There is a well-documented “gender profit gap” for small and medium enterprises in developing countries. Yet, observable differences between men and women entrepreneurs explain only a small portion of this earnings differential. This suggests that gender discrimination may be an important, yet understudied, factor inhibiting the success of female entrepreneurship. If women face discrimination in the evaluation of their businesses, this may reduce female entrepreneurs’ ability to obtain capital, which in turn reduces the performance of their business. Policy solutions that reduce such discrimination, such as gender blinding, could both increase gender equity and potentially improve efficiency in the allocation of capital, unleashing the high-growth potential of many businesses in Sub-Saharan Africa with downstream effects on job creation, economic growth, and innovation.
In addition, there is evidence that algorithms can be used to help avoid human prejudice in decisions prone to biases. On the other hand, by improving predictive accuracy, automated decision-making using available data could replicate or exacerbate underlying inequalities. For example, when approving small businesses for loans, an algorithm might systematically limit women’s access to credit if women-run businesses tend to receive lower returns on their investments. In the context of a business plan competition with the Entrepreneurship Development Center in Ethiopia and the World Bank, the research team, with additional funding from the DCO to study algorithms as well as loan-officer decisions, compares how four different methods of evaluating businesses affects the equity and efficiency of capital allocation decisions. To test both human-based and algorithm-based decision-making, they will compare decisions made by (i) loan officers with applicants’ gender information; (ii) loan officers without gender information (“gender-blinded”); (iii) an algorithm designed solely to predict business success and (iv) an algorithm designed to predict business success which incorporates discrimination-aware methods. Results forthcoming.
Copyright 2019. All Rights Reserved
Design & Dev by Wonderland Collective