# Introduction ounting poverty level in developing countries is a posing formidable threat to the very sustainability of the economies. The poverty indicators such as Gross Domestic Product (GDP) per capita, low life expectancy, low educational enrolment, many people living below $1.25 a day, nutrition, etc., point to gravity of poverty situation in developing countries. Country specific development programmes are underway in these countries to alleviate poverty. Micro and Small-scale Enterprises (MSEs) play a pivotal role in the socio-economic development particularly in developing economies. MSEs have greater economic benefits than large firms in terms of employment generation and growth since they use more of what a country is endowed with and less of what it lacks (Admassie and Matambalya, 2002;Habtamuet al., 2013). Unlike large-scale enterprises, which are often capital-intensive and import-dependent for raw materials and machinery, MSEs mostly use locally available resources. By creating employment opportunities for the semi-skilled and unskilled labor, MSEs could increase the household income of the labor force at the micro level and reduce the level of poverty at the macro level, apart from creating the basis for a more sustained industrial development. Moreover, MSEs' nurture of indigenous entrepreneurial and managerial talents which foster economic development, poverty reduction and employment generation (Eshetu and Zeleke, 2008). Needless to say that promoting MSEs has become a preferred development strategy in many developing countries. Statistics reveal the facts about concrete contribution of MSEs in developing countries. MSEs as well as medium enterprises account for about 30 per cent of employment and 17 per cent of GDP (Beck and Demirguc-kunt, 2005). In developed countries, the share of the enterprises is even larger; about of 50 per cent to GDP and about 60 per cent of employment. As economies grow, the share and contribution of MSEs willnaturally increase. In these economies, the expansion of these enterprises is significantly important as they are closely associated to the livelihoods of the poor and disadvantaged groups that include women and youth (Robu, 2013).The MSEs in Ethiopia contributes to about 3.4 per cent of the GDP, about 33 per cent of the overall industrial production and 52 per cent of the manufacturing output (Habtamuet al., 2013). Government of Ethiopia has acknowledged the role of these enterprises in the economic growth and transformation. The Growth and Transformation Plan (GTP I), for instance, has envisaged that micro and small scale enterprises create employment opportunities for about three million people and thereby enhance household income, domestic saving, reduce unemployment and poverty, particularly benefiting the women and the youth (Mo FED, 2014).However, MSEs are faced with a number of problems. In the first place, there seems to be a bias against MSEs in Ethiopian as in many other developing countries in terms of the support provided to them. Product diversification sometimes does not bring the expected results such as sales volume, revenue and profits. The most often reason of need of knowledge and skills in the field of change management, insufficient training, technical know-how (Luxenber, 2004) Researcher experiencesas well as Trade and Industry statistical abstract showed that MSEs at Wolaita Zone (The study location in Ethiopia) continue to be capital-starved and remain low in their job creation potential. Furthermore, the Zone has not yet exploited their potential very well to contribute towards economic development, job creation and poverty reduction. Their contribution to the local economy, capital accumulation and employment generation remains much low. This is due to lack of basic entrepreneurial and quality management systems, such as management of financial and customer focused activities, in order to enhance enterprises competitiveness. The managerial inefficient leads MSEs to incur unnecessary cost and wastage resulting in low returns on invested capital. As a result, less competitive MSEs are unable to compete effectively in the market impacting the entire MSEs sector performance in the country as well as in the study area (Fikirte and Endrias, 2013). Product diversification among MSEs is known to contribute to risk mitigation and stable MSE sector for job creation. Besides, product diversification is a good business development tool for MSEs. It brings new opportunities through new product lines and services, as well as makes easier the consumption of its products by producing complementing goods or offering complementing services. Furthermore, diversifying small enterprises may be looking for synergies or the sharing of co-specialized innovative assets between different lines of business (Baptista, et al, 2010). Hence, the government initiated various support programmes with the aim to improve MSEs' competitiveness and performance through enhancing innovation and product development capabilities, competence and technology such as upgrading existing product quality, improving design and packaging, and training (Geberyesus, 2009). Product diversification is essential for MSEs to become competitive, build business volume and graduate to medium sized enterprise status, thus, creating new employment opportunities (Ernst, 2004and Geberyesus, 2009). Successful implementation of product diversification enhances MSEs to become competitive in a market, increases its market share and provides opportunities to penetrate existing and new markets (Luxenber, 2004). Systematic literature survey conducted by the researchers revealed that no study has been undertaken so far in the study area to investigate the aforementioned issue. However, there are a number of features which make the present study different from the existing empirical studies. First, few number of the studies probed into the issue of MSEs product diversification, employing econometric techniques. In the econometric analysis, the proposed study employed Heckman two-stage regression analysis to unravel the issue of choice and level of diversification which accounts for the problem of censoring. Second, the study used more objective measure of the product diversification in small business. Third, the study dealt with endogeniety problem by using two stages least square estimation techniques by separating the existing instrumental variables. This study was conducted to identify determinants of product diversification decision and level of diversification in Micro and Small Enterprises in Wolaita Zone, Ethiopia. # II. # Research Methodology # b) Sampling Technique Study unit for the research was MSEs. Multistage sampling technique was employed to select representative units of MSEs. At the first stage, Wolaita Zone was purposively selected due to largest number of MSEs (WOZTID, 2015). Besides having unexploited potential for MSE development, Wolaita Zone was the jurisdiction for the research and development program of Wolaita Sodo University. At the second stage, three administrative towns (Sodo, Boditi and Areka) were selected purposively from the target Zone as they housed largest numbers of MSEs and members. At third stage stratified random sampling technique was used as the MSE population was heterogeneous. It was necessary to classify the population into two strata. These were Micro and Small Enterprises that were major components of sectors in the study area. # c) Sample Size Determination To determine appropriate sample size simplified formula which was developed by Yamane (1967) was used. Where, n=required sample size; e =degree of error margin (at 0.05); and ??= total population of MSEs. The sample size was computed from the population of 790 Micro and 112 Small enterprises. Accordingly, 265 and 87 from micro and small enterprises respectively were selected. The total sample size was 352 drawn from the three administrative towns in proportions to the number of MSEs. Thus, 228 units from Sodo, 69 units from Areka and 55 units from Boditi were included in the sample. Systematic sampling technique was employed to draw sampling units from each stratum. # ?? # d) Data Sources and Methods of Data Collection Both qualitative and quantitative data were collected from primary and secondary sources. Quantitative data from primary sources were collected through interview schedule while qualitative data were collected through key informant interview, focus group discussions and personal observations. The relevant data were collected from 352 sample MSE managers. An interview schedule was prepared in English and translated into Amharic to ease communication during data collection. The interview-schedule was pre-tested before actual data collection and necessary corrections were effected in the final version. Five enumerators were recruited based on their proficiency in local language, educational background and prior exposure to data collection. Training was given to enumerators on the content of the interview schedule and procedures to follow while conducting interview. The survey team gathered data on socioeconomic, individual and firm related, institutional and linkage related factors. Secondary data were collected from Wolaita Zone trade and industry promotional department, Micro finance institution, Journals, and Central Statistical Authority (CSA) publications, published and unpublished documents of national, regional and zonal offices. # e) Method of Data Analysis The general hypothesis of the modeling approach is to check whether the business firm is linked directly to the firms' decisions on diversification choices of product. In product development, users make their decisions on product diversification choices in the context of their own strategies or rules, which affect the decision of the firms from the expansion of existing/new business product to another as well as the preservation of existing product in its current condition. The general structure of the regression equations is expressed in a simple form by Where, D i represents the Entropy index of richness, X represents a vector of business firms affecting factors, ? i stands for unobserved factors,b and c are the parameters to be estimated. With a view to assess the degree of diversification in the business sector, the entropy index is constructed as: Where, the index is 0 when the business firm hasno product diversification (diversification is absent).The assumption in that given x i as sales group of business firm i then i?. n, there is the total sales? x i j = X and?? = x ?? where, P is the quota of the sales. D i = ? P i j (ln1 Several estimation problems will be encountered in estimating inter specific diversification strategies. First, a sample selection problem occur because the diversification index for business firm i exist only when the business firms expand or adds related product for more revenue. Second, a large proportion of business firms that not expand or add related product (without diversification) so that richness indicate to censor at zero. To overcome the sample selection bias arising from estimations out of observed variable in the sample, Heckman (1979) proposed a two-step estimation method. Application of Heckman's two-step procedure used a Probit in the first stage (probability of diversification decision). In the second step, the level of product diversification or diversification equation (Entropy index) was analyzed. The Inverse of Mills Ratio (IMR) is as regressors in this function in order to correct selection bias. Based on these specifications, Heckman specified: Step 1: First, a probit model for diversification decision or selection equation was estimated. The dependent variable in this application assumes either a value of 1 or 0 depending on whether a firm has decided to diversify or not. A probit model was used in estimation given the binary nature of the dependent variable. The binary probit is suitable to deal with latent (or unobservable) dependent variable and usually is expressed as a linear function of a set of explanatory variables as follows: Probability equation: P i * = ? i X i + ?? ?? = ?X'+?? ?? ; ?? ?? ~ N (0, 1)... ......... (4) Threshold index equation: Di = Di * if Pi = 1-------------------------(5) 0 if Pi = 0 Where: P* is a latent index variable that denotes binary censoring, X 1i is a vector of variables that affect diversification decision, ?? ?? is an error term, Piis a binary variable (1 if diversification is observed; and zero otherwise), representing the diversification decision (propensity to diversify). To be specific, it takes 1 if an enterprise produces multiple products and the diversification richness index (MI index) is positive; and it is zero otherwise. In other ways (6) (7) The parameters are estimated using conventional nonlinear optimization algorithms. The efficient score tests suggested by Chesher and Irish (1987) are undertaken to diversification choice of the reported specifications in terms of homoscedastic errors, and a normal distribution of the generalized residuals. The estimated probit coefficients can be interpreted by reference to their effect on the standardized probit index but it is generally more convenient to translate them into marginal and impact effects. The marginal effects are denoted for continuous variables as ø(z)?, where ø(?) denotes the probability distribution (or density) function for the standard normal (8) distribution, ? is the estimated probit coefficient for the corresponding k th continuous variable, and is the standardized probit index computed at the sample mean values of the characteristics. The impact effects are computed as Where ? j is the corresponding probit coefficient for the j th dummy variable and the remainder is as defined above. The asymptotic sampling variances for the marginal and impact effects are computed using the delta method. Now the second stage decision, the intensity of diversification, can be represented as follows: Determinants of Product Diversification Among Micro and Small Enterprises in Wolaita Zone, Ethiopia: An Econometric Analysis # Volume XVI Issue IV Version I Where is a column vector of realizations on an explanatory variables including a constant for business firm i; and ? is a column vector of i unknown parameters. The values of the latent dependent variable are measured on the real line and in this case reflect the underlying propensity of MSEs to have diversification decision. The error term is assumed normally distributed with a mean of zero and a constant variance? 2 . A threshold (assumed zero in this case) is used to delineate whether the firm has diversification choice or not. The probability of the event occurring can be linked to the latent dependent variable as follows: Where P* is the dichotomous realization of the latent dependent variable: with decision: Diversification (Value 1) and No diversification (Value 0). In this study ø (?) denotes the cumulative distribution function operator for the standard normal distribution, and for identification purposes it is conventional to normalize ? =1. # The log-likelihood function is defined as Step2: Intensity of diversification (outcome equation): Entropy index equation: D i * = ? 2 X 2i + ? j ? i ~ N (0, ? 2 ) -----------------------------------(9)D i = Dif P i = 1 -------------------------------------------------(10)0 if P i = 0 In this specification, separate sets of factors are assumed to influence the decisions to participate in business diversification versus the positive entropy index of richness (D i ). Hence, X 1i and X 2i are vectors of explanatory variables that affect equation ( 4) and equation ( 9), respectively. Both variables are also assumed to be uncorrelated with their respective error terms, ? i and ? i assumed to have a correlation rho (?) and their joint distribution is normal bivariate. The ? 1 and ? 2 are the corresponding vectors of parameters. P i is the observed value representing the individual's firm participation decision (i.e., if 1, it means the respondent is reporting a positive amount of entropy index of richness (P* i > 0), else 0). Hence, the actual observed D i equals the unobserved latent value D i * only when a positive entropy index of richness is reported; otherwise, it takes the value of 0. In this specification, the error terms are assumed to be normally and independently distributed in equating ( 4) and ( 9), implying that there is no dependence between the diversification participation and level of diversification (i.e., the two decisions are made independently). Assuming that the error terms in (4) and ( 9) were independent with mean zero, that µ~ N (0,1), and that ? i ~ N (0 , ? 2 ), and that corr(µ, ?) = ?, the stochastic specification in (12) Where ? denotes IMR, ? is the normal probability density function (PDF), ? (.) is the standard normal cumulative density function (CDF), X 1 is a vector of factors known to influence a business firm decision to participate. A significant coefficient of the ?indicates that the selection model must be used to avo.id inconsistency. Then, the new ? is used in Equation ( 12) as an explanatory variable. If ? = 0, then there is no evidence of the selection bias and the regression reverts to 2SLS. When ??0, standard regression techniques applied to the first equation ( 4) correlated with X 1 , yield biased results, which is corrected by including IMR in the second regression. It can be shown that the expected value of D i * when D is observed which is given by Equation (13). The new equation for the second stage regression (level of business diversification degree) equation is then given by: E (Di \X 1 , Pi = 1) = ??X2+ ? ? (?X1) +?j (13) Where, E is the expectation operator, D i is the extent (continuous) of diversification (entropy index of richness), X 2 is a vector of independent variables that affect D i and ? is the vector of the corresponding coefficients to be estimated, ?is the correlation between unobserved determinants of probability to diversify u and unobserved determinants of level of diversification ?, ? is a vector of unknown parameters. Equation (13) gives the expected level of diversification D i , given vectors of observable factors X 2 and given that the household has already made the decision to diversify. This can be explained by vector of observable characteristics X 2 and the IMR evaluated at ? (?X 1 ). To the extent that ? (?X 1 ) is correlated with X 2 , the regression equation ( 9) resulting estimates is biased unless ? = 0. # Results and Discussions a) Descriptive Analysis The study was conducted to identify determinants of product diversification among MSEs atWolaita Zone in Ethiopia. Out of the 14 explanatory variables 10 variables were significant determined diversification decision. These variables were household size, age enterprises, start-up capital, access to market, promoters training, own concentration, business plan, information, communication technology, risk management, enterprises opportunities and strategic location that are discussed below in detail. Promoters Age and Diversification: Comparing diversified and non-diversified MSEs, the average age of diversified and non-diversified MSEs promoters were 34 and 32 years respectively. The result showed that diversified and non-diversified MSEs were found to be young. This implies that younger MSEs promoters are expected to be more adventurous, accept technologies, less risk averse than the older ones and possess comparative advantage with respect diversification and product innovation. The t-test results revealed significant relationship between enterprises age and diversification choice of enterprises at less than five per cent significant level (t =4.91, P=0.027). Enterprise Opportunities and Diversification: Product diversification can be a matter of business choice for improving living standards and accumulating wealth of the promoters (Ellis, 2000). As shown in Table 2, within diversified enterprises,23 per cent of the MSEs promoters employed business opportunities created by economic agents such as flow of market information; government projects and other business related information that facilitated product diversification while the rest 77 per cent did not employ business opportunities. Similarly, from non-diversified MSEs promoters, 14 per cent had employed business opportunities created by economic agents whereas the rest 86 per cent did not employed business opportunities. Chi-square results revealed significant relationship between MSEs opportunities and diversification choice of enterprises at less than five per cent significant level ((?2 = 4.099, P=0.043). Risk Maagement and Diversification: MSEs promoter semploy past work experience of market failure (both supply and demand driven) and decide to diversify as risk mitigation. In general, from diversified MSEs promoters, 66 per cent foreseen/experienced market risk and suitably managed it while the rest 34 per cent did not face market failure. Similarly, from nondiversified MSEs promoters, 82 per cent foreseen/experienced market risk and suitably managed it whereas the rest 18 per cent did not face market failure. This also implies that product diversification by MSEs promotes by perceived or real market risk. Chisquare result revealed significant relationship between risk management and diversification choice of enterprises at less than one per cent significant level (?2 = 8.363, P=0.001). # Market Concentration and Diversification: The market concentration is one of the major structural market characteristics. Market concentration was used as a measure of competitiveness in the market. Market concentration in MSEs referred to the direct involvement of the entrepreneur in the market effectively ensure market share. As shown in Table 2, within diversified MSEs promoters only 41 per cent followed market concentration approach. This implies that about41 per cent of MSE promoters followed the strategy of price or quality control rather than diversifications stay competitive. However, 59 per cent of MSEs did not follow market concentration approach. Similarly, from non-diversified MSEs promoter 30 per cent followed market concentration approach whereas 70 per cent did not follow this approach. This implies that MSEs prefers product diversification to market concentration as approach to mitigate risks (Fikirte and Enderias, 2013; Kale, 2005).Chi-square results revealed significant relationship between market concentration and diversification choice of enterprises at less than five per significant level (?2 = 4.95, p= 026). # Start-Up Capital and Diversification: The average startup capital for diversified enterprises was 19,473Birr whereas for non-diversified enterprises was 21,448 Birr. MSE operators have been starting MSEs businesses by raising financial capital as low as below 25,000 Birr (though this is totally negligible amount these days). This implies that majority of the operators are 'Necessity' promoters (pushed to start enterprises out of sheer poverty/economic necessity) rather than 'Opportunity' promoters. In order to remit the problems of micro financing in study area hence, the shortage of financial resources, the government should exert extra efforts to encourage and effectively attract private MFIs to the market and offer sufficient micro finance for MSEs. The t-test results revealed significant relationship between start-up capital and diversification choice of enterprises at less than one per significant level (t = 23.98, p= 0.000). # Investment in ICT and Diversification: The technology particularly Information Technology (IT) increases the resource use in diversifying of any business. Comparing diversified and non-diversified MSEs, average invested capital in ICT gadgets/instruments were about Birr 4,349 and 6,603 respectively. This implies wider gap prevailing among enterprises in harnessing ICT to its full potential. Hence, MSEs could use computers for book keeping, documentation like files maintenances, and networking, communication, etc. The t-test revealed that average invested capital utilization on ICT significant relationship between diversified and not diversified enterprises at less than one per cent level (t-test =27.66, p= 0.000). Strategic location and Diversification: Location affects diversification of the MSEs product by determining demand for goods and services (Gebreeyesus, 2009 andBelay, 2012). Enterprise location decides cost of raw materials and marketing of finished goods thereby affecting competitiveness (Belay, 2012).In the study, within diversified MSEs promoters, about 71 per centlocated proximity to commercial area makes the enterprises prosper whereas 29 per cent located outside of commercial area. Similarly, non-diversified enterprises found that about 62 per cent investigated were located at commercial sites and 38 per cent outside of commercial site. The chi-square test revealed significantrelationship between diversification choice and strategic location of enterprises at less than five per cent level (?2 =3.35, p= 0.042). Business plan and Diversification: practice of business plan among MSEs promoters was found to affect diversification decision and level of diversification. Regular and updated records enable MSE promoters to track the cash inflow and outflow, thereby minimizing the operational risks and optimizing the profit. In the study, within diversified MSEs promoters, about 96 per cent had business plan whereas 4 per cent follow without business plans. Similarly, non-diversified enterprises found that about 88 per cent investigated had a business planwhereas22 per cent follow without business plan. However, the plans so prepared were sketchy, had improper business projections behind them and consequently were unacceptable to formal financial institution for getting the credit. Hence, Planning practice and record keeping ensures diversification and profitability of MSEs (MUCD, 2013).Chi-square result revealed significant relationship between business plan and diversification choice of enterprises at less than 5 per significant level (?2 = 6.57, p= 0.037). Work Experience and Diversification: Previous work experience would provide knowledge of organizational routines and necessary skills enabling the promoters to apply them to the current business (Delmar and Shane, 2006; Belay, 2012). As presented in Table 3, comparing diversified and non-diversified MSEs, the average work experience of diversified and non-diversified MSEs were 3.48 and 2.64 years respectively. The result showed that business experience gives a person the required technical skill necessary to start and run the current business efficiently. The t-test revealed significant relationship between diversification choice and previous work experienceat less than 5 per cent significant level (t =19.98, p= 0.000). # Family size and Diversification (FASIZE): Comparing diversified and non-diversified MSEs, the average sampled respondents were4.57 and 4.43 respectively. The results showed that average size of the sampled household was comparable with the national average (4.8). This justifies managing a large family requires a substantial financial commitment and in times of economic hardship this may make business owners with more households more risk averse and less likely to choose diversification. However, different studies identified that household size positively affected product diversification. This could imply that diversification of the households was to meet different needs of the family (Weiss and Briglauer, 2000;Benin et al., 2004;Rehimaet al, 2015).The t-test revealed significant relationship between diversification choice and family size at less than one per cent significant level (t =45.78, p= 0.000). Fortune (2003) and Eshetu and Zeleke, (2008), MSEs often collapsed due to inefficiency in financial management caused by lack of business plans. In the study area,a sizeable proportion of MSEs do not have the ability to produce plans for taking advantage of institutional credit. Irregular and unorganized plan/ record keeping practice lead to weak risk monitoring and cash flow tracking on income and expenditure. This study was consonant with those of Eshetu and Zeleke, 2008: Belay, 2012. RISK Management (RSKMGT): As expected, this variable positively affected diversification decision and level of product diversification at less than one per cent significant level in both MSEs together and micro enterprises category. The result indicated that, all other variables being constant, having risk management would increase the probability of diversification decision for both MSEs together and micro-enterprises category by about22.5 and 23.8 per cent. It would increase the level of diversification for both MSEs together and micro enterprises category by 0.635 and 0.670 respectively for having management while other variables are kept constant. The result indicated that risk management favored the probability of diversification decision in related product. Higher product diversification greater would be the risk mitigation. The tendency among the MSE promoters to mitigate risk would encourage them to diversify; after all they would not like to put all their eggs in one basket. This study result was in line with those of Fikirte and Enderias, (2013); Kale, 2005;Santarelli, and Tran, (2013). Access to market (MARKAC): Access to market (indirectly measured in terms of walking time taken to reach market) positively affected diversification decision at less than ten per cent significant level for microenterprises. The result indicated thata one minute walk increase to the nearest market increased diversification decision for micro enterprises category about by 5.1 per cent, assuming all other variables remaining constant. This implies that MSEs incurred higher transaction costs (transport, market information, difficulty in searching new market, etc.) when getting to sell or to buy their product in far off market and that cost consequently could have served as a deterrent in diversification decision. The primary motive of MSEs perhaps would be risk mitigation rather than asset accumulation. Diversification entails higher frequency of interaction with market and MSEs would refrain from diversification in poor market access scenario. A business firm far away from a market was positively related to product diversification which entailed higher transaction costs leading to weak market integration (Joshi et al., 2004;Alpízar, 2007). In addition, according to Admasu (2012), marketing problems included inadequacy of market, difficulty of searching new market, absence of market intelligence and of interaction with organization/association that conduct marketing research. MSEs spatially away from market would have limited market information, negatively affecting diversification decision (Alpízar, 2007;Rehimaet al., 2015 ). # ( E ) # Information, Communication and Technology (CAICT): ICT ownership positively and significantly affected diversification decision at less than ten per cent significant level for both MSEs together and less than five per cent significant level for small enterprises. It also positively and significantly affected level of product diversification for small enterprises at less than five per cent significant level. The marginal effect indicted that a Birr increase investment in ICT would increase the probability of diversification decision for both MSEs together and small enterprises category by 5*10 -5 . It would increase level of diversification for small enterprises category by 0.640while assuming all other independent variables being constant. This implies that investment in ICT enables MSEs to enhance diversification decision and level of diversification. Investment in ICT also enables MSEs to manage price variation through enhanced access to information in product and input markets. Therefore, Additional investment in ICT could, thus, be expected to have a greater influence on diversification decision and level of diversification. This study was consonant with the study of Admassie and Matambalya, 2002;Radamet al, 2008. Enterprises location (LOC): MSEs located far away from market place (outside of commercial area) negatively determined diversification decision at less than ten per cent significant level inmicro enterprises category and less than five per cent significant level in micro enterprises category. The result indicated that one minute increase in walking time from commercial site (indicative of the market distance and location) would decrease diversification decision by 13.4 per cent for micro enterprises category and would decrease level of product diversification for micro enterprises category by 0.142.This implies that location affects diversification of the enterprises as demand for goods and services depended on location (Gebreeyesus, 2009 andBelay, 2012). Enterprise location decides cost of raw materials and marketing of finished goods thereby affecting competitiveness (Belay, 2012). Competing enterprises concentrated on the close geographical area would face stiff competition to serve a given segment of clients and eventually realize lower profit. Distantly spaced enterprises would grow faster, enjoying the patronage of the clients with less competition. Similarly, proximity to commercial area makes the enterprises prosper. It further implies that MSEs located nearer to the commercial place (better market access) enjoys support services apart from incurring less transaction costs, and enhancing economies of scale and product diversification. Contrary to commercial area, MSEs located far away from the market (commercial center) were less inclined for diversification decision and level of diversification (Joshi et al., 2004;Alpízar, 2007) # IV. Conclusion and Recommendation The study was conducted to identify determinants of diversification decision and level of diversificationin micro and small enterprises at Wolaita Zone in Ethiopia. The result showed that participation in diversification and level of diversification in micro and small enterprise s was significantly determined by startup capital, market access, managerial training, age of promoters, enterprises opportunities, own concentration, economic size, business plan, risk management and ICT. Out of the 10 significant explanatory variables e ntrepreneurs' prior experience in risk management, enterprise economic size, and investment in ICT determined diversification decision and level of diversification positively and significantly. Promoter's age, business location, enterprises opportunities, business plan and enterprises own concentration determined diversification choice and level of diversification significantly and negatively. Access to market was negatively and significantly related with diversification decision while managerial training was related positively and significantly with level of product diversification. Based on the findings of product diversification the following policy recommendations were made. Development practitioners should create awareness among members and encourage the use of family planning in order to limit household size. This can be achieved through integrated health and education services. Businesses promoters who are participating in related product diversification invested considerably on ICT for coordination of inputs, industry knowledge, production skills, special technology and distribution channel. Therefore, government body and business promoters should promote for ICT infrastructure. Age matters in diversification. Hence, the government should strengthen training system to train the older enterprises and entrepreneurs as they were found to have less inclination for diversification which is desirable for risk mitigation and sector stability. .Prior experience in business and enterprises wealth contributed to higher product diversification through reinvestment of higher income generated out of the MSEs. Again better enterprise training would compensate for lack of experience among budding entrepreneurs. Access to startup capital has significant and negative effect on level of product diversification. Moreover, the functionality of MSEs promoter is also constrained by shortage of start-up capital. An effective and sustainable MSEs movement requires overcoming major start-up capital constraints. It is imperative that the government should provide support with respect to timely and adequate supply of affordable and timely start-up credit facilities for Micro and Small enterprises significantly to make them become competitive in the local, regional as well as national market. Banks should allocate some resources and develop innovative ways of lending to small businesses and enterprises by following the successful example of the Grameen Bank of Bangladesh. Product diversification helps MSEs to mitigate production, income and price risks because of spreading their investments in different related products. Thus, policies need to foster product diversification. Hassle free loan and enterprise training are among those interventions that would go a long way in promoting diversified product by MSEs in the country. ![.................................. (1)](image-2.png "") ![Di = b i + Xc i + ? i .................................. (2)](image-3.png "") ![D ? 0........................... (3)](image-4.png "??)") ![....... .......... .......... .......... .......... .......... ....... .......... .......... .......... ....... .......... .......... .......... ..........](image-5.png "") a) Study AreaWolaita Zone is one of 14 zones in SouthernNations, Nationalities and Peoples' Region (SNNPR) ofEthiopia. 2Variable CodeDescription, Type of Data and Operational MeasurementExpectedsign ( +/ -)AGEAge of Promoters in years-INCAPAmount of start-up capital in Birr+MARKACMarket access dummy ( 1= if accessed and 0 otherwise)+PWEXPPrevious work experience in years+BUSPLBusiness plan-dummy (1= if MSE promoters owned business plan and 0+otherwise)MTRIADDuration of skill training provided in months+FOPPOpportunities employed by firms created by economic agents, dummy(1= haveemployed opportunity(information or market) and 0 otherwise)+LOCLocation of the MSEs (1= if in commercial area and 0 otherwise)+OWNCONOwnership concentrated market-dummy (1= if MSEs ownership concentrated,-0= otherwise)FASIZEThe family size of the MSEs owners in number-STRHHADiversifying product strengthening household assets base Dummy-(= diversifying+product that strengthens household asset and 0=otherwiseFWELTHEnterprise Capital size dummy(1 = increased, 0 = decreased)+CAICTICT Investmentcost in Birr+RISKMTProper Risk management Dummy-(1= managing risks and 0, otherwise).+III. 3Micro(n=265)Small(n= 85)Both MSEs?2/ t-valueVariableCategoryDiversifyNot DiversifyDiversifyDiversifyNotDiversify(n=136)Diversify(n=216)NotAge of theAverage3432333234324.27**promoters(7.3)(6.3)(6.6)(6.2)(6.8)(6.2)WorkAverage3.252.683.532.423.482.6419.98***experience(2.6)(2.4)(3.0)(2.3)(3.1)(2.4)Family sizeAverage4.484.454.494.34.574.4345.78***(1.2)(1.2)(1.6)(1.6)(1.6)(1.6)ICTAverage45174925609047845050481527.67***investment(2147)(3532)(5396)(3453)(3691)(3080)Start-upAverage18,74919,45326,58426,98719,47321,44823.98***capital(13,623)(14322)(21,355)(21,360)(16,053)(16,299)EnterprisesYes25286331(23)31(14)4.099*opportunitiesNo801322553105(77)185(86)RiskYes68129224890(66)177(82)8.363***ManagementNo37319846(34)39(18)MarketYes4248141656(41)64(30)4.95*concentrationNo63112174080(59)152(70)StrategicYes77100203497(71)134(62)3.35*locationNo2860112239(29)82(38)Business planYes1011392951130(96)190(88)6.57**No421256(4)26(12)n= sample size, ***, ** and * indicate that statistically significant difference at less than 1%, 5% and 10% significant level,respectively. The figures in Parenthesisare percentages and standard deviation. Source; Computed from Field Survey data,2015/16b) Determinants of Diversification Decision and Level ofwealth (FWELTH), investment costin ICT(CAICT),Product DiversificationamongMSEsManagerial training (MTRIAD), Strategic location (LOC)Soundness of the model was established byand business plan (BUSPL).The second stages ofWald test. The chi-square of the model regression in theHeckman-Two-Stage model result are also presented inZone indicated overall goodness of fit (showing a strongTable 4. The level of product diversification representedexplanatory power) of the model with statistical significance at a probability of one per cent. The Wald test of the business enterprises ? 2 (14) 2 (14) = 131.49), ? = 121.98), and ? 2 (14) = 26.35), for both MSEs together, Micro and Small enterprises respectively confirmed that the coefficients of the level of diversification equation was significantly different from zero. As a result the model fulfilled conditions of good fit.by Entropy Index, which was significantly determined byFOPP, OWNCON, CAICT, RISKMGT, LOC,BUSPL, FWELTH and INCAP Age of the promoters (AGE): Age of the household head negatively affected the productdiversification decision at less than five per cent significance level. Keeping all others variables constant, one-yearincrease in age of enterprises promoters caused decrease in probability of product diversification decision of both MSEscategoryTable 3 presents the probit model's estimatesby 0.5 per cent. From this, one would expect olderunderlyingtheHeckman-Two-Stepestimationpromoters to be less likely to engage in related productprocedure. It clearly shows the binary dependentdiversification. This showed that older promoters mightvariable: one (1) if the business enterprises diversifiedbe less receptive to technology/adventurous/risk averse.and zero (0) otherwise. Ten variables significantly explained the probability of product diversification.This result concurred with those of Weiss and Briglauer's (2000); Fikirte and Endrias, (2013;These are market access (MARKAC), promoters ageMashimbaandKihll(2014).(AGE), opportunities (FOPPD), ownership concentrationEnterprises opportunities (FOPPD): Contrary to the(OWNCON), RSKMGT (risk management), enterprisesexpectation,thisvariablenegativelyaffected Promoters training (MTRIAD): As expected, promoterstrainingpositivelyandsignificantlyaffecteddiversification decision at less than ten per centsignificant level in both MSEs together. The resultindicated that one month increase in promoters' trainingwould increase product diversification decision for bothMSEs together by4.5per cent, assuming all othervariables remaining unaltered. This implies that trainingopportunity minimized the risk of failure involved inrelated product diversification. The provision of trainingto entrepreneurs who wanted to start new businesses orrelated product gave businesses a better chance ofexpansion or product diversification. Training furtherenhances individual's access to information andtechnology thereby contributing to a wider array ofbusinesses. Several studies identified importance oftraining in product diversification (Santarelli,. and Tran.2013 and Ibrahim et al. 2009). 4VariablesBoth MSE altogetherMicro-Enterprises CategorySmall-Enterprise CategoryCoefMarginalCoefMarginalCoefMarginalMARKAC0.1410.0530.275*0.051-0.1010.030(0.172)(0.065)(0.217)(0.082)(0.334)-(0.120)AGE-0.013**-0.0050.0135-0.0050.01810.006(0.006)(0.004)(0.014)(0.005)(0.027)(0.010)PWEXP-.092-0.034-0.010-0.003-0.474-0 .171(.257)(0.097)(0.301)(0.114)(0.584)(0.210)INTCAP-5.07e-06-1.92e-06-1.51e-06-5.74e-07-9.31e-06-3.36e-06(4.89e -06 )( 1.84e-06)(6.60e-06)(2.52 e-06)(9.05e-06)(3.27e-06)LOC-0.208-0.078-0.352*-0.1340.258-0.093(0.162)(0.061)(0.188)(0.071)(0.378)(0.137)MTRIAD0.119*0.0450.1690.0640.0960.034(0.06)(0.023)(0.212)(0.080)(0.407)(0.147)OWNCON-0.390**-0.147-0.310*-0.118-0.745*-0.269(0.160)(0.060)(0.187)(0.071)(0.376)(0.135)FOPPD-0.341*-0.129-0.226-0.086-1.16**-0.420(0.184)(0.069)(0.203)(0.077)(0.553)(0.200)RSKMGT0.595***0.2250.626***0.2380.5190.187(0.167)(0.063)(0.191)(0.072)(0.4009)(0.145)STHHAB-0.551-0.208-0.456-0.174-1..049-0.379(0.398)(0.150)(0.488)(0.186)(0.872)(0.311)FWELTH0.345*0.1300.3040.1150.5170.187(0.209)(0.079)(0.237)(0.090)(0.549)(0.198)CAICT0.00003*0.00001-0.00002-7.84e-060.00005**0.00002(0.00002)(8.00e-06 )(0.00004)(0.00001)(0.00002)(0.0001)FASIZE-0.037-0.014-0.041-0.015-0.030-0.010(0.051)(0.019)(0.061)(0.023)(0.112)(0.040)BUSPL-0.677***-0.251-0.963***-0.366-0.280-0.101(0.268)(0.101)(0.348)(0.132)(0.558)(0.201)CONS1.261.632.85(0.991)(1.15)(2.74)n=352,LR 5VariablesBoth-EnterprisesMicro-EnterprisesSmall-EnterprisesCoef.St. errorCoef.St. errorCoefSt. errorsCONS-0.2880.526-0.0560.636-1.2601.110LN(AGE)-0.4190.3130.5300.3670.1860.705MARKACC-0.1530.161-0.2540.205-0.1080.307LOC-0.1910.161-0.142**0.1140.1230.361LN(MTRIDA)0.1530.1810.176 .0.2090.2890.404OWNCON-0.406***0.154-0.297*0.178-0.833**0.354FOPPD-0.317*0.183-0.1960.200-0.9220.514RSKMGT0.635***0.1680.670***0.1930.5580.392FWELTH0.327*0.2060.3220 .2330.4510.496LN(INTCAP)-0.092**0.2320.1420.1140.467**0.197LN(CAICT)0.1960.1300.0730 .1580.640**0.289BUSPL-0.692***0.266-0.978***0.3410.2830.521Number of obs= 352 ,Number of obs= 265 ,Number of obs= 87 ,Censored obs = 216Censored obs = 160Censored obs = 56Uncensored obs=136Uncensored obs=105Uncensored obs=31Wald chi 2 = 131.49Wald chi 2 = 121.98Wald chi 2 = 26.35Prob>F=0.000Prob>F=0.000Prob>F=0.0233Rho= 0.679Rho= 0.898Rho= 1.00Sigma= 0.225Sigma= 0.225Sigma= 0.438Lambda= 0.153Lambda= 0.2023Lambda= 0.438 © 2016 Global Journals Inc. (US) s © 2016 Global Journals Inc. (US) © 2016 Global Journals Inc. (US) s 26 ( E ) © 2016 Global Journals Inc. (US) s 28 ( E ) * Technical Efficiency of Small and Medium Enterprises: Evidence from a Survey of Enterprises in Tanzania AAdmassie FMatambalya 2002 MUSE Project * Risk coping strategies and rural household production efficiency: Quasiexperimental evidence from El Salvador. A PhD Dissertation Presented in the Graduate School of the Ohio State University CAAlpízar 2007 209 USA * Factors affecting the performance of micro and small enterprises in Arada and Lideta Sub-Cities Admasuabera 2012 Addis Ababa Presented to the School of Graduate Studies of Addis Abeba University Msc Thesis * Small and Medium-Size Enterprises: Access to Finance as a Growth Constraint TBeck ADemirguc-Kunt Journal of Banking & Finance 30 11 2005 * Determinants of Microenterprise Success in the Urban Informal Sector of Addis Ababa: Multidimensional Analysis. Thesisto obtain the degree of Doctor from the Erasmus University Rotterd GBelay 2012 Netherlands * Area and production of crops Statistical bulletin 417 2007 CSA (Central Statistical Agency) of Ethiopia * Residual analysis in the grouped and censored normal linear model AChesher MIrish Journal of Econometrics 34 1-2 1987 * Rural livelihoods and diversity in developing countries FEllis 2000 Oxford University Press Oxford * Factors that affect the long-term survival of micro, small, and medium enterprises in Ethiopia ZelekeworkuEshetubekele South African Journal of Economics 10 2 2008 * Measuring on farm diversity and determinants of barley diversity in Tigray AFetien ABjornstad MSmale Ethiopia.MEJS 1 2 2009 * Determinants of Growth and Diversification of Micro and Small Enterprises: the case of DirDawa town WFikirte GEndrias 2013 Ethiopia The master thesis * Is Ethiopia in a Poverty Trap? EFortune Ethiopian Business Journal 4 158 2003 * Innovation and Growth of Micro-Enterprises: Empirical evidence from Ethiopia. Ethiopian Development Research Institute (EDRI) MGebreeyesus 2009. First Draft -February, 2009 * Growth and Determinants of Micro and Small Enterprises: Evidence from Northern Ethiopia Habtamu AregawiAndaberaGebremichael Nigus Journal of Economics and Sustainable 4 9 2013 Development * Sample selection bias as a specification error JJHeckman Journal of Econometrical 47 1 1979 * Growth, Diversification, and Business Group formation in Entrepreneurial firms DIacobucci Rosa Small Business Economics 25 1 2005 * Income and crop diversification among farming households in a rural area of north central Nigeria HIbrahim SARahman EEEnvulus SOOyewole J. of Tropical Agriculture, Food, Environment and Extension 8 2 2009 * Jobs Gender and Small Enterprises in Africa Ethiopian Women Entrepreneurs: Going for Growth May 2003 ILO Subregional Office, Addis Ababa, Ethiopia and Women's Affairs Department, Ministry of Trade & Industry (MTI/WAD) in association with InFocus Programme on Boosting Employment through Small Enterprise Development International Labour Office 2003 ILO Geneva International Labor Office * Agriculture diversification in South Asia: Patterns, determinants, and policy implications PKJoshi AGulati SBirthal LTewari Economic and Political Weekly 39 4 2004 * Organizational Strategy and Development: New arenas for action SKale Journal of Management 18 2005 * The dynamics of micro and small enterprises in developing countries MeadLiedholm DC Journal of Finance 26 1 2008 * Diversification Strategy Raises Doubts.National Real Estate Investor Stan2004Luxenber Germany * Performance of Micro and Small Scale Enterprises in Tanzania.Growth Hazard of Fruits and Vegetables Processing Vendors SMashimba RKihll Journal of Applied Economics and Business Research 4 2 2014 JAEBR * Ethiopia's Growth and Transformation Plan Ministry of Finance and Economic Development Addis Ababa, Ethiopia; Ethiopia Addis Ababa 2014. 2014 * Ministry of Urban and Construction Development, 2013.Survey on Micro and Small Enterprises (MSEs) * Volume XVI Issue IV Version I * Technical Efficiency of Small and Medium Enterprise in Malaysia: A Stochastic Frontier Production Mode: Institute of Agricultural and Food Policy Studies ARadam MAbu AAbdullah Selected Major Cities of Ethiopia.MUCD: Addis Ababa Ethiopia 26 Serdang, Selango 2008. 2008 43400 University Putra Malaysia * Determinants of Crop Diversification in Ethiopia: Evidence from Oromia Region BelayRehimamussema DawitalemuKassa RashidShahidur Ethiopia Journal of Agricultural science 25 2 2015. 2015 * The dynamic and importance of SMEs in economy MRobu 2013 * Diversification Strategies and Firm Performance: A ample selection Approach. Quaderni-working paper. DSE No 896 ESantarelli HTran 2014 * Determinants and dynamics of farm diversification.Department of Food Economics and Consumption Studies CRWeiss WBriglauer EWP 0002 2000 20 University of Kiel Working Paper * Data Collection, organization and Dissemination Work Process Annual Abstract 2015 WolaitaSodo Ethiopia Wolaita Zone Trade and Industry Development