Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Correlates of business survival: empirical evidence on youth-owned micro and small enterprises in Urban Ethiopia

Correlates of business survival: empirical evidence on youth-owned micro and small enterprises in... woldehanna@gmial.com Department of Economics, Addis This paper investigates the effects of person-, firm-, industry-, and business strategy- Ababa University, Addis Ababa, specific characteristics on the survival of youth-owned urban micro and small Ethiopia Full list of author information is enterprises in Ethiopia. It employs nonparametric and semi-parametric methodologies available at the end of the article using a retrospective data. The hazard rate reaches the highest point at business 2 years for micro enterprises and 4 years for small enterprises. Owner-, firm-, and industry-specific characteristics are important factors for micro and small enterprises’ (MSEs’) survival. Marketing and financial management strategies are playing a crucial role on extending MSEs’ survival duration. The study implies there should be more effective and longer period of support for micro than small enterprises. JEL classification: C14, C41, D21, J23, L26, M21 Keywords: Youth self-employment, Micro- and small-scale enterprises, Hazard model 1 Introduction The co-existence of micro and small enterprises (MSEs) can be interesting; concomi- tantly, the performance of MSEs in terms of survival and growth needs to be assessed in detail. Identifying the growth and survival factors of MSEs is important as it estab- lishes the base for preparing a policy framework and strategy that safeguards the success of MSE operators. However, there are few detailed studies, particularly survival analysis, which identify the factors influencing the survival of MSEs, in general, and youth-owned MSEs, in particular. Moreover, there are hardly research outputs intended to identify factors influencing the survival of MSEs in Ethiopia using firm- specific data. Thus, the existed knowledge gap and the focus for the sectoral develop- ment motivated us to empirically analyze the factors associated with the risk of dropout of MSEs in Ethiopia. The paper intends to identify how the socio-demographic and economic characteris- tics of the owner- and firm-level attributes at start-up phase are associated with the survival of youth-owned MSEs. Specifically, the objectives of the paper are (1) to investigate the effect of owner- and firm-specific attributes on MSEs’ survival, (2) to analyze how strategic thinking and systematic decision-making are partly responsible for the failure or success of entrepreneurs, and (3) to identify owner, firm, industry, © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 2 of 26 and business strategy characteristics that can explain the difference among micro and small enterprises’ survival rate. The present paper uses a nationwide but urban-based representative cross-sectional survey data which is rich with retrospective information. We apply nonparametric and semi-parametric survival models to study and assess the relative strength of the associ- ated factors with survival of the MSEs and estimate the survival rate. There are major concerns why firms eventually fail to become established firms. Ex- tensive theoretical and empirical literature in developed economies have identified micro-level factors, determining the emergence and success of enterprises. Most of the studies have shown that owner- and firm-related characteristics are the basic factors that determine the success of a firm (see, e.g., Bruderl et al. 1992; Storey 1994; Coleman et al. 2010). These characteristics are typically identifiable prior to the start-up and comprise personal and psychological characteristics, such as gender, education and training, entrepreneur’s mind setup, previous work experience, and ethnicity. Most studies indicated that women-owned firms perform less in terms of different performance measures, such as number of employees (Rosa et al. 1996), revenue (Rosa et al. 1996; and Brush et al. 2003), and profitability and survival (Fairlie and Robb 2008). Some empirical analyses suggested that owner’s/manager’s age tends to be nega- tively related to risk of dropout (see Bates 1998; van Praag 2003). In contrast, there are empirical evidences that prove the negative association of owner’s/manager’s age with the growth of small firms (see Boswell 1973; Davidsson 1991; Amran 2011). The owner’s human capital quality, which can be measured by education, prior work, and managerial experience, is also pointed out as a positive factor for the performance of firms (see, e.g., Schiller and Crewson 1997; Honig 1998; Chandler and Hanks 1998; Kangasharju and Pekkala 2002; Pena 2002). Cowling (2009) has found a positive effect of school-based entrepreneurial training on job creation by small firms in the UK. He also found a positive relation between entrepreneurs who received government-backed entrepreneurial training and exporting intensity of small firms. The size of a start-up has been considered as a key factor for self-employment sur- vival, and it is widely documented that larger size entry is a signal of commitment and self-confidence, which reduces the occurrence of an entry mistake and the risk of a fail- ure due to diseconomies of scale (Mata et al. 1995; Agarval and Audretsch 2001). Ex- tensive empirical works also focus on assessing the impact of industry-specific features on start-up success and reveal the presence of relationship between firm’s success and the type of the sector the firm is operating, level of competition within the industry, linkage, market, and technological change and innovation (see Bruderl et al. 1992; Audretsch and Mahmoud 1995; Mata and Machado 1996; Headd 2000; Bates 2005; Carter and Van Auken 2006). For example, Coleman et al. (2010) indicated that firms involved in retail and service sectors are more prone to failure due to low barriers to entry and high levels of competition. In the contemporary, global, dynamic business environment, strategic thinking and systematic decision-making are also important for the growth and survival of small enterprises (see Dobbs and Hamilton 2007; Skrt and Antoncic 2004). The characteristics of business management strategies, including marketing and finan- cial management, are found to have a strong positive relationship with business performance (O'Dwyer et al. 2009). Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 3 of 26 A small business in which the owner reinvests retained earnings has the propensity to grow more than businesses that are not willing to reinvest (Fadahunsi 2012). Con- sistently, internal financing is presumed as the main reason for the success of most firms in the USA (Carpenter and Petersen 2002). The study by Read (1998) showed that over half of her total sample of small firms in the UK suffered from being under- capitalized. Moreover, evidences highlighted that internal control, including financial report and inventory management, allows a business to enhance effectiveness and gain competitive advantage (see Moores and Mula 1993; Mongare and Nasidai 2014). Currently, MSEs in Ethiopia are becoming important vehicles to create new jobs for millions of unemployed youth (FeMSEDA 2014). Despite the focus given to the sector, a large number of MSEs are expected to eventually close or stagnated at starting phase. As reports and studies documented, there are internal and external factors which per- vasively hamper the productivity improvement and survival of the MSEs; shortage of working capital, a lack of marketing skills, poor location of business sites, failure to take risk, and over emphasis on short-term profit are internal factors affecting the survival rate of MSEs. In addition, a lack of access to efficient infrastructure and factor of production, a nontransparent regulatory environment, bureaucracy burden, high inci- dence of bribing, and a lack of appropriate skills and training are also documented as external factors hindering the success of MSEs (see World Bank 2014; Dayanandan 2012; Bigsten et al. 2010 and FeMSEDA 2014). Hence, it is worth to investigate the explanatory factors that influenced the survival and exit of businesses using a micro dataset which is representative. The paper is organized as follows: Section 2 presents the econometric model specification employed in the paper. Section 3 presents the data and descriptive statistics. Section 4 reports the empirical results and interprets the esti- mated parameters. Section 5 presents the conclusion remark and the policy implications. 2 Empirical model specification: survival models Survival rate is often the most practical index to measure the success of a firm. The analysis examines the elapsed time since the entry to the state at time t = 0. The present paper specifies a survival model that accounts for the sequential nature of the data and is able to handle censoring. Let T represents survival time and is a random variable with a cumulative distribu- tion function F(t) = Pr(T ≤ t) and a probability density function f(t)= dF(t)/dt. The reverse cumulative distribution function of T provides a new function called survival function S(t), and it is defined as StðÞ ¼ PrðÞ T > t¼ 1−FtðÞ: ð1Þ The hazard rate, the instantaneous risk of demise at time t conditional on the firm having survived up to that point in time, is the motive to be estimated in our survival analysis. The hazard function is an absolute slope of log survival function and it is defined as follows ftðÞ d lnStðÞ htðÞ ¼ ¼ − ð2Þ 1−FtðÞ dt and the cumulative hazard function H(t)is Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 4 of 26 HtðÞ ¼ − lnStðÞ: ð3Þ The survival function describes the probability of surviving past time t. The hazard function describes the relative likelihood of the event occurring at time t, conditional on the subject’s survival up to time t. The hazard rate shows the instantaneous rate of failure at time t and ignores the accumulation of hazard up to time t (see Jenkins 2004 for details). The present paper utilizes both nonparametric and parametric models to analyze the survival rate of MSEs in urban Ethiopia. We estimate the nonparametric survival and cumulative hazard analysis using the Kaplan-Meier method. The paper uses the Kernel-smoothed hazard estimation method for smoothed hazard analysis, as it effi- ciently detects the times where the most rapid changes of the hazard function occur. 2.1 Nonparametric approach The above modeled survival and hazard functions can be estimated without making as- sumptions about the shape of the relevant functions. The Kaplan-Meier estimator is a nonparametric estimator that does not make any assumptions about the distribution of exit times and how covariates shift the hazard functions (Jenkins 2004). Hence, the pro- portion of surviving to survival time t and the corresponding integrated hazard func- tion is estimated by S t ¼ 1− ð4Þ jjt ≤ t ∧ ∧ H t ¼ − ln S t ð5Þ j j where d denotes the number of failed enterprises at t and n is the number of enter- j j j prises at risk of failure immediately prior to t and defines as the sum of those who have a censored or completed spell of length t or longer. The Kernel-smoothed estimator is a weighted average of Δ H ðtÞ over event times that are within a bandwidth distance b ∧ ∧ of t. The Kernel-smoothed estimator considers the jumps of H ðtÞ and variance of H ð tÞ at the event times as follows: ∧ ∧ ∧ Δ H t ¼ H t − H t j j j−1 ∧ ∧ ∧ ∧ ∧ ∧ ð6Þ V H t ¼ V H t − V H t j j j−1 where t =0. As the effect of an enterprise’s initial size on its survival probability is one major ob- jective of our paper, we grouped the dataset into micro and small enterprises. Accord- ingly, for ease of interpretation, we grouped enterprises into 6-month intervals and execute the so-called Lifetable estimator. In essence, the Lifetable estimator processes the same idea as the Kaplan-Meier method, but it was explicitly designed to handle situations where there is grouped survival time data. 2.2 Semi-parametric hazard model The focal point in our survival analysis is the hazard function, allowing for time invari- ant covariate effects as an econometric framework. The Cox proportional hazard model Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 5 of 26 has considerable flexibility and is the most widely used hazard regression model in em- pirical studies. This model, proposed by Cox (1972), enables us to estimate the relation- ship between the hazard rate and covariates without having to make any assumptions about the shape of the baseline hazard function, referring to the model as a semi- parametric model. A survival model examines the relationship between the survival distribution and the covariates and mostly entails the specification of a linear-like model for the log hazard. logh ðÞ t ¼ α þ β x þ β x þ … þ β x j j1 j2 jk 1 2 k ð7Þ h ðÞ t ¼ e α þ β x þ β x þ … þ β x j j1 j2 jk 1 2 k where x’s are the covariates and α represents a log-baseline hazard because if all x’s are zero, then h (t)= e . The Cox model leaves the baseline hazard function α(t) = log h (t) j 0 unspecified. Thus, the hazard function is parameterized as follows: β x þβ x þ…þβ x ðÞ j1 j2 jk 1 2 k h ðÞ t ¼ h ðÞ t e ð8Þ j 0 where h (t) is the so-called baseline hazard which depends only on t. The change in the explanatory variables, therefore, induces multiplicatively proportional shift in the haz- ard rate relative to the baseline hazard. This implies that the model assumes that the hazard functions of any individuals with different values on one or more covariates dif- fer only by a factor of proportionality. Based on Eq. (8), the hazard ratio for two individuals, j and i, is defined as follows: h ðÞ t ¼ exp β x −x þ β x −x þ … þ β x −x ð9Þ j1 i1 j2 i2 jk ik 1 2 k h ðÞ t where the hazard ratio is independent of time t. In the way to compute the hazard ra- tio, the baseline hazard h cancels out of the computation, implying that the relation- ship between the hazard rate and the explanatory variables is estimated without taking h into account. Consequently, the Cox model becomes a proportional hazard model, but the proportional hazard assumption should be tested. Instead of the maximum-likelihood method, the Cox model can be estimated by the method of partial likelihood (PL), because it requires less assumption and so is more robust (Fox 2002). The Cox PH estimation function is “partial” likelihood function because it only considers probabilities for failed events explicitly, referring that the baseline hazard function is removed from the estimation process. The estimation process started by sorting the ordered failure times and PL is derived from the product of the conditional probability of a failure time, given the risk set at that time and that one failure is to happen. Hence, the PL only considers the contribution from failure times, not from the right-censored times (we are referring the cases). This implies that the Cox model is conspicuously ignoring all information available at times when no failure occurs. Cox (1972) has found little efficiency loss due to ignoring spells in which no failure occurs. 3 Data and descriptive statistics 3.1 Data The dataset of many empirical studies on the survival of enterprises is derived from panel and cohort surveys and/or a database administered by a recognized office, usually Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 6 of 26 government offices. However, there are two basic problems using the formal registration database of MSEs collected by the Federal Micro and Small Enterprises Development Agency (FeMSEDA) in this particular study. Firstly, the existing registry of MSEs is based on information obtained during the formal registration of the enterprises, and yet most of the enterprises are informal and/or not registered by the agency. Secondly, it mainly pro- vides details of the business, such as registered address, setup date, industry type, owner- ship type, start-up capital, to date capital, and firm size during the start-up. Thus, it was found to be inadequate in terms of providing all the required information and current status of the MSEs to undertake a survival analysis. To this end, this study employs information from the “youth self-employment” survey collected by the Association of Ethiopian Microfinances Institutions (AEMFI), which has details on the parameters required for a survival analysis of MSEs in Ethiopia. The youth self-employment research project primary concern was to obtain fresh informa- tion and evidence on strategies and policies that enhance youth self-employment within the framework of MSE development in Ethiopia. The survey collected quantitative in- formation of the sampled MSEs on their current status; profiles; retrospective informa- tion about the initial start-up data, including owner- and firm-level characteristics; and ongoing and past business strategies since start-up. The stratum was regions (Tigray, Amehara, Oromia, SNNP, and Hareri) and two city administrators (Addis Ababa and Dire Dawa). The survey focused on all regional cap- ital cities and the two city administratives (Addis Ababa and Dire Dawa). In addition, two towns from each big-four region were selected based on their economic activity and population size that they had proportion to the total urban population. Hence, 15 major urban centers were included directly in the sampling strategy given that they constitute about 91% of the Ethiopian urban population, and these are the major pri- vate sector hubs throughout the country. Regional sample size was disaggregated to the urban areas by purposeful proportion al- location; regional capital cities had 50% of the total sample size allocated for the region while the two towns accounted for 25% each. We hypothetically drew a 200-m-radius grid, which was an enumeration area (EA) for the survey, from the city midpoint. We pre- ferred urban centers based on a mere assumption that household members in these areas are engrossed in self-employment. Hence, households within the EA were the data frame from this particular survey. Randomly selected households were screened through listing questionnaire before a mem- ber is selected for the main survey. The listing questionnaire had a filter question to classify the members for existed or closed enterprise sampling. The key filter questions included the following: whether there is a member who is currently operating a micro or small enterprise and whether there is a member who had started but closed a business for any reason but due to business challenges before today. In case of multiple MSE operators and/or dropouts within a selected household, only one household member was selected for the interview. The survey employed a population sampling process, implying both existing enterpri- ses—the start-ups observed at the time of the survey—and closed enterprises—the start-ups that have been already dropped out at the time of the survey—were inter- viewed. Closed firms, in this particular study, refer to those that have discontinued their operations, which included enterprises sold, bankrupted, and merged with other enter- prises due to business challenges. Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 7 of 26 A multi-stage sampling technique (with stratification by gender, enterprise size, and industry type) was applied to randomly selected 1109 enterprises from existing and closed MSEs owned by the youth (ages ranging from 18 to 34 years). The enterprises operating during the survey were 909, and the already closed were 200. Sampled enter- prises were distributed across different industries such as agriculture, manufacturing, and services. To make the sample representative of all sectors, attempts were made to take proportional samples from enterprises who are engaged in manufacturing, service, trade, and urban agriculture. Moreover, stratification was made by gender of the owner and the enterprise size (micro and small). The federal and regional MSE development agency/bureaus played a key role in facilitating the entire sampling strategy. Existing enterprise owners and those who closed their business were asked about the specific date of business start-up and exist date, in case the start-up had already closed. This information allowed the researchers to establish a business operation spell for each firm, and the spell might be either completed or right-censored at the time of survey. The established spell is the dependent variable with a value of 1, if the spell is right-censored, and 0, if the spell is completed. The survey had also de- tailed information on the socio-demographic and economic background of the owner/manager, as well as firm, industry, and business strategy characteristics. Moreover, the sample survey had detailed information regarding the feature of the owner as well as the firm during the setting up of the business, the quality of the support program, the respective gaps and challenges experienced by MSE opera- tors, and the labor market status of the youth before and after self-employment, which are important variables in the survival analysis. Retrospective information determines the validity of a research, since it can be sub- scribed to self-justification bias easily. To account for potential survey bias and retro- spective bias, the data used in this paper consists 941 enterprises which were established from 2008, and thereafter, expecting the start-ups’ owners can remind and provide information with less inaccuracy later than this period. Thus, the sample includes 941 enterprises; 569 enterprises are grouped as micro and 372 are as small enterprises. The group classification is based on registration certificate and their num- ber of workers . Moreover, out of the total sample, we observe 18% completed business operation spell and the remaining 82% is right-censored at time of interview. 3.2 Descriptive analysis Table 1 presents the hypothesized variables that were identified at the time of start-up and on business activity, along with their description using summery statistics. Part A and part B of the table describe the socio-demographic and economic characteristics of the owner, and firm- and industry-specific characteristics, respectively. Part C of the table highlights the description of business strategy variables. About 26% of the enterprises were owned by female (i.e., as a solo and group) entre- preneurs, implying the dominance of male entrepreneurs in MSEs in Ethiopia. The average age of the owners at the time of the start-up was 24. The owners’ human cap- ital quality is measured by formal education attainment and work and business experi- ence prior to the start-up. Accordingly, during the start-up period, 25% of the entrepreneurs had only elementary school attainment, while about 17% of the Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 8 of 26 Table 1 Definition of variables and their corresponding mean and standard deviation value Covariates Mean Std.dev Part A: owner-specific characteristics Dummy for female 0.255 0.436 Age of the firm owner at the time of establishment 24 4.42 Dummy for elementary education attainment 0.249 0.432 Dummy for college diploma and above attainment 0.166 0.372 Dummy for TVET education 0.001 0.300 Dummy for active in a labor market prior to the new business 0.514 0.500 Dummy for self-employment experience prior to the new business 0.122 0.328 Dummy for apprentice experience prior to the new business 0.329 0.154 Dummy for business experience prior to the new business 0.501 0.500 Dummy for entrepreneurship training prior to the new business 0.143 0.351 Dummy for short-term training prior to the new business 0.350 0.477 Dummy for opportunity motivation 0.690 0.463 Part B: firm and industry specific characteristics Firm size (initial number of workers) 8 10.6 Dummy for large start-up capital (more than 100,000) 0.055 0.229 Dummy for micro start-up capital (less than 25,000) 0.833 0.373 Dummy for sole proprietorship start-up 0.383 0.486 Dummy for cooperative start-up 0.262 0.440 Dummy for metal and wood work 0.258 0.438 Dummy for leather products 0.015 0.121 Dummy for textile and cloth 0.048 0.214 Dummy for construction 0.197 0.398 Dummy for service 0.114 0.318 Dummy for product provider 0.705 0.456 Dummy for support beneficiary 0.661 0.474 Dummy for registered during establishment 0.810 0.393 Part C: business strategy-specific characteristics Marketing management Dummy for competitive pricing strategy 0.718 0.450 Dummy for mark-up pricing strategy 0.426 0.016 Dummy for at home placing strategy 0.238 0.426 Dummy for at industrial site placing strategy 0.084 0.277 Dummy for salesmanship promotion strategy 0.258 0.438 Dummy for product innovation strategy 0.733 0.442 Dummy for market place innovation strategy 0.228 0.420 Dummy for niche market strategy 0.734 0,442 Dummy for high degree of competition 0.758 0.429 Dummy for external market linkage (proxied by siting cluster areas) 0.104 0.306 Financial management Capital investment (in Birr) 22,280 10,205 Dummy for retained earning investment 0.875 0.331 Dummy for internal financing (proxied by working capital financing from internal source) 0.693 0.462 Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 9 of 26 Table 1 Definition of variables and their corresponding mean and standard deviation value (Continued) Covariates Mean Std.dev Dummy for business financing (proxied by working capital financing from suppliers) 0.058 0.008 Dummy for external financing (proxied by investment financing from external source) 0.319 0.466 Dummy for undercapitalization challenge 0.366 0.482 Internal control management Dummy for inventory system (proxied by using detail and modern list of input stocks at 0.244 0.430 any time) Dummy for annual budget 0.079 0.269 Dummy for poor accounting system (proxied by a significant difference between the 0.188 0.391 book value and the actual value of revenue, cost, and profit) Human capital management Dummy for owners training after establishment 0.363 0.481 Dummy for workers training while they are on duty 0.169 0.012 Internet technology Dummy for Internet connection in the premises 0.029 0.170 Number of observation 941 Note: The left side for the owner’s education covariate is the secondary school attainment; for start-up capital, covariate is small-sized capital (between 25,000 and 100,000); for the legal form of the firm, covariate is partnership; for pricing strategy, covariate is negotiation with the customer; and for placing strategy, covariate is shopping district entrepreneurs had at least a college diploma and only 10% of the entrepreneurs had Technical and Vocational Education and Training (TVET). There were few owners who had self-employment experience prior to the start-up. Regarding previous experi- ence, 33 and 50% of the owners had apprenticeship and business experience prior to the start-up, respectively. The entrepreneurs who started the business out of push factor, such as to escape from being unemployed, lack of skill and proper education, and those who are forced by friends and family factors, are categorized as necessity-driven enterprises. On the other hand, those owners who started business as result of observing the enabling busi- ness environment and believe that self-employment is better than being employed by others are classified as opportunity-driven enterprises. Accordingly, the motivation be- hind being self-employed for about 69% of the owners was opportunity driven. The average initial number of workers, which is a proxy for initial size of the firm, was eight, implying the dominance of micro-level enterprises in our sample. About 83% raised a start-up capital of less than or equal to Birr 25,000 (USD 925), while only 6% raised a start-up capital of more than Birr 100,000 (USD 3700) during establishment. Regarding industry type, most of the enterprises are product providers (almost 71%), of which about 37 and 28% of the enterprises were engaged in metal and woodwork and construction sector, respectively. Part C of the table shows that about 72% of the firms set price for their product and service conditioning on the competitors’ position. This suggests that the firms’ pricing objective is status quo; MSEs seek to keep their product prices consistent with or con- ditioned on the competitors’ position. With respect to production innovation, about 73% of the owners/managers undertook screening and evaluation of the existing prod- uct mix line and made improvements and/or developed new products. Most enter- prises, about 76% of enterprises, were experiencing a high degree of competition in Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 10 of 26 recent times due to the expansion of similar businesses. Niche marketing is found to be a significant marketing strategy in the sector, as 73% of the owners/managers focused their marketing on serving a preferably small niche market. The average value of capital investment made by the enterprises since establishment was less than Birr 25,000 (USD 925). Most owners/managers, 88%, opted to reinvest their earnings in the business. Not few, almost 37%, enterprises experienced a shortage of capital investment funds during their business operations. It turns out that 36% of the owners improved themselves through attaining short- term trainings after establishment. Moreover, almost 14% of the enterprises arranged skills training for their workers. Last but not the least, for adoption of information and Internet technology, only 3% of the enterprises had an Internet connection in their business premises. 3.3 Differences between micro and small enterprises In order to reveal differences between owner- and firm-related characteristics among micro and small enterprises, a t test mean equality is conducted. The reported P values show whether the means were equal at a significance level of 5%. Table 2 compares micro- and small-sized firms using business operation survivors and dropout samples. The result indicates that the majority of female entrepreneurs were apt at establishing a micro-sized enterprise. The share of female entrepreneurs was significantly higher among the micro enterprises (almost 47%) within the dropout MSE groups. The share of owners with apprentice experience within the dropout sam- ple was significantly higher among the micro enterprises (almost 37%) than the small enterprises (almost 20%). In the survivor sample, a significantly higher share of owners with entrepreneurship training at the start-up period found among the small (almost 18%) than the micro en- terprises (almost 13%). Opportunity-driven entrepreneur’s share is significantly lower among the small (almost 38%) than the micro enterprises (almost 61%) in the dropout sample. In the survivor sample, small business owners were likely to invest, on average, higher amounts compared to the micro enterprises. In contrast, the share of owners who invested less than Birr 25,001 was significantly higher among the mi- cro enterprises than the small enterprises in the dropout sample. About 70% of start-ups were engaged in the manufacturing sector (product providers). This share was significantly higher among small than micro enterprises in the business oper- ation survivor group and lower in the dropout group. The proportion of micro enterprises engaged in the service sector was higher compared to small enterprises in both survivor and dropout samples. 4 Estimation results 4.1 The nonparametric estimation results Figure 1 shows the estimated survival and hazard rates. Figure 1a shows that survival rates declined over time. On average, almost 88% of MSEs survived their first 2 years in business. However, the survival function sloped downward strongly after and the prob- ability of survival reached almost 68% after 6 years of business operation. As indicated Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 11 of 26 Table 2 Mean separation test in the survival rate of firms between micro and small enterprises Survival Drop-out Variable All Micro Small P value Micro Small P value MSEs Socio-demographic and economic Dummy for female 0.255 0.279 0.157 0.0000 0.466 0.212 0.0004 Age of the firm owner at the time of establishment 24 24 24 0.4566 26 26 0.2673 Dummy for elementary education attainment 0.249 0.247 0.231 0.3198 0.291 0.273 0.3979 Dummy for college diploma and above attainment 0.166 0.172 0.170 0.4750 0.0126 0.167 0.2327 Dummy for TVET education 0.100 0.109 0.118 0.3624 0.029 0.030 0.4826 Dummy for active in a labor market prior to the 0.514 0.543 0.502 0.1405 0.447 0.470 0.3852 new business Dummy for self-employment experience prior to 0.122 0.105 0.134 0.1113 0.155 0.136 0.3682 the new business Dummy for apprentice experience prior to the new 0.024 0.350 0.314 0.1500 0.370 0.200 0.0087 business Dummy for business experience prior to the new 0.501 0.521 0.471 0.0836 0.485 0.515 0.3541 business Dummy for entrepreneurship training prior to the 0.143 0.131 0.183 0.0242 0.126 0.076 0.1512 new business Dummy for short-term training prior to the new 0.350 0.264 0.418 0.0000 0.398 0.561 0.0194 business Dummy for opportunity motivation 0.690 0.781 0.644 0.0000 0.612 0.379 0.0015 Firm specific Dummy for large start-up capital (more than 0.055 0.024 0.111 0.0000 0.039 0.045 0.4172 100,000) Dummy for micro start-up capital (less than 25,000) 0.833 0.906 0.729 0.0000 0.874 0.742 0.0146 Dummy for sole proprietorship start-up 0.383 0.541 0.183 0.0000 0.466 0.061 0.0000 Dummy for cooperative start-up 0.262 0.122 0.418 0.0000 0.175 0.667 0.0000 Dummy for metal and wood work 0.258 0.294 0.271 0.2470 0.155 0.106 0.1825 Dummy for leather products 0.015 0.017 0.013 0.3266 0.000 0.03 0.0382 Dummy for textile and cloth 0.048 0.051 0.059 0.3307 0.000 0.045 0.0145 Dummy for construction 0.197 0.146 0.271 0.0000 0.136 0.303 0.0040 Dummy for service 0.114 0.127 0.069 0.0048 0.204 0.091 0.0255 Dummy for product provider 0.705 0.670 0.807 0.0000 0.544 0.727 0.0083 Dummy for support beneficiary 0.661 0.556 0.794 0.0000 0.592 0.894 0.0000 Dummy for registered during establishment 0.81 0.755 0.912 0.0000 0.718 0.864 0.0137 Number of observation 466 306 103 66 Backgrounds are found from the retrospective information. Numbers are shares unless stated otherwise. P value is the lowest level of significance at which the null hypothesis (the difference in mean is equal to zero) is rejected in Fig. 1b, there was a sharp rise in cumulative hazard at the beginning of the establish- ment time and then decelerated, reflecting higher hazard rate during the start-up period. As Fig. 1c highlights, the risk of dropout increased during the start-up phase of the enterprise and reached its peak at 24 months of business operation duration. For group analysis, the paper employs Lifetable estimator for survival function esti- mation. Figure 2 presents the nonparametric survival and hazard rates. Figure 2a shows that the survival probability among micro enterprises decreased rapidly after the first 6 months of business operation. The estimate for micro/small groups shows that about 92% of the small enterprises and about 90% of the micro enterprises survived the first Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 12 of 26 Fig. 1 The nonparametric survival and hazard rate estimation. a Kaplan-Meier survival estimate. b Cumulative hazard estimate. c Kernel-smoothed hazard estimate 2 years of business operation, which is almost the same. However, after 4 years of business duration, the survival rates among the groups were notably growing apart: 77% for the small enterprises and 66% for the micro enterprises (see Appendix 1:Table 4). Thus, the results reveal strong evidence that a start-up’s initial firm size affects sur- vival probability. Entrepreneurs who establish new small-sized firms survive more com- pared to those who set up micro-sized enterprise in Ethiopia. To test for equality of the survival functions, we run the log rank test. The null hypothesis of no survival difference between the groups was rejected at a 10% level of significance (P = 0.0707). Figure 2b depicts that dropout risk was higher among micro than small enterprises at a given business operation duration. Moreover, it has an inverted U shape for the micro enterprises. Initially, there was an increased risk of dropout during the first 2 years of business operation, and then, the level of the risk remained high for the next two con- secutive years before it declined over time. In contrast, we observed an increased drop- out risk only during the first business operation year, which then decreased among small-sized enterprises. Fig. 2 The survival and hazard rate deferential between micro and small enterprises. a Survival rates for micro and small enterprises. b Kernel-smoothed hazard estimate Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 13 of 26 Overall, the nonparametric survival analysis suggests two basic facts on MSEs’ oper- ation in Ethiopia: (i) there is a survival differential among micro and small enterprises; (ii) the risk of dropout is high during the first two business operation years, particularly for micro enterprises. This implies that there is dire need to develop efficient and inclu- sive support interventions for micro business founders and mostly during the start-up phase of MSEs in Ethiopia. However, there is also a need to investigate the factors underpinning the survival of MSEs and what factors account for the deferential in survival rates among micro/small enterprise. 4.2 The Cox proportional hazard estimation To find out what factors derive the survival of MSEs and the associated factors for the deferential in survival rates between micro and small enterprises, we employ semi-parametric analysis using the Cox proportional hazard estimation. The tests for the assumption of proportional hazard reveal strong evidence of proportional hazard for each covariate and for the global test (see Appendix 1: Tables 5 and 6). The chi-square statistics soundly failed to reject accommodating proportional haz- ard for model A and model B as a whole; P values are equal to 0.988 and 0.374, respectively. Similarly, the graphical diagnostic tests of the Schoenfeld residuals also supported the assumption of proportional hazard for all covariates as there was non-systematic departure of covariates from their smoothing-spline fitted line over time. We observe non-trend pattern with time in the plots for all covariates (see Appendix 2: Figures 3 and 4). The survival rate of an enterprise may be influenced by various factors, such as owner, firm, and business strategy characteristics and the overall socio-business en- vironment the enterprise is operating. Hence, the present paper specifies three dif- ferent models to consider the effect of each category, separately. The first specification (model A) only presents and discusses the effect of the characteristics that are typically identifiable at the time of establishing the enterprise on MSEs’ survival. We re-estimate the semi-parametric Cox proportional model (model B) which focuses on whether the business strategic characteristics of the MSEs are associated with the survival or failure of the enterprises. In order to control the effects of person-, firm-, and industry-specific background characteristics, we in- clude three covariates in model B as a control variable, namely female, sole propri- etorship, and manufacturing. The last specification (model C) estimates all the hypothesized variables using a Cox analysis. Table 3 illustrates the estimated coefficients from the Cox proportional models and has three parts with sub-columns. Section A constitutes three sub-columns that present estimates of model A from the pooled sample and the grouped sam- ples. Section B constitutes the next three sub-columns of the table, and they present estimates of model B from the pooled sample and the grouped samples. Section C is the last column of the table, and it highlights the estimates of all the hypothesized variables. The results reveal that gender has a significant positive association with risk of dropout. Enterprises owned by women had a higher risk of dropout than men en- trepreneurs, which is consistent with the results from Coad and Tamvada (2008). Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 14 of 26 Table 3 Partial likelihood estimates for the Cox proportional hazard models Model A Model B Model C Covariates MSEs Micro Small MSEs Micro Small MSEs Dummy for female 0.683*** 0.907*** 0.370 0.610*** 0.882*** 0.381 0.493* (3.838) (3.916) (1.142) (3.510) (3.925) (1.119) (2.566) Age of the owner 3.244*** 3.635*** 2.714** 3.073*** (6.426) (5.705) (3.256) (5.968) Dummy for elementary − 0.111 − 0.065 − 0.034 − 0.255 education attainment (− 0.608) (− 0.261) (− 0.109) (− 1.332) Dummy for college diploma 0.037 − 0.025 0.252 0.456 and above attainment (0.154) (− 0.078) (0.689) (1.802) Dummy for TVET education − 0.523 − 0.462 − 0.612 − 0.481 (− 1.121) (− 0.764) (− 0.827) (− 1.004) Dummy for active in a − 0.388* − 0.400 − 0.332 − 0.346 labor market prior to the (− 2.014) (− 1.557) (− 1.043) (− 1.656) new business Dummy for self-employment 0.309 0.494 0.047 0.316 experience prior to the new (1.169) (1.485) (0.099) (1.142) business Dummy for business 0.191 − 0.042 0.625 0.242 experience prior to the (1.053) (− 0.180) (1.951) (1.256) new business Dummy for apprentice 0.306 0.574* − 0.281 0.639** experience prior to the (1.582) (2.417) (− 0.749) (2.950) new business Dummy for opportunity − 0.837*** − 0.712** − 0.974** − 0.776*** motivation (− 4.583) (− 3.002) (− 3.192) (− 3.785) Dummy for entrepreneurship − 0.543* − 0.178 − 1.210* 0.032 training prior to the new (− 2.112) (− 0.577) (− 2.440) (0.110) business Firm size (initial number of − 0.198* − 0.029 workers) (− 2.073) (− 0.293) Large start-up capital (more − 0.824 0.358 − 1.433* − 0.485 than 100,000) (− 1.866) (0.556) (−2.177) (− 1.013) Micro start-up capital (less − 0.272 − 0.134 − 0.377 − 0.016 than 25,000) (− 1.162) (− 0.360) (− 0.935) (− 0.061) Dummy for sole proprietorship − 0.898*** − 1.016*** − 1.026 − 1.079*** − 1.014*** − 1.961** − 1.187*** start-up (− 3.771) (− 3.880) (− 1.687) (− 5.220) (− 4.063) (− 3.260) (− 4.542) Dummy for cooperative 0.312 0.055 0.478 − 0.196 start-up (1.483) (0.177) (1.487) (− 0.796) Dummy for product provider − 0.598*** − 0.712** − 0.470 − 0.627*** − 0.674** − 0.539 − 0.643*** (− 3.445) (− 3.295) (− 1.433) (− 3.506) (− 2.939) (− 1.738) (− 3.411) Dummy for support beneficiary − 0.092 − 0.219 0.238 0.040 (− 0.466) (− 0.960) (0.544) (0.189) Dummy for registered during − 0.503* − 0.485* − 0.411 − 0.101 establishment (− 2.495) (− 2.005) (− 1.048) (− 0.446) Dummy for competitive − 0.337* − 0.145 − 0.582* − 0.421* pricing strategy (− 1.994) (− 0.630) (− 2.035) (− 2.353) Dummy for mark-up pricing − 0.090 − 0.036 0.031 − 0.040 strategy (− 0.514) (− 0.157) (0.110) (− 0.218) Dummy for at home placing − 0.245 − 0.019 − 0.822* − 0.213 strategy (− 1.149) (− 0.078) (− 1.515) (− 0.963) Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 15 of 26 Table 3 Partial likelihood estimates for the Cox proportional hazard models (Continued) Model A Model B Model C Covariates MSEs Micro Small MSEs Micro Small MSEs Dummy for at industrial site − 0.960* − 0.234 − 2.091** − 0.980* placing strategy (− 2.554) (− 0.510) (− 2.801) (− 2.433) Dummy for salesmanship − 0.221 − 0.228 − 0.080 − 0.138 promotion strategy (− 0.970) (− 0.724) (− 0.230) (− 0.581) Dummy for product innovation − 0.258 − 0.102 − 0.664* − 0.261 strategy (− 1.470) (− 0.446) (− 2.187) (− 1.452) † † Dummy for market place − 0.407 − 0.292 − 0.754* − 0.424 innovation strategy (− 1.844) (− 0.981) (− 2.143) (− 1.887) Dummy for high degree of − 0.223 − 0.095 − 0.575 − 0.099 competition (− 1.244) (− 0.396) (− 1.946) (− 0.530) Dummy for niche market − 0.149 − 0.134 0.004 − 0.143 strategy (− 0.818) (− 0.579) (0.011) (− 0.724) Dummy for external market − 0.997* − 0.468 − 1.455* − 1.168** linkage (− 2.454) (− 0.755) (− 2.509) (− 2.776) Log capital investment − 0.150*** − 0.146*** − 0.170*** − 0.158*** (− 7.936) (− 5.651) (− 5.613) (− 7.824) Dummy for retained earnings − 0.738*** − 0.861** − 0.472 − 0.688*** investment (− 3.779) (− 3.102) (− 1.546) (− 3.292) Dummy for internal financing − 0.850* − 1.009* − 0.873 − 1.001** (− 2.520) (− 2.403) (− 1.259) (− 2.810) Dummy for business financing − 0.328 − 0.596 − 0.254 − 0.294 (− 1.105) (− 1.509) (− 0.481) (− 0.973) Dummy for external financing − 0.577* − 0.716* − 0.426 − 0.642* (− 2.124) (− 2.124) (− 0.802) (− 2.245) Dummy for undercapitalization 0.392* 0.583* 0.162 0.220 challenge (2.189) (2.471) (0.527) (1.151) Dummy for inventory system − 0.670** − 0.768* − 0.614 − 0.752** (− 2.695) (− 1.998) (− 1.731) (− 2.869) Dummy for annual budget − 0.708 − 1.398 − 0.743 − 0.703 (− 1.513) (− 1.369) (− 1.345) (1.459) Dummy for poor accounting 0.160 0.313 − 0.097 0.074 system (0.675) (1.034) (− 0.235) (0.310) Dummy for owners training 0.423* 0.415 0.378 0.203 after establishment (2.035) (1.551) (0.982) (0.884) Dummy for workers training − 0.644* − 0.992* − 0.625 − 0.534 while they are on duty (− 2.459) (− 2.155) (− 1.570) (− 1.955) † † Dummy for Internet connection − 1.760 − 2.011 in the premises (− 1.732) (− 1.952) LHR test 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Wald test 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Log rank test 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Note: Figures in parentheses are P values, and †, *, **, and *** indicate statistically significant at the 10, 5, 1, and 0.1% level, respectively. Values in parentheses are the Z values Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 16 of 26 The log form of the owner’s age was applied in the model. As Davidsson (1991) and Amran (2011) found, there was a significant negative relationship between the business founders’ age and business survival. This suggests that entering a self-employment at matured age is associated with a higher dropout rate than younger entrepreneurs. Education qualification does not appear to be a significant for a survival of MSEs. As Taylor (1999) argued, this suggests that to enter self-employment and survive, entrepreneurs do not need to have a high level of formal education. Moreover, there is no significant positive effect of TVET on the probability of a firm’s survival. The fitted model proves that previous labor market experience was an important fac- tor for survival. Consistent with the result of Taylor (1999), active labor market partici- pation prior to the start-up reduced the risk of dropout. But, as Kalleberg and Leicht (1991) found, there was no significant relationship between self-employment experience prior to the start-up and business survival. Previous general business and apprentice ex- perience did not have a significant impact on a firm’s survival. In line with Caliendo and Kritikos (2009), opportunity motivation significantly re- duced the risk of dropout, implying that the opportunity-driven entrepreneur’s prob- ability of survival was significantly higher than the necessity entrepreneurs. Entrepreneurial education and training prior to start-up appear to be important factor to stimulate entrepreneurial activity and business survival, as it had a significant nega- tive effect on dropout rate. Concerning firm and industry characteristics, start-up size has the expected positive effect on MSEs’ survival, implying the larger the size of the setup, the lower is the prob- ability of dropout. We capture the impact of financial resource at the time of start-up using the size of the start-up capital, and it turns out that larger amount of start-up capital reduced the risk of dropout. Entrepreneurs, who invested more than Birr 100,000 (USD 3700), had a significantly higher probability of survival than the others, suggesting that an adequate amount of start-up capital is a key factor for a firm’s sur- vival. Unlike Kalleberg and Leicht (1991), a solo type of business setup had a significant positive effect on MSEs’ survival than cooperatives. This might be associated with the complete control and decision-making power that the owner enjoys, as well as being free from corporate tax. Youth-owned MSEs engaged in the manufacturing sector had a lower risk of dropout. This is consistent with Taylor’s(1999) results, implying that the service sector was ex- posed to a higher risk of dropout. Unpredictably, supports obtained by the owner at time of start-up did not appear to have significant association with the survival of MSEs. Last but not the least, our analysis proves that formal system is a significant de- terminant of MSEs’ survival, and registration during the start-up phase reduced the probability of dropout. Overall, gender, age, previous labor experience, motivation, entrepreneurial education and training, initial size, legal form, size of start-up capital, industry type, and formal system, therefore, are important factors in determining the probability of MSEs’ survival in Ethiopia. As section B of Table 3 presents, in line with the results of Chittithaworn et al. (2011), marketing strategy was an important factor in explaining MSEs’ survival rate in Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 17 of 26 Ethiopia. Concerning marketing mix strategy, the pricing and place strategies appeared to be significant factors in explaining the probability of MSEs’ survival. Competitive pri- cing strategy had a significant negative association with MSEs’ dropout rate. This re- flects that it is better to develop status quo pricing objective for MSEs to survive and be successful since most of them are operating under high degree of competition eman- ating from similar business; about 76% of the MSEs reported that they were experien- cing high competition due to an increase in the number of similar businesses. The location of enterprises is one of the elements of a business strategy that can address the challenges of competition by delivering the right product to the right customer at the right time. The study reveals that an enterprise which was based in an industrial area was more likely survive than those at residential places. This reflects that industrial areas offer better access to customers, raw materials, labor, and transportation. Marketing place innovation was statistically significant at the 10% level and had a positive association with survival rate, suggesting that reallo- cating for innovative marketing places was an important factor for attracting new customers and expanding market shares. Moreover, external market linkages had a significant positive association with the probability of business survival, suggesting that operating under linked marketing environment significantly reduced the risk of dropout. In line with the results from McMahon (2001b), our fitted model proves that financial management does appear to be an important factor for MSEs’ survival. There was a negative association between capital investment and a risk of dropout, implying that the more the manager/owner invests in capital equipment, the higher the probability of survival the business had. Consistent with the results of Fadahunsi (2012), a manager/owner who is willing to reinvest retained earnings had a propensity to survive more than businesses that were not willing. This sug- gests that a less costly and internal source of financing (retained earnings) is a key factor of MSEs’ survival since access to external formal sources of financing was a major challenge for MSEs in Ethiopia; almost 32 and 26% of managers/owners reported that a lack of access to finance was a major constraint to revenue growth and business improvement, respectively. Like Carpenter and Petersen (2002), the present paper proves that the MSEs which used retained earnings to finance their working capital had a propensity to survive more compared to the MSEs which used other type of sources. The study also proves that there existed a positive association between external source of financing capital investment and MSEs’ survival, implying that those firms which finance their capital investment from external source had higher survival rate. Consistent with the results from Read (1998), the study shows that undercapitalization in- creased the risk of dropout, reflecting that MSEs which suffered from a shortage of funds to finance their investment survived less than those which did not. Like Mongare and Nasidai (2014), the study proves that the survival of MSEs was highly associated with the inventory system of the business, reflecting that efficient and modern inventory systems reduced risk of demise. As Savery and Luks (2004) found, the study reveals that training for workers was an important factor for MSEs’ survival, suggesting that building human capital had a pro- pensity to reduce the risk of dropout. In contrast, there exist inconsistent results Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 18 of 26 concerning the effect of owners’ training on business survival, as it appeared to be posi- tively associated with dropout rate. Moreover, the fitted model attests that business sur- vival was positively associated with Internet technology adoption, which is consistent with the result of Warren (2004). This suggests that MSEs with Internet connection in their business premises had a higher probability of survival as they were more con- nected with the world and had better access to information. Overall, the study confirms that the survival of an enterprise is also highly associated with business strategy characteristics of MSEs. Marketing management and financial management strategies play a significant role on reducing the risk of dropout of MSEs in Ethiopia. Moreover, human capital development and access to Internet are also important elements to increase the MSEs’ probability of survival. So far, the Cox proportional hazard model is estimated for two different data setups; model C estimates all the hypothesized variables to see if there is a sig- nificant discrepancy on the covariate effect on the hazard rate. As section C highlights consistent with the first model, gender, age, labor market experience, motivation, legal form, and industry type appeared to be significant factors for business survival. Moreover, unlike the first part model, entrepreneurship educa- tion, size of the enterprise, level of the start-up capital, and legal registration ap- peared to be insignificant factors in this model for the survival of enterprises. In line with the results from the second model, this fitted model proves that mar- keting management, financial management, controlling system, human capital development, and Internet technology appeared to be important factors for mi- cro- and small-scale enterprises’ survival. 4.3 The group survival rate analysis The second and third column of section A highlights the results from group analysis with model A specification. As the descriptive analyses highlight, female entrepreneurs were apt at establishing a micro-sized enterprise, but the results from the survival esti- mates show that female entrepreneurs who owned micro enterprises experienced a sig- nificant higher dropout rate. This suggests that gender is one of the factors behind less survival rate among the micro-level enterprises compared to small enterprises. Unex- pectedly, apprenticeship experience prior to start-up had a significant positive effect on the dropout rates for micro enterprises. The justification for this result demands further study on the subject matter, but the probable explanation could be the gap between the level and quality of apprenticeship experience with respect to the nature of and demand for micro enterprises. Entrepreneurship training drives small enterprise owners to experience low level of dropout risk compared to micro enterprise founders. This confirms that entre- preneurial training is one of the explanations for survival deferential between micro and small enterprises. The analysis also depicts that large value of start-up capital had a significant and pronounced negative effect on the risk of dropout from the small enterprises’ sample. This implies that it is an important factor accounted for survival rate deferential between micro and small enterprises. The relationship be- tween engaging in a manufacturing sector and the risk of dropout is significant Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 19 of 26 and negative for micro enterprises. We also found that formality had a significant negative association with the risk of dropout for micro enterprises. The model estimation identifies what owner, firm, and industry characteristics identi- fiable at start-up period explained deferential of survival rate between micro and small enterprises. Accordingly, we found that gender, entrepreneurial training, and size of start-up capital were the major factors accounted for the better success of small enter- prises compared to micro enterprises. From sub-columns 2 and 3 of model B section, the estimation for group analysis shows that pricing and placing strategy, product and marketplace innovation, competi- tion, and market linkage appeared to be the majors accounting for the better success of small enterprises compared to micro enterprises. Status quo pricing objective led small firms to survive more compared to micro enterprises. Locating a business at residential and industrial place reduced the risk of dropout for small enterprises. Like Ibidunni et al. (2014), the paper depicts that product innovation reduced the risk of dropout for small enterprises, calling the firms to undertake a progressive product innovation process to meet the demand of their customers and expand their market. Moreover, high degree of competition had a significant positive association with the survival of small enterprises. This suggests two things: (i) during a high competition period, managers/owners probably pressured from their competitors to enter new mar- kets, renovate their products, or introduce new products so that their market increases; (ii) high degree of competition in a specific product line might imply the high demand for the product. External market linkage had a negative association with the risk of dropout for the small enterprises. Overall, our study proves that marketing manage- ment strategy was one of the explanations for survival deferential between micro and small enterprises. Regarding financial management characteristics, capital investment, retained earnings investment, external financing, and inventory system had a significant negative associ- ation with the probability of dropout for micro enterprises. Like the results from the pooled sample, shortage of investment funds had a significant positive effect on micro enterprises’ dropout rate. The fitted model also confirmed that workers’ training played a key role on reducing the probability of dropout for micro enterprises. The results from model B specification identify what business strategy characteristics explained why the survival rate deferred between micro and small enterprises. Pricing and placing strategy, innovation, competition, and external market linkage were the probable major factors accounted for the better success of small enterprises compared to micro enterprises. 5 Conclusions This paper examines the survival of youth-owned MSEs and the associated factors based on a data obtained from 941 youth-owned MSEs in urban Ethiopia. The study uses nonparametric and semi-parametric hazard models to estimate the survival rate and its associated factors. The nonparametric estimation results reveal that MSEs’ risk of dropout was high during the first two to four business operation years in Ethiopia. The analysis also confirms that there was survival differential between micro and small enterprises. Particularly, the risk of dropout reached its maximum at 4 years after Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 20 of 26 establishment for micro enterprises while it was after 2 years for small enterprises. This could be due to inclination of most support programs towards small enter- prises or overlooking of the importance of micro enterprises in the process of cre- ating viable and sustainable medium and large industries from both sides—the MSE promoters and micro scale enterprise owners. Gender, age, previous work experience, motivation, entrepreneurship training, initial size, ownership type, type of industry, and legal status (registration) were important factors in explaining the probability of MSEs’ survival in Ethiopia. The survival of MSEs was also highly influenced by business strategy features. Market- ing and financial management strategies played a significant role on reducing the risk of dropout. Moreover, human capital development and use of Internet technology were also important elements to increase the probability of MSEs’ survival. The group-based survival analysis (micro vs. small enterprises) reveals that gender, entrepreneurial training, start-up capital size, pricing and placing strategy, product and market place innovation, and external market linkage appeared to increase small enter- prises’ survival rate compared to micro enterprises. Endnotes Regional government offices provide a certificate for new startups upon registration, and it includes information about the initial size of the business. Micro enterprise is define as an enterprise employing less than five workers, while small is defined as an enterprise employing more than five but less than 30 workers. Appendix 1 Table 4 Survival and hazard rate from the Lifetable estimator for small/micro enterprises Duration Small Micro Survival rate difference (%) At risk Survival Hazard At risk Survival Hazard 0–6 372 1 0.0027 569 1 0.0024 0 6–12 358 0.9837 0.0034 533 0.9856 0.0055 − 0.19 12–18 320 0.9636 0.0072 430 0.9534 0.0103 1.02 18–24 280 0.9227 0.0044 348 0.8963 0.0037 2.64 24–30 251 0.8987 0.0094 286 0.8767 0.0103 2.20 30–36 210 0.8494 0.0043 233 0.8243 0.0064 2.51 36–42 181 0.8280 0.0040 183 0.7932 0.008 3.48 42–48 152 0.8083 0.0024 151 0.7561 0.0063 5.22 48–54 124 0.7967 0.0059 113 0.728 0.0153 6.87 54–60 103 0.7691 0.0018 83 0.6641 0.0046 10.5 60–66 80 0.7607 0.0024 63 0.6461 0.0000 11.46 66–72 60 0.7499 0.007 47 0.6461 0.0000 10.38 72–78 35 0.7190 0.0056 30 0.6461 0.0000 7.29 78–NA 25 0.6954 0.0000 19 0.6461 0.0000 4.93 Note: The survival rate difference is the difference of cumulative proportion of small and micro entrepreneurs’ survival rate up to specific time interval NA not applicable Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 21 of 26 Table 5 Proportional hazard test using chi-square statistics: Cox model analysis—model A Covariates P value Female 0.112 Age (at start-up) 0.559 Elementary school 0.523 College diploma and above 0.347 TVET 0.967 Active in labor market 0.959 Self-employment experience 0.333 Business experience 0.487 Apprentice experience 0.552 Opportunity 0.824 Entrepreneurship training 0.988 Initial number of workers 0.471 Sole proprietorship 0.329 Cooperative 0.731 Large start-up capital 0.954 Micro start-up capital 0.286 Product provider 0.611 Support 0.985 Registered 0.591 Global 0.951 Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 22 of 26 Table 6 Proportional hazard test using chi-square statistics: Cox model analysis—model B Covariates P value Female 0.193 Manufacturing 0.554 Sole proprietorship 0.378 Competitive pricing strategy 0.373 Mark-up pricing strategy 0.376 Place at home (shop) 0.351 Place at industrial place (shop) 0.328 Salesmanship promotion 0.476 Product innovation 0.326 Market place innovation 0.825 Degree of competition 0.385 Niche market 0.288 Market linkage 0.481 Capital investment 0.069 Retained earnings investment 0.290 Internal financing (working capital) 0.260 Business financing (working capital) 0.375 External financing (investment) 0.021 Undercapitalization 0.713 Inventory system 0.906 Annual budget 0.508 Poor accounting system 0.492 Owners training 0.155 Workers training 0.516 Internet 0.790 GLOBAL 0.378 Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 23 of 26 Appendix 2 Fig. 3 Graphical diagnostic Schoenfeld residual test: model A Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 24 of 26 Fig. 4 Graphical diagnostic Schoenfeld residual test: model B Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 25 of 26 Acknowledgements The authors would like to thank Martha Melese, the anonymous reviewers, and the editor of the journal for the important suggestions. We would also like to acknowledge the comments received at the 2015 CSAE workshop held at Oxford University, the Center for the Study of African Economies. However, the authors are responsible for any error that may have found. Responsible editor: David Lam This paper draws from research project on Youth Self-Employment in Ethiopia: Promoting Micro and Small Enterprises (MSEs) Through Government Support Programmes carried out under the auspices of the Association of Ethiopian Microfinance Institutions (AEMFI) with the financial support from Canada’s International Development Research Centre (IDRC). The views expressed in this brief are those of the authors and do not necessarily reflect those of IDRC and AEMFI. Competing interests The IZA Journal of Development and Migration is committed to the IZA Guiding Principles of Research Integrity. The authors declare that they have observed these principles. Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Author details 1 2 Department of Economics, Addis Ababa University, Addis Ababa, Ethiopia. Association of Ethiopian Microfinance Institutions, Addis Ababa, Ethiopia. Received: 15 May 2017 Accepted: 25 January 2018 References Agarval R, Audretsch DB. Does entry size matter? The impact of the life cycle and technology on firm survival. J Ind Econ. 2001;49:21–43. Amran NA. The effect of owner’s gender and age to firm performance: a review on Malaysian public listed family businesses. J Glob Bus Econ. 2011;2(1):104–16. Audretsch, D. and Mahmood, T. (1995) New firm survival: New results using a hazard function. The Review of Economics and Statistics, 77(1), pp. 97–103. Bates T. Survival patterns among newcomers to franchising. J Bus Ventur. 1998;13:113–30. Bates T. Analysis of young, small firms that have closed: delineating successful from unsuccessful closures. J Bus Ventur. 2005;20:343–58. Bigsten A, Kimuyu P, Söderborn M. Chapter 10: the manufacturing sector. In: Adam CS, Collier P, Ndung’uNS, editors. Kenya policies for prosperity. Oxford and Nairobi: Oxford University Press and Central Bank of Kenya; 2010. p. 243–66. Boswell J. The rise and decline of small firms. London: George Allen & Unwin; 1973. Bruderl J, Preisendorfer P, Ziegler R. Survival changes of newly founded business organizations. Am Sociol Rev. 1992; 57(2):227–42. Brush CG, Carter NM, Gatewood E, Greene PG, Hart MM. Venture capital access in the new economy: is gender an issue? In: Hart D, editor. The emergence of entrepreneurship policy: governance, start-ups and growth in the knowledge economy. Cambridge: Cambridge University Press; 2003. p. 141–55. Caliendo, M., and Kritikos, A. (2009). “I want to, but I also need to”: start-ups resulting from opportunity and necessity. IZA Discussion Paper No. 4661. Carpenter R, Petersen B. Capital market imperfections, high-tech investment, and new equity financing. Economic Journal. 2002;112(477):F54–72. Carter R, Van Auken H. Small firm bankruptcy. J Small Bus Manag. 2006;44(4):493–512. Chandler GN, Hanks SH. An examination of the substitutability of founders human and financial capital in emerging business ventures. J Bus Ventur. 1998;13:353–69. Chittithaworn C, Islam A, Keawchana T, Yusuf DHM. Factors affecting business success of small & medium enterprises (SMEs) in Thailand. Asian Social Science. 2011;7(5):180–90. Coad, A. and Tamvada, J. P. (2008). The growth and decline of small firms in developing countries. Papers on Economics and Evolution No. 0808. Coleman, S., Cotei, C. and Farhat, J. (2010). Factors affecting survival, closure and M&A exit for small businesses. Available at SSRN: http://ssrn.com/abstract=1768728. Cowling, M. (2009). The impact of entrepreneurship training and small business experience on future entrepreneurial activity in the UK. Brighton: Institute for Employment Studies. Cox, D. (1972). Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B (Methodological), 34(2), 187-220. Retrieved from http://www.jstor.org/stable/2985181. Davidsson P. Continued entrepreneurship: ability, need and opportunity as determinants of small firm growth. J Bus Ventur. 1991;6:405–29. Dayanandan R. Sustainability of micro and small enterprises: imperatives, myths and realities. Int J Entrepreneurship Bus Environ Perspect. 2012;1(2):129–42. Dobbs M, Hamilton R. Small business growth: recent evidence and new directions. Int J Entrepreneurial Behav Res. 2007;13:296–322. Fadahunsi A. The growth of small business: towards a research agenda. Am J Econ Bus Adm. 2012;4(1):105–15. Fairlie, R. W. and Robb, A. M. (2008). Gender differences in business performance: evidence from the characteristics of business owners survey. IZA Discussion Paper Series No. 3718. Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 26 of 26 FeMSEDA. Micro and small enterprises sector development yearly statistical bulletin. Addis Ababa: Federal micro and small enterprises development agency; 2014. Fox, J. (2002) Cox proportional-hazards regression for survival data. Available at: http://cran.r-project.org/doc/contrib/ FoxCompanion/appendix-cox-regression.pdf. (Accessed: 17 Apr 2017). Headd, B. (2000). Business success: factors leading to surviving and closing successfully. U.S. Small Business Administration, office of advocacy. Honig B. What determines success? Examining the human, financial, and social capital of Jamaican micro entrepreneurs. J Bus Ventur. 1998;13:371–94. Ibidunni O, Oluwole I, Ayodotun S, I. Product innovation, a survival strategy for small and medium enterprises in Nigeria. Europ Sci J. 2014;10(1):194–209. Jenkins SP. Survival analysis. Unpublished manuscript, Institute of Social and Economic Research. Colchester: University of Essex; 2004. Downloadable from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.176.7572&rep= rep1&type=pdf Kalleberg AL, Leicht KT. Gender and organizational performance: determinants of small business survival and success. Acad Manag J. 1991;34(1):136–61. Kangasharju A, Pekkala S. The role of education and self-employment success in Finland. Growth and Change. 2002;33:216–37. Mata J, Machado J. Firm start-up size: a conditional quantile approach. Eur Econ Rev. Elsevier. 1996;40(6):1305–23. Mata J, Portugal P, Guimaraes P. The survival of new plants: start-up conditions and post-entry evolution. Int J Ind Organ. 1995;13:459–82. McMahon, R. G. (2001b), Business Growth and Performance and the Financial Reporting Practices of Australian Manufacturing SMEs. Journal of Small Business Management, 39: 152–164. Mongare ME, Nasidai SE. The impact of information communication technology on inventory control systems in transport organization: a case study of Kenya ferry services. Eur J Logistic Supply Chain Manage. 2014;2(1):17–41. Moores K, Mula J. Contextual and strategic impacts on the form of management accounting systems. School of Business Discussion Papers, paper 49. Robina: Bond University; 1993. O'Dwyer M, Gilmore A, Carson D. Innovative marketing in SMEs. Eur J Mark. 2009;43(1/2):46–61. Pena I. Intellectual capital and business start-up success. J Intellect Cap. 2002;3(2):180–98. Read L. The financing of small business: a comparative study of male and female business owners. London: Routledge; 1998. Rosa P, Carter S, Hamilton D. Gender as a determinant of small business performance: insights from a British study. Small Bus Econ. 1996;8(6):463–78. Savery LK, Luks JA. Does training influence outcomes of organizations? Manage Dev. 2004;23(2):119–23. Schiller BR, Crewson PE. Entrepreneurial origins: a longitudinal inquiry. Econ Inq. 1997;35(3):523–31. Skrt B, Antoncic B. Strategic planning and small firm growth: an empirical examination. Manag Glob Transitions. 2004; 2(2):107–22. Storey D. Understanding the small business sector. London: Routledge; 1994. Taylor MP. Survival of the fittest? An analysis of self-employment duration in Britain. Econ J. 1999;109:C140–55. Van Praag C. Business survival and success of young small business owners. Small Bus Econ. 2003;21(1):1–17. Warren M. Farmers online: drivers and impediments in adoption of internet in UK agricultural businesses. J Small Bus Enterprise Dev. 2004;11(3):371–81. World Bank. Youth employment in sub-Saharan Africa. Washington, D.C.: World Bank; 2014. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png IZA Journal of Migration Springer Journals

Correlates of business survival: empirical evidence on youth-owned micro and small enterprises in Urban Ethiopia

Loading next page...
 
/lp/springer-journals/correlates-of-business-survival-empirical-evidence-on-youth-owned-LaoD7IRUTD
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Economics; Population Economics; Labor Economics; Migration; Demography
eISSN
2520-1786
DOI
10.1186/s40176-018-0122-x
Publisher site
See Article on Publisher Site

Abstract

woldehanna@gmial.com Department of Economics, Addis This paper investigates the effects of person-, firm-, industry-, and business strategy- Ababa University, Addis Ababa, specific characteristics on the survival of youth-owned urban micro and small Ethiopia Full list of author information is enterprises in Ethiopia. It employs nonparametric and semi-parametric methodologies available at the end of the article using a retrospective data. The hazard rate reaches the highest point at business 2 years for micro enterprises and 4 years for small enterprises. Owner-, firm-, and industry-specific characteristics are important factors for micro and small enterprises’ (MSEs’) survival. Marketing and financial management strategies are playing a crucial role on extending MSEs’ survival duration. The study implies there should be more effective and longer period of support for micro than small enterprises. JEL classification: C14, C41, D21, J23, L26, M21 Keywords: Youth self-employment, Micro- and small-scale enterprises, Hazard model 1 Introduction The co-existence of micro and small enterprises (MSEs) can be interesting; concomi- tantly, the performance of MSEs in terms of survival and growth needs to be assessed in detail. Identifying the growth and survival factors of MSEs is important as it estab- lishes the base for preparing a policy framework and strategy that safeguards the success of MSE operators. However, there are few detailed studies, particularly survival analysis, which identify the factors influencing the survival of MSEs, in general, and youth-owned MSEs, in particular. Moreover, there are hardly research outputs intended to identify factors influencing the survival of MSEs in Ethiopia using firm- specific data. Thus, the existed knowledge gap and the focus for the sectoral develop- ment motivated us to empirically analyze the factors associated with the risk of dropout of MSEs in Ethiopia. The paper intends to identify how the socio-demographic and economic characteris- tics of the owner- and firm-level attributes at start-up phase are associated with the survival of youth-owned MSEs. Specifically, the objectives of the paper are (1) to investigate the effect of owner- and firm-specific attributes on MSEs’ survival, (2) to analyze how strategic thinking and systematic decision-making are partly responsible for the failure or success of entrepreneurs, and (3) to identify owner, firm, industry, © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 2 of 26 and business strategy characteristics that can explain the difference among micro and small enterprises’ survival rate. The present paper uses a nationwide but urban-based representative cross-sectional survey data which is rich with retrospective information. We apply nonparametric and semi-parametric survival models to study and assess the relative strength of the associ- ated factors with survival of the MSEs and estimate the survival rate. There are major concerns why firms eventually fail to become established firms. Ex- tensive theoretical and empirical literature in developed economies have identified micro-level factors, determining the emergence and success of enterprises. Most of the studies have shown that owner- and firm-related characteristics are the basic factors that determine the success of a firm (see, e.g., Bruderl et al. 1992; Storey 1994; Coleman et al. 2010). These characteristics are typically identifiable prior to the start-up and comprise personal and psychological characteristics, such as gender, education and training, entrepreneur’s mind setup, previous work experience, and ethnicity. Most studies indicated that women-owned firms perform less in terms of different performance measures, such as number of employees (Rosa et al. 1996), revenue (Rosa et al. 1996; and Brush et al. 2003), and profitability and survival (Fairlie and Robb 2008). Some empirical analyses suggested that owner’s/manager’s age tends to be nega- tively related to risk of dropout (see Bates 1998; van Praag 2003). In contrast, there are empirical evidences that prove the negative association of owner’s/manager’s age with the growth of small firms (see Boswell 1973; Davidsson 1991; Amran 2011). The owner’s human capital quality, which can be measured by education, prior work, and managerial experience, is also pointed out as a positive factor for the performance of firms (see, e.g., Schiller and Crewson 1997; Honig 1998; Chandler and Hanks 1998; Kangasharju and Pekkala 2002; Pena 2002). Cowling (2009) has found a positive effect of school-based entrepreneurial training on job creation by small firms in the UK. He also found a positive relation between entrepreneurs who received government-backed entrepreneurial training and exporting intensity of small firms. The size of a start-up has been considered as a key factor for self-employment sur- vival, and it is widely documented that larger size entry is a signal of commitment and self-confidence, which reduces the occurrence of an entry mistake and the risk of a fail- ure due to diseconomies of scale (Mata et al. 1995; Agarval and Audretsch 2001). Ex- tensive empirical works also focus on assessing the impact of industry-specific features on start-up success and reveal the presence of relationship between firm’s success and the type of the sector the firm is operating, level of competition within the industry, linkage, market, and technological change and innovation (see Bruderl et al. 1992; Audretsch and Mahmoud 1995; Mata and Machado 1996; Headd 2000; Bates 2005; Carter and Van Auken 2006). For example, Coleman et al. (2010) indicated that firms involved in retail and service sectors are more prone to failure due to low barriers to entry and high levels of competition. In the contemporary, global, dynamic business environment, strategic thinking and systematic decision-making are also important for the growth and survival of small enterprises (see Dobbs and Hamilton 2007; Skrt and Antoncic 2004). The characteristics of business management strategies, including marketing and finan- cial management, are found to have a strong positive relationship with business performance (O'Dwyer et al. 2009). Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 3 of 26 A small business in which the owner reinvests retained earnings has the propensity to grow more than businesses that are not willing to reinvest (Fadahunsi 2012). Con- sistently, internal financing is presumed as the main reason for the success of most firms in the USA (Carpenter and Petersen 2002). The study by Read (1998) showed that over half of her total sample of small firms in the UK suffered from being under- capitalized. Moreover, evidences highlighted that internal control, including financial report and inventory management, allows a business to enhance effectiveness and gain competitive advantage (see Moores and Mula 1993; Mongare and Nasidai 2014). Currently, MSEs in Ethiopia are becoming important vehicles to create new jobs for millions of unemployed youth (FeMSEDA 2014). Despite the focus given to the sector, a large number of MSEs are expected to eventually close or stagnated at starting phase. As reports and studies documented, there are internal and external factors which per- vasively hamper the productivity improvement and survival of the MSEs; shortage of working capital, a lack of marketing skills, poor location of business sites, failure to take risk, and over emphasis on short-term profit are internal factors affecting the survival rate of MSEs. In addition, a lack of access to efficient infrastructure and factor of production, a nontransparent regulatory environment, bureaucracy burden, high inci- dence of bribing, and a lack of appropriate skills and training are also documented as external factors hindering the success of MSEs (see World Bank 2014; Dayanandan 2012; Bigsten et al. 2010 and FeMSEDA 2014). Hence, it is worth to investigate the explanatory factors that influenced the survival and exit of businesses using a micro dataset which is representative. The paper is organized as follows: Section 2 presents the econometric model specification employed in the paper. Section 3 presents the data and descriptive statistics. Section 4 reports the empirical results and interprets the esti- mated parameters. Section 5 presents the conclusion remark and the policy implications. 2 Empirical model specification: survival models Survival rate is often the most practical index to measure the success of a firm. The analysis examines the elapsed time since the entry to the state at time t = 0. The present paper specifies a survival model that accounts for the sequential nature of the data and is able to handle censoring. Let T represents survival time and is a random variable with a cumulative distribu- tion function F(t) = Pr(T ≤ t) and a probability density function f(t)= dF(t)/dt. The reverse cumulative distribution function of T provides a new function called survival function S(t), and it is defined as StðÞ ¼ PrðÞ T > t¼ 1−FtðÞ: ð1Þ The hazard rate, the instantaneous risk of demise at time t conditional on the firm having survived up to that point in time, is the motive to be estimated in our survival analysis. The hazard function is an absolute slope of log survival function and it is defined as follows ftðÞ d lnStðÞ htðÞ ¼ ¼ − ð2Þ 1−FtðÞ dt and the cumulative hazard function H(t)is Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 4 of 26 HtðÞ ¼ − lnStðÞ: ð3Þ The survival function describes the probability of surviving past time t. The hazard function describes the relative likelihood of the event occurring at time t, conditional on the subject’s survival up to time t. The hazard rate shows the instantaneous rate of failure at time t and ignores the accumulation of hazard up to time t (see Jenkins 2004 for details). The present paper utilizes both nonparametric and parametric models to analyze the survival rate of MSEs in urban Ethiopia. We estimate the nonparametric survival and cumulative hazard analysis using the Kaplan-Meier method. The paper uses the Kernel-smoothed hazard estimation method for smoothed hazard analysis, as it effi- ciently detects the times where the most rapid changes of the hazard function occur. 2.1 Nonparametric approach The above modeled survival and hazard functions can be estimated without making as- sumptions about the shape of the relevant functions. The Kaplan-Meier estimator is a nonparametric estimator that does not make any assumptions about the distribution of exit times and how covariates shift the hazard functions (Jenkins 2004). Hence, the pro- portion of surviving to survival time t and the corresponding integrated hazard func- tion is estimated by S t ¼ 1− ð4Þ jjt ≤ t ∧ ∧ H t ¼ − ln S t ð5Þ j j where d denotes the number of failed enterprises at t and n is the number of enter- j j j prises at risk of failure immediately prior to t and defines as the sum of those who have a censored or completed spell of length t or longer. The Kernel-smoothed estimator is a weighted average of Δ H ðtÞ over event times that are within a bandwidth distance b ∧ ∧ of t. The Kernel-smoothed estimator considers the jumps of H ðtÞ and variance of H ð tÞ at the event times as follows: ∧ ∧ ∧ Δ H t ¼ H t − H t j j j−1 ∧ ∧ ∧ ∧ ∧ ∧ ð6Þ V H t ¼ V H t − V H t j j j−1 where t =0. As the effect of an enterprise’s initial size on its survival probability is one major ob- jective of our paper, we grouped the dataset into micro and small enterprises. Accord- ingly, for ease of interpretation, we grouped enterprises into 6-month intervals and execute the so-called Lifetable estimator. In essence, the Lifetable estimator processes the same idea as the Kaplan-Meier method, but it was explicitly designed to handle situations where there is grouped survival time data. 2.2 Semi-parametric hazard model The focal point in our survival analysis is the hazard function, allowing for time invari- ant covariate effects as an econometric framework. The Cox proportional hazard model Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 5 of 26 has considerable flexibility and is the most widely used hazard regression model in em- pirical studies. This model, proposed by Cox (1972), enables us to estimate the relation- ship between the hazard rate and covariates without having to make any assumptions about the shape of the baseline hazard function, referring to the model as a semi- parametric model. A survival model examines the relationship between the survival distribution and the covariates and mostly entails the specification of a linear-like model for the log hazard. logh ðÞ t ¼ α þ β x þ β x þ … þ β x j j1 j2 jk 1 2 k ð7Þ h ðÞ t ¼ e α þ β x þ β x þ … þ β x j j1 j2 jk 1 2 k where x’s are the covariates and α represents a log-baseline hazard because if all x’s are zero, then h (t)= e . The Cox model leaves the baseline hazard function α(t) = log h (t) j 0 unspecified. Thus, the hazard function is parameterized as follows: β x þβ x þ…þβ x ðÞ j1 j2 jk 1 2 k h ðÞ t ¼ h ðÞ t e ð8Þ j 0 where h (t) is the so-called baseline hazard which depends only on t. The change in the explanatory variables, therefore, induces multiplicatively proportional shift in the haz- ard rate relative to the baseline hazard. This implies that the model assumes that the hazard functions of any individuals with different values on one or more covariates dif- fer only by a factor of proportionality. Based on Eq. (8), the hazard ratio for two individuals, j and i, is defined as follows: h ðÞ t ¼ exp β x −x þ β x −x þ … þ β x −x ð9Þ j1 i1 j2 i2 jk ik 1 2 k h ðÞ t where the hazard ratio is independent of time t. In the way to compute the hazard ra- tio, the baseline hazard h cancels out of the computation, implying that the relation- ship between the hazard rate and the explanatory variables is estimated without taking h into account. Consequently, the Cox model becomes a proportional hazard model, but the proportional hazard assumption should be tested. Instead of the maximum-likelihood method, the Cox model can be estimated by the method of partial likelihood (PL), because it requires less assumption and so is more robust (Fox 2002). The Cox PH estimation function is “partial” likelihood function because it only considers probabilities for failed events explicitly, referring that the baseline hazard function is removed from the estimation process. The estimation process started by sorting the ordered failure times and PL is derived from the product of the conditional probability of a failure time, given the risk set at that time and that one failure is to happen. Hence, the PL only considers the contribution from failure times, not from the right-censored times (we are referring the cases). This implies that the Cox model is conspicuously ignoring all information available at times when no failure occurs. Cox (1972) has found little efficiency loss due to ignoring spells in which no failure occurs. 3 Data and descriptive statistics 3.1 Data The dataset of many empirical studies on the survival of enterprises is derived from panel and cohort surveys and/or a database administered by a recognized office, usually Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 6 of 26 government offices. However, there are two basic problems using the formal registration database of MSEs collected by the Federal Micro and Small Enterprises Development Agency (FeMSEDA) in this particular study. Firstly, the existing registry of MSEs is based on information obtained during the formal registration of the enterprises, and yet most of the enterprises are informal and/or not registered by the agency. Secondly, it mainly pro- vides details of the business, such as registered address, setup date, industry type, owner- ship type, start-up capital, to date capital, and firm size during the start-up. Thus, it was found to be inadequate in terms of providing all the required information and current status of the MSEs to undertake a survival analysis. To this end, this study employs information from the “youth self-employment” survey collected by the Association of Ethiopian Microfinances Institutions (AEMFI), which has details on the parameters required for a survival analysis of MSEs in Ethiopia. The youth self-employment research project primary concern was to obtain fresh informa- tion and evidence on strategies and policies that enhance youth self-employment within the framework of MSE development in Ethiopia. The survey collected quantitative in- formation of the sampled MSEs on their current status; profiles; retrospective informa- tion about the initial start-up data, including owner- and firm-level characteristics; and ongoing and past business strategies since start-up. The stratum was regions (Tigray, Amehara, Oromia, SNNP, and Hareri) and two city administrators (Addis Ababa and Dire Dawa). The survey focused on all regional cap- ital cities and the two city administratives (Addis Ababa and Dire Dawa). In addition, two towns from each big-four region were selected based on their economic activity and population size that they had proportion to the total urban population. Hence, 15 major urban centers were included directly in the sampling strategy given that they constitute about 91% of the Ethiopian urban population, and these are the major pri- vate sector hubs throughout the country. Regional sample size was disaggregated to the urban areas by purposeful proportion al- location; regional capital cities had 50% of the total sample size allocated for the region while the two towns accounted for 25% each. We hypothetically drew a 200-m-radius grid, which was an enumeration area (EA) for the survey, from the city midpoint. We pre- ferred urban centers based on a mere assumption that household members in these areas are engrossed in self-employment. Hence, households within the EA were the data frame from this particular survey. Randomly selected households were screened through listing questionnaire before a mem- ber is selected for the main survey. The listing questionnaire had a filter question to classify the members for existed or closed enterprise sampling. The key filter questions included the following: whether there is a member who is currently operating a micro or small enterprise and whether there is a member who had started but closed a business for any reason but due to business challenges before today. In case of multiple MSE operators and/or dropouts within a selected household, only one household member was selected for the interview. The survey employed a population sampling process, implying both existing enterpri- ses—the start-ups observed at the time of the survey—and closed enterprises—the start-ups that have been already dropped out at the time of the survey—were inter- viewed. Closed firms, in this particular study, refer to those that have discontinued their operations, which included enterprises sold, bankrupted, and merged with other enter- prises due to business challenges. Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 7 of 26 A multi-stage sampling technique (with stratification by gender, enterprise size, and industry type) was applied to randomly selected 1109 enterprises from existing and closed MSEs owned by the youth (ages ranging from 18 to 34 years). The enterprises operating during the survey were 909, and the already closed were 200. Sampled enter- prises were distributed across different industries such as agriculture, manufacturing, and services. To make the sample representative of all sectors, attempts were made to take proportional samples from enterprises who are engaged in manufacturing, service, trade, and urban agriculture. Moreover, stratification was made by gender of the owner and the enterprise size (micro and small). The federal and regional MSE development agency/bureaus played a key role in facilitating the entire sampling strategy. Existing enterprise owners and those who closed their business were asked about the specific date of business start-up and exist date, in case the start-up had already closed. This information allowed the researchers to establish a business operation spell for each firm, and the spell might be either completed or right-censored at the time of survey. The established spell is the dependent variable with a value of 1, if the spell is right-censored, and 0, if the spell is completed. The survey had also de- tailed information on the socio-demographic and economic background of the owner/manager, as well as firm, industry, and business strategy characteristics. Moreover, the sample survey had detailed information regarding the feature of the owner as well as the firm during the setting up of the business, the quality of the support program, the respective gaps and challenges experienced by MSE opera- tors, and the labor market status of the youth before and after self-employment, which are important variables in the survival analysis. Retrospective information determines the validity of a research, since it can be sub- scribed to self-justification bias easily. To account for potential survey bias and retro- spective bias, the data used in this paper consists 941 enterprises which were established from 2008, and thereafter, expecting the start-ups’ owners can remind and provide information with less inaccuracy later than this period. Thus, the sample includes 941 enterprises; 569 enterprises are grouped as micro and 372 are as small enterprises. The group classification is based on registration certificate and their num- ber of workers . Moreover, out of the total sample, we observe 18% completed business operation spell and the remaining 82% is right-censored at time of interview. 3.2 Descriptive analysis Table 1 presents the hypothesized variables that were identified at the time of start-up and on business activity, along with their description using summery statistics. Part A and part B of the table describe the socio-demographic and economic characteristics of the owner, and firm- and industry-specific characteristics, respectively. Part C of the table highlights the description of business strategy variables. About 26% of the enterprises were owned by female (i.e., as a solo and group) entre- preneurs, implying the dominance of male entrepreneurs in MSEs in Ethiopia. The average age of the owners at the time of the start-up was 24. The owners’ human cap- ital quality is measured by formal education attainment and work and business experi- ence prior to the start-up. Accordingly, during the start-up period, 25% of the entrepreneurs had only elementary school attainment, while about 17% of the Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 8 of 26 Table 1 Definition of variables and their corresponding mean and standard deviation value Covariates Mean Std.dev Part A: owner-specific characteristics Dummy for female 0.255 0.436 Age of the firm owner at the time of establishment 24 4.42 Dummy for elementary education attainment 0.249 0.432 Dummy for college diploma and above attainment 0.166 0.372 Dummy for TVET education 0.001 0.300 Dummy for active in a labor market prior to the new business 0.514 0.500 Dummy for self-employment experience prior to the new business 0.122 0.328 Dummy for apprentice experience prior to the new business 0.329 0.154 Dummy for business experience prior to the new business 0.501 0.500 Dummy for entrepreneurship training prior to the new business 0.143 0.351 Dummy for short-term training prior to the new business 0.350 0.477 Dummy for opportunity motivation 0.690 0.463 Part B: firm and industry specific characteristics Firm size (initial number of workers) 8 10.6 Dummy for large start-up capital (more than 100,000) 0.055 0.229 Dummy for micro start-up capital (less than 25,000) 0.833 0.373 Dummy for sole proprietorship start-up 0.383 0.486 Dummy for cooperative start-up 0.262 0.440 Dummy for metal and wood work 0.258 0.438 Dummy for leather products 0.015 0.121 Dummy for textile and cloth 0.048 0.214 Dummy for construction 0.197 0.398 Dummy for service 0.114 0.318 Dummy for product provider 0.705 0.456 Dummy for support beneficiary 0.661 0.474 Dummy for registered during establishment 0.810 0.393 Part C: business strategy-specific characteristics Marketing management Dummy for competitive pricing strategy 0.718 0.450 Dummy for mark-up pricing strategy 0.426 0.016 Dummy for at home placing strategy 0.238 0.426 Dummy for at industrial site placing strategy 0.084 0.277 Dummy for salesmanship promotion strategy 0.258 0.438 Dummy for product innovation strategy 0.733 0.442 Dummy for market place innovation strategy 0.228 0.420 Dummy for niche market strategy 0.734 0,442 Dummy for high degree of competition 0.758 0.429 Dummy for external market linkage (proxied by siting cluster areas) 0.104 0.306 Financial management Capital investment (in Birr) 22,280 10,205 Dummy for retained earning investment 0.875 0.331 Dummy for internal financing (proxied by working capital financing from internal source) 0.693 0.462 Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 9 of 26 Table 1 Definition of variables and their corresponding mean and standard deviation value (Continued) Covariates Mean Std.dev Dummy for business financing (proxied by working capital financing from suppliers) 0.058 0.008 Dummy for external financing (proxied by investment financing from external source) 0.319 0.466 Dummy for undercapitalization challenge 0.366 0.482 Internal control management Dummy for inventory system (proxied by using detail and modern list of input stocks at 0.244 0.430 any time) Dummy for annual budget 0.079 0.269 Dummy for poor accounting system (proxied by a significant difference between the 0.188 0.391 book value and the actual value of revenue, cost, and profit) Human capital management Dummy for owners training after establishment 0.363 0.481 Dummy for workers training while they are on duty 0.169 0.012 Internet technology Dummy for Internet connection in the premises 0.029 0.170 Number of observation 941 Note: The left side for the owner’s education covariate is the secondary school attainment; for start-up capital, covariate is small-sized capital (between 25,000 and 100,000); for the legal form of the firm, covariate is partnership; for pricing strategy, covariate is negotiation with the customer; and for placing strategy, covariate is shopping district entrepreneurs had at least a college diploma and only 10% of the entrepreneurs had Technical and Vocational Education and Training (TVET). There were few owners who had self-employment experience prior to the start-up. Regarding previous experi- ence, 33 and 50% of the owners had apprenticeship and business experience prior to the start-up, respectively. The entrepreneurs who started the business out of push factor, such as to escape from being unemployed, lack of skill and proper education, and those who are forced by friends and family factors, are categorized as necessity-driven enterprises. On the other hand, those owners who started business as result of observing the enabling busi- ness environment and believe that self-employment is better than being employed by others are classified as opportunity-driven enterprises. Accordingly, the motivation be- hind being self-employed for about 69% of the owners was opportunity driven. The average initial number of workers, which is a proxy for initial size of the firm, was eight, implying the dominance of micro-level enterprises in our sample. About 83% raised a start-up capital of less than or equal to Birr 25,000 (USD 925), while only 6% raised a start-up capital of more than Birr 100,000 (USD 3700) during establishment. Regarding industry type, most of the enterprises are product providers (almost 71%), of which about 37 and 28% of the enterprises were engaged in metal and woodwork and construction sector, respectively. Part C of the table shows that about 72% of the firms set price for their product and service conditioning on the competitors’ position. This suggests that the firms’ pricing objective is status quo; MSEs seek to keep their product prices consistent with or con- ditioned on the competitors’ position. With respect to production innovation, about 73% of the owners/managers undertook screening and evaluation of the existing prod- uct mix line and made improvements and/or developed new products. Most enter- prises, about 76% of enterprises, were experiencing a high degree of competition in Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 10 of 26 recent times due to the expansion of similar businesses. Niche marketing is found to be a significant marketing strategy in the sector, as 73% of the owners/managers focused their marketing on serving a preferably small niche market. The average value of capital investment made by the enterprises since establishment was less than Birr 25,000 (USD 925). Most owners/managers, 88%, opted to reinvest their earnings in the business. Not few, almost 37%, enterprises experienced a shortage of capital investment funds during their business operations. It turns out that 36% of the owners improved themselves through attaining short- term trainings after establishment. Moreover, almost 14% of the enterprises arranged skills training for their workers. Last but not the least, for adoption of information and Internet technology, only 3% of the enterprises had an Internet connection in their business premises. 3.3 Differences between micro and small enterprises In order to reveal differences between owner- and firm-related characteristics among micro and small enterprises, a t test mean equality is conducted. The reported P values show whether the means were equal at a significance level of 5%. Table 2 compares micro- and small-sized firms using business operation survivors and dropout samples. The result indicates that the majority of female entrepreneurs were apt at establishing a micro-sized enterprise. The share of female entrepreneurs was significantly higher among the micro enterprises (almost 47%) within the dropout MSE groups. The share of owners with apprentice experience within the dropout sam- ple was significantly higher among the micro enterprises (almost 37%) than the small enterprises (almost 20%). In the survivor sample, a significantly higher share of owners with entrepreneurship training at the start-up period found among the small (almost 18%) than the micro en- terprises (almost 13%). Opportunity-driven entrepreneur’s share is significantly lower among the small (almost 38%) than the micro enterprises (almost 61%) in the dropout sample. In the survivor sample, small business owners were likely to invest, on average, higher amounts compared to the micro enterprises. In contrast, the share of owners who invested less than Birr 25,001 was significantly higher among the mi- cro enterprises than the small enterprises in the dropout sample. About 70% of start-ups were engaged in the manufacturing sector (product providers). This share was significantly higher among small than micro enterprises in the business oper- ation survivor group and lower in the dropout group. The proportion of micro enterprises engaged in the service sector was higher compared to small enterprises in both survivor and dropout samples. 4 Estimation results 4.1 The nonparametric estimation results Figure 1 shows the estimated survival and hazard rates. Figure 1a shows that survival rates declined over time. On average, almost 88% of MSEs survived their first 2 years in business. However, the survival function sloped downward strongly after and the prob- ability of survival reached almost 68% after 6 years of business operation. As indicated Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 11 of 26 Table 2 Mean separation test in the survival rate of firms between micro and small enterprises Survival Drop-out Variable All Micro Small P value Micro Small P value MSEs Socio-demographic and economic Dummy for female 0.255 0.279 0.157 0.0000 0.466 0.212 0.0004 Age of the firm owner at the time of establishment 24 24 24 0.4566 26 26 0.2673 Dummy for elementary education attainment 0.249 0.247 0.231 0.3198 0.291 0.273 0.3979 Dummy for college diploma and above attainment 0.166 0.172 0.170 0.4750 0.0126 0.167 0.2327 Dummy for TVET education 0.100 0.109 0.118 0.3624 0.029 0.030 0.4826 Dummy for active in a labor market prior to the 0.514 0.543 0.502 0.1405 0.447 0.470 0.3852 new business Dummy for self-employment experience prior to 0.122 0.105 0.134 0.1113 0.155 0.136 0.3682 the new business Dummy for apprentice experience prior to the new 0.024 0.350 0.314 0.1500 0.370 0.200 0.0087 business Dummy for business experience prior to the new 0.501 0.521 0.471 0.0836 0.485 0.515 0.3541 business Dummy for entrepreneurship training prior to the 0.143 0.131 0.183 0.0242 0.126 0.076 0.1512 new business Dummy for short-term training prior to the new 0.350 0.264 0.418 0.0000 0.398 0.561 0.0194 business Dummy for opportunity motivation 0.690 0.781 0.644 0.0000 0.612 0.379 0.0015 Firm specific Dummy for large start-up capital (more than 0.055 0.024 0.111 0.0000 0.039 0.045 0.4172 100,000) Dummy for micro start-up capital (less than 25,000) 0.833 0.906 0.729 0.0000 0.874 0.742 0.0146 Dummy for sole proprietorship start-up 0.383 0.541 0.183 0.0000 0.466 0.061 0.0000 Dummy for cooperative start-up 0.262 0.122 0.418 0.0000 0.175 0.667 0.0000 Dummy for metal and wood work 0.258 0.294 0.271 0.2470 0.155 0.106 0.1825 Dummy for leather products 0.015 0.017 0.013 0.3266 0.000 0.03 0.0382 Dummy for textile and cloth 0.048 0.051 0.059 0.3307 0.000 0.045 0.0145 Dummy for construction 0.197 0.146 0.271 0.0000 0.136 0.303 0.0040 Dummy for service 0.114 0.127 0.069 0.0048 0.204 0.091 0.0255 Dummy for product provider 0.705 0.670 0.807 0.0000 0.544 0.727 0.0083 Dummy for support beneficiary 0.661 0.556 0.794 0.0000 0.592 0.894 0.0000 Dummy for registered during establishment 0.81 0.755 0.912 0.0000 0.718 0.864 0.0137 Number of observation 466 306 103 66 Backgrounds are found from the retrospective information. Numbers are shares unless stated otherwise. P value is the lowest level of significance at which the null hypothesis (the difference in mean is equal to zero) is rejected in Fig. 1b, there was a sharp rise in cumulative hazard at the beginning of the establish- ment time and then decelerated, reflecting higher hazard rate during the start-up period. As Fig. 1c highlights, the risk of dropout increased during the start-up phase of the enterprise and reached its peak at 24 months of business operation duration. For group analysis, the paper employs Lifetable estimator for survival function esti- mation. Figure 2 presents the nonparametric survival and hazard rates. Figure 2a shows that the survival probability among micro enterprises decreased rapidly after the first 6 months of business operation. The estimate for micro/small groups shows that about 92% of the small enterprises and about 90% of the micro enterprises survived the first Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 12 of 26 Fig. 1 The nonparametric survival and hazard rate estimation. a Kaplan-Meier survival estimate. b Cumulative hazard estimate. c Kernel-smoothed hazard estimate 2 years of business operation, which is almost the same. However, after 4 years of business duration, the survival rates among the groups were notably growing apart: 77% for the small enterprises and 66% for the micro enterprises (see Appendix 1:Table 4). Thus, the results reveal strong evidence that a start-up’s initial firm size affects sur- vival probability. Entrepreneurs who establish new small-sized firms survive more com- pared to those who set up micro-sized enterprise in Ethiopia. To test for equality of the survival functions, we run the log rank test. The null hypothesis of no survival difference between the groups was rejected at a 10% level of significance (P = 0.0707). Figure 2b depicts that dropout risk was higher among micro than small enterprises at a given business operation duration. Moreover, it has an inverted U shape for the micro enterprises. Initially, there was an increased risk of dropout during the first 2 years of business operation, and then, the level of the risk remained high for the next two con- secutive years before it declined over time. In contrast, we observed an increased drop- out risk only during the first business operation year, which then decreased among small-sized enterprises. Fig. 2 The survival and hazard rate deferential between micro and small enterprises. a Survival rates for micro and small enterprises. b Kernel-smoothed hazard estimate Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 13 of 26 Overall, the nonparametric survival analysis suggests two basic facts on MSEs’ oper- ation in Ethiopia: (i) there is a survival differential among micro and small enterprises; (ii) the risk of dropout is high during the first two business operation years, particularly for micro enterprises. This implies that there is dire need to develop efficient and inclu- sive support interventions for micro business founders and mostly during the start-up phase of MSEs in Ethiopia. However, there is also a need to investigate the factors underpinning the survival of MSEs and what factors account for the deferential in survival rates among micro/small enterprise. 4.2 The Cox proportional hazard estimation To find out what factors derive the survival of MSEs and the associated factors for the deferential in survival rates between micro and small enterprises, we employ semi-parametric analysis using the Cox proportional hazard estimation. The tests for the assumption of proportional hazard reveal strong evidence of proportional hazard for each covariate and for the global test (see Appendix 1: Tables 5 and 6). The chi-square statistics soundly failed to reject accommodating proportional haz- ard for model A and model B as a whole; P values are equal to 0.988 and 0.374, respectively. Similarly, the graphical diagnostic tests of the Schoenfeld residuals also supported the assumption of proportional hazard for all covariates as there was non-systematic departure of covariates from their smoothing-spline fitted line over time. We observe non-trend pattern with time in the plots for all covariates (see Appendix 2: Figures 3 and 4). The survival rate of an enterprise may be influenced by various factors, such as owner, firm, and business strategy characteristics and the overall socio-business en- vironment the enterprise is operating. Hence, the present paper specifies three dif- ferent models to consider the effect of each category, separately. The first specification (model A) only presents and discusses the effect of the characteristics that are typically identifiable at the time of establishing the enterprise on MSEs’ survival. We re-estimate the semi-parametric Cox proportional model (model B) which focuses on whether the business strategic characteristics of the MSEs are associated with the survival or failure of the enterprises. In order to control the effects of person-, firm-, and industry-specific background characteristics, we in- clude three covariates in model B as a control variable, namely female, sole propri- etorship, and manufacturing. The last specification (model C) estimates all the hypothesized variables using a Cox analysis. Table 3 illustrates the estimated coefficients from the Cox proportional models and has three parts with sub-columns. Section A constitutes three sub-columns that present estimates of model A from the pooled sample and the grouped sam- ples. Section B constitutes the next three sub-columns of the table, and they present estimates of model B from the pooled sample and the grouped samples. Section C is the last column of the table, and it highlights the estimates of all the hypothesized variables. The results reveal that gender has a significant positive association with risk of dropout. Enterprises owned by women had a higher risk of dropout than men en- trepreneurs, which is consistent with the results from Coad and Tamvada (2008). Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 14 of 26 Table 3 Partial likelihood estimates for the Cox proportional hazard models Model A Model B Model C Covariates MSEs Micro Small MSEs Micro Small MSEs Dummy for female 0.683*** 0.907*** 0.370 0.610*** 0.882*** 0.381 0.493* (3.838) (3.916) (1.142) (3.510) (3.925) (1.119) (2.566) Age of the owner 3.244*** 3.635*** 2.714** 3.073*** (6.426) (5.705) (3.256) (5.968) Dummy for elementary − 0.111 − 0.065 − 0.034 − 0.255 education attainment (− 0.608) (− 0.261) (− 0.109) (− 1.332) Dummy for college diploma 0.037 − 0.025 0.252 0.456 and above attainment (0.154) (− 0.078) (0.689) (1.802) Dummy for TVET education − 0.523 − 0.462 − 0.612 − 0.481 (− 1.121) (− 0.764) (− 0.827) (− 1.004) Dummy for active in a − 0.388* − 0.400 − 0.332 − 0.346 labor market prior to the (− 2.014) (− 1.557) (− 1.043) (− 1.656) new business Dummy for self-employment 0.309 0.494 0.047 0.316 experience prior to the new (1.169) (1.485) (0.099) (1.142) business Dummy for business 0.191 − 0.042 0.625 0.242 experience prior to the (1.053) (− 0.180) (1.951) (1.256) new business Dummy for apprentice 0.306 0.574* − 0.281 0.639** experience prior to the (1.582) (2.417) (− 0.749) (2.950) new business Dummy for opportunity − 0.837*** − 0.712** − 0.974** − 0.776*** motivation (− 4.583) (− 3.002) (− 3.192) (− 3.785) Dummy for entrepreneurship − 0.543* − 0.178 − 1.210* 0.032 training prior to the new (− 2.112) (− 0.577) (− 2.440) (0.110) business Firm size (initial number of − 0.198* − 0.029 workers) (− 2.073) (− 0.293) Large start-up capital (more − 0.824 0.358 − 1.433* − 0.485 than 100,000) (− 1.866) (0.556) (−2.177) (− 1.013) Micro start-up capital (less − 0.272 − 0.134 − 0.377 − 0.016 than 25,000) (− 1.162) (− 0.360) (− 0.935) (− 0.061) Dummy for sole proprietorship − 0.898*** − 1.016*** − 1.026 − 1.079*** − 1.014*** − 1.961** − 1.187*** start-up (− 3.771) (− 3.880) (− 1.687) (− 5.220) (− 4.063) (− 3.260) (− 4.542) Dummy for cooperative 0.312 0.055 0.478 − 0.196 start-up (1.483) (0.177) (1.487) (− 0.796) Dummy for product provider − 0.598*** − 0.712** − 0.470 − 0.627*** − 0.674** − 0.539 − 0.643*** (− 3.445) (− 3.295) (− 1.433) (− 3.506) (− 2.939) (− 1.738) (− 3.411) Dummy for support beneficiary − 0.092 − 0.219 0.238 0.040 (− 0.466) (− 0.960) (0.544) (0.189) Dummy for registered during − 0.503* − 0.485* − 0.411 − 0.101 establishment (− 2.495) (− 2.005) (− 1.048) (− 0.446) Dummy for competitive − 0.337* − 0.145 − 0.582* − 0.421* pricing strategy (− 1.994) (− 0.630) (− 2.035) (− 2.353) Dummy for mark-up pricing − 0.090 − 0.036 0.031 − 0.040 strategy (− 0.514) (− 0.157) (0.110) (− 0.218) Dummy for at home placing − 0.245 − 0.019 − 0.822* − 0.213 strategy (− 1.149) (− 0.078) (− 1.515) (− 0.963) Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 15 of 26 Table 3 Partial likelihood estimates for the Cox proportional hazard models (Continued) Model A Model B Model C Covariates MSEs Micro Small MSEs Micro Small MSEs Dummy for at industrial site − 0.960* − 0.234 − 2.091** − 0.980* placing strategy (− 2.554) (− 0.510) (− 2.801) (− 2.433) Dummy for salesmanship − 0.221 − 0.228 − 0.080 − 0.138 promotion strategy (− 0.970) (− 0.724) (− 0.230) (− 0.581) Dummy for product innovation − 0.258 − 0.102 − 0.664* − 0.261 strategy (− 1.470) (− 0.446) (− 2.187) (− 1.452) † † Dummy for market place − 0.407 − 0.292 − 0.754* − 0.424 innovation strategy (− 1.844) (− 0.981) (− 2.143) (− 1.887) Dummy for high degree of − 0.223 − 0.095 − 0.575 − 0.099 competition (− 1.244) (− 0.396) (− 1.946) (− 0.530) Dummy for niche market − 0.149 − 0.134 0.004 − 0.143 strategy (− 0.818) (− 0.579) (0.011) (− 0.724) Dummy for external market − 0.997* − 0.468 − 1.455* − 1.168** linkage (− 2.454) (− 0.755) (− 2.509) (− 2.776) Log capital investment − 0.150*** − 0.146*** − 0.170*** − 0.158*** (− 7.936) (− 5.651) (− 5.613) (− 7.824) Dummy for retained earnings − 0.738*** − 0.861** − 0.472 − 0.688*** investment (− 3.779) (− 3.102) (− 1.546) (− 3.292) Dummy for internal financing − 0.850* − 1.009* − 0.873 − 1.001** (− 2.520) (− 2.403) (− 1.259) (− 2.810) Dummy for business financing − 0.328 − 0.596 − 0.254 − 0.294 (− 1.105) (− 1.509) (− 0.481) (− 0.973) Dummy for external financing − 0.577* − 0.716* − 0.426 − 0.642* (− 2.124) (− 2.124) (− 0.802) (− 2.245) Dummy for undercapitalization 0.392* 0.583* 0.162 0.220 challenge (2.189) (2.471) (0.527) (1.151) Dummy for inventory system − 0.670** − 0.768* − 0.614 − 0.752** (− 2.695) (− 1.998) (− 1.731) (− 2.869) Dummy for annual budget − 0.708 − 1.398 − 0.743 − 0.703 (− 1.513) (− 1.369) (− 1.345) (1.459) Dummy for poor accounting 0.160 0.313 − 0.097 0.074 system (0.675) (1.034) (− 0.235) (0.310) Dummy for owners training 0.423* 0.415 0.378 0.203 after establishment (2.035) (1.551) (0.982) (0.884) Dummy for workers training − 0.644* − 0.992* − 0.625 − 0.534 while they are on duty (− 2.459) (− 2.155) (− 1.570) (− 1.955) † † Dummy for Internet connection − 1.760 − 2.011 in the premises (− 1.732) (− 1.952) LHR test 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Wald test 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Log rank test 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Note: Figures in parentheses are P values, and †, *, **, and *** indicate statistically significant at the 10, 5, 1, and 0.1% level, respectively. Values in parentheses are the Z values Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 16 of 26 The log form of the owner’s age was applied in the model. As Davidsson (1991) and Amran (2011) found, there was a significant negative relationship between the business founders’ age and business survival. This suggests that entering a self-employment at matured age is associated with a higher dropout rate than younger entrepreneurs. Education qualification does not appear to be a significant for a survival of MSEs. As Taylor (1999) argued, this suggests that to enter self-employment and survive, entrepreneurs do not need to have a high level of formal education. Moreover, there is no significant positive effect of TVET on the probability of a firm’s survival. The fitted model proves that previous labor market experience was an important fac- tor for survival. Consistent with the result of Taylor (1999), active labor market partici- pation prior to the start-up reduced the risk of dropout. But, as Kalleberg and Leicht (1991) found, there was no significant relationship between self-employment experience prior to the start-up and business survival. Previous general business and apprentice ex- perience did not have a significant impact on a firm’s survival. In line with Caliendo and Kritikos (2009), opportunity motivation significantly re- duced the risk of dropout, implying that the opportunity-driven entrepreneur’s prob- ability of survival was significantly higher than the necessity entrepreneurs. Entrepreneurial education and training prior to start-up appear to be important factor to stimulate entrepreneurial activity and business survival, as it had a significant nega- tive effect on dropout rate. Concerning firm and industry characteristics, start-up size has the expected positive effect on MSEs’ survival, implying the larger the size of the setup, the lower is the prob- ability of dropout. We capture the impact of financial resource at the time of start-up using the size of the start-up capital, and it turns out that larger amount of start-up capital reduced the risk of dropout. Entrepreneurs, who invested more than Birr 100,000 (USD 3700), had a significantly higher probability of survival than the others, suggesting that an adequate amount of start-up capital is a key factor for a firm’s sur- vival. Unlike Kalleberg and Leicht (1991), a solo type of business setup had a significant positive effect on MSEs’ survival than cooperatives. This might be associated with the complete control and decision-making power that the owner enjoys, as well as being free from corporate tax. Youth-owned MSEs engaged in the manufacturing sector had a lower risk of dropout. This is consistent with Taylor’s(1999) results, implying that the service sector was ex- posed to a higher risk of dropout. Unpredictably, supports obtained by the owner at time of start-up did not appear to have significant association with the survival of MSEs. Last but not the least, our analysis proves that formal system is a significant de- terminant of MSEs’ survival, and registration during the start-up phase reduced the probability of dropout. Overall, gender, age, previous labor experience, motivation, entrepreneurial education and training, initial size, legal form, size of start-up capital, industry type, and formal system, therefore, are important factors in determining the probability of MSEs’ survival in Ethiopia. As section B of Table 3 presents, in line with the results of Chittithaworn et al. (2011), marketing strategy was an important factor in explaining MSEs’ survival rate in Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 17 of 26 Ethiopia. Concerning marketing mix strategy, the pricing and place strategies appeared to be significant factors in explaining the probability of MSEs’ survival. Competitive pri- cing strategy had a significant negative association with MSEs’ dropout rate. This re- flects that it is better to develop status quo pricing objective for MSEs to survive and be successful since most of them are operating under high degree of competition eman- ating from similar business; about 76% of the MSEs reported that they were experien- cing high competition due to an increase in the number of similar businesses. The location of enterprises is one of the elements of a business strategy that can address the challenges of competition by delivering the right product to the right customer at the right time. The study reveals that an enterprise which was based in an industrial area was more likely survive than those at residential places. This reflects that industrial areas offer better access to customers, raw materials, labor, and transportation. Marketing place innovation was statistically significant at the 10% level and had a positive association with survival rate, suggesting that reallo- cating for innovative marketing places was an important factor for attracting new customers and expanding market shares. Moreover, external market linkages had a significant positive association with the probability of business survival, suggesting that operating under linked marketing environment significantly reduced the risk of dropout. In line with the results from McMahon (2001b), our fitted model proves that financial management does appear to be an important factor for MSEs’ survival. There was a negative association between capital investment and a risk of dropout, implying that the more the manager/owner invests in capital equipment, the higher the probability of survival the business had. Consistent with the results of Fadahunsi (2012), a manager/owner who is willing to reinvest retained earnings had a propensity to survive more than businesses that were not willing. This sug- gests that a less costly and internal source of financing (retained earnings) is a key factor of MSEs’ survival since access to external formal sources of financing was a major challenge for MSEs in Ethiopia; almost 32 and 26% of managers/owners reported that a lack of access to finance was a major constraint to revenue growth and business improvement, respectively. Like Carpenter and Petersen (2002), the present paper proves that the MSEs which used retained earnings to finance their working capital had a propensity to survive more compared to the MSEs which used other type of sources. The study also proves that there existed a positive association between external source of financing capital investment and MSEs’ survival, implying that those firms which finance their capital investment from external source had higher survival rate. Consistent with the results from Read (1998), the study shows that undercapitalization in- creased the risk of dropout, reflecting that MSEs which suffered from a shortage of funds to finance their investment survived less than those which did not. Like Mongare and Nasidai (2014), the study proves that the survival of MSEs was highly associated with the inventory system of the business, reflecting that efficient and modern inventory systems reduced risk of demise. As Savery and Luks (2004) found, the study reveals that training for workers was an important factor for MSEs’ survival, suggesting that building human capital had a pro- pensity to reduce the risk of dropout. In contrast, there exist inconsistent results Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 18 of 26 concerning the effect of owners’ training on business survival, as it appeared to be posi- tively associated with dropout rate. Moreover, the fitted model attests that business sur- vival was positively associated with Internet technology adoption, which is consistent with the result of Warren (2004). This suggests that MSEs with Internet connection in their business premises had a higher probability of survival as they were more con- nected with the world and had better access to information. Overall, the study confirms that the survival of an enterprise is also highly associated with business strategy characteristics of MSEs. Marketing management and financial management strategies play a significant role on reducing the risk of dropout of MSEs in Ethiopia. Moreover, human capital development and access to Internet are also important elements to increase the MSEs’ probability of survival. So far, the Cox proportional hazard model is estimated for two different data setups; model C estimates all the hypothesized variables to see if there is a sig- nificant discrepancy on the covariate effect on the hazard rate. As section C highlights consistent with the first model, gender, age, labor market experience, motivation, legal form, and industry type appeared to be significant factors for business survival. Moreover, unlike the first part model, entrepreneurship educa- tion, size of the enterprise, level of the start-up capital, and legal registration ap- peared to be insignificant factors in this model for the survival of enterprises. In line with the results from the second model, this fitted model proves that mar- keting management, financial management, controlling system, human capital development, and Internet technology appeared to be important factors for mi- cro- and small-scale enterprises’ survival. 4.3 The group survival rate analysis The second and third column of section A highlights the results from group analysis with model A specification. As the descriptive analyses highlight, female entrepreneurs were apt at establishing a micro-sized enterprise, but the results from the survival esti- mates show that female entrepreneurs who owned micro enterprises experienced a sig- nificant higher dropout rate. This suggests that gender is one of the factors behind less survival rate among the micro-level enterprises compared to small enterprises. Unex- pectedly, apprenticeship experience prior to start-up had a significant positive effect on the dropout rates for micro enterprises. The justification for this result demands further study on the subject matter, but the probable explanation could be the gap between the level and quality of apprenticeship experience with respect to the nature of and demand for micro enterprises. Entrepreneurship training drives small enterprise owners to experience low level of dropout risk compared to micro enterprise founders. This confirms that entre- preneurial training is one of the explanations for survival deferential between micro and small enterprises. The analysis also depicts that large value of start-up capital had a significant and pronounced negative effect on the risk of dropout from the small enterprises’ sample. This implies that it is an important factor accounted for survival rate deferential between micro and small enterprises. The relationship be- tween engaging in a manufacturing sector and the risk of dropout is significant Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 19 of 26 and negative for micro enterprises. We also found that formality had a significant negative association with the risk of dropout for micro enterprises. The model estimation identifies what owner, firm, and industry characteristics identi- fiable at start-up period explained deferential of survival rate between micro and small enterprises. Accordingly, we found that gender, entrepreneurial training, and size of start-up capital were the major factors accounted for the better success of small enter- prises compared to micro enterprises. From sub-columns 2 and 3 of model B section, the estimation for group analysis shows that pricing and placing strategy, product and marketplace innovation, competi- tion, and market linkage appeared to be the majors accounting for the better success of small enterprises compared to micro enterprises. Status quo pricing objective led small firms to survive more compared to micro enterprises. Locating a business at residential and industrial place reduced the risk of dropout for small enterprises. Like Ibidunni et al. (2014), the paper depicts that product innovation reduced the risk of dropout for small enterprises, calling the firms to undertake a progressive product innovation process to meet the demand of their customers and expand their market. Moreover, high degree of competition had a significant positive association with the survival of small enterprises. This suggests two things: (i) during a high competition period, managers/owners probably pressured from their competitors to enter new mar- kets, renovate their products, or introduce new products so that their market increases; (ii) high degree of competition in a specific product line might imply the high demand for the product. External market linkage had a negative association with the risk of dropout for the small enterprises. Overall, our study proves that marketing manage- ment strategy was one of the explanations for survival deferential between micro and small enterprises. Regarding financial management characteristics, capital investment, retained earnings investment, external financing, and inventory system had a significant negative associ- ation with the probability of dropout for micro enterprises. Like the results from the pooled sample, shortage of investment funds had a significant positive effect on micro enterprises’ dropout rate. The fitted model also confirmed that workers’ training played a key role on reducing the probability of dropout for micro enterprises. The results from model B specification identify what business strategy characteristics explained why the survival rate deferred between micro and small enterprises. Pricing and placing strategy, innovation, competition, and external market linkage were the probable major factors accounted for the better success of small enterprises compared to micro enterprises. 5 Conclusions This paper examines the survival of youth-owned MSEs and the associated factors based on a data obtained from 941 youth-owned MSEs in urban Ethiopia. The study uses nonparametric and semi-parametric hazard models to estimate the survival rate and its associated factors. The nonparametric estimation results reveal that MSEs’ risk of dropout was high during the first two to four business operation years in Ethiopia. The analysis also confirms that there was survival differential between micro and small enterprises. Particularly, the risk of dropout reached its maximum at 4 years after Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 20 of 26 establishment for micro enterprises while it was after 2 years for small enterprises. This could be due to inclination of most support programs towards small enter- prises or overlooking of the importance of micro enterprises in the process of cre- ating viable and sustainable medium and large industries from both sides—the MSE promoters and micro scale enterprise owners. Gender, age, previous work experience, motivation, entrepreneurship training, initial size, ownership type, type of industry, and legal status (registration) were important factors in explaining the probability of MSEs’ survival in Ethiopia. The survival of MSEs was also highly influenced by business strategy features. Market- ing and financial management strategies played a significant role on reducing the risk of dropout. Moreover, human capital development and use of Internet technology were also important elements to increase the probability of MSEs’ survival. The group-based survival analysis (micro vs. small enterprises) reveals that gender, entrepreneurial training, start-up capital size, pricing and placing strategy, product and market place innovation, and external market linkage appeared to increase small enter- prises’ survival rate compared to micro enterprises. Endnotes Regional government offices provide a certificate for new startups upon registration, and it includes information about the initial size of the business. Micro enterprise is define as an enterprise employing less than five workers, while small is defined as an enterprise employing more than five but less than 30 workers. Appendix 1 Table 4 Survival and hazard rate from the Lifetable estimator for small/micro enterprises Duration Small Micro Survival rate difference (%) At risk Survival Hazard At risk Survival Hazard 0–6 372 1 0.0027 569 1 0.0024 0 6–12 358 0.9837 0.0034 533 0.9856 0.0055 − 0.19 12–18 320 0.9636 0.0072 430 0.9534 0.0103 1.02 18–24 280 0.9227 0.0044 348 0.8963 0.0037 2.64 24–30 251 0.8987 0.0094 286 0.8767 0.0103 2.20 30–36 210 0.8494 0.0043 233 0.8243 0.0064 2.51 36–42 181 0.8280 0.0040 183 0.7932 0.008 3.48 42–48 152 0.8083 0.0024 151 0.7561 0.0063 5.22 48–54 124 0.7967 0.0059 113 0.728 0.0153 6.87 54–60 103 0.7691 0.0018 83 0.6641 0.0046 10.5 60–66 80 0.7607 0.0024 63 0.6461 0.0000 11.46 66–72 60 0.7499 0.007 47 0.6461 0.0000 10.38 72–78 35 0.7190 0.0056 30 0.6461 0.0000 7.29 78–NA 25 0.6954 0.0000 19 0.6461 0.0000 4.93 Note: The survival rate difference is the difference of cumulative proportion of small and micro entrepreneurs’ survival rate up to specific time interval NA not applicable Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 21 of 26 Table 5 Proportional hazard test using chi-square statistics: Cox model analysis—model A Covariates P value Female 0.112 Age (at start-up) 0.559 Elementary school 0.523 College diploma and above 0.347 TVET 0.967 Active in labor market 0.959 Self-employment experience 0.333 Business experience 0.487 Apprentice experience 0.552 Opportunity 0.824 Entrepreneurship training 0.988 Initial number of workers 0.471 Sole proprietorship 0.329 Cooperative 0.731 Large start-up capital 0.954 Micro start-up capital 0.286 Product provider 0.611 Support 0.985 Registered 0.591 Global 0.951 Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 22 of 26 Table 6 Proportional hazard test using chi-square statistics: Cox model analysis—model B Covariates P value Female 0.193 Manufacturing 0.554 Sole proprietorship 0.378 Competitive pricing strategy 0.373 Mark-up pricing strategy 0.376 Place at home (shop) 0.351 Place at industrial place (shop) 0.328 Salesmanship promotion 0.476 Product innovation 0.326 Market place innovation 0.825 Degree of competition 0.385 Niche market 0.288 Market linkage 0.481 Capital investment 0.069 Retained earnings investment 0.290 Internal financing (working capital) 0.260 Business financing (working capital) 0.375 External financing (investment) 0.021 Undercapitalization 0.713 Inventory system 0.906 Annual budget 0.508 Poor accounting system 0.492 Owners training 0.155 Workers training 0.516 Internet 0.790 GLOBAL 0.378 Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 23 of 26 Appendix 2 Fig. 3 Graphical diagnostic Schoenfeld residual test: model A Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 24 of 26 Fig. 4 Graphical diagnostic Schoenfeld residual test: model B Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 25 of 26 Acknowledgements The authors would like to thank Martha Melese, the anonymous reviewers, and the editor of the journal for the important suggestions. We would also like to acknowledge the comments received at the 2015 CSAE workshop held at Oxford University, the Center for the Study of African Economies. However, the authors are responsible for any error that may have found. Responsible editor: David Lam This paper draws from research project on Youth Self-Employment in Ethiopia: Promoting Micro and Small Enterprises (MSEs) Through Government Support Programmes carried out under the auspices of the Association of Ethiopian Microfinance Institutions (AEMFI) with the financial support from Canada’s International Development Research Centre (IDRC). The views expressed in this brief are those of the authors and do not necessarily reflect those of IDRC and AEMFI. Competing interests The IZA Journal of Development and Migration is committed to the IZA Guiding Principles of Research Integrity. The authors declare that they have observed these principles. Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Author details 1 2 Department of Economics, Addis Ababa University, Addis Ababa, Ethiopia. Association of Ethiopian Microfinance Institutions, Addis Ababa, Ethiopia. Received: 15 May 2017 Accepted: 25 January 2018 References Agarval R, Audretsch DB. Does entry size matter? The impact of the life cycle and technology on firm survival. J Ind Econ. 2001;49:21–43. Amran NA. The effect of owner’s gender and age to firm performance: a review on Malaysian public listed family businesses. J Glob Bus Econ. 2011;2(1):104–16. Audretsch, D. and Mahmood, T. (1995) New firm survival: New results using a hazard function. The Review of Economics and Statistics, 77(1), pp. 97–103. Bates T. Survival patterns among newcomers to franchising. J Bus Ventur. 1998;13:113–30. Bates T. Analysis of young, small firms that have closed: delineating successful from unsuccessful closures. J Bus Ventur. 2005;20:343–58. Bigsten A, Kimuyu P, Söderborn M. Chapter 10: the manufacturing sector. In: Adam CS, Collier P, Ndung’uNS, editors. Kenya policies for prosperity. Oxford and Nairobi: Oxford University Press and Central Bank of Kenya; 2010. p. 243–66. Boswell J. The rise and decline of small firms. London: George Allen & Unwin; 1973. Bruderl J, Preisendorfer P, Ziegler R. Survival changes of newly founded business organizations. Am Sociol Rev. 1992; 57(2):227–42. Brush CG, Carter NM, Gatewood E, Greene PG, Hart MM. Venture capital access in the new economy: is gender an issue? In: Hart D, editor. The emergence of entrepreneurship policy: governance, start-ups and growth in the knowledge economy. Cambridge: Cambridge University Press; 2003. p. 141–55. Caliendo, M., and Kritikos, A. (2009). “I want to, but I also need to”: start-ups resulting from opportunity and necessity. IZA Discussion Paper No. 4661. Carpenter R, Petersen B. Capital market imperfections, high-tech investment, and new equity financing. Economic Journal. 2002;112(477):F54–72. Carter R, Van Auken H. Small firm bankruptcy. J Small Bus Manag. 2006;44(4):493–512. Chandler GN, Hanks SH. An examination of the substitutability of founders human and financial capital in emerging business ventures. J Bus Ventur. 1998;13:353–69. Chittithaworn C, Islam A, Keawchana T, Yusuf DHM. Factors affecting business success of small & medium enterprises (SMEs) in Thailand. Asian Social Science. 2011;7(5):180–90. Coad, A. and Tamvada, J. P. (2008). The growth and decline of small firms in developing countries. Papers on Economics and Evolution No. 0808. Coleman, S., Cotei, C. and Farhat, J. (2010). Factors affecting survival, closure and M&A exit for small businesses. Available at SSRN: http://ssrn.com/abstract=1768728. Cowling, M. (2009). The impact of entrepreneurship training and small business experience on future entrepreneurial activity in the UK. Brighton: Institute for Employment Studies. Cox, D. (1972). Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B (Methodological), 34(2), 187-220. Retrieved from http://www.jstor.org/stable/2985181. Davidsson P. Continued entrepreneurship: ability, need and opportunity as determinants of small firm growth. J Bus Ventur. 1991;6:405–29. Dayanandan R. Sustainability of micro and small enterprises: imperatives, myths and realities. Int J Entrepreneurship Bus Environ Perspect. 2012;1(2):129–42. Dobbs M, Hamilton R. Small business growth: recent evidence and new directions. Int J Entrepreneurial Behav Res. 2007;13:296–322. Fadahunsi A. The growth of small business: towards a research agenda. Am J Econ Bus Adm. 2012;4(1):105–15. Fairlie, R. W. and Robb, A. M. (2008). Gender differences in business performance: evidence from the characteristics of business owners survey. IZA Discussion Paper Series No. 3718. Woldehanna et al. IZA Journal of Development and Migration (2018) 8:14 Page 26 of 26 FeMSEDA. Micro and small enterprises sector development yearly statistical bulletin. Addis Ababa: Federal micro and small enterprises development agency; 2014. Fox, J. (2002) Cox proportional-hazards regression for survival data. Available at: http://cran.r-project.org/doc/contrib/ FoxCompanion/appendix-cox-regression.pdf. (Accessed: 17 Apr 2017). Headd, B. (2000). Business success: factors leading to surviving and closing successfully. U.S. Small Business Administration, office of advocacy. Honig B. What determines success? Examining the human, financial, and social capital of Jamaican micro entrepreneurs. J Bus Ventur. 1998;13:371–94. Ibidunni O, Oluwole I, Ayodotun S, I. Product innovation, a survival strategy for small and medium enterprises in Nigeria. Europ Sci J. 2014;10(1):194–209. Jenkins SP. Survival analysis. Unpublished manuscript, Institute of Social and Economic Research. Colchester: University of Essex; 2004. Downloadable from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.176.7572&rep= rep1&type=pdf Kalleberg AL, Leicht KT. Gender and organizational performance: determinants of small business survival and success. Acad Manag J. 1991;34(1):136–61. Kangasharju A, Pekkala S. The role of education and self-employment success in Finland. Growth and Change. 2002;33:216–37. Mata J, Machado J. Firm start-up size: a conditional quantile approach. Eur Econ Rev. Elsevier. 1996;40(6):1305–23. Mata J, Portugal P, Guimaraes P. The survival of new plants: start-up conditions and post-entry evolution. Int J Ind Organ. 1995;13:459–82. McMahon, R. G. (2001b), Business Growth and Performance and the Financial Reporting Practices of Australian Manufacturing SMEs. Journal of Small Business Management, 39: 152–164. Mongare ME, Nasidai SE. The impact of information communication technology on inventory control systems in transport organization: a case study of Kenya ferry services. Eur J Logistic Supply Chain Manage. 2014;2(1):17–41. Moores K, Mula J. Contextual and strategic impacts on the form of management accounting systems. School of Business Discussion Papers, paper 49. Robina: Bond University; 1993. O'Dwyer M, Gilmore A, Carson D. Innovative marketing in SMEs. Eur J Mark. 2009;43(1/2):46–61. Pena I. Intellectual capital and business start-up success. J Intellect Cap. 2002;3(2):180–98. Read L. The financing of small business: a comparative study of male and female business owners. London: Routledge; 1998. Rosa P, Carter S, Hamilton D. Gender as a determinant of small business performance: insights from a British study. Small Bus Econ. 1996;8(6):463–78. Savery LK, Luks JA. Does training influence outcomes of organizations? Manage Dev. 2004;23(2):119–23. Schiller BR, Crewson PE. Entrepreneurial origins: a longitudinal inquiry. Econ Inq. 1997;35(3):523–31. Skrt B, Antoncic B. Strategic planning and small firm growth: an empirical examination. Manag Glob Transitions. 2004; 2(2):107–22. Storey D. Understanding the small business sector. London: Routledge; 1994. Taylor MP. Survival of the fittest? An analysis of self-employment duration in Britain. Econ J. 1999;109:C140–55. Van Praag C. Business survival and success of young small business owners. Small Bus Econ. 2003;21(1):1–17. Warren M. Farmers online: drivers and impediments in adoption of internet in UK agricultural businesses. J Small Bus Enterprise Dev. 2004;11(3):371–81. World Bank. Youth employment in sub-Saharan Africa. Washington, D.C.: World Bank; 2014.

Journal

IZA Journal of MigrationSpringer Journals

Published: Jul 25, 2018

References