Digital Business Training Programs: Evidence from Jordan

The core training bundle includes the following:

1- On a password-protected website: users will have access to videos covering the story of a young Jordanian woman and her journey to open a business, in addition to accompanying short lectures, animations, and clips on real businesses by Jordanian women. The material covers 3 modules, and participants can get a certificate at the end of each module. The 3 modules cover the following themes:
- Motivation and ideation
- Financing, accounting, and HR
- Growth and marketing

2- On WhatsApp, users will be sent the following:
- a suggested schedule and reminders to watch the training
- prompts asking participants to complete reflection exercises related to the training
- “flashcards” containing summaries / rules of thumb from the 3 modules

The social media exposure arm implements the WhatsApp part of the training bundle through groups of 5-6 participants, as opposed to individual communication, and sends each group a curated social media feed of business profiles and tutorials.

2021-07-15

2021-08-15

The main outcomes of this experiment are as follows:

1) take-up of training: this consists of survey and administrative data on respondents' participation in the training, including access, progress, and completion.

2) business creation, survival, and performance: willingness to start a business, the percent of fully and partially operating businesses, the type of women who select into operating a business across treatment arms. Additionally, willingness and ability to participate in markets / bazars for the business, willingness and ability to be featured on the MFI website as operating a business. While profits and sales will be measured and reported, these are expected to be too noisy to be picked up by this experiment.

3) knowledge and implementation of business practices: variety and type of products sold, accounting and financial practices, management and HR practices, digital & non-digital marketing practices, variety and type of clients.

4) empowerment and psychological impact: confidence and resilience, decision-making at home, locus of control, values and opinions on women's work, ability and frequency to leave the house, perceived share of women working in community / around Jordan

5) networks and learning: sources of learning and support (peers vs. online resources to learn business practices), number of friends that the respondent can talk to about business / personal problems

As the microfinance I am working with has a limit on the number of trainees it can register per year, I am recruiting and randomizing interested women into one of the below treatment arms:
1- control arm: their access to the training is postponed by a year
2- treatment arm: will get access to the training bundle, which includes individual reminders on WhatsApp
3- treatment and social media exposure arm: will access the training bundle + get placed in WhatsApp groups of 5-6 participants. These groups are meant to increase participants' exposure to existing and aspiring business owners in two ways. First, directly through networking in these chatting groups; and second, indirectly through a "curated social media feed" of business profiles and tutorials on social media. In practice, this feed entails sending participants highlights of business ideas, business skill tutorials, and businesses pages as publicly available on Facebook, Instagram, and Youtube.

Randomization into the above treatments will be stratified by the following:
- business ownership: whether women currently own a business, have owned a now closed business in the past two years, or neither
- reported struggle to afford Internet: while all women have access to a smartphone and internet, some of them suffer more than others to pay their internet bills. Since the training is virtual, I plan to stratify by reported struggles to pay Internet
- willingness to participate in the social media exposure groups: participants have the option to participate in the training while opting out of the social media exposure group, in which case they would just be randomized into control vs. training.
In total, these constitute 12 strata.

randomization done in office by a computer

individual level randomization: each participant is randomly placed into control, training, or training + social media exposure. The social media exposure arm will contain WhatAapp groups of 5-6 people each, for a total of 34 groups in that arm.

No

While randomization is at the individual level, we might want to cluster at the whatsapp group level in the third treatment arm where individuals are assigned to groups, if we think that outcomes are likely to be correlated within group. If the sample size remains the same by the endline, this would deliver 649 clusters (clustering at the group-level for the 34 WhatsApp groups, and at the individual-level for the remaining 615 participants).

787 participants, including existing business owners and individuals aspiring to become business owners.

While all participants applied to the training, they were not required to participate in WhatsApp groups. The social media exposure training was randomly assigned only among those willing to participate in groups. The resulting sample therefore includes:

Control group not receiving the training: 388 participants, including:
- 332 who agreed to participate in whatsapp groups
- 56 who opted out of whatsapp groups

Treatment group receiving the training: 399 participants, including:
- 170 who agreed to participate in whatsapp groups but were randomly assigned to receive individual communication
- 172 who agreed to participate in whatsapp groups and were randomly assigned to one of 34 groups of 5-6 participants each
- 57 who opted out of whatsapp groups