Effects of gender on the performance of micro and small enterprises in Malawi

Abstract

Women are increasingly venturing into ownership of micro and small enterprises, either on their own or in partnership with male entrepreneurs. Using national survey data from Malawi, this study compares the performance of enterprises owned by females with those owned by males. The results show that the relationship between gender and business performance is complex. While there are no significant differences in profit margins, female-owned enterprises tend to grow more rapidly in terms of employment than male-owned ones. Gender-based regression results also show that while there are common factors that affect the performance of both kinds of enterprise, there are also differential effects in which education is a critical factor for the success of female-owned enterprises.

1. INTRODUCTION

The role of women in economic development cannot be understated. Jiggins (1989) notes that about 30 per cent of rural households in the world are headed by women, and that women contribute about 80 per cent of agricultural labour, produce almost 60 per cent of the food that is consumed by rural households and generate more than one-third of all household incomes, mainly through small-scale agro-industry, trading, craft work and casual labour. In Malawi, about 59 per cent of female proprietors said 50 per cent or more of their household income came from their enterprises (Daniels & Ngwira, 1993). Concern about women's participation in economic development is relatively new. Lele (1986) notes that the frequently debated questions are whether women have adequate opportunities to participate in the productive processes or whether they are just beasts of burden, the primary victims of exploitation. However, Heilbrunn (2004) asserts that, over time, the numbers of female-owned businesses all over the world have been rising, and that in the past decade women have begun to be recognised as successful entrepreneurs.

The micro and small enterprises (MSEs) sector plays an important role in Malawi. The official definition of an MSE is based on employment and annual turnover (Government of Malawi, 1999). Micro enterprises employ up to four persons and have a turnover of up to MWK0.12 million, while small enterprises employ between six and 20 persons and have a turnover of between MWK0.12 million and MWK4 million. (The exchange rate in August 2007 was MWK139.5 = US$1.) In 1992 it was estimated that there were 587 283 MSEs generating about 1.07 million self-employment and paid employment opportunities. In 2000, there were 557 848 MSEs, now generating 1.01 million of these opportunities (Ebony Consulting International & National Statistical Office [ECI & NSO], 2000). The decrease in the number of MSEs is attributed to the high closure rates among small enterprises, with HIV/AIDS exacerbating this.

However, the role of women in MSEs has significantly improved in Malawi. At independence, the economy was dominated by Europeans specialising in manufacturing and Asians specialising in commerce. In the post-independence era there have been many policy changes, and these have contributed to the development of local and female entrepreneurship. In a study of small and agro-business enterprises in 1987, it was found that only 7 per cent of the sampled enterprises were owned and operated by women (Malawi/United States Agency For International Development, 1987). However, the increasing importance of women in business enterprises was revealed in later studies. In the 1992 survey of micro, small and medium-scale enterprises, about 46 per cent of proprietors in Malawi were women (Daniels & Ngwira, 1993). A recent Growth and Equity through Microenterprise Investments and Institutions (GEMINI) study further reveals that 34 per cent of MSEs were owned by women, 35 per cent by men and 30 per cent by married couples (ECI & NSO, 2000). Because of its potential for reducing poverty, the MSE sector is receiving increased focus in development policies. In the Malawi Poverty Reduction Strategy Paper, it is singled out as one of the sectors that could achieve pro-poor growth in which women will play a significant role (Government of Malawi, 2002).

The paper is organised as follows. The next section outlines the problem statement and research questions. Section 3 reviews literature on the relation between gender and business performance, Section 4 outlines data sources and specifies a model for the relationship between enterprise performance and gender while controlling for other variables, and Section 5 reports and discusses the empirical results. Section 6 provides concluding remarks, and the final section presents policy implications and recommendations.

2. PROBLEM STATEMENT AND RESEARCH QUESTIONS

The present study assesses the relative performance of MSEs operated by women and men in non-agricultural enterprises in Malawi. It attempts to answer two questions. First, what are the factors that explain the performance of MSEs and what role does the gender of the owner play? Secondly, do female-owned enterprises face different constraints from those faced by male-owned enterprises and are there differences between the factors that affect female-owned and male-owned enterprises? The study therefore tests the following propositions:

• P1:Female-owned MSEs perform worse than male-owned ones.

• P2:Female-owned and male-owned MSEs face different constraints affecting business performance.

3. LITERATURE REVIEW

The marginalisation of women in economic activities is well documented. Moser (1989) argues that women perform three roles in society: reproductive, productive and community management – but for a long time only the reproductive role has been emphasised, because women have been seen primarily as home-makers. They have been marginalised by unequal economic opportunities and inequalities of access to productive resources such as capital and control over their own labour, and by differences in human capital leading to differences in managerial and technical skills (Loscocco et al., 1991). Abor and Biekpe (2006∶105) provide evidence from Ghana that suggests female-owned small and medium-sized enterprises are less likely than their male counterparts to employ debt financing due to differential access to loan facilities, and McCormick et al. (1997∶1102) find that women entrepreneurs are over-represented in mini-manufacturing enterprises, under-represented in contract workshops and totally absent in mass production in the garment industry, but that there is a more balanced gender mix in custom tailoring. Others argue that women are likely to operate in low-risk and low-technology industries such as petty trading, and that the gender division of labour and gender stereotypes tend to push women into low-status and low-income business activities (for example, Von Massow, 1999). As a result women tend to be employed in small, home-based low-capitalised enterprises that supply goods and commodities for low-income consumer communities, particularly in urban areas.

Loscocco et al. (1991) argue that the differences between men's and women's socialisation, training and other experience may lead to different outcomes in business performance – female-owned enterprises generally perform worse than male-owned ones. In a study of small businesses in the US, Merrett and Gruidl (2000) find that rural female entrepreneurs face more obstacles to business success than their male or female urban counterparts; and in a summary of the literature on gender and performance, Zinger et al. (2005) note that female-owned MSEs' levels of performance are more modest than those of male-owned ones. Other studies also tend to support the view that female-owned enterprises perform worse than male-owned ones in terms of sales revenue, assets, profit margins and likelihood of survival (Loscocco et al., 1991; Rosa et al., 1996; McPherson, 1996; Daniels & Mead, 1998; Mead & Liedholm, 1998; Daniels, 1999; Kimuyu, 2002). The relatively poor performance of female-operated enterprises is attributed to many factors: inaccessibility to credit from the formal financial system, lack of capital, poor technical and managerial know-how, poor access to markets and raw material procurement problems, unfavourable legal systems, competition from state enterprises, diversion of business capital to men, poor government policies and inadequate institutional framework (Berger, 1989; Jiggins, 1989; Daniels & Ngwira, 1993).

4. METHODOLOGY AND DATA

4.1 Model specification and definition of variables

The empirical models that have been used in the literature suggest four main categories of factors that affect the performance of business enterprises, with gender being one of the factors. These categories are human capital, personal characteristics, family characteristics and business characteristics (Loscocco et al., 1991; McPherson, 1996; Daniels & Mead, 1998). Following these studies, we specify econometric models to analyse the effect of gender on business performance and analyse whether the factors that influence the performance of female-owned and male-owned enterprises are different. An earlier study in Malawi (ECI & NSO, 2000) analyses the data using mainly statistical methods, frequencies, cross-tabulation and bivariate comparisons, without testing for the statistical significance of the resulting differences. One problem with such a statistical approach is that it does not take into account the complexity of the relationships between the variables. We use two performance indicators, profit margin and growth in employment, and estimate three models for each: a general model that includes the gender variable, a model for female-owned enterprises only and a model for male-owned enterprises only. The general model tests the hypothesis that female-owned enterprises perform worse than male-owned enterprises. Regressions based on the subsample of female-owned and male-owned enterprises tests whether there are gender differences in the factors that affect the way enterprises perform. The regression model, based on the ordinary least squares (OLS) method, is specified as follows:

where P _i is the performance indicator (profitability and employment growth) for enterprise I; SEX is the gender of the owner of the enterprise; ECH is a vector of entrepreneur characteristics, including age, education, entrepreneurial skills training and business experience; BCH is a vector of business characteristics, including diversification, location of the business, main markets for the products and sector; and CVA is a vector of control variables such as access to credit, membership of business associations, stratum of the enterprise (urban, small town or rural) and hours the business is open during the year; μ _i the error term.

Two dependent variables are used in the analysis: profit margin and employment growth. The first dependent variable, the profit margin, is the annualised ratio of net profits to total sales. Since data on expenses and sales were collected with reference to ‘last week’, the method used to calculate annualised expenses and sales assumed that the ratio of expenses last week to sales last week was constant. For each month, information on high, average and low sales months with corresponding average monthly sales values were obtained to compute annual sales, and the constant expenses:sales ratio was used to generate annual expenses. The second dependent variable, employment growth, is calculated using the McPherson (1996) method, in which growth in employment is defined as the ratio of the difference in the logarithms of the current employment and the initial employment to the age of business. We define ‘employees’ in this study as working owners, paid workers, unpaid workers and trainees, and full-time and part-time workers.

The first category of independent variables is the gender of the proprietors. The gender of the owner of the enterprise is the central variable in our models. Three dummy variables are created, representing female-owned, male-owned and mixed-owned enterprises. Theories of gender based on marginalisation, unequal access to physical and human capital and socialisation postulate that female-owned enterprises are likely to perform worse than male-owned ones. Studies by Kimuyu (2002) in Kenya, by McPherson (1996) in five southern African countries, and by Daniels and Mead (1998) in Kenya find that female-owned enterprises generate less revenue, grow at a slower rate and earn lower profits than male-owned ones, respectively. Assuming the mixed-gender-owned enterprises as the base category, we expect female-owned enterprises to perform much worse than male-owned ones, using the three measures of the performance.

The second category of explanatory variables captures entrepreneur characteristics. Loscocco et al. (1991) argue that family situations, such as marital status, may have both a positive and a negative impact on the business's success, and that personal characteristics embody entrepreneurial traits for risk-taking behaviour. Five entrepreneur characteristics are included in the model: age, marital status, education, business skills training and business experience. The age of the entrepreneur is measured in years. Marital status is represented by a dummy equal to one if the entrepreneur is married, and zero otherwise. The entrepreneurs' education, business skills training and business experience capture the human capital elements embodied in entrepreneurship. Education is represented by six dummies: no education (the base category), some primary education, completed primary education, completed junior secondary education, completed senior secondary and completed higher education. Business skills training is represented by three dummies: no business training (base category), informal business training (learning from relatives and friends) and formal business skills training. The entrepreneurs' business experience is captured by number of years of business, and business experience squared is included to capture the non-linear relationship between performance and experience. The human capital theory postulates that the more educated and experienced the individual, the higher the degree of success in economic activities, so we expect to find a positive relationship between human capital variables and business performance.

The literature also suggests the importance of business characteristics in determining the performance of MSEs (McPherson, 1996; Zinger et al., 2005). We include the number of businesses operated by the entrepreneur, as a measure of diversification, the location of the MSE, the major market for the MSEs' products and the industry sector in which the MSE operates. The number of businesses operated may have both positive and negative effects on performance. On one hand, diversification may be used to spread the risk and this may lead to improved performance; but on the other hand, the allocation of labour by the entrepreneur may lead to crisis management that may negatively affect the performance. Five location dummies are included in the model: home or near home as the base category, traditional market place, mobile market, roadside or footpath, and commercial or industrial area. McPherson (1996) finds evidence that MSEs located in commercial areas grow more rapidly than home-based ones, a result attributed to access to high-income customers. Mukhtar (1998) finds evidence that female-owned enterprises are concentrated in services, wholesaling and retailing and manufacturing sectors. MSEs sell their products either to direct consumers or to other small, medium and large enterprises and institutional organisations. Institutional buyers may provide a reliable market and more contractual binding orders that may be used as a platform for growth. We construct two dummies representing the institutional market (as base category) and the consumer market.

Finally, several control variables are included in all three models. These include access to credit, membership of a business association, the stratum in which the MSE operates and the number of hours the business operates in a year. Access to credit is captured by a dummy variable on whether credit was obtained to finance the business. Marlow and Patton (2005∶717) argue that access to finance is critical for start-up and consequently affects the performance of any enterprise. Membership of a business association represents social capital and the importance of networks in business activities, and we expect enterprises in a network to perform better than those that work independently. Okten and Osili (2004) find that family and community networks play an important role in facilitating access to credit in Indonesia. The stratum in which the enterprise operates is represented by three dummy variables: rural MSEs (base category), urban MSEs and small-town MSEs. We expect urban MSEs to perform better than small-town ones, and these in turn to perform better than rural ones, since there are large markets and more business in the urban areas. We also include the number of hours the business is open during the year, in order to control for variations in working hours.

4.2 Data

This study uses secondary data from a national survey of MSEs in Malawi conducted in 2000 (ECI & NSO, 2000), using the census sampling frame. Data were collected using the GEMINI methodology. The major objective of the ECI & NSO (2000) study was to provide an overview of the MSE sector in Malawi, so as to capture business start-up and profitability, employment growth, reasons for business closure, the impact of HIV/AIDS and the contribution of the sector to the national income. (The analysis in the EIC & NSO (2000) study is more descriptive, providing a profile of MSEs, their performance, problems and constraints and how the sector had changed between 1992 and 2000.) The data were drawn from a stratified random sample of 22 000 households that was used to identify business enterprises. The country was stratified into seven strata, and in each stratum the enumeration areas were randomly selected for the administration of questionnaires. Each household and business unit in the selected enumeration area was visited and MSEs were identified for questionnaire interviews on existing and closed business enterprises. The survey covered enterprises with fewer than 50 employees, including working proprietors, but excluded businesses with multiple branches even where one of the branches qualified according to the employment criterion. Although the data also cover agricultural enterprises, this study uses only data for non-agricultural business enterprises. A sample of 3074 non-farm MSEs usable for this study was generated from the survey data.

5. EMPIRICAL RESULTS AND ANALYSIS

Table 1 presents the descriptive statistics of the variables used in the regression models. The average profit margin is 56.8 per cent, and female-owned enterprises generate more profits (57.7 per cent) than male-owned ones (56.6 per cent) while mixed-gender-owned enterprises generate the lowest profits (55.6 per cent), although these differences are marginal. Similarly, female-owned enterprises tend to grow faster in terms of employment (11.6 per cent per year) than male-owned ones (6.5 per cent per year) and mixed-gender-owned ones (6.9 per cent per year).

Table 1: Descriptive statistics of variables in the model^a

Variable	All MSEs	Gender of owners
		Female	Male	Mixed
		Dependent variables
Profit margin (%)	56.813	57.738	56.572	55.593
	[25.96]	[23.88]	[27.49]	[25.67]
Employment growth	0.084	0.116	0.065	0.069
	[0.44]	[0.63]	[0.30]	[0.20]
Independent variables
Gender
Female-owned (1, 0)	0.359	–	–	–
Male-owned (1, 0)	0.475	–	–	–
Mixed-owned (1, 0)	0.166	–	–	–
Entrepreneur characteristics
Age in years	33.046	34.027	31.803	34.483
	[11.62]	[11.25]	[11.77]	[11.58]
Married (1, 0)	0.764	0.652	0.777	0.972
Some primary education (1, 0)	0.460	0.438	0.489	0.422
Completed primary education (1, 0)	0.236	0.218	0.253	0.230
Completed junior secondary education (1, 0)	0.112	0.109	0.108	0.130
Completed senior secondary education (1, 0)	0.069	0.058	0.071	0.084
Higher education (1, 0)	0.022	0.024	0.018	0.031
Informal business skills training (1, 0)	0.135	0.079	0.185	0.114
Formal business skills training (1, 0)	0.142	0.084	0.164	0.200
Business experience (years)	6.342	5.236	7.228	6.207
	[7.66]	[6.69]	[8.38]	[7.15]
Business experience squared	98.833	72.061	122.385	89.520
	[269.7]	[212.3]	[316.2]	[227.3]
Business characteristics
Ownership of multiple MSEs (1, 0)	0.138	0.171	0.084	0.222
Traditional market place (1, 0)	0.140	0.081	0.162	0.202
Mobile market (1, 0)	0.073	0.050	0.098	0.049
Roadside/path market (1, 0)	0.268	0.188	0.352	0.200
Commercial/industrial (1, 0)	0.119	0.063	0.155	0.138
Consumer market (1, 0)	0.984	0.992	0.982	0.974
Manufacturing: food processing (1, 0)	0.079	0.136	0.039	0.073
Manufacturing: beer brewing/spirit distilling (1, 0)	0.066	0.140	0.002	0.086
Manufacturing: textiles and leather products (1, 0)	0.071	0.043	0.102	0.041
Manufacturing: wood and pottery products (1, 0)	0.046	0.021	0.072	0.024
Manufacturing: other manufactures (1, 0)	0.024	0.010	0.038	0.016
Retail: fish and forest-based products (1, 0)	0.071	0.069	0.066	0.088
Retail: food, beverages and tobacco (1, 0)	0.092	0.134	0.075	0.049
Retail: garments and footwear (1, 0)	0.079	0.024	0.130	0.055
Retail: general merchandise/grocery (1, 0)	0.231	0.149	0.264	0.314
Bars, restaurants and hotels (1, 0)	0.039	0.034	0.023	0.094
Services (1, 0)	0.076	0.034	0.116	0.049
Control variables
Credit access (1, 0)	0.105	0.149	0.077	0.094
Business association (1, 0)	0.029	0.028	0.027	0.037
Urban (1, 0)	0.568	0.572	0.616	0.424
Small town (1, 0)	0.225	0.193	0.211	0.332
Hours business is open	246 74	237.55	249.20	259.70
	[111.1]	[118.2]	[91.74]	[140.9]
Number of enterprises	3074	1104	1460	509
Note: ^aData in parentheses represent dummy variables equal to one for the reference category, otherwise equal to zero, and the means are proportions for the reference category. Figures in brackets are standard deviations for continuous variables.

There are gender differences in entrepreneur characteristics, business characteristics and other control variables. For instance, most female entrepreneurs lack business skills training (informal and formal) and have less business experience than male entrepreneurs, while differences in education are marginal. The descriptive statistics also show that a higher proportion of female entrepreneurs (17.1 per cent) own more than one enterprise than male entrepreneurs (8.4 per cent). In terms of location of business, a higher proportion of female-owned MSEs are home-based than male-owned ones, which tend to be located in high-demand environments. Female entrepreneurs also tend to operate more in food processing, beer brewing, retail of food and beverages, and bars and restaurants than their male counterparts. For instance, 13.6 per cent of female entrepreneurs are in the food processing industry compared with only 3.9 per cent of male ones; while 13 per cent of male entrepreneurs operate in the garments and footwear retail trade compared with only 2.4 per cent of female ones.

Gender biases are also revealed in access to credit, with 14.9 per cent of female entrepreneurs having accessed credit to support their business operations, compared with only 7.7 per cent of male entrepreneurs. The reason for this positive bias in access to credit is that most micro-finance institutions that operate in urban, peri-urban and rural areas tend to lend to women. (This is in contrast to the findings of Abor and Biekpe (2006), who find that financial services are biased against women in Ghana.) There are, however, no significant differences in the proportion of male and female entrepreneurs belonging to business associations. Overall, only 3 per cent of entrepreneurs belong to these.

5.1 Impact of gender of entrepreneur on performance

Table 2 presents OLS estimates of factors that influence profit margins and growth in employment for all MSEs in the sample. The gender variables in the regression results test the hypothesis that female-owned MSEs perform worse than male-owned MSEs. With respect to profitability, the gender of ownership is not statistically significant, although male-owned enterprises tend to perform better than female-owned ones. Both female-owned and male-owned MSEs perform better than MSEs jointly owned by females and males. The insignificance of gender in the profitability equation contrasts with the findings of other studies in Africa, in which female-owned MSEs perform significantly worse than male-owned enterprises (Kimuyu, 2002).

Table 2 OLS estimates on determinants of profit margins and employment growth: all MSEs

Variable	Profit margin		Employment growth
	Coefficient	t -ratio	Coefficient	t -ratio
	Gender
Female-owned	1.1678	0.86	0.0471	2.04**
Male-owned	2.2285	1.64	0.0157	1.15
Entrepreneur characteristics
Age in years	0.0515	1.06	0.0007	0.51
Married	0.3996	0.38	0.0143	0.76
Some primary education	1.3610	0.88	–0.0021	–0.08
Completed primary education	1.5610	0.93	0.0075	0.28
Completed junior secondary education	4.0617	2.14**	0.0033	0.10
Completed senior secondary education	1.7602	0.75	0.0670	1.34
Higher education	7.9491	2.19**	0.0296	0.62
Informal business skills training	2.4987	1.64	–0.0020	–0.12
Formal business skills training	1.9205	1.31	–0.0297	–1.62
Business experience	–0.1797	–1.22	–0.0149	–4.69*
Business experience squared	–0.0005	–0.12	0.0003	4.63*
Business characteristics
Ownership of multiple MSEs	–2.3826	–1.79^†	0.0149	0.52
Traditional market place	3.1906	2.02**	–0.0053	–0.22
Mobile market	0.5808	0.30	–0.0694	–3.31*
Roadside/path market	4.4351	3.61*	–0.0068	–0.23
Commercial/industrial	–0.1994	–0.12	–0.0115	–0.42
Consumer market	–2.4828	–0.66	0.0144	0.50
Manufacturing: food processing	–0.3013	–0.18	0.0494	1.99^†
Manufacturing: beer brewing/spirit distilling	–0.4350	–0.22	–0.0138	–0.65
Manufacturing: textiles and leather products	–18.188	–8.08*	0.0183	0.79
Manufacturing: wood and pottery products	–23.777	–8.77*	0.0658	1.66^†
Manufacturing: other manufactures	–16.767	–4.62*	0.0081	0.32
Retail: fish and forest-based products	–1.1681	–0.57	0.1045	1.38
Retail: food, beverages and tobacco	4.8415	3.10*	0.0419	0.94
Retail: garments and footwear	6.6680	3.58*	0.0432	1.47
Retail: general merchandise/grocery	1.8951	1.29	0.0262	1.31
Bars, restaurants and hotels	–17.099	–5.07*	0.0548	1.13
Services	–24.271	–10.17*	0.0574	1.79^†
Control variables
Credit access	1.1753	0.82	0.1118	2.31**
Business association	4.2701	1.46	0.0193	0.52
Urban	7.5394	5.51*	0.0056	0.15
Small town	5.1327	3.37*	0.0429	1.49
Hours business is open	–0.0019	–0.36	0.0001	0.70
Constant	52.070	10.76*	0.0051	0.06
R ^2	0.187		0.033
F-statistic	18.07		3.20
Probability > F	0.00		0.00
N	3074		3074
Notes: The t-statistics are based on heteroskedastic-consistent standard errors. *Statistically significant at the 1% level. **Statistically significant at the 5% level. ^†Statistically significant at the 10% level.

The only entrepreneur characteristic statistically significant at the 1 per cent level that affects profitability is education. Although the coefficients of the education dummies are positive, as expected, profitability is statistically significantly higher for entrepreneurs with higher education (7.9 per cent above those without education), followed by those who completed junior secondary education (4.1 per cent above those with no education). Among the business characteristics, operation of multiple enterprises, location of the business and the industry in which the MSEs operate have a significant influence on profitability. The coefficient of the ownership of multiple enterprises is negative and statistically significant at the 10 per cent level. The negative relationship reflects the inefficiency in time allocation among the various enterprises that may lead to managerial inefficiencies.

MSEs that operate in a traditional market place and by the roadside or footpath tend to generate more profits than home-based ones, with the coefficients being statistically significant at the 1 per cent and 5 per cent levels, respectively. These locations are important because they are places of high demand and the transaction costs (such as transport) are minimal compared with mobile markets or commercial areas. Profitability also tends to be statistically significantly (at the 1 per cent level) higher for MSEs operating in industrial sectors such as retail of food and beverages and retail of garments and footwear, while profitability is lower in the manufacture of textiles and leather products, wood and pottery products and other manufactured products, bars and restaurants and services. There is evidence that MSEs located in high-income strata (urban and small towns) generate more profits than those in rural areas, and the coefficients are statistically significant at the 1 per cent level. The profit margin is higher by 7.5 per cent in urban areas and by 5.1 per cent in small towns than the profitability generated by MSEs in rural enterprises.

With respect to growth in employment, we find evidence that, after controlling for other factors that influence growth, female-owned MSEs tend to grow at a faster rate than male-owned ones. This is in contrast to the hypothesis that female-owned MSEs perform worse than male-owned ones. The coefficient of female-owned enterprises is statistically significant at the 5 per cent level and shows that female-owned MSEs grow 4.7 per cent faster than MSEs with mixed gender. The coefficient of male-owned MSEs is positive and its value below that of female-owned MSEs, but it is statistically insignificant. These results contrast with the negative employment growth associated with female ownership in some southern African countries (McPherson, 1996). Hence, in the case of Malawi, the contention that female entrepreneurs in Africa tend to be low-risk investors does not receive support in this study. The only entrepreneur characteristic that is statistically significant at the 1 per cent level is business experience. We find a curvilinear (U-shaped) relationship between employment growth and business experience. Thus, MSEs belonging to new entrepreneurs tend to contract or have retarded growth, and growth reaches a minimum as a U-shaped relationship. Thus businesses with more than 25 years experience tend to grow faster, and thereafter a positive relationship is observed as entrepreneurs gain more experience.

In terms of business characteristics, only mobile market location, manufacturing in food processing, manufacture of wood and pottery products and service industries are statistically significant variables at least at the 10 per cent level. MSEs that are mobile experience a reduction in growth in employment compared with home-based industries, possibly reflecting the transaction costs of mobility and the potentially high waste of supplies. MSEs in the manufacturing of food products and wood and pottery products and those that operate in the service sector tend to grow at a faster rate (at least 4.9 per cent faster) than the base category (retail of farm products). The important role that access to credit can play in business expansion is revealed by the positive and statistically significant relationship between credit and employment growth. The results show that MSEs with access to credit grew 11.2 per cent faster than those that did not have access. Since a higher proportion of female-owned MSEs had access to credit, it is not surprising that we find a significant relationship between female ownership of MSEs and growth in employment.

5.2 Impact of gender-specific constraints on performance

The literature suggests that female-owned and male-owned enterprises face different constraints in the operation of their businesses, which in turn result in the differential impact on business performance. We use gender-based regressions to test the hypothesis that the various factors affect female-owned and male-owned MSEs differently. Table 3 presents gender-based results on profit performance, using the OLS method. Factors that are statistically significant in both female-owned and male-owned MSEs are type of industry – including manufacture of textiles and leather products, manufacture of wood and wood products, bars and restaurants, and services – and whether the MSEs are in the urban areas or small towns. All of the industry sector dummy variables that are significant are negatively associated with profit margins, while urbanisation has a positive influence on profit margins. The marginal effects of industry variables are higher for male-owned than female-owned enterprises, suggesting that male-owned ones perform worse than female-owned ones, with the exception of manufacture of textiles and leather products. Differences in the location of the enterprises show that male-owned MSEs are more profitable than female-owned ones that operate in urban and small town centres. Completing junior secondary education and higher education is positively associated with profitability and only statistically significant in the female-owned MSE model, while education does not seem to play a major role in determining the profitability of male-owned MSEs.

Table 3 OLS estimates on determinants of profit margins: female-owned and male-owned MSEs

Variable	Female-owned		Male-owned
	Coefficient	t -ratio	Coefficient	t -ratio
	Entrepreneur characteristics
Age in years	0.186	2.6*	–0.005	–0.1
Married	0.488	0.3	1.073	0.7
Some primary education	1.924	0.9	–0.036	0.0
Completed primary education	3.966	1.6	–1.054	–0.3
Completed junior secondary education	5.298	1.8^†	2.188	0.6
Completed senior secondary education	2.888	0.7	0.892	0.2
Higher education	10.96	1.9^†	6.813	1.2
Informal business skills training	1.952	0.7	1.557	0.7
Formal business skills training	0.964	0.3	2.088	1.0
Business experience	–0.022	–0.1	0.051	0.2
Business experience squared	–0.011	–1.7^†	–0.002	–0.4
Business characteristics
Ownership of multiple MSEs	–3.815	–2.0**	–2.261	–0.9
Traditional market place	1.758	0.6	5.682	2.2**
Mobile market	–0.296	–0.1	1.418	0.5
Roadside/path market	1.130	0.6	6.346	3.1*
Commercial/industrial	–1.106	–0.3	0.739	0.3
Consumer market	–3.979	–0.6	–0.091	0.0
Manufacturing: food processing	–0.581	–0.3	–1.962	–0.6
Manufacturing: beer brewing/spirit distilling	–0.243	–0.1	–0.093	0.0
Manufacturing: textiles and leather products	–14.04	–3.4*	–23.37	–7.2*
Manufacturing: wood and pottery products	–33.51	–7.1*	–23.20	–6.5*
Manufacturing: other manufactures	–9.000	–1.0	–20.16	–4.4*
Retail: fish and forest-based products	0.029	0.0	–3.437	–1.0
Retail: food, beverages and tobacco	7.275	3.4*	–0.056	0.0
Retail: garments and footwear	18.20	6.3*	1.302	0.5
Retail: general merchandise/grocery	4.733	2.0**	–3.442	–1.5
Bars, restaurants and hotels	–9.532	–1.8^†	–21.80	–3.3*
Services	–21.06	–3.8*	–28.54	–8.9*
Control variables
Credit access	3.918	2.1**	–0.195	–0.1
Business association	5.285	1.1	–1.400	–0.3
Urban	6.784	3.4*	11.27	4.7*
Small town	4.729	2.1**	5.462	2.1**
Hours business is open	–0.004	–0.5	0.005	0.6
Constant	49.27	6.7*	52.22	7.2*
R ^2	0.171		0.239
F-statistic	8.98		12.85
Probability > F	0.00		0.00
N	1104		1460
Notes: The t-statistics are based on heteroskedastic-consistent standard errors. *Statistically significant at the 1% level. **Statistically significant at the 5% level. ^†Statistically significant at the 10% level.

Ownership of multiple enterprises is statistically significant at the 5 per cent level in the female-owned MSE model and has a negative influence on profitability. Business location factors are only statistically significant among male-owned MSEs, in which location at a traditional market place or roadside or footpath tends to boost enterprise profitability. Female-owned enterprises operating in the retail of food and beverages, retail of garments and footwear, and retail of general merchandise generate more profits than the base category of retail of farm produce. Profit margins among female-owned MSEs are highest in retail of garments and footwear (18.2 per cent higher than the base category), followed by retail of food and beverages (7.3 per cent), while male-owned enterprises in other manufacturing tend to generate less profit than the base category.

Table 4 presents OLS results with respect to employment growth. The performance of the regression models is rather poor, with low explanatory power and few significant variables. The results, however, show that different factors affect the performance of female-owned and male-owned enterprises. The common factors that significantly affect both kinds of enterprise include business experience and operating a mobile MSE. The relationship between employment growth and business experience in both cases is U-shaped, with experience initially leading to slow growth rates of up to 13 years for female-owned and 20 years for male-owned enterprises, and thereafter leading to higher growth rates. However, the marginal effects are lower in male-owned enterprises than in female-owned ones. If the business location is a mobile market, both female-owned and male-owned enterprises tend to have a retarded growth rate in employment. Operating in another manufacturing sector is negatively associated with growth in female-owned MSEs but operating in the service sector is positively associated with growth in employment. Operating in the manufacture of food products is associated with high growth rates in employment among male-owned MSEs. Access to credit is an important factor in explaining growth in female-owned MSEs, but it is not important in the growth of male-owned MSEs. This suggests that access to credit by female entrepreneurs has the greatest potential for generating employment among MSEs.

Table 4 OLS estimates on determinants of employment growth: female-owned and male-owned MSEs

Variable	Female-owned		Male-owned
	Coefficient	t -ratio	Coefficient	t -ratio
	Entrepreneur characteristics
Age in years	0.004	1.3	–0.002	–1.5
Married	0.038	1.1	–0.005	–0.2
Some primary education	0.044	1.1	–0.043	–0.8
Completed primary education	0.037	0.9	–0.029	–0.5
Completed junior secondary education	0.057	1.3	–0.036	–0.6
Completed senior secondary education	0.230	1.7^†	–0.028	–0.5
Higher education	0.119	1.3	–0.094	–1.5
Informal business skills training	–0.029	–0.7	0.005	0.2
Formal business skills training	–0.101	–2.2**	0.002	0.1
Business experience	–0.026	–3.0*	–0.008	–3.3*
Business experience squared	0.001	2.8*	0.0002	2.8*
Business characteristics
Ownership of multiple MSEs	0.008	0.1	0.016	0.3
Traditional market place	–0.019	–0.4	0.006	0.2
Mobile market	–0.081	–2.0**	–0.077	–2.9*
Roadside/path market	–0.001	0.0	–0.023	–1.0
Commercial/industrial	–0.027	–0.6	–0.015	–0.6
Consumer market	0.046	0.7	–0.006	–0.2
Manufacturing: food processing	0.028	0.7	0.085	2.1**
Manufacturing: beer brewing/spirit distilling	–0.005	–0.2	0.351	1.2
Manufacturing: textiles and leather products	–0.009	–0.2	0.013	0.4
Manufacturing: wood and pottery products	0.104	1.1	0.054	1.2
Manufacturing: other manufactures	–0.111	–2.0**	–0.004	–0.1
Retail: fish and forest-based products	0.217	1.2	0.009	0.2
Retail: food. beverages and tobacco	0.031	0.4	0.053	1.0
Retail: garments and footwear	0.127	0.8	0.016	0.6
Retail: general merchandise/grocery	0.048	1.0	0.001	0.0
Bars. restaurants and hotels	0.066	0.8	0.097	1.0
Services	0.029	0.5	0.055	1.3
Control variables
Credit access	0.163	1.7^†	0.080	1.6
Business association	–0.037	–0.8	0.035	0.7
Urban	–0.002	0.0	–0.009	–0.5
Small town	0.018	0.3	0.036	1.1
Hours business is open	0.0001	0.4	0.0001	1.1
Constant	–0.118	–0.8	0.166	1.9^†
R ^2	0.047		0.047
F-statistic	1.53		1.92
Probability > F	0.03		0.00
N	1104		1460
Notes: The t-statistics are based on heteroskedastic-consistent standard errors. *Statistically significant at the 1% level. **Statistically significant at the 5% level. ^†Statistically significant at the 10% level.

6. CONCLUSIONS

The development of MSEs is seen as one instrument for addressing the problems of poverty in developing countries. Women are also increasingly participating in the ownership of MSEs in developing countries. In Malawi, 34 per cent of MSEs are owned by women, compared with 35 per cent owned by men and 31 per cent owned by mixed gender, implying that women are involved in 65 per cent of MSEs (ECI & NSO, 2000). Using profitability and employment growth as indicators of enterprise performance, the evidence for the first proposition – that female-owned MSEs perform worse than male-owned ones – is mixed and sensitive to the measure of performance of MSEs in Malawi. In terms of profitability performance, we find no evidence of significant differences in the performance of female-owned and male-owned MSEs; hence, we reject the proposition that they perform differently. However, the proposition that female-owned and male-owned MSEs perform differently is accepted in terms of employment growth. We find statistically significant gender-based differences in employment growth, in which female-owned enterprises grow at a faster rate than male-owned ones. This is partly due to the relative access to credit facilities from microfinance institutions that mostly target women entrepreneurs and partly due to the high marginal impact of education.

With respect to the second proposition – whether there are gender-based differences in factors influencing performance – the study provides support for the view that female-owned and male-owned MSEs face different constraints on business performance. One interesting result is that the impact of education on performance is more pronounced in female-owned businesses than in male-owned ones, although male entrepreneurs are slightly better endowed in human capital than female ones. Completion of junior secondary education and higher education are positively related to profitability among female-owned enterprises, while none of the education variables are statistically significant among male-owned enterprises. Similarly, at least one education category (completion of senior secondary education) is statistically significant among female entrepreneurs in employment growth, compared with none among male entrepreneurs. We also find that access to credit is more productive in female-owned enterprises. Female-owned enterprises that have access to credit tend to perform better in terms of profitability and employment growth than male-owned ones. This suggests that gender biases against women's access to capital and finance may impede the growth of MSEs, which supports the proposition of the gender and development theorists. Credit in male-owned enterprises is not used productively, and, although this is not statistically significant, it does reduce the profitability of male-owned MSEs.

The others factors that lead to differential performance of female-owned and male-owned MSEs are the economic sectors within which the enterprises operate. Sectors that are gender neutral and in which profit margins are lower include textile and leather manufacturing, wood and pottery products manufacturing, bars, restaurants and hotels, and services. However, female-owned enterprises achieve higher rates of profits than male-owned ones in sectors such as retailing of food, beverages and tobacco, retailing of garments and footwear and retailing of general merchandise. We also find no significant differences in the extent to which business problems affect the gender-based performance of MSEs. All of the business problems retard the revealed growth in sales, although these problems are more pronounced in the case of female-owned enterprises. The most important problem that affects the probability of revealed decrease in sales for both female-owned and male-owned enterprises relates to marketing, followed by competition in female-owned enterprises and finance in male-owned enterprises.

7. RECOMMENDATIONS

The results in the present study point to several policy issues. First, with respect to differential performance between female-owned and male-owned MSEs, the results suggest the need to promote female entrepreneurship as a way of generating paid employment. Second, the differential factors that affect the performance of female and male MSEs suggest that interventions in the MSE sector may require gender-specific interventions, since different factors affect the performance of female-owned and male-owned enterprises in varying ways. For instance, the relative importance of education in female-owned enterprises suggests the need to increase human capital investments in women. Investments in female education have the effect of not only improving the profitability of their enterprises, but also generating paid employment opportunities. Similarly, increasing female entrepreneurs' access to credit facilities is more productive than increasing this access for male entrepreneurs. There is a need therefore to promote microfinance institutions that target financing of non-farm economic activities, with a deliberate bias towards providing credit to women entrepreneurs.

Acknowledgments

This study was funded by the Organisation of Social Science Research in East and Southern Africa (OSSREA), Addis Ababa, Ethiopia, under the 15th OSSREA Gender Issues Research Competition Grant, for which the author acknowledges the financial support. The author also acknowledges the comments from two anonymous referees. However, the author takes sole responsibility for any remaining errors.