Digital training on entrepreneurship, financial inclusion and formality for micro-firms

We provide firm owners and managers with access to a digital-based training app (Expertienda) on entrepreneurship skills and financial literacy. The training course is certified upon completion by Rosario University and Minuto de Dios University, and it could be the basis to access further interventions. Firms are located nearby Minuto de Dios University campus, which includes both residential and commercial areas of he cities.

2021-04-01

2021-09-30

Survival after 1 year of the intervention start
Three indexes: formalization, management practices and financial inclusion.
Profits self-reported category
Number of employees
Reported an investment in the business

Variables included in the indexes
Formalization: the firm is registered with the appropriate institutions, have the required operation permits, have a book keeping system, and the proportion of employees who are under a formal contract (pays health and pension insurances).
Management practices: A set of variables from the literature including practices related to management of stocks , sales, providers and financial planning.
Financial inclusion: the firm has its own bank account, asked for a loan in the last year, has an insurance product, knows about electronic wallets, beliefs that banks can be useful for firms

Expanded management practices and knowledge variables

There are 10 neighbourhoods for the study. In each case, we made a rapid census of the study area and locate all small businesses. We did a baseline survey (Nov, Dec 2019) in order to develop the design. Units (firms) which wanted to participate in the study, were open, and have someone available -after several attempts- (52% of the rapid census), were grouped graphically and assigned to three blocks:
Arm 1: All units in the cluster are treated
Arm 2: 50% of the units are treated
Arm 3: None of the units are treated
As clusters are located contiguously within neighbourhoods, contamination due to externalities is possible. Therefore, we also consider 4 groups constructed with the proportion of the nearest neighbours (NN) within 50 meters which received treatment.
Group 1: Unit received treatment, and at least 50\% of the NN received treatment.
Group 2: Unit received treatment, and less than 50\% of the NN received treatment.
Group 3: Unit did not received treatment, but at least 50\% of the NN received treatment.
Group 4: Unit did not received treatment, and less than 50% of the NN received treatment (pure control group).
If the effect for group 1 and 2 is the same, and the effect for groups 3 and 4 as well, we would conclude that there are no spillovers. In such case, a simple treated vs control exercised can be performed.

Two-stage randomization

Randomization with Stata randomize
Level one: geographic cluster, blocks by two global regions, and commercial density of the location of the firm
Level two: firm (for arm 2), same blocks

Yes

146 sets of firms, defined by geographic proximity

1172 firms

Arm 1: 50 clusters; 400 treated units
Arm 2: 48 clusters; 189 treated (26 clusters), 189 control units (22 clusters)
Arm 3: 48 clusters; 394 control units

Considering the three arms, the minimum detectable difference is of 0.05. That is considering a mean of 0.42 for the main total score with a standard deviation of 0.16, and a intra-cluster correlation of 0.19 (information from baseline survey). With a power of 80%, an average cluster size of 8, with 49 cluster per arm. Yet, simulation exercises will be done for the four groups analysis.