Determinants of Micro and Small Scale Enterprises’ financing preference in line with POH and Access to Credit: Empirical Evidence from Entrepreneurs in Ethiopia, Benishangul-Gumuz Regional State

Purpose: The purpose of this study was to examine factors that determine micro and small scale enterprises’ financing preference in line with pecking order theory and access to credit in Benishangul-Gumuz Regional State of Ethiopia. Design / Methodology / Approach : The study used primary data collected using cross sectional survey questionnaire followed by mixed research approach. The sample of this study was 296 enterprises selected using proportional stratified random sampling technique. The data was analyzed using descriptive and logistic regression analysis. Findings : The results of logistic regression analysis revealed that business experience, collateral, gender, motivation and enterprises’ sectoral engagement affect financing preference of enterprises in line with pecking order hypothesis. On the second step, only enterprises that need to raise capital through credit were considered to investigate access to credit determinants. Accordingly, the logistic regression result revealed that business experience, size, sectoral engagement, collateral, interest rate, loan repayment period, financial reporting, preparation of business plan, location and educational background of entrepreneurs affect access to credit of enterprises. Research limitations/Implications : More evidence is on enterprises’ financing preference and access to credit determinants before any generalization of the results can be made. In addition, the empirical tests were conducted only on 296 entrepreneurs since 2019. Therefore, the results of the study cannot be assumed to extend beyond this group of entrepreneurs to different study periods. Originality/value : This paper adds the on the determinants of micro and small scale enterprises’ in


Introduction
In spite of their contribution to employment, Micro and small scale enterprises (MSSEs, hereafter) face various constraints while operating their business Nguyen, Gan, & Hu, 2015;Belanova, 2013;Ali, Godart, Gorg, & Seric, 2016). Moreover, access to finance is by far the most critical bottleneck of MSSEs in Ethiopia (Assefa, Zerfu, & Tekle, 2014;Menkir, 2016;Tarfasa, Ferede, Kebede, & Behailu, 2016). In order to promote MSSEs as engines of growth, however, it is essential to understand the bottlenecks surrounding their access to finance (Mersha & Ayenew, 2017). Otherwise, the financial constraints they face in their operations will daunt their development negatively that could limit their potential to drive the national economy as expected (Manaye & Tigro, 2017).
In literature, there are various factors that may influence access to finance. Some argue that the fundamental reasons behind MSSEs' lack of access to funds can be found in their peculiar characteristics, while others argue that MSSEs suffer from financing gaps because of supply side factors (Mazanai & Fatoki, 2012). Moreover, MSSEs may face financing gaps probably because of combination of factors originating from both the supply and demand sides (Stijn & Tzioumis, 2006). However, irrespective of its source, whether raised internally by MSSEs owners or externally through credit, finance is needed to start and expand business. To identify literature based variables, systematic review was carried out on studies made by Alhassan & Sakara, 2014;Awlachew & Motumma, 2017;Elly & Kaijage, 2017;Fufa, 2016;Gamage, 2013;Gebru, 2009;Kebede, Tirfe, & Abera, 2014;Kira & He, 2012;Kung'u, 2011;Kuruppu & Azeez, 2016;Makina, Fanta, Mutsonziwa, Khumalo, & Maposa, 2015;Manaye & Tigro, 2017;Marwa, Micro, 2014;Mashenene, 2015;Mersha & Ayenew, 2017;Mole & Namusonge, 2016;Nguyen & Wolfe, 2016;Nega & Hussein, 2016;Nguyen, Gan, & Hu, 2015;Nkuah, Tanyeh, & Gaeten, 2013;Osano & Languitone, 2016;Waari & Mwangi, 2015;Zarook, Rahman, & Khanam, 2013) With regard to financing MSSEs, the most competing theories of financing decisions are static tradeoff theory and pecking order theory (Johnsen & McMahon, 2005). On the one side, firm's optimal debt ratio is usually viewed as determined by a tradeoff of the costs and benefits of borrowing, holding the firm's assets and investment plans constant as of trade off theory. The firm is portrayed as balancing the value of interest tax shields against various costs of bankruptcy or financial embarrassment (Myers, 1984). On the other side, the pecking order model of corporate financing says that when a firm's internal cash flows are inadequate for its real investment, the firm issues debt (Shyam-Sunder & Myers, 1999). However, an application of firm value optimization technique under static trade-off theory demands substantial reliable data. But, MSSEs usually have weakly organized accounting system (Gebru, 2009). Therefore, the purpose of this study was to examine factors that determine micro and small scale enterprises' financing preference to establish whether these enterprises had followed the pecking order hypothesis and access to credit in their operation in Benishangul-Gumuz Regional State of Ethiopia using binary logistic regression. The remainder of this paper is structured as follows: Section two discusses about research methodology followed by section three that presents empirical results and discussion. Finally, section four provides the conclusion thereafter forwards recommendation.

Research Methodology Research Design and Approach
The research design was correlational explanatory research design based on a deterministic philosophy in which causes probably determine outcome followed by mixed approach.

Data Type And Source
The study used primary data collected from selected micro and small scale enterprises since 2019.
The main instrument for data collection in this research was questionnaire which incorporated both open ended and close ended questions. The questionnaire was prepared in English language.
Reliablity and validity of the instrument was also checked. It is evidenced in literature that reliability can be checked using test-retest measurements of the same construct administered to the same sample at two different points in time. Besides, validity can be assessed based on correlational coefficient of pilot test data in quantitative research (Bhattacherjee,2012). In this study, the survey instrument was first reviewed by lecturers of accounting and finance department in Assosa University for validity and then pre-tested to evaluate its suitability on 30 piloted entrepreneurs. Thereafter, a test-retest method was used to examine the reliability of the instrument and the instrument was administered twice to the same group of subjects at an interval of one month and gave a correlation coefficient of 0.724 that indicates high reliability of the instrument for the fact that coefficient of 0.5 and above is deemed reliable as of Kothari (2004).

Population, Sample Size And Sampling Technique
There were 1,140 micro and small scale enterprises according to the data obtained from Benishangul Gumuz Regional State of MSSEs agency during 2019. The target population of the study was, therefore, all micro and small business enterprises in the study area. Moreover, geographical and sectoral population distribution is given below in Table 1. Indeed, sample size was determined based on the whole zonal active enterprises using a simplified formula which is developed by Yamane (1967).
Where, level of precision= 5% Therefore, representative of 296 enterprises were used from the target enterprises. With regard to sampling technique of the survey, the study used a combination of cluster, stratified and purposive sampling methods. First, four clusters (Assosa Wereda, Bambasi Wereda, Menge Wereda, and Homosha Wereda) were selected purposely due to their relative higher enterprises' density.
Thereafter, the enterprises were stratified in to manufacturing sector, construction sector, service sector, agricultural sector and trade to create sectoral homogeneity in each. Accordingly, proportional representative enterprises were selected through snowballing to come up with a total of 296 respondents as depicted in Table 2 below. Variable Description And Model Specification The choice of dependent and independent variables with their measurement is a matter of no choice while specifying an empirical model. In line with this, dependent variables of this study were financing preference and access to credit by micro and small scale enterprises. Besides, independent variables were combination of owner's characteristics, firm related characteristics and creditor related variables which are described in Table 3. In this study binary logistic regression model was used to examine the relationship between independent variables and dependent variables (Financing preference and access to credit of MSSEs).
The basis for selecting the binary logistic regression model was the nature of both the dependent variables which are financing preference and access to credit. The first dependent variable, financing preference, was rated and reported by the respondents with a discrete scale of 1 and 0, where 1 is financing preference from internal sources and 0 is for financing preference from external sources.
Access to credit was also rated and reported by the respondents with a discrete scale of 1 and 0, where 1 is access to credit and 0 is for not accessing credit.
According to Gujarati (2004), the cumulative logistic probability distribution model for this study could be econometrically specified as follows: Where 'e' is the base of natural logarithm; P i is the probability of success; X i represents the i th explanatory variables; α & β i are regression parameters to be estimated If P i, the probability of success, is given by Eq. 1 above, then (1-p i ), probability of failure could therefore be: Thus, for ease of interpretation of the coefficients, a logistic model could be written in terms of the odds and log of odd by taking natural logarithm of Eq. 2. Indeed, we can rewrite the above equation as: Now, Pi/ (1-P i ) is the odd ratio which is the ratio of the probability of success (P i ) to the probability of failure. Now if we take the natural logarithm of Eq. 3, we could have the following result.
If the stochastic error term, U i is taken into account for estimation purpose, the logit model becomes: The logistic regression can therefore be specified as: In this study, two dependent variables which are financing preference and access to credit were used.
In both cases, MSSEs are assumed to have two choices which are "preference of internal financing" or "preference of debt financing" in the first scenario and MSSEs' "access credit" and "no access to credit" in the second case. Therefore, the binary choice logistic regression model that assumes dichotomous dependent variable which takes either 1 or 0 value was used. Accordingly, probability of success, P i, represents probability of preferring internal source in the first case and probability of having access to credit for the second. On the other side, probability of failure, (1-Pi), represents probability of preferring external source in the first scenario and probability of not having access to credit for the second.

Empirical Result And Discussion
Both descriptive and logistic regression analysis were applied using STATA software version 13 for statistical treatment. In this section, descriptive statistics is carried out.   Source: Author's computation based on firm survey (2019) The survey revealed that majority of the respondents with 70.94% had operated their businesses for a period of fewer than three years followed by 26.69% with business experience that ranges between 4 and 5 years while those who had been in operation for more than five years shared the least percentage which is 2.37%. Therefore, majority of the enterprises are at their start up stage.  (2019) The results of the questionnaire indicated that 40.54%) were female and the remaining 59.46% were male.  (2019) Among the enterprises, with regard to motivation of owners, almost 57.43% of the owners did not join to the sector by their choice and the rest 42.57% of the owners joined to their business with interest. Among the enterprises, majority of the respondents in the survey (55.41%) use their own land for operation, 15.88% operated their businesses on rented houses while the remaining percentage (28.72%) operate their business using premises from the government. Source: Author's computation based on firm survey (2019) The survey revealed that 62.96% of the respondents replied that their request for loan was accepted by lenders whereas 37.04% of the respondents reported rejection of their request.  (2019) The study revealed that among the enterprises that prefer external debt as means of financing, 5.56% are engaged on agricultural sector, 17.28% are engaged on trade sector, 29.63% are engaged on construction, 23.46% are engaged on service sector, and the remaining 24.04% are engaged on Manufacturing sector. Therefore, relatively, more of the enterprises engaged in construction sector Prefer external source of financing.
On the one side, according to the survey, 40.74% of owners of micro and small enterprises responded that they prepare business plan which could assist the operation of their businesses. On the other hand, 59.26% of the owners answered that they do not have business plan for their business and described that they faced number of problems one of which being lack of access to loan because business plan is a proposal that describes a business opportunity to financing. Among the enterprises that need to finance their operation from external sources through credit, 75.31% were operated in Assosa Wereda, 12.35% Bambasi Wereda, 8.02% in Menge and 4.32% were operating in Oda.

Econometric Analysis
In this section regression analysis is carried out using odds ratio and marginal effect. In this study, the econometric analysis section consists of two step analysis. The first step is used to analyze financing preference of entrepreneurs to test pecking order theory of capital structure by binary logistic regression model. On the second step, only enterprises that need to raise capital through credit are considered. In this section discussion of financing preference determinants for micro and small scale enterprises is carried out. Prior to running logistic regression, explanatory variables were checked for the existence of multicollinearity problem by using Variance Inflation Factor (VIF). The study reveals that the VIF values are all well below 10. Therefore, it can be safely conclude that there is no collinearity problem within the data.
The association between financing preference and its determinants was explained by using binary logistic regression model. In the logistic regression model, thus, the dependent variable, financing preference, was rated and reported by MSSEs' owners with a discrete scale of 0 and 1, where 1 indicates financing preference from internal sources like personal saving and retained earnings and 0 indicates preferring external sources. The Hosmer and Lemeshow test for the goodness of fit in the logistic model has Prob > chi2 = 0.8971 which indicates that it is not statistically significant and therefore the model is quite a good fit. Table 12, econometric results from binary logistic regression on financing preference of entrepreneurs' show that 5 of the 8 variables are highly influential on financing choice which includes business experience, collateral, gender, motivation and enterprises' sector. The other variables including firm size, education and access to land failed to show significant influence. Ceteris paribus, as described in Table 12 above, female owned micro and small scale enterprises was found to have positive relation with financing decision of MSSEs in light with pecking order theory and statistically significant at 1 percent. The odds ratio of the variable "gender of owner" indicates the probability of choosing internal financial source of MSSEs that are owned by female owners is 2.78 times higher than male owned counterparts. The marginal effect of this variable shows that the probability of choosing internal financing for female owned MSSEs increase by 24.32 percent as compared to male owned MSSEs. Therefore, it is evident that female owned MSSEs more likely prefer internal financial sources as compared to male owned MSSEs which is in favor of pecking order hypothesis of financing decision.

As indicated in
Entrepreneurial interest of micro and small scale enterprises' owners was found to have positive relation with financing decision of MSSEs following pecking order theory and statistically significant at 10 percent. The odds ratio of the variable indicates the probability of choosing internal financial source of MSSEs that are established by interested owners is 1.65 times higher than those who join to the business without interest due to lack of other choices. In addition, the marginal effect of this variable shows that the probability of choosing internal financing for motivated MSSEs' owners increase by 12. 37 percent as compared to their counterparts. Therefore, it is found that motivated entrepreneurs more likely prefer internal financial sources supporting pecking order hypothesis.
This result contradicts with the finding of a study made by Gebru (2009) in MSSEs of Tigray regional state that Business experience, measured by year of establishment, has a positive effect on financing decision at 1percent significant level statistically in favor of pecking order theory of financing decision. The odds ratio in favor of internal financing is increased by 1.44 times as the business experience increases by one year. In addition, the marginal effect implies that, ceteris paribus, the probability of choosing internal financial source increases by 9 percent as business experience increases by a year.
In the following logistic regression result, only those enterprises with external financing preference were considered. Therefore, 'access to credit' refers to those respondents who were able to receive credit from a lending institution, taking a dichotomous response variable of 'yes indicated by 1' or 'no indicated by 0' for those who had credit and those who did not respectively. As indicated in Table 13 above, the Hosmer and Lemeshow test for the goodness of fit in the logistic model has Prob > chi2 = 0.9990 which indicates that it is not statistically significant and therefore the model is quite a good fit. Business experience has a positive effect on access to credit at 1percent significant level statistically. The odds ratio in favor of access to credit is increased by 25.02 times as the business experience increases by one year. In addition, the marginal effect implies that, ceteris paribus, the probability of accessing credit increases by 33.6 percent as business experience increases by a year.
Enterprise size has a positive and statistically significant effect on MSSEs' access to credit at 1% level of significance. Holding other factors constant, the odds ratio in favor of access to credit is increased by 1.11 times as the number of employee's increases by one. The marginal effect shows that the probability of accessing credit increases by 11.1 percent for MSSEs as employee number by a unit.
The logistic regression results indicates that holding other factors constant, the probability of accessing credit for MSSEs that are engaged in service sector is 16.95 times higher than those engaged in manufacturing sector at 5 percent significance level statistically. The marginal effect shows that the probability of accessing credit increases by 19.17 percent for those MSSEs that are engaged in service sector than those engaged in manufacturing sector.
The output of binary regression revealed that the variable collateral, represented by possession of fixed asset, has a positive and statistically significant effect on MSSEs' access to credit at 1% level of significance level statistically. The odds ratio indicates that MSSEs which have fixed asset are 30.81 times more likely to access credit than their counter parts. The marginal effect, in addition, shows that the probability of accessing credit increases by 33.44 percent for those MSSEs that have collateral than their counter parts.
Interest rate has a negative and statistically significant association with MSSEs 'access to credit at 5% significance level statistically. The odds ratio indicates that MSSEs with negative attitude about interest rate are 0.13 times less likely access credit than their counter parts. The marginal effect, in addition, shows that the probability of accessing credit decreases by 21.19 percent for those MSSEs' owners who believe that interest rate of loan is higher than their counter parts.
Loan repayment period has a negative and statistically significant association with MSSEs 'access to credit at 10% significance level statistically. The odds ratio indicates that MSSEs with negative attitude about loan repayment period are 0.13 times less likely access credit than their counter parts. This is to mean that opinion about loan repayment period does not affect the probability of MSSEs owners' access to credit.
The output of binary regression revealed that financial reporting has a positive and statistically significant effect on MSSEs access to credit at 1% level of significance statistically. The odds ratio indicates that MSSEs that prepare financial reporting are 29.81 times more likely to access credit than their counter parts. The marginal effect, in addition, shows that the probability of accessing credit increases by 26.41 percent for those MSSEs that prepare financial reporting than their counter parts.
The odds ratio shows that the probability of MSSEs' access to credit with business plan is 35.49 times higher than their counter parts. The marginal effect also indicated that the probability of MSSEs' access to credit with business plan increased by 34.9% compared to MSSEs that don't have business plan at 1 percent level of significance statistically.
Location of enterprises has a positive and significant influence on MSSEs' access to finance. The odds ratio indicates that, holding other factors constant, probability of access to finance for MSSEs operate at Bambasi Wereda is 23.93 times higher than those operated in Assosa Wereda.
Educational level of the MSSEs' owner/manager has a positive and statistically significant effect on MSSEs' access to credit at 5% level of significance statistically. The odds ratio indicates that MSSEs who have not formal education are 0.026 times less likely to access credit compared with those who attend TVET and above. The odds ratio for primary school education is 0.003. This indicates that compared to MSSE owners/ mangers who have attended TVET and above, those with who attend primary school are 0.003 times less likely to get credit at 1% level of significance statistically. In the same vein, the odds ratio indicates that MSSEs who attend education in secondary school are 0.01 times less likely to access credit compared with those who attend TVET and above.

Conclusion And Recommendation
The purpose of this study was to examine factors that determine micro and small scale enterprises' financing preference in line with pecking order theory and access to credit in their operation in Benishangul-Gumuz Regional State of Ethiopia using logistic regression using two stage analysis.
The first step is used to analyze financing preference of entrepreneurs and its determinants to test whether the firms followed pecking order hypothesis in their financing decision by binary logistic regression model using data collected from 296 enterprises. The evidence revealed that almost 54.73% (162) of the enterprises prefer to finance their businesses by external source of fund through credit. On the other side, 45.27% (134) prefer to raise fund internally from their own source. Therefore, majority of the entrepreneurs did not follow a financing decision suggested by pecking order hypothesis. The results of logistic regression analysis revealed that business experience, collateral, gender, motivation and enterprises' sectoral engagement affect financing preference of enterprises. On the second step, only enterprises that need to raise capital through credit (162 enterprises) were considered to investigate the determinants of access to credit. Accordingly, the survey revealed that 62.96% (102) of the respondents replied that their request for loan was accepted by lenders whereas 37.04% (60) of the respondents reported rejection of their request. The logistic regression result revealed that business experience, size, sectoral engagement, collateral, interest rate, loan repayment period, financial reporting, preparation of business plan, location and educational status of entrepreneurs affect access to credit of enterprises. Therefore, it is recommended that MSSEs' owners should share business experience, improve entrepreneurial interest, prepare business plan and produce financial report for the fact that these variables are powerful in explaining outcomes of the study.

Availability of data and materials
The data that support the findings of the study can be obtained from the author based on request.

Competing interests
The author declare that there is no competing interest

Funding
This research was sponsored by Assosa University.

Author's contributions
The author personally undertook this research paper. The author also read and approved the final manuscript.

Ethics approval and consent to participate
'Not applicable' 'Not applicable'

Authors' information
'Not applicable'

Consent for publication
The author personally approve that the paper should be published in your journal.