World Bank From the SelectedWorks of Mohammad Amin December, 2010 Efficiency, Firm-size and Gender: The Case of Informal Firms in Latin America Mohammad Amin Available at: https://works.bepress.com/mohammad_amin/28/
Efficiency, Firm size and Gender: The Case of Informal Firms in Latin America Mohammad Amin* December, 2010 A number of studies show that relative to male owned businesses, female owned businesses are less efficient and smaller in size the female under performance hypothesis. The present paper tests this hypothesis for a sample of informal or unregistered firms in Argentina and Peru. The results do not reject the stated hypothesis. The gender based differences in firm efficiency and firm size do not appear to be driven factors such as industry and country, quality of human capital of the firm, use of physical capital, age of the firm and use of family labor. Keywords: Informality, Gender, Firm size, Latin America JEL: J16, L26, L25, L21, L20 *Enterprise Analysis Unit, World Bank, Washington DC, 20433. Email: mamin@worldbank.org.
1. Introduction Existing studies show that relative to male owned businesses, female owned businesses tend to underperform the female under performance hypothesis. Various measures of performance are used in these studies including profits, firm size (sales, employment) and factor productivity. The relatively poor performance among female owned businesses could be due to the disadvantages that women face in, for example, getting credit, having less education, fulfilling their role of primary care givers in the family and discrimination that they face in the more lucrative but male dominated industries. The literature mentioned above is mainly focused on the developed countries and more importantly, on the formal or the registered sector. For example, Robb and Wolken (2002) analyze small formal businesses in the U.S. and find that women owned businesses generate only 78 percent of the profits generated by male owned businesses. Similar results are reported by Sabarwal and Terrell (2008) for firms in the formal sector in 26 transition countries. Focusing on firm size, Brush et al. (2006) among others find that in the U.S., average revenue of female owned formal firms equaled USD 151,130, about 26 percent of the level for male owned businesses. However, a substantial proportion of economic activity occurs in the informal economy, especially in the developing countries. Recent estimates suggest that for the world as a whole, between 22.5 and 34.5 percent of all economic activity occurs in the informal economy; for countries in the lowest quartile of GDP per capita, the estimates range between 29 and 57 percent (La Porta and Shleifer, 2008). The informal economy is particularly important for gender related issues since compared with men, women prefer the informal over the formal economy The present paper uses data on informal firms in two developing countries, Argentina and Peru, to analyze how labor productivity (a measure of firm efficiency) and firm size vary by the gender of the owner of the business. The results strongly confirm the female under performance hypothesis. That is, compared with male owned businesses, female owned business show significantly lower levels of labor 1
productivity and firm size. The results also reveal that the female under performance cannot be easily explained away by differences in physical and human capital of the firm and other firm characteristics. 2. Data and main variables The data we use is a random sample of close to 750 informal or unregistered firms in Argentina and Peru. The survey was conducted by the World Bank s Enterprise Analysis Unit in 2010 and restricted to two regions in Argentina (Buenos Aires and El Chaco) and two regions in Peru (Lima and Arequipa). The sample includes a roughly equal mix of manufacturing and service firms, young and old firms and operating from inside vs. outside of household premises. It is important to note that due to lack of a proper sampling frame of the informal economy, the survey is not necessarily representative of the informal economy in the two counties or the cities covered. 1 Hence, the results discussed below relate to the sample of firms surveyed rather than the informal economy per se. An extension of the results to the broader informal economy requires due caution. To test the female under performance hypothesis, three separate measures of performance are used. The first measure relates to firm efficiency as measured by the average productivity of labor. It equals the log of total monthly sales (in USD) of a firm in a regular month divided by the total number of workers employed at the firm (including the owner, if applicable) in a regular month (Labor productivity). The mean value of labor productivity (log values) equals 5.6 and the standard deviation is 0.96. The average value is significantly higher in Peru than in Argentina (5.8 vs. 5.3). The second and third measures relate to firm size as measured by the log of total sales in a regular month expressed in USD (Sales) and the log of total number of workers employed at the firm in a regular month (Employment). Monthly sales (log values) average 5.9 in the full sample, varying between 6.0 in Peru and 5.6 in Argentina. The standard deviation of Sales equals 1.0. The mean value of Employment equals 0.27 1 Data and methodology are available at www.enterprisesurveys.org. 2
in the full sample, 0.26 in Argentina and 0.27 in Peru. The standard deviation of the variable in the full sample equals 0.44. The main explanatory variable is a dummy variable equal to 1 if the largest owner of the firm is female and 0 otherwise (Female). In the full sample, 52.8 percent of the firms have a female largest owner, varying between 49.3 percent in Argentina and 55 percent in Peru. For convenience, a firm with a largest female (male) owner will be referred to as a female owned (male owned) firm. There is little distinction in our sample between the owner and manager since for about 99 percent of the firms the owner is also the manager of the firm. Without much loss of generality, we focus on labor productivity. Results for sales and employment similar and discussed briefly in a separate section. It is possible that the observed relationship between efficiency and gender could be spuriously driven by other firm characteristics that are correlated with both, the gender of the owner as well as firm efficiency. We control for a number of firm characteristics to eliminate such possibilities of spurious correlation and to also rule out some of the possible explanations for the female under performance hypothesis. In the main specification we control for the (log of) number of employees to account for diminishing marginal productivity of labor. Cultural factors associated with gender or the self selection of individuals by gender into high and low productive industries could lead to the omitted variable bias problem with our main results. Country and industry fixed effects (20 industries within manufacturing and services) are used to guard against the problem. The next set of controls includes proxy measures for the human capital of the owner. Many studies have documented lower level of education and also managerial experience among women entrepreneurs compared with men. Human capital can also impact firm efficiency implying the omitted variable bias problem. The controls for human capital include age of the owner (logs), age of the firm (logs), number of years of experience the manager has in 3
running the business (logs) and fixed effects for the level of education of the owner (no education, primary education, secondary education, vocational training and university education). The last two controls in the main specification relate to physical capital and include a dummy variable equal to 1 if the firm uses machinery in the production process and 0 otherwise and another similar dummy for the use of vehicles. For the robustness of the results, a number of additional controls were added to the main specification. These controls are discussed in the next section. Estimation Regression results for the main specification are provided in Table 1, columns 1 5. Without any other controls, labor productivity for a female owned firm is lower by 0.46 log points compared with a maleowned firm (column 1, Table 1). The difference is significant at less than the 1 percent level. Converted to levels (that is, without logs), the estimate implies that sales per worker in a male owned firm is about 1.58 times that in a female owned firm. Controlling for country and industry fixed effects hardly changes the estimated coefficient of Female from above and the same holds for the remaining controls discussed above (columns 2 5, Table 1). For example, with all the controls in the main specification in place, the estimated coefficient value of Female equals 0.42, significant at less than the 1 percent level (column 5, Table 1) compared with 0.46 without any controls. The estimate of 0.42 with all the controls implies that in level terms, sales per worker in a male owned firm equal 1.52 times that in a female owned firm, not too different from the figure of 1.58 without any controls. For the remaining variables, the results confirm a large positive effect on productivity of machinery and vehicle use. The results also confirm declining labor productivity with the level of employment at the firm level and higher productivity among the relatively younger firm owners. The estimated coefficient value of Female hardly changes when we include the additional controls in the specification. These controls are motivated by existing studies and anecdotal evidence 4
suggesting that they could vary across gender and may also have a direct effect on firm efficiency. The controls include a dummy variable indicating if the firm perceived corruption to be a severe obstacle to business and a similar dummy variable for crime, a dummy variable indicating if the firm operates from inside or outside of household premises, a dummy variable indicating if the firm has a bank account for running the business, a dummy variable indicating if the firm uses family labor or not, the number of owners of the firm, a dummy variable indicating if the owner is single (as opposed to married, widowed), a dummy variable indicating if the firms produces under contract from another business or not and a dummy indicating if either of the owners parents owned a business (currently or in the past). With all these additional controls, the estimated coefficient value of Female remains negative, significant at less than the 1 percent level and roughly unchanged in magnitude equaling 0.41 (column 8, Table 1) compared with 0.42 above (column 5, Table 1). Firm size and gender Regression results with and without the various controls mentioned above for sales as the dependent variable are provided in Table 2. These results show that irrespective of the set of the controls, firm size is significantly lower for female owned compared with male owned firms. For example, without any other controls, monthly sales (without logs) of a male owned firm equal 1.78 times that for a femaleowned firm (based on the estimate in column 1, Table 2). The ratio declines somewhat to 1.55 when the controls discussed above are included in the specification (based on the estimate in column 7, Table 2), but the gender based gap in sales is still large and statistically significant at less than the 1 percent level. Last, regression results using employment at the measure of firm size are similar to the ones discussed above for monthly sales. Without any other controls, a male owned firm has 1.13 times the number of employees (without logs) that a female owned firm has. This ratio declines to 1.06 with all the controls discussed above in place. The decline is mainly due to the dummy variable indicating the use of family 5
labor by the firm. Note that with the employment variable as the dependent variable and for all the specification in Table 2, the estimated coefficient of Female is negative and significant at less than the 5 percent level. Conclusion The paper extends the female under performance hypothesis to a sample of informal firms in Argentina and Peru. Empirical results strongly confirm that the average productivity of labor is lower among female owned compared with male owned firms. Similarly, firm size as measured by monthly sales and number of workers is also significantly smaller for female owned firms. More work is needed to ascertain or reject the generality of these results to other developing countries. References [1] Brush, C., N. Carter, E.J. Gatewood, P. Greene and M. Hart, (2006), Growth Oriented Women Entrepreneurs and Their Businesses (New Horizons in Entrepreneurship). Cheltenham, UK and Northampton, MA: Edward Elgar. [2] Carter, Sara and Eleanor Shaw (2006), Women s Business Ownership: Recent Research and Policy Development, Report to the Small Business Service, U.K. [3] La Porta, Rafael and Andrei Shleifer (2008), The Unofficial Economy and Economic Development, Tuck School of Business Working Paper No. 2009 57. [4] Robb A., and J. Wolken (2002), Firm, Owner and Financing Characteristics: Differences between Male and Female owned Small Businesses, Working Paper, Federal Reserve Board of Governors. [5] Sabarwal, Shwetlana and Katherine Terrell (2008), Does Gender Matter for Firm Performance? Evidence from Eastern Europe and Central Asia, IZA Discussion Paper Series No. 3758. 6
Table 1: Labor productivity and gender (1) (2) (3) (4) (5) (6) (7) (8) Dependent variable: Labor productivity Female 0.46*** 0.45*** 0.47*** 0.45*** 0.42*** 0.42*** 0.40*** 0.41*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] City fixed effects Yes Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Employment (logs) 0.27*** 0.33*** 0.39*** 0.38*** 0.38*** 0.38*** [0.005] [0.001] [0.000] [0.000] [0.001] [0.001] Education fixed effects Yes Yes Yes Yes Yes Managerial experience (logs) 0.1 0.09 0.09 0.09 0.09 [0.158] [0.175] [0.193] [0.186] [0.204] Age of the firm (logs) 0.05 0.05 0.05 0.05 0.05 [0.392] [0.377] [0.353] [0.367] [0.351] Age of the owner (logs) 0.51*** 0.49*** 0.48*** 0.47*** 0.50*** [0.000] [0.000] [0.000] [0.000] [0.000] Firm uses machinery 0.24*** 0.24*** 0.26*** 0.26*** [0.003] [0.004] [0.002] [0.002] Firm uses vehicles 0.40*** 0.40*** 0.40*** 0.39*** [0.003] [0.003] [0.003] [0.004] Corruption is a severe obstacle 0.01 0.01 0.01 [0.895] [0.889] [0.912] Crime is a severe obstacle 0.08 0.1 0.1 [0.366] [0.262] [0.273] Firm operates from outside household premises 0.13* 0.13* [0.082] [0.089] Firm has a bank account to run the business 0.18 0.16 [0.198] [0.230] Firm uses family labor 0.03 0.03 [0.786] [0.728] Number of owners 0.02 0.02 [0.776] [0.772] Largest owner is single 0.07 [0.372] Firm produces or sells under contract for another business 0.05 [0.571] Either of the largest owner s parents own/owned a business 0.02 [0.737] Observations 744 744 744 744 744 744 744 744 R squared 0.06 0.17 0.18 0.23 0.25 0.25 0.26 0.26 p values in brackets. Significance level is denoted by *** (1% level or less) ** (5% or less) and * (10% or less). All regressions are based on OLS specification with Huber White robust standard errors. 7
Table 2: Monthly sales and gender (1) (2) (3) (4) (5) (6) (7) Dependent variable: Sales(monthly, logs, USD) Female 0.58*** 0.56*** 0.53*** 0.48*** 0.48*** 0.43*** 0.44*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] City fixed effects Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Education fixed effects Yes Yes Yes Yes Yes Managerial experience (logs) 0.11 0.1 0.1 0.09 0.09 [0.143] [0.157] [0.172] [0.178] [0.212] Age of the firm (logs) 0.04 0.04 0.05 0.05 0.06 [0.478] [0.463] [0.427] [0.394] [0.343] Age of the owner (logs) 0.55*** 0.52*** 0.52*** 0.49*** 0.51*** [0.000] [0.000] [0.000] [0.000] [0.000] Firm uses machinery 0.30*** 0.30*** 0.31*** 0.30*** [0.000] [0.000] [0.000] [0.000] Firm uses vehicles 0.53*** 0.54*** 0.49*** 0.48*** [0.000] [0.000] [0.000] [0.001] Corruption is a severe obstacle 0.05 0.02 0.03 [0.525] [0.792] [0.744] Crime is a severe obstacle 0.09 0.12 0.12 [0.320] [0.158] [0.162] Firm operates from outside household premises 0.16* 0.16** [0.051] [0.044] Firm has a bank account to run the business 0.32** 0.30** [0.033] [0.043] Firm uses family labor 0.29*** 0.28*** [0.000] [0.001] Number of owners 0.11 0.12 [0.231] [0.194] Largest owner is single 0.06 [0.447] Firm produces or sells under contract for another business 0.1 [0.258] Either of the largest owner s parents own/owned a business 0.01 [0.850] Observations 747 747 747 747 747 747 747 R squared 0.08 0.19 0.25 0.28 0.28 0.31 0.31 p values in brackets. Significance level is denoted by *** (1% level or less) ** (5% or less) and * (10% or less). All regressions are based on OLS specification with Huber White robust standard errors. 8