The first randomised control trial examining the effect of capital on firms was conducted by De Mel et al. (2008). Working in Sri Lanka, they distributed grants of US$100 or $200 to a random subset of a sample of microenterprises. The grants allow them to isolate a shock to the supply of capital that is independent of any other characteristic of the firm.[1] Using this supply shock, they estimate that enterprises have average returns of 5-6% per month, representing annual returns of more than 60%. In De Mel et al. (2012), the authors report on a longer-term follow-up of the enterprises, showing that the returns on the initial grants are sustained five to six years later. Fafchamps et al. (2014) replicate the experiment in Ghana, again finding that marginal returns significantly exceed loan rates. In a similar experiment in Mexico, McKenzie and Woodruff (2008) find average returns of 20-30% per month.
An additional notable finding from these studies is that capital has a much more positive effect in male-owned enterprises than in female-owned enterprises. De Mel et al. (2008) estimate that marginal returns are near zero in female-owned enterprises while they are almost twice the full sample average in male-owned businesses. Fafchamps et al. (2014) find similar results for cash grants, but find positive returns in female-owned businesses for grants made in-kind (that is, through the purchase of assets for the business directly). De Mel et al. (2009) examine the results for female-owned business in Sri Lanka more closely, finding evidence of positive returns for female-owned businesses in households where women have more say in decision-making, and higher returns in sectors in which both males and females work, such as retail.
McKenzie (2017) provided much larger grants to somewhat larger enterprises in Nigeria, in an experiment designed around a business plan competition (YouWiN!) conducted by the Nigerian government with the support of the UK Department for International Development and the World Bank. The business plans submitted by applicants were rated by judges, with the 475 entrants rated highest nationally or in their region declared winners and awarded a grant of US$50,000. Another 1,841 entrants rated just below this highest group were ‘semi-finalists’, and 729 of these were randomly selected to also receive the grant. The random allocation of grants within the group of semi-finalists allowed McKenzie to assess the impact of a substantial relaxation of the credit constraint conditional on being good, but not great, in the eyes of the judges.[2]
Since the competition attracted both existing businesses and proposed new ventures, the results are differentiated into these two groups. McKenzie’s discussion focuses more on employment than on the effect of the grants on profit. Among start-ups, receipt of the grant led to a 37 percentage point increase in the likelihood of having a business three years later, and a 23 percentage point increase in the likelihood of having a business employing 10 or more workers. Among existing businesses, grant winners were 20 percentage points more likely to be in business, and 21 percentage points more likely to have 10 or more employees, three years later. This implies a cost per job created of US$9,600. McKenzie notes that measured profits are extremely noisy, so it is difficult to estimate the marginal returns on the grants. The regressions for both new and existing businesses show significant effects on profits only in the second follow-up survey, with the return a little under 1% per month for both groups. It is likely that these returns are lower than the interest rates that these sorts of firms would pay. Given the uniqueness of the very large grants provided, it is hard to know how to generalise the results apart from the estimated returns to capital they generate. It seems likely that the employment outcomes, for example, would have been smaller if the capital had been provided as loans rather than grants.
[1] Nevertheless, the grant is not a perfectly clean injection of capital into the enterprise, because owners may either use their own resources to make complementary investments or de-capitalise the investment, for example by drawing down inventories. De Mel et al. also show that owners respond by increasing the hours they work in the enterprise, at least in the short run.
[2] McKenzie’s design also allows him to ask whether judges can predict which firms will grow faster, conditional on receiving the grant. We return to this comparison below.
