Analysis of Firm Growth in Ethiopia: An Exploration of the High-growth Firms

Transcription

1 Analysis of Firm Growth in Ethiopia: An Exploration of the High-growth Firms 1 Guta LEGESSE Department of Economics, Addis Ababa University guta.legesse@au.edu.et Abstract The inquiry about determinants of firm growth and the role of firms in an economy has been vastly studied in the literature. The concept of high-growth firms (HGFs), their role and determinants, however, is a recent phenomenon. This research is undertaken with the aim of identifying the incidence of HGFs in Ethiopia with the corresponding business obstacles and growth determinants. The research was based on data from the World Bank s Enterprise Survey dataset of The survey covered 848 firms distributed over the six major regions of the nation namely Addis Ababa, Oromia, Amhara, SNNP, Tigray and Dire Dawa. The analysis was done using OLS and QR. Concerning the incidence of HGFs, they are found to be overpopulated in the capital city and the service sector while the medium sized firms dominate the population of HGFs. Alike the non HGFs, access to finance is the biggest perceived obstacle to HGFs. For the HGFs, tax rates are found to be the biggest obstacle next to finance compared to informal sector activities for the non- HGFs. Region-wise, access to finance is the key problem only for firms operational in Addis Ababa and Tigray while practices of informal sector dominates in Oromia region. In Amhara region, corruption is found to be a top ranked obstacle. The econometric estimation results show that firm growth is negatively related to firm size, and export engagement while it is associated positively with firms product and process innovation, firm resources and ownership of website. The research failed to show any significant difference among firms growth based on gender of ownership, competition, capacity utilization and nationality of ownership. The heterogeneity in business obstacles across regions and firm growth performance can be taken as important lessons for policy intervention. Key words: High-growth firms, business obstacles & quantile regression 1 The researcher would like to thank the World Bank Enterprise Survey team for granting him access to the data. 0

2 1. Introduction The process of firms growth has long attracted the attention of economists. According to Sutton (1997) the work of Robert Gibrat was the first formal model dealing with the dynamics of firm size and industry structure. According to this law, the rate of firm growth is independent of its size and is framed as the law of proportionate effect (LPE) (Gibrat, 1931) cited in (Sutton, 1997). This law stipulates that the capacity to grow is the same for all firms, regardless of their initial size. Several empirical works have been done in this regard with somewhat inconclusive findings. Following the well documentation of the role of entrepreneurial firms in employment creation and wealth generation, more recent studies have turned their attention to the prevalence and determinants of high-growth firms in addition to measurement and definition issues. Several alternative measures have been suggested to classify firms as high-growth firms by researchers with employment being the most studied output variable although, productivity, sales, wages, and revenue were also used as indicators (Daunfeldt, Elert, & Johansson, 2013a). Attempts to identify prevalence of HGFs in different countries and industries have shown that HGFs are only a small percent of all firms and are found in all countries across all industries. A meta- analysis by Henrekson and Johansson (Henrekson and Johansson, 2009), for instance, failed to show any evidence in support of the view that HGFs are overrepresented in high-technology industries. In their survey, they noted that there are more HGFs in service industries relative to other sectors such as manufacturing. Furthermore, understanding of the persistence and incidence of HGFs has become an important task to policy makers as better insights into the existence, characteristics and stimulating factors of high-growth firms could be a key breakthrough to sustain economic growth. To the shareholders, the concern is to know stimulates the growth of their firm while to the policy makers it is the issue of sustaining firm growth and capitalize on incidences of HGFs. A new initiative of research has been undertaken to know if HGFs can be sustained. The aim of these research works is to find out if firm growth can be sustained for a long period of time and whether firm growth is a random process. They wanted to know if the probability of repeating the high-growth rates was high. We know that governments spend considerable amount of money to support specific types of firms based on either firm size and/or industry type to encourage them grow. It would be difficult to target policies towards certain groups of firms if growth is 1

3 unsustainable. The two dominant empirical works in this regard are the works of Daunfeldt & Halvarsson (2014) & Hölzl (2014) who argue that high-growth firms are one hit wonders and the probability of repeating the high-growth rates is very low. Despite such findings, the role played by HGFs is well documented. The role of business environment in deterring firm performance is not a well-studied phenomenon. Firms have heterogeneous abilities and entrepreneurs could perceive environmental challenges differently. For firms operating in different regions and sectors, the effect of the obstacles could vary and this is another dimension of the current research. The purpose of this research is to provide an insight on the incidence of high-growth firms (HGFs) in Ethiopia by firm characteristics (such as size, age, location ownership type, and industry type). Furthermore, the research tries to explore the perceived obstacles to firm performance and growth determinants of these firms. In general, HGFs have attracted considerable attention from researchers, policy-makers and also from practitioners. This research would add to the literature by investigating the incidence of highgrowth firms & business obstacles by region, industry type and the size growth relationship using the ES data on Ethiopia. As far as the knowledge of the researcher is concerned, this research is the first of its kind in Ethiopia. The rest of the paper is organized as follows. Section 2 presents review of related literature while Section 3 presents data description. Method issues are discussed in Section 4. Results are discussed in Section 5 while summary and conclusion are presented in Section Literature Review Firms have long been recognized as one of the determinants of economic growth and the factors affecting their performance have attracted lots of researchers among which the work of Robert Gibrat is recognized as the first formal model dealing with the dynamics of firm size and industry structure (Sutton, 1997). His work has been called Gibrat s Law and it states that the rate of firm growth is independent of its firm size although empirical studies conducted later have predominantly rejected it. In his work, Gibrat challenged the traditional economic theory which argues that firm size is inversely related to its growth rate. The fundamental belief in such arguments is that large 2

4 firms already operate close to the optimum level and may not need to grow while small firms would be far below the optimum size and would need to grow faster (Awoyemi, 2011). Firm growth shall be viewed as resulting from the continuous discovery and use of productive knowledge which requires institutional framework that determines the incentives to acquire and utilize knowledge (Henrekson & Johansson, 2010) The role and prevalence of high-growth firms There are an increased interest form academicians and policy makers on the prevalence of HGFs in an economy. Some of the questions they have tried to address include predominant size, age, industry type and region of HGFs. The role of HGFs in the job creation process has been examined in a large number of empirical studies and most of them showed that job creation is very much accounted by few firms. Several recent researches have strongly verified the role played by HGFs in terms of job creation (Storey, 1994; Autio et al., 2000; Schreyer, 2000; Davidsson and Henrekson, 2002; Delmar et al., 2003; Acs et al., 2008; Henrekson and Johansson, 2010; Nesta 2011 & 2009; Anyadike-Danes et.al, 2013; Daunfeldt et al. 2013b; Coad et al. 2014, Moreno & Coad, 2015). Coad et al. (2014), for instance, presents the disproportionate job creation role of HGFs as a stylized fact. The recent work by Daunfeldt et al. (2013b) showed that the 6% fastest growing firms in the Swedish economy contributed to 42% of the jobs in Sweden during Nesta (2009) documented that the 6% HGFs in UK generated 49.5% of all new jobs created by operational firms in UK during while Storey (1994) found that 4% of firms create 50% of the jobs. Although the role of HGFs may depend on how they are measured, Daunfeldt et al. (2013a) found that they play key role in the economy as sources of economic growth, employment growth, sales & productivity growth. In terms of employment growth, for instance, Daunfeldt et al. (2013a, P.353) The one percent fastest-growing firms in terms of absolute employment growth, absolute sales growth, and absolute value added growth contribute to more than 100 % of the net growth in total employment. 3

5 2.2. Determinants of firm growth Several researches have been done to address the question of what determines firm growth. Moreno & Coad, (2015) presented two types of theoretical explanations of firm growth determinants where one relates to dynamic strategic choices within the firm while the other considers growth as pure random. Other recent students have tried to classify determinants of firm growth in to firm size firm age, firm innovation and capabilities, entrepreneurship characteristics, and resources. Proponents of strategic choice theory argue that firm s output will depend on the owner s behaviour, which is determined by their knowledge, skills and ability to access and capitalize on key resources. This theory relates to the contribution of human capital in the form of formal education, experience (industry, managerial and/or prior business experience). It is proposed that human capital and firm resources together with entrepreneur specific capabilities will allow some entrepreneurs to enter profitable niches and enjoy sustained superior performance compared to others (Moreno & Coad, 2015). According to this explanation, HGFs can be seen as skilled firms with the ability to identify entrepreneurial opportunities to create a competitive advantage. The second argument about determinant of firm growth argues that growth is a product of random events. They argue that patterns that are identified in stochastic methods are confused and used to fit a specific theory of convenience and hence believe that it would be difficult to fully understand the systematic drivers of sustained superior performance unless the effect of randomness is known in a large population of firms (Henderson et al., 2012) Empirical work of Goedhuys & Sleuwaegen (2009) who classified the determinants of firm growth in to firm characteristics, technological characteristics and resources could fall under this theory. Firm characteristics refers to variables such as firm age & size, sex of entrepreneur, education of top management etc. while resources refer to firm level resources to deal with constraints arising from poor infrastructure, insecurity, and financial constraints. In this study they found that firm size and age are inversely related to employment growth. They also found that entrepreneurial characteristics such as education and country of origin have significant effect of firm growth. Generator ownership and own transportation are found to have positive association with growth while product innovation had effects on the upper quantiles of growth distribution. 4

6 2.3. Business environment and firm performance The role of business environment for firm growth and improved performance has been an interest for policy makers and the entrepreneurs. The world Banks publication of Doing Business has been widely used to give a general picture about business environment of an economy. Policy makers have been advocating reforms that would improve their country s ranking. Nguimkeu (2013) investigated the main barriers of doing business in Cameroon using ES data of 2009 on retailing firms identified. His findings show that, taxation, illicit trade, lack of infrastructure, lack of access to credit, administrative delays and incompetence of labour are the major obstacles in Cameroon for retailing firms. Using a structural econometric analysis, he was able to show that business climate factors reduce the annual gross margins of domestic traders significantly. From their study on prevalence and determinants of high-growth enterprises in 11 SSA countries, Goedhuys & Sleuwaegen (2009) show that electricity and access to finance are reported as the major constraints in all surveyed countries among the listed elements of business environment. Hallward-Driemeir & Aterido (2007) did a comprehensive study on the role of business environment in sub-saharan Africa relative to the rest of the developing world using the World Bank s ES data of They found that employment growth in the region is relatively concentrated in the smallest firms. According to their findings, medium and large firms grow less rapidly compared to other parts of the world. This could be due to the fact that, firms in Africa are found to face greater challenges in accessing finance, reliable infrastructure services and other public services deemed crucial for growth which may hinder growth of large firms relative to the small firms. 3. Methods 3.1. Defining and measuring high-growth firms Analysis of prevalence and determinants of HGFs could not be done without setting working definition of HGFs. There are several approaches used in this regard although the following four and their derivatives are widely used in the literature. i. Top 1% or 5% firms in terms of revenue, employment, profit, labor productivity as measured in growth rates, absolute change, log changes, index etc. 5

7 ii. Firms with 20 or more employees for the period under investigation (GEM, 2007) iii. Firms with annualized growth rate of at least 20% over 3 years period and at least 10 iv. employees ( Eurostat- OECD, 2007) Establishments which have achieved a minimum of 20 per cent sales growth each year over the interval, starting from a base-year revenue of at least $100,000 Birch (1987) cited in (Du, Gong & Temour, 2013, Henrekson & Johansson, 2009). Earlier estimates of high-growth firms in the literature used to define HGFs as the share of firms with the highest growth during a particular period, for instance, the 1% or 5% of firms with the highest growth rate. The problem with this approach is that it is difficult to create a consistent time series data of high-growth firms because the threshold that defines the top firms is higher during the expansion phase of the business cycle than during the contraction phase. It is also inconvenient to compare the share of HGFs across time or across countries. Later on, the original proposition by Birch has been dropped and a new index called the Birch Index has been introduced as an alternative measure of firm growth (Coad et.al. 2014, Hölzl, 2011, Schreyer, 2000). The Birch index tries to correct the inherent bias of using absolute and relative measures of growth since several studies have documented that small firms exhibit larger relative growth rates of employment while bigger firms show larger absolute growth rates. The Birch-Index considers both the relative and the absolute employment growth rate and is based on a multiplicative combination of the absolute growth rate and the relative growth rate. The value of this index for is calculated as (Coad et.al. 2014, Hölzl, 2011); [1] BI= Employ t( 2014) Employ t( 2010) ( ) ( ) [2] BI= l ETHl2a Under this index, firms can be classified as HGFs by deciding on the cut-off point to be used such as firms with BI values of top 1%, 5% and 10%. Some studies define the 10% of firms with the highest Birch index as high-growth firms (Schreyer, 2000; Garcia and Puente, 2012). For the investigation at hand, although one or a combination of these approaches can be used, customizing the criteria is required due to availability of data and the economic situation of the country under investigation. Application of the GEM approach does not really show potential of growth of firms since it ignores number of years required to reach the threshold employment level. On the other hand, the threshold levels of growth rates and initial employment recommended by 6

8 OECD needs to be adjusted by considering the fact that there are limited numbers of entrepreneurial firms in Ethiopia. According to the findings by Daunfeldt et al. (2013), the OECD criteria would exclude close to 95 % of all surviving firms in Sweden over the period and about 40 % of all created private jobs. Similarly, based the ES data on Ethiopia, the standard Eurostat- OECD definition of HGFs will exclude more than 95% of the firms in the sample. Based on the works of Goedhuys & Sleuwaegen (2009), the threshold level of initial size is firms with at least five employees and the growth rate is calculated for four years owing to data availability problem from while the threshold is set to be a minimum of 10 percent average growth rate per annum. Accordingly, high-growth firms are firms with annualized growth rate in excess of 10 percent, over the period & with at least 5 employees in For the Birch Index measure of HGfs in this study, owing to the low incidence of HGFs in Ethiopia and in order to generate comparable number of HGFs to the Eurostat- OECD, top 20 % firms were used. The ES of the World Bank reports sales data of all firms only for two years (i.e. the year 2012 & 2014) leading to too narrow measurement of firm growth in terms of sales. Therefore, growth of an establishment measured by sales growth is ignored for this study as the survey does not report sales data for the year Using the relative measure of growth, 137 firms were classified as HGFs while only 109 firms emerge as HGFs using BI. The number of HGFs further decreases to 86 and 56 if one adopts top 15 % and 10% cutoff points in the BI, respectively. Similar to the Eurostat- OECD, a 10 % cut-off point on the BI will exclude some 90 % of the sample firms. In this analysis, fast growing firms selected with the modified Eurostat-OECD definition as (HGF) and firms selected on the basis of the modified Birch Index as (BHGF) were used Measuring Business Obstacles Concerning business obstacles, there are two groups of questions presented in the questionnaire. The first group of questions asks about severity of an obstacle on a Likert scale question format by listing each obstacle separately. Establishments are then asked to express their perceptions about the magnitude of the obstacle paused by the elements of business environment with 0 score implying it is not an obstacle at all while a score of 5 implies that it is a very severe obstacle. The 7

9 second type of question, however, asks firms to select the single most important obstacle among a list of possible challenges. In the second approach, they are expected to compare obstacles and select the one they believe is the biggest obstacle relative to all listed obstacles while in the first approach they are expose to one challenge at a time and asked to state if it is an obstacle or not. Since the sampling design for the World Bank Enterprise Surveys is a stratified random sampling, individual observations should be properly weighted when making inferences about the population. Under stratified random sampling, unweighted estimates are biased unless sample sizes are proportional to the size of each stratum. This is important because individual observations may not represent equal shares of the population. In order to identify key business obstacles, the analysis is based on percent of firms who reported the listed elements as major or severe obstacle (score of 3 or4) from the first group of questions. In an effort to identify the number one perceived obstacle among the given list of challenges, the frequency with which a given obstacle is selected by firms as its biggest obstacle is computed the (i.e. response to question m1a in the questionnaire) Modeling determinants of firm growth The data analysis involves the use of both descriptive and econometric techniques. The descriptive analysis is used to explore the distribution of high-growth firms (HGFs) in Ethiopia using firm characteristics and other relevant factors. Although several researchers have modeled the determinants of firm growth differently, the empirical model for this research is based on the works of Goedhuys & Sleuwaegen (2009) who modeled firm growth as a function of firm age and size after controlling for other relevant factors which they classified in to three major categories as firm characteristics, technological characteristics and firm resources. Firm characteristics refers to variables such as firm age & size, sex of entrepreneur, education of top management etc. while resources refer to firm level resources to deal with constraints arising from poor infrastructure, insecurity, and financial constraints. Furthermore, the nature of a firm concerning export status, licensing technology from foreign-owned companies, ownership of a website & delivery of training were used as a proxy for firm technological characteristics. 8

10 Owing to data availability and high rate of non-response in some of these variables, some of these characteristics have been dropped and other new variables were included. The model estimated is given by equation [4] below. [3] Firm growth = f(firm age,firm size & firm resources,technological & market characteristics & other dummies) [4] Firm growth = a 0 + a 1 (Employment 2010 ) + a2 (Employment 2010 )² + a3 (Firm age) + a4 (Firm Age)² + a5 (Employment 2010 )*(Firm age) + Σ b (Entrepreneur characteristics) +Σ c (Technological & Market characteristics) + Σ d (Resources) + Σ f (Industry dummies) + ε i Given several approaches of measuring high-growth firms (HGFs), in this research, the two most frequently used ones were adopted. These are the modified Eurostat-OECD definition and the modified Birch Index. To measure firm growth using the modified Eurostat-OECD definition the logarithmic difference in the number of employees over a 4-year period is used i.e. [5] 4=ln, ln (, ) where 4 is the growth rate for firm i, and, &, are firm size measured by the number of employees in 2014 and 2010 respectively. Quantile regression is preferred to OLS to estimate the results in this study due to the fact that the OLS estimates how the mean of the (conditional) distribution of firm growth rates changes systematically with its covariates assuming a well-shaped normal distribution of growth around the mean. In other words, it provides the marginal effect of the explanatory variables at the mean of the growth distribution (Goedhuys & Sleuwaegen, 2009). Quantile regression, on the other hand, estimates the effects of the different explanatory variables at the different quantiles of the growth distribution. Since the HGFs are located in the extreme tail of the conditional growth distributions, factors that affect the upper deciles can be considered as factors that generate significant number of high-growth firms. Use of quantile regression avoids regression to the mean and shows the marginal effects at various deciles of the growth distribution. 9

11 4. Data 4.1. Data Source This research is based on the World Bank s Enterprise Survey (ES) data on Ethiopia for the year 2015 which was a sample survey conducted using stratified random sampling with industry, establishment size, and region representing the three levels of stratification used. The survey covered 848 firms which include micro, small, medium and large firms. For the purpose of this study the 26 micro firms have been excluded owing to their insufficient representation ending up with a total of 822 firms. Further cleaning of the data by considering firms with positive employment history in 2010 (to calculate growth rate over four years), dropping firms with no/error response to employment size and defining outliers in employment data as observations that are more than three standard deviations away from the mean in 2014 to purge out the effect of few outliers, leading to 547 firms. After removing the outliers, nearly 97% of the enterprises had employees. A bunch of questions were asked in the questionnaire to capture important dimensions of firm performance, infrastructure availability and business obstacles they face. The questionnaire has 14 major components with relevant sub sections for each. It starts by asking respondents control information (biography) on firm size, size of locality, industry classification and region of operation. The general information section asks issues related to ownership type and sex of top manager while the next section raises questions related to infrastructure and services. Questions related to sales and supplies, degree of competition, innovation, capacity utilization, land and permits, incidence and cost of crime, sources of finance, business-government relations, labor, business environment and firm performance are all integral part of both questionnaires. The questionnaires distributed to manufacturing firms and the service sectors have comparable contents with some minor differences. The survey covered firms operating in the six major geographic regions of the nation namely Addis Ababa, Oromia, Amhara, SNNP, Tigray and Dire Dawa while the size stratification was defined in to small if employment is between 5 to 19 employees, medium if employment is between 20 to 99 employees, and large if the firm has more than 99 employees. Half of the sample firms are operational in Addis Ababa with Oromia and Tigray hosting 15% of the sample firms each. Dire Dawa represents the smallest number of firms while Amhara and SNNPR accounted for about 8% of the sample firms each. 10

12 The survey was conducted for all categories of businesses in the sample region. Two questionnaires were used in the survey with common questions (core module) and additional questions to capture manufacturing and services sector specific issues. The distribution of the sample by industry classification shows that the highest number of enterprises are from wholesale (16%) followed by food industry (11%). The retail trade sector accounted for the 3 rd highest number of firms in the sample with 11 %. In terms of gross classification in to service and manufacturing, 56% of the firms belong to the service sector while the remaining 44% to the manufacturing sector. Considering their size, the small firms account for just over half (51%) while the remaining half is accounted for by medium (33%) and large (16 %) firms. The key difference between data on employment (labour) and sales is that while the employment data is reported for four years period ( ) the sales data is available only for two periods ( ). Among the firms with positive employment data in 2010 & 2014, the number of firms with invalid response to the sales question is very high (128 firms for the 2014 & 2012 year data) decreasing the number of eligible firms further to 419 which is less than 50 % of the sampled establishments. 5. Empirical Results 5.1. The prevalence of HGFs Using the two measures, two cohorts of HGFs have been identified. The Eurostat- OECD classified 137 firms as HGFs while from the BI there are 109 HGFs. Compared to the BI, the Eurostat- OECD measure identified some 25% of the surveyed firms as HGFs while the BI shows that some 20% of the firms can be considered as HGFs in Ethiopia as presented by Table 1 below. The Relaxation of the assumptions in the Eurostat- OECD measure could lead to different level and type of HGFs. Using the standard Eurostat-OECD definition of 20% annualized growth rate and minimum of 10 employees at the start of study period, only 6% of the sample firms emerge as HGFs. These results are consistent with works of Petersen & Ahmad (2007) and Goedhuys & Sleuwaegen (2009). 11

13 Table 1: Comparison of high-growth firms by measurement type (%) BHGF HGF 0 1 Total Total Source: Enterprise Survey 2015 (World Bank) Irrespective of the type of measurement, 369 firms (over two-third of the establishments) are nonhigh-growth firms. On the other hand, more than 50% of the HGFs identified through the relative criteria remain HGFs when evaluated using the Birch Index while some 86% of the HGFs identified using the Birch Index remains in the same category when the Eurostat- OECD measure is used. This result is consistent with previous research findings claim that different measures of HGFs lead to different firms being selected as high-growth firms. The two cohorts of HGFs identified in Ethiopia according to this study are nearly the same in terms of age with mean age of around 12 years compared to the mean age of the non-high-growth firms which is found to be close to 15 years and 14 years for the whole firms. Under both measures, HGFs are found to be younger by 3 years on average than the non- HGFs. Concerning ownership structure, the Eurostat- OECD measure identifies around 53 % of the HGFs from the sole ownership type while 25% of them are operational under the limited partnership form of ownership. The Birch Index, on the other hand, shows that some 70 % of the BHGFs come from sole ownership and limited partnership with each contributing half of the proportion. All these results were found to be statistically significant. Characterizing the HGFs using their age and ownership style, under both measures, HGFs are found to be younger by 3 years on average than the non- HGFs, although, the search for Gazelles, firms which are HGFs and younger than 5 years old was unsuccessful as there were nearly no firms of these sort in the economy. On the other hand, persistence of high-growth firms was not studied due to data problem. Studies by Daunfeldt and Halvarsson (2014) and Hölzl (2014) show that high-growth firms are one hit 12

14 wonders and the probability of repeating the high-growth rates is very low. In Ethiopia, since most of the firms are small sized firms, there is high tendency for firms to fall below the threshold level of employment. A research by Ayenew (2015) which was based on CSA data of large and medium sized manufacturing firms showed that on average 22% of firms are new entrants while 19% of them leave the category in the same year with the exit reaching as high as 46%. This makes it difficult to analyze persistent of high growth firms. Table 2: Distribution of HGFs by sector and by growth measure Distribution of HGFs by sector and by growth measure Sector Industry Screener Sector Proportion of HGFs Proportion of BHGFs Services of motor vehicles (G) 26.92% 22.13% Construction Section (F) 20.57% 21.60% Service Sector Wholesale (G) 19.08% 29.10% Retail (G) 15.07% 3.42% Transport Section I: (60-64) 6.48% 5.49% Hotel and restaurants (H) 5.30% 5.00% Subtotal 93.41% 86.75% Manufacturing Sector Non -metallic mineral products (D) 1.71% 2.25% Food products and beverages (D) 1.32% 4.01% Plastics & rubber (D) 1.04% 1.97% Subtotal 4.07% 8.23% The rest of the sectors 2.50% 5.02% Source: Enterprise Survey 2015 (World Bank) Looking at the industry type, the two measures refer to nearly the same types of firms where the service sector is overrepresented in the HGFs classification with a share of over 90% & 85% under the Eurostat- OECD and Birch Index measure, respectively. The Eurostat- OECD measure shows that services of motor vehicles (section G) has the highest incidence of HGs (around 27%) followed by construction sector (around 27%) with both belonging to the service sector while under the BI, wholesale business represents the highest incidence of HGFs (29%) followed by services of motor vehicles (section G) which accounts for 22 % of the BHGFs. Under the two measures, services of 13

15 motor vehicles, wholesale business, and construction section represent the top three dominant sources of HGFs. From the manufacturing sector, only food, non - metallic mineral products, and plastics & rubber accounted for noticeable proportion of HGFs. The three accounted for some 4 % of the HGFs using the Eurostat- OECD measure while the percentage doubles to 8% using the BI. The domination of HGFs in the service sector in Ethiopia is consistent with the findings of Henrekson and Johansson (2009) that did a meta-analysis of the role of HGFs. The incidence of high-growth firms in manufacturing sector is very low in Ethiopia with only 4-8% of the HGFs coming from this sector. Coming to the size of the firms, both measures show somewhat similar cohorts of HGFs since medium sized firms (with employees between 20 & 99) dominate the proportion of HGFs. Under the Eurostat- OECD measure they constitute 60% of the HGFs while in the BI they account for 75.5% of the HGFs. The essential difference between the two measures, however, is that the Eurostat- OECD measure shows that the incidence of HGFs tend to be the least for large firms (only 2.4%) while it is the least in small firms under the BI (less than 1%). This finding could be due to the inherent bias of relative growth measures such as the Eurostat- OECD measures towards the small firms while the BI controls for such bias (Coad et.al. 2014, Hölzl, 2011). Table 3: Distribution of HGFs by firm size and by growth measure Size screener Proportion of HGFs using Eurostat- OECD measure (%) Proportion of HGFs using BI score (%) Small Medium Large Total Source: Enterprise Survey 2015 (World Bank) 14

16 Incidence of HGFs in Ethiopia by Firm Size Eurostat- OECD measure of HGF (%) HGFs using Birch Index Score ( in %) Small Medium Large Figure1: Incidence of HGFs by firm size classification Another indicator of prevalence of HGFs used in the research is their regional distribution. Nearly all of the HGFs are concentrated in Addis Ababa regardless of the type of measurement used (over 90 % of them) while Oromia region is the second largest host of HGFs (around 4.5%) under the BI and 2.4 % under the Eurostat- OECD measure. Regions showed higher share of HGFs when the BI was used relative to the Eurostat- OECD measure. This result is not surprising as such owing to the fact that Addis Ababa accounts for over 80% of the sampled establishments and significant percentage of them being medium sized firms (35% of them) with high incidence of HGFs in the survey and the differences were found to be statistically significant. Table 4: Distribution of HGFs by region and by growth measure Incidence of HGFs in Ethiopia by Firm Region and Measurement Type Sampling region Percent of HGFs Percent of BHGFs Addis Ababa 93.9% 89.9% Amhara 0.6% 1.5% Dire Dawa 0.2% 0.2% Oromia 2.4% 4.5% SNNPR 1.3% 2.7% Tigray 1.6% 1.5% Total 100.0% 100% Source: Enterprise Survey 2015 (World Bank) 15

17 Table 5 below presents average statistics on firm performance for the two cohorts of firms. The table shows that high-growth firms have growth rate which is three times that of the non-highgrowth firms under the two measures on average. The HGFs also show higher number of employees on average with nearly twice the number of employees of the non HGFs using the BI. They also have high proportion of export engagement and significantly large proportion of firms owned foreigners belong to the HGFs. Table 5: Descriptive Statistics (average values) in 2014 for HGFs and non-hgfs Static HGFs in terms of Eurostat- OECD Non- HGFs in terms of Eurostat- OECD HGFs measured as top 20% on BI score (i.e. BHGF) Non- HGFs measured using BI score Employee growth from Sales growth from % 7.2% 26.5% 8.1% 12.4% 13.5% 18.5% 12% Firm size 11 employees 9 employees 17 employees 9 employees R&D Engagement Export Engagement Innovation activity Domestic Ownership Foreign Ownership Female Ownership 4.8% 2.6% 6.5% 2.5% 4.5% 2.6% 6.5% 2.5% 45% 55% 32.5% 67.5% 27.6% 72.4% 15.9% 84.1% 60.5% 39.5% 70.5% 29.5% 38% 62% 20.8% 79.2% Source: Enterprise Survey 2015 (World Bank) 5.2. Perceived business obstacles by establishments The analysis of business obstacles is based on the two interrelated group of questions asked in the questionnaire. Measuring the proportion of firms who reported element of a business environment 16

18 as major obstacle or very severe obstacle, 33% of all firms reported supply of electricity as major or severe obstacle making it a top ranked obstacle to doing business followed by corruption and tax rates. Corruption is perceived to be a top obstacle by around 29 % of the establishments and some 28% of them ranked tax rates as either major or very severe obstacle. Problems related to tax administration and informal sector competition is found to be the 4 th and 5 th major or severe obstacle to doing business in Ethiopia. From this data, it can be seen that both the tax rate and its administration pose severe threat to doing business. Figure 1 below presents the details of the perceived obstacles by firms. Proportion of Enterprises Reporting Obstacle as Major or Very Severe Obstacle Business Licensing & Permits Courts Ttransport Finance Customes & Trade Regulation Land Informal Sector Tax Adminstration Tax rates Corruption Electricity 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% Figure 2: Percentage of firms reporting Business Obstacles as major or very severe obstacle Source: Enterprise Survey 2015 (World Bank) 17

19 Inadequately educated workforce Transport labour regulations 60.0% Electricity 50.0% Corruption 40.0% 30.0% 20.0% Tax rates 10.0% 0.0% Tax administration Ethiopia SSA Middle East & North Africa High Income OECD All Countries Access to Finance Customs & trade regulation Informal sector Figure 3: Global picture of the perception about business obstacles by firms Source: Enterprise Survey 2015 (World Bank) The World Bank Enterprise Surveys which covers 139 countries and over 125,000 firms (World Bank, 2015) presents an excellent opportunity to do global comparison of the business environment in which firms operate. Figure 3 shades light on the global picture of business obstacles that firms believe hinder the growth. The radar locates Ethiopia close to the center next to the high income OECD countries using most of the indicators which shows that firms in Ethiopia work under better environment relative to most of the countries surveyed. Compared to SSA, for example, Ethiopia is superior in nearly all of the indicators. Furthermore, the ES also asks establishments to identify the biggest obstacle among a given list of 15 obstacles some of which are identified above. Looking at the responses to the question that asks for the single most important obstacle, over 40% of the establishments selected access to finance as number one problem while customs & trade regulations and electricity supply are rated as biggest obstacle by 12 % and 10% of the establishments, respectively. Tax administration and practices of informal sector were reported as biggest obstacles by approximately 8 % and 6 % of the establishments each, respectively. Figure 4 presents the details. Comparing the results to the two types of soliciting information, access to finance is the key obstacle relative to other perceived challenges. When establishments are asked to rate each obstacle 18

20 independently, supply of electricity, corruption and tax rates represent the three biggest obstacles to doing business in Ethiopia. Business Environment (biggest obstacle among the list in %) Access to finance Customs & Trade Regulations Electricity Tax Adminstration Informal Sector Corruption Transport Access to Land Inadequately educated workforce Crime/ Theft Tax Rates Political Instability Courts Labor Regulations Business Licensing & Permits 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% Figure 4: Single most important obstacle to doing business in Ethiopia (%) Source: Enterprise Survey 2015 (World Bank) Analysis of the business obstacles by firm size found to be statistically insignificant using chi2 test of independence. They all suffer from similar problems of lack of access to finance, problems emanating from customs and trade regulations, electricity and tax administration. Decomposition of the analysis of biggest obstacles using firm growth achievement, it is possible to see that the perceived business obstacles are not the same and it depends on whether a firm is a HGF or not. Access to finance is perceived as the biggest obstacle by both cohorts of firms with the problem being more severe for the non HGFs. For the HGFs, tax rates and customs and trade regulations represent the 2 nd and 3 rd biggest obstacles to firms while electricity and corruption completes the list of top five. For the non HGFs, informal sector, electricity, tax administration & customs and trade regulations are among the top five biggest obstacles (see figure 5 below for details). These findings show that access to finance can be considered as dominant challenge 19

21 affecting significant number of firms irrespective of the growth nature of the firms. The difference in perceived obstacles by the two groups of firms has been tested using chi2 test of independence and the result confirms the presence of statistically significant difference at 5% significance level. Top five Obstacles of HGFs(%) 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Top five Obstacles of Non - HGFs (%) Figure 5: Top five most important obstacles in Ethiopia by firm growth category. Source: Enterprise Survey 2015 (World Bank) 63% 80% 60% 40% 20% 0% 42% 13% 29% 13% 47% 18% 20% 17% 21% 24% 19% 29% 15% 17% 12% 14% 10% Addis Ababa Oromia Tigray Amhara SNNPR Dire Dawa Figure 6: Top Business Obstacles by region of establishment Source: Enterprise Survey 2015 (World Bank) 20

22 Analysis of business obstacles using region of operation as a reference point reveals that there is a systematic difference among regions as presented in figure 6 above. Looking at these problems from regional perspective, firms operating in different regions perceive different obstacles and the differences have been found to be statistically significant. For establishments in Addis Ababa, for example, 45% of the firms believe that the biggest obstacle is access to finance while only 21 % firms operating in Oromia consider finance as biggest obstacle and is not reported in the list of top three problems for firms operating in Amhara region and SNNPR. For firms in these regions, corruption tops the list in Amhara while electricity is reported as biggest obstacle in SNNPR. Establishments in Oromia reported informal sector activities as their biggest obstacle (29%) while those operating in Tigray reported finance as a key problem (42%). The implication is that regions should take in account these differences in their attempt to improve business environment A Test of the Gibrat s law Gibrat s law of proportionate effect proposes that firm growth is independent of its size. This law can be easily tested by plotting the log size of a firm at a point. In the figure, the normal line is presented by the dashed line while the unbroken line represents the kernel density curve. Looking at figure 8, the natural logarithm of size does not follow normal distribution. The distribution has a peak around 8 employees and is skewed to the right. This is a somehow an indirect proof against the law because small firms (as presented by the high density around 8 employees) grow faster than their medium and large counterparts. Density Natural log of employment in 2014 ==Ll1 Figure: 8 log normality plot of firm size using number of employees in Source: Enterprise Survey 2015 (World Bank) 27

23 5.4. Econometrics Analysis Econometric estimation was done using OLS and Quantile regression and the results are presented in Table 6 below. The first column gives the results of the OLS estimation while columns 2 10 present the results of the QR which shows the marginal effects at various deciles of the distribution. The reference group consists of firms in Addis Ababa, active in hotel and restaurants, solely owned by male domestic entrepreneurs. Analysis of the results from the OLS estimation shows that firm growth is negatively related to firm size and positively related to the squared term of firm size. The square term captures nonlinear relationship between the two under OLS. The result of the QR also supports OLS estimation. From the Quantile regression, the size effect is highly significant and negatively related to firm growth at each decile. The negative relationship shown here suggests that small firms grow faster than larger firms and this result is consistent with many global studies on the nexus between firm size and growth. The log normality plot of firm size introduced above is also in line with this finding. Furthermore, using CSA data on Ethiopian manufacturing firms, Bigsten & Gebreeyesus (2007) found similar results. Similarly, concerning the association between firm age and growth, the analysis shows that there is negative and significant relationship between the age and growth under the OLS estimation with the relationship being positive for the square term. The QR also shows similar relationship between the two but the relationship is found to be significant at the 60 th, 70 th & 80 th growth deciles. The nonlinear relationship between age and growth is presented by the significance of the squared term. Other important variables of interest in the analysis are the role played by gender and nationality of the owner of the establishments on firm growth. The OLS regression reports negative effect of female and foreign ownership on growth relative to male domestic ownership although both are found to be statistically insignificant. Hence, there is no statistically significant difference in growth of firms based on gender and nationality of the owner. The result is generally the same when evaluated using QR. The QR shows that foreign ownership is negatively associated with growth across the growth distribution, though not significant. More importantly, female ownership has been found to have a statistically significant negative effect on growth for the 30 th and 40 th growth deciles at the conventional significance level. 22

24 Concerning technological and market factors that are hypothesized to determine growth, the OLS regression shows that, firm level product and process innovation and ownership of website have positive and significant effect on firm growth. A unit increase in ownership of website or product innovation leads to 4 percentage point increase in employment while the effect of process innovation is a bit lower (close to 3 percentage point). Other explanatory variables such as degree of competition, experience of top management and export engagement are found to have negative but insignificant effect on firm growth. Training and degree of capacity utilization, however, are found to have positive but insignificant effect on firm growth. Analysis of the QR conveys more or less similar story. From the findings of the QR, the role of training for firm growth is still insignificant except for the poorest quantile while the role played by innovation (both product & process) are found to be significant at all deciles of the growth distribution. Both process innovation and product innovation could contribute for a maximum of 5 percentage point increase in firm growth. The negative effect of top management experience and export engagement becomes significant for some quantiles. Competition tends to associate with lower firm growth, though insignificant, which could be due to the dominance of small sized firms in the survey (50% of the analyzed firms are small firms) which could worry about survival than growth. Competition does not play a major role in determining the employment growth of firms. This result is consistent with the findings of Essmui et.al (2014). The association of firm s capacity utilization with its growth is found to be positive but insignificant in the Quantile regression. Exporting firms are found to have lower growth rates using the QR and the results are significant for most of the growth deciles. Exporting firms growth might be better measured by other measures of growth such as sales or revenue growth. Goedhuys & Sleuwaegen (2009) also found similar relationships in their study. From the resource dummies used in the regression, the OLS regression shows a positive relationship between ownership of generator and access to overdraft facility with growth. The mean growth is predicted to grow by 2 and 3 percentage points for firms with generator and access to overdraft facility, respectively. The quantile regression shows that the role of these resources is not the same for all firms. Ownership of generator enhances growth for firms that fall in the 60 and 70 growth percentile while access to overdraft facility could increase firm growth by 6 percentage points for the top growing firms. 23

25 Analysis of ownership type and region of operation dummies provide an interesting insight. Sole ownership seems to have the upper hand in growth performance for some of the growth deciles against all other forms of ownership although the OLS estimation found it to be insignificant. Hence, the role of ownership on firm growth is not well established. Similarly, establishments whose business operations were located in the capital city, as expected, are found to outperform others. The differences are found to be significant for firms in Oromia, SNNP and Tigray regions under the OLS estimation. The QR estimates confirm these findings although the top growing firms (i.e. firms in the 90 th percentile) did not show statistically significant difference across regions. For firms in the Amhara and Dire Dawa region, both estimation techniques fail to show any statistically significant difference from firms in Addis Ababa. Concerning the relationship between sector of establishment and growth, the OLS estimation shows that there is no significant difference among firms except the construction sector in which firms have a statistically significant superior growth performance relative to those in hotel and tourism. With the exception of firms operating in wholesale and retail, all other firms seem to have a slightly higher but statistically insignificant growth relative to firms in hotel industry. From the QR, only firms in the construction sector have a higher growth performance across most of the growth distribution with the exception of the fastest growing firms from garment and textile industry. These firms outperform the reference group in the 90 deciles. 24