Research Paper No.

2009/51
Innovation and Microenterprises
Growth in Ethiopia
Mulu Gebreeyesus*
November 2009

Abstract

This paper addresses two prominent issues on the development of small enterprises in Africa.
Which factors inhibit or foster innovation activities in small enterprises? Do innovators create
more jobs? We use a large set of microenterprises survey data from Ethiopia that comprise 1000
observations with ten and fewer workers. The analysis shows that firms larger in size and in
manufacturing are more likely to engage in innovative activities. Among the human capital
variables vocational training is found to have a strong effect on the innovation activity.
However, firms owned by female and old entrepreneurs are less likely to get involved in
innovation. In an extended model of firm growth determinants that includes innovation
indicators we found strong evidence that innovators grow faster than non-innovators. Firm
growth is also affected by other factors such as the firm’s initial size, age, access to finance,
sector, and owner character. Our estimation results provide supporting evidence to the stylized
fact that the smaller, younger, and less capital constrained firms grow faster than their
counterparts. Firms in manufacturing also grow faster than other sectors.

Keywords: micro and small enterprises, firm growth, innovation, developing countries, Ethiopia

JEL classification: L25, L26, O31, O55

Copyright © UNU 2009

* UNU-MERIT, email: [email protected]
This paper was presented at the UNU-WIDER and UNU-MERIT Research Workshop on
Entrepreneurship, Technological Innovation, and Development, held in Maastricht, the Netherlands,
30–31 October 2008. This paper is also published as UNU-MERIT Working Paper 2009-053.
UNU-WIDER gratefully acknowledges the financial contributions to the project by the Finnish Ministry
for Foreign Affairs, and the financial contributions to the research programme by the governments of
Denmark (Royal Ministry of Foreign Affairs), Finland (Finnish Ministry for Foreign Affairs), Sweden
(Swedish International Development Cooperation Agency—Sida) and the United Kingdom (Department
for International Development).
ISSN 1810-2611 ISBN 978-92-9230-230-6
Abbreviations

CSA Central Statistical Agency

DWH Durbin-Wu-Hausman endogeneity test
EDRI Ethiopian Development Research Institute
GMM Generalized method of moments
ILO International Labour Organization
IV Instrumental variable
LIML Limited information maximum likelihood
MSE Micro and small enterprise
OLS Ordinary least squares
R&D Research and development
RHS Right-hand side
SME Small and medium enterprises
SSA Sub-Saharan Africa

Tables appear at the end of this paper.

The World Institute for Development Economics Research (WIDER) was

established by the United Nations University (UNU) as its first research and
training centre and started work in Helsinki, Finland in 1985. The Institute
undertakes applied research and policy analysis on structural changes
affecting the developing and transitional economies, provides a forum for the
advocacy of policies leading to robust, equitable and environmentally
sustainable growth, and promotes capacity strengthening and training in the
field of economic and social policy making. Work is carried out by staff
researchers and visiting scholars in Helsinki and through networks of
collaborating scholars and institutions around the world.

www.wider.unu.edu [email protected]

UNU World Institute for Development Economics Research (UNU-WIDER)

Katajanokanlaituri 6 B, 00160 Helsinki, Finland

Typescript prepared by Lisa Winkler at UNU-WIDER

The views expressed in this publication are those of the author(s). Publication does not imply
endorsement by the Institute or the United Nations University, nor by the programme/project sponsors, of
any of the views expressed.
1 Introduction

In developing countries the informal sector that mainly constitutes microenterprises is

the major source of employment and income for the urban population. According to
ILO (2002) estimations, the share of informal employment (outside agriculture) to the
total non-agricultural employment accounts for nearly half or more in all regions of the
developing world and about 72 per cent in sub-Saharan Africa (SSA). They are also
very important part of the developing world economy. For example, in SSA the
contribution of the informal sector in non-agriculture GDP is about 41 per cent. Hence,
their efficiency matters in determining overall economic performance and poverty
reduction.

Despite their potential to improve economic growth, micro and small enterprises
(MSEs) in developing countries lack expectations. They produce largely for the low
income group and employ lower levels of techniques. Many microenterprises are the
self-employed type with a low graduation rate into higher size categories and their
innovative activities are limited (Kiggundu 2002). This is largely due to the harsher
environment they operate in. Unreliable enforcement of contracts, excessive regulatory
and administrative requirements, limited access to finance, and inadequate infrastructure
services all impose disproportionately high transaction costs on MSEs for doing
business generally, and for innovative activity in particular (Ernst 2004).

The promotion of MSEs is becoming a popular development tool. Accordingly,

governments and donors in the developing countries have shown increasing interest in
promoting innovations and entrepreneurship. They have initiated various support
programmes with the aim to improve MSEs’ competitiveness through enhancing
technology and innovation capabilities such as upgrading product quality, improving
design and packaging, and training to improve competitiveness (Pyke 1994). The notion
is that innovation is essential for MSEs to become and remain competitive, move to
higher return activities, and to grow and graduate to small and medium sized enterprise
status, thus, creating new employment opportunities (Ernst 2004). Improving
competitiveness is even more crucial in the context of liberalization and increasing
integration into the world market. Lack of adaptation and upgrading spells defeat, while
firms that keep up or even initiate their own original improvements can be expected to
perform well (Romijn 2002).

The efficacy of such interventions, however, depends on identifying key factors that
foster or inhibit innovation by MSEs and targeting the potentially successful
entrepreneurs. Small business entrepreneurs are hardly homogeneous in objective and
capability. Many are self-employed while others have high vigour to innovate and grow.
They also differ in terms of socioeconomic background and access to resources such as
financial capital. The type of activities they are in is also widespread. What types of
entrepreneurs/firms are more likely to engage in innovative activity? Do innovators
grow faster and create more jobs than non-innovators as it is claimed? Understanding
the attributes of innovators and their impact on employment is crucial in order to
formulate effective policies.

Despite the high profile of the issue in current policy formulations in Africa, there is
little empirical evidence on innovativeness and its impact on firm performance in
MSEs. The existing few studies in Africa mainly examined the determinants of

1
innovative activity and attributes of innovativeness (for example, Van Dijk 2002;
Oyelaran-Oyeyinka 2006; Robson et al. 2008). Van Dijk (2002) examines the
importance of enterprise clusters and cooperation on innovation in the informal sector in
Ghana, Burkina Faso, and Zimbabwe. Oyelaran-Oyeyinka (2006) analyses the impact of
inter-firm collaboration on innovation in Kenya, Nigeria, and Zimbabwe using 200
manufacturing firms. Robson et al. (2008) investigate the determinants of innovation in
Ghanaian small enterprises that employ between four and 50 workers. The lack of
empirical evidence is even more apparent when it comes to the effect of innovation
activity on firm growth. Mahemba and de Bruijn (2003) reported only weak association
between innovativeness in small firms and growth in Tanzanian manufacturing sector.
Thus, innovativeness and small firms’ growth relation has not yet empirically confirmed
in Africa.

In this paper we seek to address two inter-related issues; the determinants of innovative
activity and if innovative enterprises grow faster than non-innovators in African MSEs.
We use a large set of microenterprises survey data from Ethiopia that comprise 1000
observations from six selected major towns including the capital city Addis Ababa. Like
other developing countries, in Ethiopia, the informal sector plays a significant role in
the economy. According to the 1999 survey by the Central Statistical Agency (CSA) the
urban informal sector comprises about 50.6 per cent of the 2.88 million total urban
employments.1 Women employment accounts for about 58 per cent of the employment
in the informal sector.

Recognizing the significance of this sector, the Ethiopian government issued the
National Micro and Small Enterprises Strategy in 1997 and established the Federal
Micro and Small Enterprises Development Agency in 1998. The country’s industrial
policy in 2003 and the poverty reduction strategy in 2006 have singled out MSEs as
major instruments to create a productive and vibrant private sector and reduce poverty
among urban dwellers. These documents reiterated the importance of MSEs promotion
through the provision of finance, training, and infrastructure services among others.
However, in our data there are only a few enterprises (no more than eight per cent)
reported that have received some support from government or NGOs. This implies that
the innovation activity of the microenterprises is expected to be a result of the decision
of the owner. Our analysis will, therefore, emphasize upon the entrepreneurs behaviour
and resource availability to the enterprises as a major determinant of innovativeness and
firm growth.

This paper contributes to the thin literature on innovations in African MSEs in the
following ways. First, it analyses not only the determinants of innovation but also the
impact of innovation on firm employment growth. Second, it exclusively relies on the
lower bottom of size category, firms with ten and fewer workers usually termed as
microenterprises. By doing so, this study tries to address the bias that might arise from
pooling a heterogeneous group in the previous studies as a result of broader definition of
small enterprises, i.e. up to 100 or so workers. Third, unlike to most previous studies it

1 CSA defines urban informal activity as those unincorporated enterprises with fewer than ten
employees, no book accounts, and no license—basically microenterprises. Enterprises with less than
ten workers are also customarily classified as microenterprises in other countries, for example the
European Community defines micro as firms that have zero to nine workers and small firms with 10-
99 workers.

2
covers not only manufacturing but also other major sectors such as service and trading
activities.

This paper is structured as follows. The next section gives data and some descriptive
analysis. Section 3 discusses the determinants of innovative activities. Section 4
examines the relation between innovation and firm growth, and the last section
concludes.

2 Data

The data source of this study is a survey conducted in 2003 by the Ethiopian
Development Research Institute (EDRI) on a 1000 microenterprises with 10 and fewer
workers. The survey was done in six selected major town: Addis Ababa, Awassa, Bahir
Dar, Jimma, Mekelle, and Nazreth. A total sample of 974 enterprises was interviewed
whereby 25 per cent of them are from Addis Ababa and almost 15 per cent each in the
other cites.2 Table 1 gives the distribution of the enterprises and characteristics of the
owners in our sample. The enterprises cover a wide variety of non-agricultural activities
such as trade, service, and manufacturing. The majority of them are engaged in trade
and service constituting 45 per cent and 36 per cent respectively. Manufacturing is also
an important component (19 per cent) of the microenterprises mainly covering
production activities such as wood and metal work, bakeries, and tailors.

Measuring the number of workers as the sum of working owners, paid and unpaid
workers in 2002 (one year before the survey), 69 per cent of the businesses have less
than five workers of which one-worker establishments constitute about 18 per cent.3
Firms that have 5–10 workers account for 30 per cent. Most of the enterprises are
young, whereby 45 per cent of them are five or less years old and 36 per cent 6–12
years old. Male-headed businesses account for 74 per cent, while only 22 per cent are
female-headed. The female-headed businesses tend to concentrate on activities such as
retail trading, beauty salon, bars and restaurants, and local drink brewing. The majority
of the owners are young, 59 per cent of them are less than 35 years old. The survey
instrument also includes the owners’ educational achievement. 32 per cent of the
owners have completed high school and 15 per cent have some college years, while 12
per cent are illiterate. About 15 per cent of the owners have also reported that they had
vocational training.

Our innovation indicator is a dichotomous variable that takes value one if the
respondent said yes for the question Did you make an important improvement/change to
your product/service recently?. As shown in the Table 2, about 34 per cent of the
enterprises said yes. Those, who responded yes were then asked to disclose what type of
improvement was involved. The lists of activities showed about 20 types. We
categorized them into main type of innovative activities such as product/service

2 The sampling frame was stratified by location and sector. Based on the population of microenterprises
six major cities were first chosen then the sample was distributed to the cities. Similar stratification
across sectors was also made based on the intensity of the sector activities such as manufacturing,
service, and trade. At last, a sample was taken randomly from each sector at each location.
3 In our calculation of employment we did not include causal workers, as about 80 per cent of the
establishments reported that they do not normally hire casual workers.

3
innovation (providing new/quality/better design or an increasing variety of products),
process innovation (machinery investment, improving or increasing business premises,
furniture, and equipment), organizational and skill improvement (improving the skill of
workers and managers), and marketing (more advertisement, shorter delivery time).
These activities are more or less incremental and consistent with the observation made
by Van Djik and Sandee (2002) on innovation in African small firms. 4

The magnitude of innovativeness differs slightly by sector. The manufacturing sector

has a higher propensity to innovate than the service sector and trade as shown by the
ratio of firms that reported had improved their products/services to the total number of
firms in each sector. But more importantly the sectors differ in the type of innovative
activity. The manufacturing sector is distinctive from trade and service in this context.
The innovation activities in the manufacturing sector are machinery improvement
(investment), better design, skill improvements. Service and trade sectors, on the other
hand, tend to concentrate on the improvement of business premises, provision of quality
and variety of products/service, and marketing.

Looking at the association between the innovative indicator and other variables defining
the characteristics of the owners and perceptions might give some guidance to the
empirical framework on the determinants of innovation. The survey instruments include
number of innovations related to owner perceptions particularly the relative status of
his/her business in terms of innovation. For example, how do you compare your main
product/services with that of your competitors in terms of quality material and
model/design? How do you characterize the enterprise’s machinery/equipment? The
survey instruments also include business environment perception and variety of owner-
firm attributes.

Table 3 presents the correlation coefficient between the innovation indicator and other
variables. Stars represent significance at 5 per cent or better. The innovative indicator is
positively associated with the owner perceptions such that his/her business has better
quality material and design than the competitors’ and use advanced machinery. The
innovators subgroup perception on business environment is also more optimistic than
the non-innovators group. The innovators’ indicator is positively associated (and
significant) with the current size, employment growth, investment, revenue increased,
have no market problem, and have planned to expand the business in the future. Among
the demographic characters (owner age, gender, and marital status), only gender is
found to be significantly associated with innovation activity and indicates women
owners are less likely to engage in innovative activity relative to male owners. We have
also tested for association between innovation activity and owner education and

4 ‘It was found that in the African case studies, everything the researcher did not expect, given the
traditional context and way of doing things can be called an innovation in the local context. This
means making a different product or a product of slightly better quality. “Innovation” would include
all the following: using different raw materials, or economizing on the use of raw materials or energy;
improving the design or introducing a new way to finance, distribute or stock products and changing
the management of a small business’.

4
experience. Owners with vocational training and some college years are more likely to
innovate, while illiterate owners less likely do so.

3 Determinants of innovation activities

3.1 The framework

Innovation was seen as a breakthrough or radical change characterized by marketing

and technological discontinuity and primarily produced by large firms and concentrated
markets as argued in the early works of Schumpeter. However, innovation is rarely a
dramatic breakthrough, rather small improvement in a new process or product—
incremental innovation (Blaug 1999). The realization of economic benefits from
‘radical’ innovations in most cases requires a series of incremental improvements. Thus,
the bulk of economic benefits come from incremental innovations and improvements
(Fagerberg 2005). Broadly defined innovative activities include introducing new
products/services, new design, and improved quality of products/services, installed new
equipment, changed sales methods, and improved working conditions. An innovation in
a small enterprise in the developing countries context is largely an adoption of a
product, process, or method that have already been adopted elsewhere but new to the
firm and not necessarily new to the world, region, country, or industry (Van Dijk 2002).

What factors determine innovativeness and innovation activities in small firms?

Hyvärinen (1990) identifies three broad (sometimes overlapping) determinants:
entrepreneurial attributes, firm level resources, and the environment in which the firm
operates.

Entrepreneurial character: In small businesses decisionmaking is concentrated in the

hands of owner-manager (Dyer and Handler 1994). Innovation activities of individuals
(owner-manager) form an important part of the innovation activities in small
enterprises. Thus, the smaller the enterprise, the nearer its innovative behaviour is to
that of an individual’s (the owner-manager) behaviour (Hyvärinen 1990). Any attempt
to investigate the innovation needs to consider an analysis of the characteristic of the
entrepreneur (Donckels and Fröhlich 1991; Hausman 2005). Rogers (1995) summarizes
characteristics of innovative entrepreneurs into three headings; socioeconomic status,
personality, and communication behaviour of which education, social status, age,
attitude toward risk and science, and density of social network in which the individual
participates, are among the long list of variables.

Various empirical studies have tested the effect of human capital and demographic
factors on innovation. Khan and Manopichetwattana (1989) and Hausman (2005)
showed that in the USA innovative firms are led by more educated executives or
owners. In Ghana, Robson et al. (2008) found that educated owners are more likely to
innovate. The experience of owners (level of skill and knowledge) is also an important
factor and has been found to affect innovation activities (Hausman 2005). Mahemba and
de Brujn (2003) and Robson et al. (2008) have also shown that training of workers is
associated with higher innovation. Khan and Manopichetwattana (1989) found that
firms led by in average younger owners, are proactive, risk taking, and more innovative.
So far, the relation between the owner’s gender and innovation activities has not been
empirically established. In the entrepreneur literature, however, there are a number of

5
evidences showing that women-headed firms grow slower than male-headed ones
(Liedholm and Mead 1993; McPherson 1996).

Firm level factors: Innovation activity occurs at firm level and the firm is a central
actor in processes of technological change (Romijn 2002). In the empirical literature
these resources are captured by firm size, age, access to finance, and network. The
relation between firm size and innovative activity is a longstanding debate since the
work of Schumpeter 1939.5 However, the empirical results so far are not conclusive
(Nootebom 1994; Ernst 2004). In this paper we are not pursuing this debate as our data
covers only the lower segment of firm size with 10 and fewer employees. But the size of
a firm could still impact innovative activities even within the microenterprises,
capturing differences in access to resources. Rogers (1995) indicated that early adopters
are the wealthier and have large sized units (farms, schools, companies, and so on).

Innovative spirit could be associated with the age of a firm in the sense that small firms
have higher innovative capacity in the first stage of a life cycle. In contrast, firm age
could also represent accumulated resource, market knowledge, and developed network
thus older firms are more likely to be involved in innovation activities. The empirical
evidence in Africa so far is mixed. Wignaraja (2002), Deraniyagaa and Semboja (1999)
found supporting evidence of positive relation between firm age and innovation, and
technological capability. Robson et al. (2008), however, found no significant relation
between firm age and innovation.

Innovation activities would take place more easily in clusters and networks (Van Djik
and Sandee 2002). Effective network that comprises lateral and vertical linkages raises
capacity for each node in the network by increasing exposure to ideas and opportunities.
They also reduce the transaction of developing and adopting innovations (Ernst 2004).
A voluminous empirical literature supports the role of clusters and networks on
innovation in Africa (Sverisson 1997; Oyelaran-Oyeyinka 2006; Chipika and Wilson
2006). Unfortunately, we do not have good approximation of network in our data thus
have not included a network variable in the empirical analysis. We believe that if such
effects exist then firm age might partly capture the impact of the network.

Based on this brief survey, the descriptive analysis in the previous section, and
availability of data we forward the following hypotheses for test.

• Hypothesis 3.1: Entrepreneurs with more formal education, technical/vocational

training, and/or longer previous experience are more likely to take up innovative
activity.

5 The debate mainly surrounds whether small or large firms are more innovative. Some argue
innovations are primarily produced by large firms and concentrated markets, while others claim that
small firms are more likely to innovate. The advantage of large firms on innovation is their deeper
level of specialization, science-based knowledge, economy of scale, larger and cheaper financial
resources, spread of risks. The strength of small firms on the other hand, lies on their flexibility,
greater motivation, tacit knowledge in unique skills, more informal communication along shorter
lines, less bureaucracy, greater proximity to market and to own production (Nooteboom 1994).

6
 • Hypothesis 3.2: Younger entrepreneurs are more likely to take up innovative
activity.

• Hypothesis 3.3: Male entrepreneurs are more likely to take up innovative

activity than female entrepreneurs.

• Hypothesis 3.4: Larger firms are more likely to innovate than smaller firms.

• Hypothesis 3.5: Older firms are more likely to innovate than younger firms.

3.2 Estimation and results

The discriminant analysis and logistic regression are widely applied for identifying the
attributes of innovative and non-innovative entrepreneurs/firms (e.g. Ostlund 1974; Kim
and Kim 1985; Oyelaran-Oyeyinka 2006; Moreno and Casillas 2007; Koellinger 2008;
Robson et al. 2008). The discriminant analysis, however, is based on a number of
assumptions that sometimes are difficult to justify. It requires assumptions such as
normal distribution, linear and homoscedastic relationships, untrancated interval or near
interval data, proper model specification, and if the dependent variable is a true
dichotomy among others. In contrast, the logistic regression requires no assumptions
regarding the distribution of the explanatory variables. It is relatively robust, flexible
and easily used, and it lends itself to a meaningful interpretation. Logistic regression is
preferred when data are not normal in distribution or group sizes are very unequal
(Pohar et al. 2004; Garson (undated)). Thus, in this paper we applied the logistic
regression to test the above hypotheses on the determinants of innovative activity.

The dependent variable ‘innovativeness’ is defined here as a categorical variable

indicating that the firm made important improvements/changes to its product/service
recently. The independent variables include firm size, firm age, and owner attributes
such as age, gender, previous business experience, general education and vocational
training. The size of the firm is measured by the current number of employees. Firm age
is the number of years since start and in logarithmic form. Business experience
represents the experience of the owner in business before starting the current business,
measured in number of months and in logarithmic form. The owner’s age is also in
logarithmic form. We made a distinction between general education and vocational
technical training. The general education is represented by a high school certificate and
some years of college education. This means the owner who did not complete high
school and the illiterate serve as the control category. Vocational training is a dummy
capturing if the owner had access to vocational training before or after the start of the
business. Two sectoral dummies representing manufacturing and service are included,
with trade as control category. Regional difference is also controlled with Addis Ababa
as a reference city.

Table 4 reports the logit estimation results. The size of the firm is positive and highly
significant. This means larger firms are more likely to participate in innovation activity.
The positive effect of size indicates the resource advantage of larger firms over smaller
ones and it is consistent with the theory of resource-based view and previous studies
(for example, Robson et al. 2008). Firm age is also positive and significant. But when

7
we include age square into the model we find a non-linear relationship (i.e. positive at
the first level and negative squared term, both significant) between innovation and firm
age (see Table 4, column 2). The concave relationship between firm age and
innovativeness suggests that innovative activity increases at an early age but tends to
decline beyond a certain age. The positive relation between age and innovation activity
at an early age might be due to accumulated business experience and market knowledge.
However, this advantage might not last long. The manufacturing dummy is positive and
significant, suggesting that manufacturing firms are more likely to engage in innovative
activity compared to the trade sector.

The coefficient of female owners is negative and significant suggesting female owned
enterprises are less likely to innovate in contrast to those owned by male. This is usually
explained by the fact that women owners are more family oriented and interested in
long term stability of the business, thus, tending to take less risk (Brush 1992), and as a
result less likely to engage in innovation activity. Women entrepreneurs also face more
operational and strategic impediments compared to male in their entrepreneurial pursuit
(Rutshobya 2001). Owner age is also negative and significant; the older the age of the
entrepreneur, the less likely they are to innovate. This is consistent with the Khan and
Manopichetwattana (1989) finding and might be explained by the fact that older
entrepreneurs are also more risk averse than young entrepreneurs. Among the human
capital variables only vocational training is found to affect innovation activity positively
and significant but neither general education nor previous experience. This lends
support to the belief that vocational (technical) training is more important than the
general education in promoting entrepreneurship.

In the last column of Table 4, we introduced employment growth measured by

cumulative change of employment in the firm between 2001 and 2003. This was to test
if there is any causation from growth to innovation. However, firm growth is not
significant while the effect of other variables remain the same. Robson et al. (2008)
have also found no clear relation from firm growth to innovation in their estimation of
innovation equation that includes firm growth. The effect might be the other way round,
i.e. from innovation to firm growth? This is the task of the next section.

4 Innovativeness and firm growth

4.1 The framework

The second objective of this study is to investigate if innovative firms grow faster than
non-innovative firms, in other words, if innovators create more jobs. Theoretically, new
technologies and processes are associated with a better utilization of resources, higher
quality of routine tasks and higher productivity. Companies that use innovative
technologies and processes can often offer qualitatively superior and/or cheaper
products, thereby enjoying higher growth potential (Minitti et al. 2006). McDaniel
(2000) also argues that those firms successfully master the timing and placement of
innovation development and innovation implementation in their respective industries
will be set to lead in profits market share and industry dominance. In the context of new
firms Geroski (1995) argues that the growth and survival prospects of firms will depend
on their ability to learn about their environment, and to link changes in their strategy
choices to the changing configuration of that environment.

8
Although, there are also counter-arguments that innovation might replace employment
most empirical studies in the developed world show that innovative firms are more
likely to grow, i.e. higher market share, profit, or employment, regardless of industry,
size, or other characteristics. Mansfield (1962) compared average annual growth rates of
innovators and non-innovators with comparable initial size in USA manufacturing and
found that the growth rate of innovators is about 4–13 percentage points higher than the
control group. Jones-Evans et al. (1996) found technologically innovative small and
medium enterprises (SMEs) in the UK have growth rates above the average regarding
assets, and expropriations. Moreover, such companies tend to have minor bankruptcy
rates. Koellinger (2008) examines the relationship between the usage of internet-based
technologies, different types of innovation, and performance at the firm level based on a
sample of 7302 European firms. He found that all considered types of innovation,
including internet-enabled and non-internet enabled product or process innovations, are
positively associated with turnover and employment growth but innovative activity is
not necessarily associated with higher profitability. Cho and Pucik (2005) conducted
research using data from the Fortune Reputation Survey and the Research Insight
Global Vantage. They found significant relationship between innovativeness and firm
growth and profitability. There is not much empirical work in Africa on the impact of
innovation on firm growth. Mahemba and de Bruijn (2003) found no clear relation
between innovativeness and firm growth of small manufacturing firms in Tanzania.

Besides innovative activity a number of other factors could also influence firm growth.
The stochastic theory of Gibrat’s law relates size distribution and firm growth and
argues growth is independent on size. However, the growing empirical literature shows
the contrary, i.e. a negative relationship between growth and firm size (Evans 1987;
Dunne and Hughes 1994; Bigsten and Gebreeyesus 2007). The inverse relation between
firm growth and size can be explained through the availability of slack resources
suggested by Penrose (1959). Such idle resources arise as a consequence of their
indivisibility. The extent to which a firm can employ the most advantageous division of
labour depends on the scale of its operations; the smaller its output the less can
resources be used in a specialized manner. The smaller the firm, the greater the
indivisibility of resources and availability of slack resources, thus higher the incentive
to expand.

In a life-cycle theory of firm Jovanovic’s (1982) passive learning model predicts

negative relation between firm age and growth. This is supported by several empirical
studies (for example Evans 1987; Dunne and Hughes 1994; Mcpherson 1996; Bigsten
and Gebreeyesus 2007). The active learning model by Ericson and Pakes (1995), on the
other hand, argue that firms investing in R&D and human and physical capital will be
more efficient and grow faster.

Lack of access to financial resources hinders firms from growing to their optimal size
(Holtz-Eakin et al. 1994; Elston 2002; Cabral and Meta 2003). Micro and small
enterprises are more likely to face liquidity problems as they are considered expensive
to be served thus less attractive to formal banks. Lack of finance is the most referred
complaint among entrepreneurs in Africa (for example Biggs and Srivastava 1996;
Bigsten et al. 2003). Other studies have also related firm growth to entrepreneurial
attributes such as owner age and gender (for example Liedholm and Mead 1993;
Mcpherson 1996; Davidson 1991; Davidson and Hoing 2003). The literature briefly
reviewed above leads to the following testable hypotheses.

9
 • Hypothesis 4.1: Innovative firms grow faster.

• Hypothesis 4.2: Growth is inversely related with the size and age of the firm.

• Hypothesis 4.3: Businesses with less access to finance grow slower.

• Hypothesis 4.4: Businesses owned by educated people grow faster than those
owned by the ones with lesser or no education.

• Hypothesis 4.5: Businesses whose owners had longer previous business

experience exhibit higher growth than those who did not have such experience.

• Hypothesis 4.6: Businesses run by younger owners grow faster than those run by
older owners.

In modelling the relation between innovation and firm growth, we start with the Evans
(1987) firm growth equation that relates growth with initial size, and age but augmented
by innovation indicator and other variables:

ΔS = β 0 + β1 ln( S 0 ) + β 2 ln At + β 3 INN + ∑ γ i X i + ut (1)

where ∆S and S0 represent the change of firm size and beginning size respectively, A
denotes firm age, INN innovation indicator (a dichotomy variable), and X indicates other
control variables (for example, owner characteristics such as education, experience, age,
and financial constraint), and u is the log-normally distributed errors term with mean
zero.

In the literature there are different measures of firm performance, such as growth in
sales, profits, market share, assets, and employment. In this analysis we confined
ourselves to employment growth basically due to the absence of sufficient sales and
assets variable in our data. The dependent variable employment growth is defined here
as the net change of employment between 2001 and 2003. The timing fairly matches
with our main explanatory variable, i.e. innovativeness that captures if the firm made
significant change/improvement in its product/service in recent years. On the right hand
side of the equation, size is measured by employment at initial year. Credit constraint is
defined as =1 if the firm reported that it needs credit but is unable to borrow due to
different reasons, and 0 otherwise. The other explanatory variables are defined in the
previous section.

4.2 Estimation and results

Estimating the above equation with ordinary least squares (OLS) might lead to
inconsistent results if one or more of right-hand side (RHS) variables are not exogenous
to the model. As we have shown in the previous section innovation is determined by
many of the variables in the model, such as entrepreneurial characteristics. We
performed a test for endogeneity of the innovation indicator using an augmented

10
regression test (DWH) and found that the innovation indicator is correlated with the
error term which makes OLS results inconsistent. There are different methods in
controlling the endogeneity problem; simultaneous equation, fixed effect, instrumental
variable (IV) method, etc. Given that our data is cross-section we use the IV method,
specifically the two-stage least square (2SLS). The disadvantage of the IV method is
that identifying a proper instrument is not easy. The requirement for proper instrument
is that it should be correlated with the instrumented (i.e. included endogenous
regressors) and at the same time uncorrelated with the error term.

Based on some experimentation of the data vocational training and owner gender are
found to satisfy the requirement for proper instrument. The first stage regression results
and test results are reported in Appendix table 1. First, the instruments are highly
correlated with the innovation variable. F-tests that the owner female and owner with
vocational school equals zero are also rejected. However, simply having an F-statistic
that is significant at the typical 5 per cent or 10 per cent level is not sufficient. Stock et
al. (2002) suggest that the F-statistic should exceed 10 for inference based on the 2SLS
estimator to be reliable when there is one endogenous regressor.6 The F-test statistic
from the first regression is 12.58 and exceeds 10—the rule of thumb (see Appendix
table 1). Hence, the instruments are strongly correlated with the endogenous regresssor.
We have also conducted a test overidentifying restriction (see Table 5).7 The Sargan
statistic of overidentification restriction can not be rejected implying the instruments are
valid. Confirming that our instruments are valid we now proceed to discuss the results.

Table 5 reports OLS, 2SLS, GMM, LIML estimation results. The 2SLS denotes a two-
stage least square estimation. GMM represents a generalized method of moment’s
estimator and generates efficient estimates of the coefficients as well as consistent
estimates of the standard error. LIML stands for limited information maximum
likelihood and the estimator may yield less bias and confidence intervals with better
coverage rates than 2SLS estimations. The OLS was included for comparison although
we showed that the innovation coefficient might be inconsistent due to its endogeneity.
The OLS result is, however, not qualitatively different from the others except the
magnitude of the innovation indicator is lower. The other estimation results are almost
identical even in terms of magnitude.

The innovativeness indicator is positive and significant in all the estimations. This
suggests that innovative firms grow faster than non-innovators, thus, innovative activity
predicts higher job creation. Other variables have also impacted employment growth.
Initial size and age of the firm are negative and significant in all estimations suggesting
that smaller and younger firms grow faster than their counterpart. Size and age often
have a non-linear relationship with firm growth. Of course, a non-linear relationship
between size and growth might not make sense in a small range of size such as the data
we have of firms with 10 and less workers. Thus, we estimated a non-linear relationship
between age and growth by introducing square of log firm age into the equation. The

6 For more discussion and examples on this see StataCorp Release 10, Reference I-P: 49.
7 The test of overidentifying restriction tests two things simultaneously: whether the instruments are
correlated with the error term and if the equation is mis-specified, i.e. one or more of the excluded
exogenous variables should in fact be included in the structural equation. Thus, a significant test
statistic could represent either an invalid instrument or incorrectly specified equation (StataCorp
Release 10, Reference I-P: 52).

11
results are reported in the last column of Table 5. There is indeed a convex relationship
between age and growth of employment with the first level taking negative sign and the
squared term positive, both significant. This means firm age is related with growth
negatively, but the negative relation diminishes with age. The negative segment
captures evidence of a learning process that was proposed by Jovanovic (1982) whereby
as a firm ages and grows more confident about its costs, the mean and variance of its
growth rate should decrease. This is also consistent with the previous findings. Bigsten
and Gebreeyesus (2007) found a convex relation in the Ethiopian manufacturing sector.
Evans (1987) reported that firm growth decreases with age for younger firms but is
roughly independent of age for older firms in US manufacturing.

Credit constraint is highly significant and negatively related with firm growth. This is
obvious given that 85 per cent of the firms in our sample have never received credit
from the formal market, such as banks and microfinance institutions. Consequently,
they largely depend on the informal network such as relatives and friends, and trade
credit. In all the estimations manufacturing is positive and highly significant. Hence,
manufacturing enterprises are more likely not only to innovate but also grow faster than
other sectors. Service gives positive coefficient but not significant.

Among the attributes of the entrepreneurs only owner age is found to be positive and
significant. However, the human capital variables such as owner previous experience
and owner education as measured by the dummy of high school certificate and above
are positive but not significant. Vocational training was also found to be insignificant
(not reported here).

5 Summary and conclusions

The aim of this paper was to address two prominent issues on the MSEs development in
Africa. The first is to show the factors that foster or constrain innovation and the second
examining if innovative enterprises create more jobs than non-innovators. We estimated
separate models of innovation and growth determinants. We used a logit estimation
method for the innovation model. In the growth equation we applied IV method to
address the endogeneity of innovation in the model. Appendix table 2 summarizes the
signs and significance level of the variables in both models.

Innovation activity is related with a number firm and entrepreneur attributes. Current
size is related positively with innovation activity. This means the larger in size the more
likely to involve in innovative activity. Resource advantage could explain why larger
firms are more innovative than smaller firms. This is consistent with most previous
studies. We found a non-linear (concave) relationship between firm age and innovation
activities. Innovative activity increases at early age but tends to decline beyond a certain
age. Our interpretation of the results is that the positive relation between firm age and
innovation activity at early age might be due to accumulated business experience and
market knowledge. However, this advantage might not last long, for example the
innovative sprit of firms might decline with age.

Among the human capital variables vocational training is found to have a strong effect
on the innovation activity. Unlike other studies (for example Hausman 2005), neither
general education (a measured by high school certificate and above) nor previous
business experience are affecting innovation in our data. This gives support to the

12
notion that technical skill is more important than general education in promoting
entrepreneurship and innovation.

Female owned firms are less likely to be involved in innovation activity. This might be
explained by the fact that women entrepreneurs are risk averse. Previous studies
indicated that women entrepreneurs are more family oriented thus less interested in
expansion of their businesses (Brush 1992). In developing countries women
entrepreneurs also face more constraints comparing to male in their entrepreneurial
pursuit. We have also found a negative relation between innovativeness and owner age
suggesting that younger owners are more likely to innovate than older ones. This is
usually an indication of the owners’ risk attitude.

The main contribution of this study to the MSEs literature in Africa is its extension in
examining the effect of innovativeness on firm growth while controlling a range of
other potential variables that could have an effect on firm growth (e.g. size, age,
financial constraint). We found strong evidence that innovators are more likely to grow
than non-innovators. On the other hand, we found no evidence of the reverse causation
(i.e. from growth to innovativeness). This supports the claim that innovations lead to
expansion of business and creation of more jobs. A focus on promoting innovation and
technological capability will, therefore, pay off not only through increasing MSEs
competitiveness but also by their ability to create more jobs.

Credit constraint affects negatively firm growth. This is obvious given that the financial
markets in Ethiopia are underdeveloped and most of the small firms rely on the informal
market for external finance. Policymakers, therefore, need to facilitate alternative
channels of access to finance for small firms. In both the innovation and growth
estimations a manufacturing dummy is found to be positive and highly significant. This
gives evidence of the superiority of the manufacturing sector as an engine of growth.

Other firm characteristics such as size and age of the firm have also been found to affect
growth. We found a negative relation between initial size (employment) and growth,
suggesting smaller firms at start tend to grow faster than larger ones. This is consistent
with the availability of the slack resources view suggested by Penrose (1959) and most
previous empirical findings. We found a non-linear (convex) relation between firm age
and firm growth. This means growth decreases with age until a certain point while the
relation turns positive beyond that. The negative segment captures evidence of the
learning process that was proposed by Jovanovic (1982) whereby as a firm ages and
grows more confident about its costs, the mean and variance of its growth rate should
decrease. This is also consistent with the previous findings (for example Bigsten and
Gebreeyesus 2007 in Africa, and Evans, 1987 in US manufacturing).

13
Appendix
Appendix table 1: Results of the first stage instrumental variables (2SLS) regression

Dependent variable dummy for innovative firms Coef.

log employment current 0.064***
log age 0.044***
log previous experience 0.002
log(owner age) -0.103**
Credit constrained 0.026
High school and above 0.001
Manufacturing 0.156***
Service 0.054*
Owner female -0.085***
Owner have vocational training 0.183***

_cons 0.409**

Adj. R-square 0.1267

F (2, 950) 12.58
Prob > F 0.00

Notes: regional location is controlled in the estimation. ***, **, and * denote level of significance at 1%,
5%, and 10% respectively.

Appendix table 2: Summary of the estimation results of innovation and growth equations

Variables Innovation Growth

Firm growth Insignificant Not included

Innovative firm dummy Not included +

log(initial employment) + -

log(firm age) + -

2
log(firm age) - +

log(previous experience) Insignificant Insignificant

log(owner age) - +

Female owned - Not included

Credit constrained Insignificant -

14
High school and above Insignificant Insignificant

Vocational training + Insignificant

Manufacturing + +

Service Insignificant Insignificant

15
References

Biggs, T., and P. Srivastrava (1996). ‘Structural Aspect of Manufacturing in Sub-

Saharan Africa: Findings from a Seven Country Enterprise Survey’. Discussion
Paper 346. Washington, DC: World Bank, Africa Technical Department Series.
Bigsten A., P. Collier, S. Dercon, B. Gauhier, M. Fafchamps, J. W. Gunning, A. Oduro,
R. Oostendorp, C. Pattillo, M. Söderbom, F. Teal, and A. Zeufack (2003). ‘Credit
Constraint in Manufacturing Enterprises in Africa’. Journal of African Economies,
12 (1): 104–25.
Bigsten, A., and M. Gebreeyesus (2007). ‘The Small, the Young, and the Productive:
Determinants of Manufacturing Firm Growth in Ethiopia’. Economic Development
and Cultural Change, 55: 813–40.
Blaug, M. (1999). ‘The Concept of Entrepreneurship in the History of Economics’. In
M. Blaug (ed.), Not only an Economist: Recent Essays. Cheltenham: Edward Elgar.
Brush, C. (1992). ‘Research on Women Business Owners: Past Trends, a New
Perspective and Future Directions’. Entrepreneurship Theory and Practice, 16 (4):
5–30.
Cabral, L., and J. Mata (2003). ‘On the Evolution of the Firm Size Distribution: Facts
and Theory’. American Economic Review, 93 (4): 1075–90.
Central Statistical Agency (CSA) (various years). Report on Urban Informal Sector
Sample Survey. Addis Ababa: various issues.
Chipika, S., and G. Wilson (2006). ‘Enabling Technological Learning among Light
Engineering SMEs in Zimbabwe through Networking’. Technovation, 26 (8), 969–
79.
Cho, H.-J., and V. Pucik (2005). ‘Relationship between Innovativeness, Quality,
Growth, Profitability, and Market Values’. Strategic Management Journal, 26 (6):
555–75.
Davidson P. (1991). ‘Continued, Entrepreneurship: Ability, Need and Opportunity as
Determinants of Small Firm Growth’. Journal of Business Venturing, 6: 405–29.
Davidson, P., and B. Honig (2003). ‘The Role of Social and Human Capital among
Nascent Entrepreneurs’. Journal of Business Venturing, 18: 301–31.
Deraniyagala, S., and H. Semboja (1999). ‘Trade Liberalization, Firm Performance and
Technology Upgrading in Tanzania. In S. Lall (ed.), The Technological Response to
Import Liberalization in Sub-Saharan Africa. London: Macmillan.
Donckels, R., and E. Fröhlich (1991). ‘Are Family Businesses Really Different?
European Experience from STRATOS’. Family Business Review, 4 (2): 149–61.
Dunne, P., and A. Hughes (1994). ‘Age, Size, Growth and Survival: UK Companies in
the 1980s’. The Journal of Industrial Economics, 42: 115–40.
Dyer, J. W., and W. Handler (1994). ‘Entrepreneurship and Family Business: Exploring
the Connections’. Entrepreneurship Theory and Practice, 19 (1): 71–84.

16
Elston, J. (2002). ‘An Examination of the Relationship between Firm Size, Growth, and
Liquidity in the Neuer Market’. Discussion Paper 15/02. Frankfurt: Deutsche
Bundesbank, Research Center.
Ericson, R., and A. Pakes (1995). ‘Markov-perfect Industry Dynamics: A Framework
for Empirical Works’. Review of Economic Studies, 62 (1): 53–82.
Ernst, U. F. W. (2004). ‘Hidden Sources of Growth? Looking at Microenterprises
through the Competitiveness Lens: A Review of Evidence’. Microreport 7.
Washington, DC: USAID.
Evans, D. E. (1987). ‘The Relationship between Firm Growth, Size and Age: Estimates
for 100 Manufacturing Industries’. The Journal of Industrial Economics, 35: 567–82.
Fagerberg, J. (2005). ‘Innovation: A Guide to the Literature. In J. Fagerberg, D.
Mowery, and R. Nelson (eds.), The Oxford Handbook of Innovation. Oxford: Oxford
University Press.
Geroski, P. A. (1995). ‘Entry, Innovation and Productive Growth’. The Review of
Economics and Statistics, 71 (4): 572–78.
Garson, G. D. (1988/2008). ‘Statnotes: Topics in Multivariate Analysis’. Available at:
http://www2.chass.ncsu.edu/garson/pa765/discrim.htm
Hausman, A. (2005). ‘Innovativeness Among Small Businesses: Theory and
Propositions for Future Research’. Industrial Marketing Management, 34 (8): 773–
82.
Holtz-Eakin, D., D. Joulfaian, and H. Rosen (1994). ‘Sticking it out: Entrepreneurial
Survival and Liquidity Constraints’. Journal of Political Economy, 102 (1): 53–75.
Hyvärinen, L. (1990). ‘Innovativeness and its Indicators in Small and Medium-sized
Industrial Enterprises’. International Small Business Journal, 9 (1): 64–79.
International Labour Office (ILO) (2002). ‘Women and Men in the Informal Economy:
A Statistical Picture’. Geneva: ILO.
Jones-Evans, D., P. Heydebreck, M. Lindqvist, E. Autio, and M. Fontes (1996).
‘Technology, Innovation and Enterprise–the European Experience’. Paper presented
at the workshop on the ICSB 41st World Conference, Stockholm, Sweden, 17–19
June.
Jovanovic, B. (1982). ‘Selection and the Evolution of Industry’. Econometrica, 50: 649–
70.
Khan, A. M., and V. Manopichetwattana (1989). ‘Innovative and Non-innovative Small
Firms: Types and Characteristics’. Management Science, 35 (5): 597–606.
Kim, L., and Y. Kim (1985). ‘Innovation in a Newly Industrializing Country: A
Multiple Discriminant Analysis’. Management Science, 31 (3): 312–22.
Kiggundu, M. N. (2002). ‘Entrepreneurs and Entrepreneurship in Africa: What is
Known and What Needs to be done. Journal of Developmental Entrepreneurship, 7
(3): 239–58.

17
Koellinger, P. (2008). ‘The Relationship between Technology, Innovation, and Firm
Performance–Empirical Evidence from E-business in Europe’. Working Paper ERS-
2008-031-ORG. Rotterdam: ERIM.
Liedholm, C., and D. Mead (1993). ‘The Structure and Growth of Microenterprises in
Southern and Eastern Africa: Evidence from Recent Surveys’. Working Paper 36.
Washington, DC: GEMINI.
Mahemba, C. M., and E. J. de Bruijn (2003). ‘Innovation Activities by Small and
Medium-sized Manufacturing Enterprises in Tanzania’. Creativity and Innovation
Management, 12 (3): 162–73.
Mansfield, E. (1962). ‘Entry, Innovation, and the Growth of Firms’. American
Economic Review, 52: 1023–51.
McDaniel, B. A. (2000). ‘A Survey on Entrepreneurship and Innovation’. The Social
Science Journal, 37 (2): 277–84.
McPherson, M. (1996). ‘Growth of Micro and Small Enterprises in Southern Africa’.
Journal of Development Economics, 48: 253–77.
Minitti, M., W. D. Bygrave, and E. Autio (2006). Global Entrepreneurship Monitor:
2005 Executive Report. Babson Park, MA: Babson College and London: London
Business School.
Moreno, A. M., and J. C. Casillas (2007). ‘High-growth SMEs versus Non-high-Growth
SMEs: A Discriminant Analysis’. Entrepreneurship & Regional Development, 19
(January): 69–88.
Nooteboom, B. (1994). ‘Innovation and Diffusion in Small Firms: Theory and
Evidence’. Small Business Economics, 6 (5): 327–47.
Ostlund, L. E. (1974). ‘Perceived Innovation Attributes as Predictors of
Innovativeness’. Journal of Consumer Research, 1 (2): 23–29.
Oyelaran-Oyeyinka, B. (2006). Learning to Compete in Africa Industry: Institutions and
Technology in Development. Farnham: Ashgate Publishing.
Penrose, E. (1959). The Theory of the Growth of the Firm. New York: Wiley.
Pohar, M., M. Blas, and S. Turk (2004). ‘Comparison of Logistic Regression and Linear
Discriminant Analysis: A Simulation Study’. Metodoloski zverzki, 1 (1): 143–61.
Robson, P. J. A., M. Haugh, and B. Obeng (2008). ‘Entrepreneurship and Innovation in
Ghana: Enterprising Africa’. Small Business Economics, 32 (3): 331–50.
Rogers, E. (1995). Diffusion of Innovations. New York: The Free Press.
Romijn, H. (2002). ‘Small Enterprise Development in Developing Countries:
Innovation or Acquisition of Technological Capability?’. In M. P. Van Djik, and H.
Sandee (ed.), Innovation and Small Enterprises in the Third World. Cheltenham:
Edward Elgar.
Rutashobya, L. (2001). ‘Female Entrepreneurship in Tanzania: Constraints and Strategic
Considerations’. Proceedings of the International Academy of African Business and
Development, 4–7 April, Washington, DC, USA.

18
Statacorp (2007). Stata Statistical Software: Release 10. College Station, TX: Statacorp.
Stock, J., J. Wright, and M. Yogo (2002). ‘A Survey of Weak Instruments and Weak
Identification in Generalized Method of Moments’. Journal of Business and
Economic Statistics, 20: 518–29.
Sverrisson, A. (1997). ‘Enterprise Networks and Technological Change: Aspects of
Light Engineering and Metal Working in Accra. In M. P. Van Dijk, and R. Rabellotti
(eds.), Enterprise Clusters and Networks in Developing Countries. Abingdon:
Routledge.
Van Dijk, M. P. (2002). ‘Innovation and Small Enterprise Development Examples from
Burkina Faso, Ghana and Zimbabwe’. In M. P. Van Djik, and H. Sandee (ed.),
Innovation and Small Enterprises in the Third World. Cheltenham: Edward Elgar.
Van Djik, M. P., and H. Sandee (2002). ‘Innovation and Small Enterprise Development
in Developing Countries’. In M. P. Van Djik, and H. Sandee (ed.), Innovation and
Small Enterprises in the Third World. Cheltenham: Edward Elgar.
Wignaraja, G. (2002). ‘Firm Size, Technological Capabilities and Market-oriented
Policies in Mauritius. Oxford Development Studies, 30 (1): 87–105.

19
 Table 1: Some characteristics of the microenterprises and owners

N %
Trade 439 45
Sector
Service 349 36
Manufacturing 186 19
1 worker 172 18
Firm size category 2–4 workers 493 51
5–10 workers 302 31
>10 workers 6 0.6
<= 5 years 439 45
Firm age category 6–12 years 347 36
13–29 141 14
above 29 47 5
Owner gender Female 226 22.3
Male 722 74
<=25 185 19
Owner age group 26–35 392 40
36–50 247 25
50 & above 150 15
Illiterate 113 12
Elementary 287 29
Owner education Some high school 119 12
High school complete 310 32
Some college years 145 15
Have vocational training 182 19

20
 Table 2: Type of innovative activities by sector

Frequency by sector
Broad category
If yes, what improvements/changes? of innovation Trade Service Manufacturing All
Improve quality product/service Product/service 20 20 6 46
Provide new products/service Product/service 9 13 3 25
Better design Product/service 9 12 18 39
Increase variety of products/services Product/service 12 5 17
Install additional machinery Process 1 4 9 14
Introduce modern machinery Process 6 7 15 28
Additional business premises/house Process 5 4 2 11
Additional utensils/furniture/equipment Process 15 16 3 34
Renovation Process 13 14 3 30
Improved production capacity Process 12 3 5 20
More advertisement Marketing 11 3 4 18
Shorten delivery time Marketing 1 3 1 5
Discount Marketing 2 2
Organization &
Accounting system skill 1 1
Organization &
Managerial skill skill 3 2 5
Organization &
Hired skilled worker skill 1 1
Organization &
Skill improvement skill 2 7 9 18
Organization &
Additional business partner skill 1 1
Organization &
Expanded the business skill 12 1 5 18

Total number of firms reported recent

important improvement 131 118 84 333
Total number of firms in the survey 439 349 186 974
% of firms reported recent important
improvement 0.30 0.34 0.45 0.34

21
 Table 3. Correlation coefficients between innovativeness and owner attributes

Correlation with improve/change products/services recently

Correlation Correlation
Variable coefficient Variable coefficient
1) Perception on innovativeness 3) Demography
Higher quality than competitors 0.1795* Owner female -0.0643*
Higher design than competitors 0.0744* Owner age -0.0174
Advanced machinery 0.0992* Married -0.0025

2) Perception on business environment 4) Education/experience

Revenue increased 0.0930* Illiterate -0.0750*
Have no market problem 0.1500* Elementary -0.0249
Have plan to expand 0.1610* High school 0.0127
Need credit but did not get any from
formal source -0.0029 High school complete 0.0087
Employment change 2001–03 0.1035* Some College 0.0770*
Current size category 0.1383* Received vocational training 0.1465*
Cumulative investment 2001–03 0.0866* Experience in business 0.0211

Note: * denote significance at 5% or better.

22
 Table 4: Determinants of innovation activities logit estimation

Dependent variable dummy for innovation activity

(1) (2) (3)
Firm attributes
∆employment (2001–03) 0.012
(0.057)
Current employment 0.153*** 0.155*** 0.153***
(0.0340) (0.034) (0.036)
log(firm age) 0.194** 0.682*** 0.687***
(0.0886) (0.198) (0.199)
2
log(firm age) -0.100*** -0.100***
(0.037) (0.037)
Owner attributes
log(owner age) -0.706*** -0.851*** -0.851***
(0.268) (0.274) (0.274)
Owner female -0.388** -0.383** -0.381**
(0.184) (0.185) (0.185)
log(previous business experience) -0.00943 -0.002 -0.002
(0.0430) (0.043) (0.043)
Owner high school complete and above -0.0505 -0.032 -0.032
(0.168) (0.169) (0.169)
Owner have vocational training 0.871*** 0.872*** 0.870***
(0.204) (0.205) (0.205)
Sector (control category trade)
Manufacturing 0.678*** 0.687*** 0.682***
(0.202) (0.203) (0.204)
Service 0.223 0.265 0.265
(0.172) (0.174) (0.174)
Constant 0.262 0.266 0.264
(0.931) (0.938) (0.938)
Region controlled Yes Yes Yes
966 966 966
Observations
Pseudo R2 0.13 0.136 0.136

Note: figures in parentheses are standard errors, and *** p<0.01, ** p<0.05, * p<0.

23
 Table 5: Firm growth and innovation

Dependent variable net change of employment (2001–03)

Variables (1) (2) (3) (4) (5)
OLS IV-2SLS8 IV-LIML IV-GMM IV-2SLS
Innovative firm dummy 0.209** 1.297** 1.322** 1.371** 1.352**
(0.0973) (0.644) (0.653) (0.664) (0.658)
log(initial employment) -0.0984 -0.163** -0.164** -0.172* -0.174**
(0.0694) (0.0820) (0.0824) (0.0942) (0.0834)
log(firm age) -0.123** -0.166*** -0.167*** -0.166** -0.427***
(0.0502) (0.0592) (0.0595) (0.0742) (0.140)
2
log(firm age) 0.0566**
(0.0250)
log(previous experience) 0.0491** 0.0434 0.0433 0.0427 0.0393
(0.0250) (0.0267) (0.0268) (0.0263) (0.0270)
log(owner age) 0.320** 0.429** 0.432** 0.432** 0.503***
(0.154) (0.175) (0.176) (0.192) (0.184)
Credit constrained -0.423*** -0.451*** -0.452*** -0.452*** -0.462***
(0.0990) (0.106) (0.107) (0.119) (0.107)
High school and above 0.151 0.107 0.106 0.0959 0.0961
(0.0958) (0.105) (0.105) (0.0996) (0.105)
Manufacturing 0.659*** 0.489*** 0.485*** 0.469*** 0.479***
(0.121) (0.166) (0.167) (0.169) (0.167)
Service 0.236** 0.165 0.163 0.153 0.146
(0.100) (0.112) (0.113) (0.109) (0.114)
Constant -0.448 -0.871 -0.881 -0.881 -0.882
(0.552) (0.641) (0.644) (0.679) (0.644)
Regions controlled yes yes yes yes yes
Observations 974 966 966 966 966
Overidentification test
a b c a
Sargan statistics (Chi-square) 0.551 0.550 0.611 0.599
p-value 0.458 0.459 0.434 0.439

Notes: Figures in parentheses are standard errors, and ** p<0.01, ** p<0.05, * p<0.1
a
The reported Chi-square and p-value are of Sargan-statistic
b
The reported Chi-square and p-value are of Anderson-Rubin-statistic
c
The reported Chi-square and p-value are of Hensen’s J-statistic..

8 Instrumented: innovation indicator; included instruments: log(size), log(age), log(experience), log(owner

age), credit constraint, education, sector, and regions; excluded instruments: owner female, vocational
training.

24