (2), we introduce the ownership variable than older rural firms, perhaps because they have
‘proprietary’, female, the knowledge source variables trouble establishing themselves in a rural setting.
‘abroad’ and ‘domestic’, export and rural location Age, it seems, has a retarding effect on growth, but
variables, and a set of industry and regional dummies. this retarding effect fades for firms above a certain
In column (3), we restrict the sample to young age threshold. Put differently, young firms grow
firms—that is, firms that are less B10 years old. In faster than older firms, but among our sample of old
column (4), we restrict the sample to firms that are firms, growth becomes less dependent on age. As
[10 years old. In column (5), we restrict the sample such, our results are similar to those obtained by
to firms that are [20 years old. Bigsten and Gebreeyesus (2007), who analyze Ethi-
In order to make the results more digestible, we opian census data on manufacturing firms with over
organize the results in the following order: we first ten employees. Bigsten and Gebreeyesus (2007)
discuss age effects, then results concerning female observe that ‘‘[g]rowth and age are inversely related
entrepreneurs, then results on exporting and internal- only in the first few years after entry and stay
ization of foreign knowledge, and finally proprietary constant for most of the age group until it starts to
ownership. Thus, the discussion of our empirical have a positive relation beyond age 50.’’ (p. 831). In
results is organized according to four main themes: our baseline regression specification (Table 2),
firm age, female entrepreneurs, exporting and over- younger firms experience higher growth rates, even
seas knowledge, and proprietary ownership. in the subsample of firms [20 years old. For the
subsample of female firms, however (Table 3), the
negative dependence of growth on age is not
5.1 Size and age significant, perhaps because these firms face formi-
dable barriers to growth irrespective of their age.
As seen in all specifications in Table 2, size and age As the significant coefficient on the variable
have a negative association with firm growth. In other ‘firewood’ in column (3) suggests, among young
words, smaller, younger firms are expected to grow firms younger than 10 years, being a firewood firm
faster. Our estimates suggest that a one standard has a positive impact on firm growth, while among
deviation increase in size results in a decrease of 2.76 ‘old’ firms[20 years, as the coefficient in columns 4,
percentage points in the growth rate.24 and 5 show, growth rates of firewood firms are
expected to be lower. It could be that for firewood
firms in particular, the advantages of experience are
not great (because of a basic production technology),
and youthful ‘spunk’ counts a lot more in determining
success in this industry.
24
This estimate is calculated as follows. From the summary The results presented in Table 2 are consistent
statistics table (Table 1), we observe that the standard deviation with the theory of ‘liability of senescence’ (Barron
of log(output) is 1.7610. Now, from column (2) of Table 2, we
et al. 1994) because we observe statistically signif-
see that the coefficient on log(output) is -0.0159. Ceteris
paribus, changing log(output) by one standard deviation changes icant negative coefficients on the knowledge source
the dependent variable by 1.761 9 -0.0159 = -0.0279999. variables ‘Abroad’ in column 4, suggesting that old
The dependent variable is loge(growth rate), and the change in firms are less capable of learning from foreign
loge(growth rate) is = -0.0279999. A log(growth rate) of
knowledge sources and translating them into superior
-0.0279999 corresponds to a (conventionally-measured)
growth rate of -0.0276115 (given that e(-0.0279999) - 1 = growth. The coefficients for the ‘export’ variable in
-0.0276115), which can then be rounded to -2.76%. columns 3 are larger than the coefficient in column 5,

123
Firm growth and barriers to growth among small firms in India 393

Table 2 Determinants of firm growth

Variables (1) (2) (3) (4) (5)
All firms All firms Young firms Old firms Oldest firms
(age B10 years) (age [10 years) (age [20 years)

Log(output) -0.0114*** -0.0159*** -0.0164*** -0.0158*** -0.0146***

(0.000129) (0.000170) (0.000275) (0.000224) (0.000341)
Log(age) -0.00876*** -0.0133*** -0.0207*** -0.00733*** -0.00675***
(0.000324) (0.000343) (0.000837) (0.000822) (0.00161)
Ownership
Proprietary -0.00358*** -0.00509*** -0.00220** -0.00111
(0.000743) (0.00119) (0.000977) (0.00145)
Female -0.00381*** -0.00403*** -0.00329*** -0.00609***
(0.000877) (0.00127) (0.00127) (0.00217)
Knowledge source
Abroad -0.00513** 0.00265 -0.0106*** -0.00136
(0.00228) (0.00368) (0.00297) (0.00489)
Domestic -0.00227*** -0.000462 -0.00367*** -0.00717***
(0.000676) (0.00108) (0.000894) (0.00140)
Firm characteristics
Rural -0.00367*** -0.00731*** -0.000985 0.000368
(0.000522) (0.000821) (0.000700) (0.00115)
Export 0.0537*** 0.0672*** 0.0462*** 0.0451***
(0.00278) (0.00460) (0.00358) (0.00515)
Power source
Coal -0.00863*** -0.0112*** -0.00856*** -0.00850***
(0.00164) (0.00319) (0.00196) (0.00270)
Oil 0.00570*** 0.00337 0.00837*** 0.00264
(0.00145) (0.00225) (0.00195) (0.00355)
LPG 0.0206*** 0.0166*** 0.0242*** 0.0202***
(0.00302) (0.00453) (0.00421) (0.00661)
Electricity 0.0166*** 0.0187*** 0.0146*** 0.0152***
(0.000721) (0.00112) (0.000970) (0.00158)
Non-conventional 0.0269*** 0.0359*** 0.0212*** 0.0211***
(0.00345) (0.00534) (0.00467) (0.00808)
Firewood 0.00266* 0.00874*** -0.000256 0.000206
(0.00152) (0.00246) (0.00198) (0.00305)
Industry dummies Yes Yes Yes Yes
Region dummies Yes Yes Yes Yes
Constant 0.301*** 0.354*** 0.380*** 0.333*** 0.314***
(0.00172) (0.00277) (0.00459) (0.00413) (0.00761)
Observations 671,159 671,159 291,466 379,693 128,150
Pseudo R2 0.0065 0.0343 0.0366 0.0328 0.0358
Notes: Standard error are given in parenthesis. *** p \ 0.01, ** p \ 0.05, * p \ 0.1. Dependent variable is growth rate over three
annual exercises (from 1999–2000 until 2001–2002). Firms experiencing greater than fourfold growth are excluded. The base
category for ownership type is ‘other’, which includes partnership firms and co-operatives; for knowledge source, it is ‘none’, for
power source, it is ‘no power’

123
394 A. Coad, J. P. Tamvada

Table 3 Determinants of firm growth (subsample of female firms)

Variables (1) (2) (3) (4) (5)
All firms All firms Young firms Old firms Oldest firms
(age B 10) (age [ 10) (age [ 20)

Log(output) -0.0140*** -0.0188*** -0.0182*** -0.0192*** -0.0198***

(0.000508) (0.000732) (0.00110) (0.00100) (0.00203)
Log(age) -0.00330** -0.0146*** -0.0183*** -0.0106*** -0.00714
(0.00129) (0.00149) (0.00323) (0.00384) (0.00993)
Proprietary -0.0174*** -0.0169*** -0.0153*** -0.00907
(0.00330) (0.00495) (0.00447) (0.00866)
Knowledge source
Abroad -0.0156* 0.0120 -0.0283** -0.0586**
(0.00929) (0.0150) (0.0117) (0.0266)
Domestic 0.00480* 0.00673 0.000592 -0.00280
(0.00281) (0.00417) (0.00385) (0.00814)
Firm characteristics
Rural 0.000593 -0.00207 0.00233 0.00354
(0.00224) (0.00321) (0.00322) (0.00744)
Export 0.0596*** 0.105*** 0.0475*** 0.0740***
(0.0110) (0.0188) (0.0134) (0.0261)
Power source
Coal -0.0321*** -0.0150 -0.0343*** -0.00293
(0.00957) (0.0167) (0.0115) (0.0223)
Oil -0.0165** -0.0207* -0.00912 -0.00591
(0.00755) (0.0113) (0.0103) (0.0242)
LPG 0.0240** 0.0352** 0.00755 0.00919
(0.0109) (0.0146) (0.0175) (0.0368)
Electricity 0.0111*** 0.0105*** 0.0116*** 0.0327***
(0.00277) (0.00386) (0.00417) (0.00978)
Non conventional 0.0120 0.0370** -0.0410** -0.0108
(0.0122) (0.0169) (0.0183) (0.0394)
Firewood 0.00768 0.00799 0.00662 -0.00719
(0.00548) (0.00797) (0.00780) (0.0169)
Industry dummies Yes Yes Yes Yes
Region dummies Yes Yes Yes Yes
Constant 0.298*** 0.391*** 0.396*** 0.379*** 0.341***
(0.00632) (0.0118) (0.0185) (0.0186) (0.0457)
Observations 54,330 54,330 30,488 23,842 6,658
Pseudo R2 0.0072 0.0469 0.0507 0.0449 0.0447
Notes: Standard errors are given in parenthesis. *** p \ 0.01, ** p \ 0.05, * p \ 0.1. Dependent variable is growth rate over three
annual exercises (from 1999–2000 until 2001–2002). Firms experiencing greater than fourfold growth are excluded. The base
category for ownership type is ‘other’, which includes partnership firms and co-operatives; for knowledge source, it is ‘none’; for
power source, it is ‘no power’

suggesting that younger exporting firms enjoy greater 30% in Table 2) as we move from young firms
growth rates than older exporting firms. The coeffi- (\10 years in column 3) to older firms ([20 years in
cient of exporting decreases dramatically (by over column 5).

123
Firm growth and barriers to growth among small firms in India 395

5.2 Female entrepreneurs 5.3 Exporting and overseas knowledge

The results in Table 2 consistently show that firms In the case of female firms, it is the younger firms (age
headed by females grow slower, a finding that is B10 years) that are better able to use both domestic
consistent with the findings from a number of other knowledge sources to their advantage (the coefficient
studies. The coefficient of the variable ‘female’ is of ‘domestic’ in column 3, Table 3, is positive and
-0.00381 in column 2 of Table 2, which corresponds significant at the 10% level). In contrast, there are a
to a decrease in the expected growth rate of 0.380 few cases in which external sources of knowledge are
percentage points (i.e., not exactly 0.381 percentage observed to have a negative association with firm
points, because our dependent variable is expressed growth in the case of old firms, as in the case of old
in terms of logarithms). female firms using knowledge from abroad (Table 3,
One possible explanation could be that women columns 4 and 5) and in old firms using domestic
self-select themselves into low-growth industries. knowledge (Table 2, columns 4 and 5).
The existing literature, however, does not favor Exporting is observed to have a positive effect on
this interpretation. Mead and Liedholm (1998) growth in all specifications. Due to data limitations,
survey the literature on female entrepreneurs in we are unable to investigate whether it is fast-growth
developing countries and conclude that female-led firms that self-select into export markets, or whether
businesses have both lower survival probabilities firms experience faster growth once they begin to
and lower expected growth rates, even after export. Evidence on the productivity growth of firms
controlling for other relevant factors. Furthermore, in four African countries, however, suggests that
we observe a negative effect of female ownership learning-by-exporting effects are predominant and
on firm growth even after controlling for other that self-selection of the more productive firms into
factors, such as industry and regional characteris- exporting plays no major role (Bigsten et al. 2004).
tics in Table 2. The estimation results in Tables 2 and 3 suggest that
However, the effect of interaction of the the coefficient on exporting is larger for younger
‘female’ variable with other variables on firm firms (in column 3) than for older firms (in columns 4
growth is not clear (see Table 2). For example, and 5), indicating that younger firms benefit more
Table 2 does not allow us to examine the impact of from exporting than their older counterparts. This is
a female business that is in a rural area, or of a consistent with learning theories of firm growth via
female business that exports. For this reason, in internationalization, which consider that entrepre-
Table 3 we estimate all of the specifications for a neurial dynamism and vision are more important
subsample of firms that are owned and managed by factors than age and experience with respect to an
women entrepreneurs. effect on export success. In this view, older firms
As the negative and statistically significant coef- suffer from cognitive inertia that makes them less
ficients of ‘abroad’ in columns 4 and 5 of Table 3 responsive to business opportunities and changing
show, old female businesses that have a foreign market conditions (Autio et al. 2000).
knowledge source actually find it harder to grow,
while this has no significant effect on younger female 5.4 Proprietary ownership
firms. As the positive and significant coefficients on
‘export’ variable in all the columns of Table 3 Our database also distinguishes between firms estab-
suggest, female firms of all ages stand to gain a lot lished as partnerships and those under proprietary
from exporting. The effect of exporting on growth is ownership. In our baseline results table (Table 2) as
stronger for female firms than for other firms when well as in the results for female firms (Table 3), we
we compare the coefficients in Tables 2 and 3. Thus, observe that proprietary firms have lower expected
once women are able to overcome the considerable growth rates. More specifically, we observe that,
difficulties associated with entering export markets, among young firms, proprietary firms face lower
female-led firms are in a position to enjoy faster expected growth rates, although proprietary owner-
growth. ship has no significant association with growth

123
396 A. Coad, J. P. Tamvada

Table 4 Barriers to growth (probit regressions)

Barrier (1) (2) (3) (4) (5) (6) (7) (8)
Lack of Working Raw Power Labor Market Equipment Management
demand capital materials shortage problems problems problems problems

Log(output) -0.0477*** -0.0194*** 0.00951** -0.0139*** 0.0438*** 0.0331*** -0.0506*** -0.0246***

(0.00316) (0.00313) (0.00396) (0.00373) (0.00479) (0.00315) (0.00474) (0.00578)
Log(age) 0.0718*** -0.0842*** 0.0543*** -0.00922 0.0383*** 0.0206*** 0.0454*** 0.0138
(0.00685) (0.00678) (0.00863) (0.00825) (0.0105) (0.00680) (0.00986) (0.0123)
Ownership
Proprietary -0.0476*** 0.0538*** -0.0654*** 0.0332** 0.0553*** -0.0233* -0.0619*** -0.103***
(0.0140) (0.0139) (0.0177) (0.0162) (0.0207) (0.0140) (0.0206) (0.0246)
Female 0.00327 0.0222 0.0860*** -0.0305 0.0144 0.0397*** 0.0207 0.0258
(0.0150) (0.0147) (0.0183) (0.0193) (0.0234) (0.0147) (0.0212) (0.0264)
Knowledge source
Abroad -0.173*** 0.120** 0.104* 0.0113 0.0513 -0.0859* 0.0917 -0.105
(0.0466) (0.0470) (0.0573) (0.0526) (0.0671) (0.0474) (0.0653) (0.0909)
Domestic -0.0520*** 0.0819*** 0.0375** 0.136*** 0.114*** 0.112*** 0.152*** 0.0192
(0.0123) (0.0122) (0.0154) (0.0141) (0.0178) (0.0122) (0.0168) (0.0219)
Firm characteristics
Rural -0.0356*** 0.00393 0.102*** 0.180*** -0.0563*** -0.0550*** 0.0967*** 0.0394**
(0.0103) (0.0102) (0.0128) (0.0126) (0.0161) (0.0102) (0.0146) (0.0186)
Export -0.0472 -0.0777* 0.00758 -0.180*** 0.227*** -0.0825* 0.194*** 0.105
(0.0459) (0.0457) (0.0583) (0.0554) (0.0568) (0.0464) (0.0640) (0.0755)
Power source
Coal -0.0810** 0.0782** 0.240*** 0.325*** 0.00288 0.0726** -0.112** -0.109*
(0.0336) (0.0343) (0.0389) (0.0550) (0.0554) (0.0334) (0.0521) (0.0627)
Oil -0.184*** -0.00836 0.0738* 0.775*** 0.0868 0.0158 0.0267 0.0664
(0.0351) (0.0362) (0.0402) (0.0442) (0.0528) (0.0351) (0.0471) (0.0585)
LPG -0.106* -0.0521 -0.102 0.666*** -0.171 0.132** 0.201*** -0.0684
(0.0556) (0.0552) (0.0704) (0.0784) (0.107) (0.0544) (0.0752) (0.104)
Electricity -0.111*** -0.0668*** -0.0949*** 1.270*** 0.0977*** -0.0946*** 0.0550*** -0.0501**
(0.0139) (0.0137) (0.0167) (0.0240) (0.0224) (0.0136) (0.0196) (0.0246)
Non-conventional -0.456*** 0.179** 0.0756 0.512*** -0.0637 -0.0864 -0.0521 -0.00519
(0.0735) (0.0758) (0.0825) (0.103) (0.117) (0.0716) (0.0992) (0.125)
Firewood -0.158*** -0.0114 0.0590* 0.332*** 0.169*** 0.160*** -0.0379 0.00881
(0.0266) (0.0260) (0.0308) (0.0459) (0.0412) (0.0259) (0.0397) (0.0466)
Industry dummies Yes Yes Yes Yes Yes Yes Yes Yes
Region dummies Yes Yes Yes Yes Yes Yes Yes Yes
Constant 0.518*** 0.654*** -1.308*** -1.670*** -2.517*** -0.888*** -0.776*** -1.557***
(0.0526) (0.0522) (0.0657) (0.0641) (0.0819) (0.0526) (0.0759) (0.0961)
Observations 84,595 84,602 84,544 84,491 84,446 84,590 84,413 84,441
Pseudo R2 0.0721 0.0382 0.0472 0.1450 0.0457 0.0375 0.0324 0.0265

Standard errors are given in parenthesis. *** p \ 0.01, ** p \ 0.05, * p \ 0.1. Dependent variable is a binary variable, taking the value 1 if
the suggested factor is identified as a growth barrier, and 0 otherwise. The sample is restricted to declining firms. Firms are considered to be
declining if their net worth in 2001–2002 is less than 50% of their net worth in 2000–2001, or if there has been a continuous delay in the
payment of interest on the loans during the last year, or if there has been continuous decline in the output in the period 2000–2002. The base
category for ownership type is ‘other’, which includes partnership firms and co-operatives; for knowledge source, it is ‘none’; for power
source, it is ‘no power’

123
Firm growth and barriers to growth among small firms in India 397

among older firms (this can be seen by comparing the In Table 4, we estimate the internal determinants
coefficients of the proprietary ownership variable in of (subjectively perceived) growth problems faced by
columns 3 and 5 in Tables 2 and 3). these firms. For the estimation results reported in
Table 4, our dependent variable takes the value of 1 if
5.5 Further results firms report suffering from a particular growth
barrier, and 0 otherwise. Our approach is similar to
In a further analysis (results not shown here; see that of Sleuwaegen and Goedhuys (2002) for Ivorian
instead Coad and Tamvada 2008) we repeat the firms and to that of Robson and Obeng (2008) who
analysis on a sample of firms who are in the energy analyze Ghanaian firms. The main problems for large
bracket of ‘no power needed.’ In this group, we see firms seem to be labor problems and market prob-
that young female firms are expected to have lower lems. In contrast, small firms are more vulnerable to a
growth, while for older female firms there is no lack of demand, problems obtaining working capital,
significant effect. Females may need to become problems of power shortages, equipment problems,
legitimized, and they may face gender-based obsta- and management problems. Old firms also suffer
cles, but once they are above a certain age, these from labor and market problems and, in addition,
gender-based liabilities seem to disappear. they are susceptible to a lack of demand and
Knowledge from domestic or foreign sources has problems with raw materials. Young firms, in
no clear beneficial effect on growth, and it even has a contrast, have problems obtaining working capital.
negative effect in some cases, such as for the Proprietary firms have many of the opposite problems
subsample of firms declaring firewood as their main to old firms, in the sense that they have no problems
source of energy. It can be anticipated that these in terms of lack of demand or raw materials, but they
firewood firms are too far from the world technology do have difficulties obtaining working capital. The
frontier and lack the ‘absorptive capacity’ required to main problems faced by female firms seem to
be able to successfully internalize new technological concern raw materials and market problems. Rural
knowledge from abroad. firms are vulnerable to problems concerning raw
materials, equipment, management, and also power
shortages, while they are relatively less exposed to
6 Barriers to growth problems of labor and lack of demand (i.e., relative to
their urban counterparts). Exporting firms, on the
In this section, we focus on a subsample of declining other hand, cite labor problems as being particularly
firms. The questionnaire contains questions that relate severe (perhaps because they need specialized skilled
specifically to declining firms and the specific labor) as well as equipment problems, while power
obstacles to growth faced by these firms. In a first shortages are relatively unproblematic.
stage, firms are identified as declining firms if their
net worth in the financial year 2001–2002 is less than
50% of their net worth in 2000–2001, if there has 7 Conclusion
been a continuous delay in the payment of interest on
the loans during the last year, or if there has been Developing countries have their own specific require-
continuous decline in the output in the period ments from small firms, and the corresponding small
2000–2002. In the database, roughly one in seven firm strategies may well be different from their
firms is classified as a ‘declining firm’. In a second counterparts in developed countries. However, in
stage, these declining firms are given a list of eight order to construct a meaningful industrial policy for
reasons for their poor performance and asked to the small firms in developing countries, reliable
identify which are the most important factors. These empirical estimates of key indicators of industrial
eight factors are the following: lack of demand, dynamics are needed. Unfortunately, however, empir-
shortage of working capital, non-availability of raw ical work on small firms has often focused on
materials, power shortage, labor problems, marketing developed countries. The little work that does exist
problems, equipment problems, and management on developing countries is all too often based on
problems. small samples taken from ad hoc questionnaires.

123
398 A. Coad, J. P. Tamvada

The aim of this paper was to complement the ‘cutting-edge innovators’—as evidenced by a number
literature that exists on the nature and performance of of firms whose main source of energy is firewood.
small businesses in developing countries. Using a Finally, we explore the most pressing issues faced
unique large-scale database on small firms in India, by different kinds of declining firms and obtain
our results are fairly representative of the structure of several interesting insights into the factors that act as
small business in India, and our results may also barriers to the growth of small firms in a developing
presumably be useful in understanding what happens country context. The results for the barriers to firm
in other developing countries. growth models suggest that the main problems faced
The presentation and interpretation of our results by female firms seem to concern raw materials and
highlight a number of issues, including size and age, market problems. Rural firms are vulnerable to
gender, exporting, and external knowledge. Small problems concerning raw materials, equipment, man-
young firms tend to grow faster than larger, older firms; agement, and power shortages, while they are rela-
this is especially true for low-tech firms using firewood tively less exposed to problems concerning labor and
as their power source. Female firms are likely to have lack of demand.
lower growth rates. Exporting has a positive effect on
growth, especially for young firms and also female Acknowledgments We are grateful to the Ministry of Small
Scale Industries, Government of India, for giving access to the
firms. Proprietary firms face lower growth on the
data, and also to Wim Naude, Ulrich Witt, and participants at
whole, especially young proprietary firms. the UNU-Wider workshop on ‘‘Entrepreneurship and
If our small firms were technological leaders, we Economic Development’’ (Helsinki, August 2008), as well as
would expect technical know-how to have a positive to two anonymous referees and the editor (Marco Vivarelli) for
many helpful comments. The usual caveat applies.
influence on output growth. Although some small
firms in our dataset are able to convert know-how
into commercial success, many others do not have
any technical knowledge, and those who do might be Appendix
unable to use it profitably. Indeed, many small firms
in our sample of Indian enterprises are hardly See Table 5.

Table 5 Variable definitions

Variable Definition

Firm Growth (for the 2-year period from financial log(Gross Output2001–2002) - log(Gross Output1999–2000)
year 1999–2000 until financial year 2001–2002)
Size log(Gross Output1999–2000)
Age2002 Log of age of the firm as of 2002 = log (2002-year of initial production)
ProprietaryOwnership Proprietary owner firm
Woman The enterprise is owned and operated by a woman
Power Source This variable takes values: No power needed, Coal, Oil, Liquified
petroleum gas (LPG), Electricity, Non-Conventional energy, Traditional
energy/Firewood
Technical Knowledge This variable takes values: No technical knowhow, technical knowhow
obtained from abroad, and technical knowhow obtained from a domestic
collaborating firm or R&D institution or specialized agency.
Industrial Sector Two Digit Industry Classification following India’s National Industrial
Classification of 1998
Urban Urban = 1 if the firm is located in an urban region
Export Export = 1 if the firm exports a positive share of its output

123
Firm growth and barriers to growth among small firms in India 399

References Coad, A. (2007). A closer look at serial growth rate correlation.

Review of Industrial Organization, 31(1), 69–82.
Acs, Z. J. (2006). How is entrepreneurship good for economic Coad, A. (2009). The growth of firms: A survey of theories and
growth. Innovations, 1(1), 97–107. empirical evidence. Cheltenham, UK: Edward Elgar.
Acs, Z. J., & Audretsch, D. B. (1990). Innovation and small Coad, A. (2010). Investigating the exponential age distribution
firms. Cambridge: MIT Press. of firms. Economics: The Open-Access, Open-Assessment
Aftab, K., & Rahim, E. (1989). ‘Barriers’ to the growth of E-Journal, 4, 2010-17.
informal sector firms: A case study. Journal of Develop- Coad, A., & Tamvada, J. P. (2008). The growth and decline of
ment Studies, 25(4), 490–507. small firms in developing countries. Papers on Economics
Aghion, P., Burgess, R., Redding, S., & Zilibotti, F. (2005). and Evolution 2008-08. Jena: Max Planck Institute of
Entry liberalization and inequality in industrial perfor- Economics.
mance. Journal of the European Economic Association, Das, S. (1995). Size, age and firm growth in an infant industry:
3(2–3), 291–302. The computer hardware industry in India. International
Allen, F., Chakrabarti, R., De, S., Qian, J., & Qian, M. (2006). Journal of Industrial Organization, 13(1), 111–126.
Financing firms in India. World Bank Policy Research Dunne, T., Roberts, M., & Samuelson, L. (1989). The growth
Working Paper 3975. Washington D.C.: World Bank and failure of US manufacturing plants. Quarterly Journal
Autio, E. (2008). High- and low-aspiration entrepreneurship of Economics, 104(4), 671–698.
and economic growth in low-income economies. Paper El-Namaki, M. S. S. (1988). Encouraging entrepreneurs in
presented at the UNU-WIDER project Workshop on developing countries. Long Range Planning, 21(4),
Entrepreneurship and Economic Development, Helsinki. 98–106.
Autio, E., Sapienza, H., & Almeida, J. (2000). Effects of age at Evans, D. S. (1987a). The relationship between firm growth,
entry, knowledge intensity, and imitability on interna- size and age: Estimates for 100 manufacturing industries.
tional growth. Academy of Management Journal, 43(5), Journal of Industrial Economics, 35, 567–581.
909–924. Evans, D. S. (1987b). Tests of alternative theories of firm
Balakrishnan, P., Parameswaran, M., Pushpangadan, K., & growth. Journal of Political Economy, 95(4), 657–674.
Babu, M. (2006). Liberalization, market power, and pro- Fields, G. S. (2005). A guide to multisector labor market
ductivity growth in Indian industry. Journal of Policy models. Social Protection Discussion Paper Series No.
Reform, 9(1), 55–73. 0505. Washington D.C.: World Bank.
Banerjee, A., & Duflo, E. (2007). The economic lives of the Fizaine, F. (1968). Analyse statistique de la croissance des
poor. Journal of Economic Perspectives, 21(1), 141. entreprises selon l’age et la taille. Revue d’e´conomie
Barron, D. N., West, E., & Hannan M. T. (1994). A time to politique, 78, 606–620.
growth and a time to die: Growth and mortality of credit Geroski, P. A. (1995). What do we know about entry? Inter-
unions in New York, 1914–1990. American Journal of national Journal of Industrial Organization, 13, 421–440.
Sociology, 100(2), 381–421. Geroski, P. A., & Gugler, K. (2004). Corporate growth conver-
Beck, T., Demirguc-Kunt, A., & Levine, R. (2005). SMEs, gence in Europe. Oxford Economic Papers, 56, 597–620.
growth, and poverty: Cross-country evidence. Journal of Giloni, A., Simonoff, J. S., & Sengupta, B. (2006). Robust
Economic Growth, 10(3), 199–229. weighted LAD regression. Computational Statistics and
Bigsten, A., Collier, P., Dercon, S., Fafchamps, M., Gauthier, Data Analysis, 50, 3124–3140.
B., Gunning, J., et al. (2004). Do African manufacturing Grossman, G., Helpman, E. (1991). Innovation and growth in
firms learn from exporting? Journal of Development the world economy. Cambridge, MA: MIT Press
Studies, 40(3), 115–141. Günther, I., & Launov, A. (2006). Competitive and segmented
Bigsten, A., & Gebreeyesus, M. (2007). The small, the young, informal labor markets. IZA Discussion Papers 2349.
and the productive: Determinants of manufacturing firm Bonn: Institute for the Study of Labor (IZA).
growth in Ethiopia. Economic Development and Cultural Harris, J. R., & Todaro, M. P. (1970). Migration, unemploy-
Change, 55(4), 813–840. ment and development: A two sector analysis. American
Blasnik, M. (1998). LMS: Stata module to perform least Economic Review, 60(1), 126–142.
median squares regression fit. Boston College Department Hisrich, R. D., & Ozturk, S. A. (1999). Women entrepreneurs
of Economics, Statistical Software Components number in a developing economy. Journal of Management
S358301. Boston, MA: Boston College Development, 18(2), 114–124.
Blau, D. M. (1985). Self-employment and self-selection in Huergo, E., & Jaumandreu, J. (2004). How does probability of
developing country labor markets. Southern Economic innovation change with firm age? Small Business Eco-
Journal, 52(2), 351–363. nomics, 22(3), 193–207.
Bottazzi, G., Coad, A., Jacoby, N., & Secchi, A. (2011). Cor- Kochhar, K., Kumar, U., Rajan, R., Subramanian, A., &
porate growth and industrial dynamics: Evidence from Tokatlidis, I. (2006). India’s pattern of development:
French manufacturing. Applied Economics, 43(1), What happened, what follows? Journal of Monetary
103–116. Economics, 53(5), 981–1019.
Clerides, S., Lauch, S., & Tybout, J. R. (1998). Is Learning by Kozan, M., Oksoy, D., & Ozsoy, O. (2006). Growth plans of
exporting important? Micro-dynamic evidence from small businesses in Turkey: Individual and environmental
Colombia, Mexico and Morocco. Quarterly Journal of influences. Journal of Small Business Management, 44(1),
Economics, 63, 903–947. 114–129.

123
400 A. Coad, J. P. Tamvada

Krishna, P., & Mitra, D. (1998). Trade liberalization, market Santarelli, E., & Vivarelli, M. (2007). Entrepreneurship and the
discipline and productivity growth: New evidence from process of firms’ entry, survival and growth. Industrial
India. Journal of Development Economics, 56(2), 447–462. and Corporate Change, 16(3), 455–488.
Leff, N. H. (1979). Entrepreneurship and economic develop- Shanmugam, K., & Bhaduri, S. (2002). Size, age and firm
ment: The problem revisited. Journal of Economic Liter- growth in the Indian manufacturing sector. Applied Eco-
ature, 17, 46–64. nomics Letters, 9(9), 607–613.
Little, I. M. D. (1987). Small manufacturing enterprises in SIDBI. (2001). SIDBI Report on Small Scale Industries Sector,
developing countries. World Bank Economic Review, Small Industries Development Bank of India. Available
1(2), 203–235. at: http://dcmsme.gov.in/publications/traderep/sidbirep.pdf.
Liu, J., Tsou, M., & Hammitt, J. (1999). Do small plants grow Accessed 21 May 2010.
faster? Evidence from the Taiwan electronics industry. Singh, S. P., Reynolds, R. G., & Muhammad, S. (2001). A
Economics Letters, 65(1), 121–129. gender-based performance analysis of micro and small
Maloney, W. F. (2004). Informality revisited. World Devel- enterprises in Java, Indonesia. Journal of Small Business
opment, 32(7), 1159–1178. Management, 39(2), 174–182.
McPherson, M. A. (1996). Growth of micro and small enter- Sleuwaegen, L., & Goedhuys, M. (2002). Growth of firms in
prises in Southern Africa. Journal of Development Eco- developing countries, evidence from Côte d’Ivoire.
nomics, 48, 253–277. Journal of Development Economics, 68(1), 117–135.
Mead, D. C., & Liedholm, C. (1998). The dynamics of micro Stinchcombe, A. (1965). Social structure and organizations,
and small enterprises in developing countries. World Chapter 4. In J. G. March (Ed.), Handbook of organiza-
Development, 26, 61–74. tions (pp. 142–193). Chicago, IL: Rand McNally &
MSME. (2009). Annual Report 2008–2009, Ministry of Micro, Company.
Small and Medium Enterprises, Government of India. Tamvada, J. P. (2007). Essays on entrepreneurship and eco-
Available at: http://msme.gov.in. Accessed 21 May 2010. nomic development. PhD thesis. Goettingen: University
Nafziger, E., & Terrell, D. (1996). Entrepreneurial human of Goettingen.
capital and the long-run survival of firms in India. World Tamvada, J. P. (2010). Entrepreneurship and welfare. Small
Development, 24(4), 689–696. Business Economics, 34(1), 65–79.
Pavitt, K., Robson, M., & Townsend, J. (1987). The size dis- Thorson, J. A. (1994). The use of least median of squares in the
tribution of innovating firms in the UK: 1945–1983. estimation of land value equations. Journal of Real Estate
Journal of Industrial Economics, 35(3), 297–316. Finance and Economics, 8, 183–190.
Phillips, B., & Kirchhoff, B. (1989). Formation, growth and Tybout, J. R. (2000). Manufacturing firms in developing
survival; small firm dynamics in the US economy. Small countries: How well do they do, and why? Journal of
Business Economics, 1(1), 65–74. Economic Literature, 38, 11–44.
Reichstein, T., & Dahl, M. (2004). Are firm growth rates Variyam, J. N., & Kraybill, D. S. (1992). Empirical evidence
random? Analysing patterns and dependencies. Interna- on determinants of firm growth. Economics Letters, 38,
tional Review of Applied Economics, 18(2), 225–246. 31–36.
Robson, P., & Obeng, B. (2008). The barriers to growth in Wennekers, S., Van Stel, A., Thurik, R., & Reynolds, P.
Ghana. Small Business Economics, 30(4), 385–403. (2005). Nascent entrepreneurship and the level of eco-
Rodrik, D., & Subramanian, A. (2005). From ‘Hindu growth’ nomic development. Small Business Economics, 24(3),
to productivity surge: The mystery of the Indian growth 293–309.
transition. IMF Staff Papers 52(2). Washington D.C.: Western, B. (1995). Concepts and suggestions for Robust
International Monetary Fund. Regression Analysis. American Journal of Political Sci-
Rogerson, C. (1996). Urban poverty and the informal economy ence, 39(3), 786–817.
in South Africa’s economic heartland. Environment and Yasuda, T. (2005). Firm growth, size, age and behavior in
Urbanization, 8(1), 167. Japanese manufacturing. Small Business Economics,
Rousseeuw, P. J., & Leroy, A. (1987). Robust regression and 24(1), 1–15.
outlier detection. Wiley series in probability and mathe- You, J.-I. (1995). Small firms in economic theory. Cambridge
matical statistics. New York: Wiley. Journal of Economics, 19, 441–462.

123