Experimental Design Details
Our experiment will involve refugee job seekers and local employers in Uganda, the third largest refugee hosting country in the world and the largest in Africa. Uganda grants refugees complete freedom of movement and full right to work. For this reason, its host policy has been praised worldwide for its openness and generosity. Refugees that migrate to Uganda come from states in turmoil such as South Sudan, the Democratic Republic of Congo, Burundi, and Somalia. Both refugee job seekers and local employers will be recruited in the capital city of Kampala.
The experiment will involve the following steps. Notice that step 1 and 2 will take place at the same time (i.e. we will interview firms and refugees interested in some specific sectors one sector at a time).
Step 1: Recruitment and certification of refugee job seekers
With the assistance of YARID, we will recruit 800 refugee job seekers. We will collaborate with refugee local leaders to identify potential beneficiaries: 18 to 25 years old job seekers and that are willing to accept a short-term (one week long) internship at some firms (in sectors we will describe below) for a salary of no more than 50,000UGX (approximately 15USD). We will first run a baseline survey, aiming at capturing: characteristics of the workers (past work experience, labor market outcomes before and after arrival in Uganda), beliefs about the Ugandan labor market, as well as cognitive (Raven Tests), non-cognitive skills (Big 5, Grit and Self-esteem), knowledge of English and Luganda, math skills.
We will test all the participants to the experiment using procedures and guidelines developed by the Uganda Directorate for Industrial Training (DIT) with the aim of providing a random selection of workers with a skills certificate called similar to the well-known “WorkersPAS” (Workers’ Practically Acquired Skills). This initiative was established several years ago by the government of Uganda with the aim of certifying non-formal and informal skills. We will offer certification in the following sectors: Bakery, Brick-laying, Hairdressing, Herbalists, Hotels, Restaurants and Catering, Phone repairing, Tailoring. We choose these sectors for the following two reasons: 1) A large number of the refugees in our previous survey expressed interest in working in these (or related) sectors and 2) The DIT has well-established assessment modules to test workers in these sectors (among others). At the time of registering this experiment, the researchers have not yet determined whether they will certify workers using the Worker's PAS or a similar one, based on the guidelines of the DIT.
Refugees will then be randomly divided in two groups: 25% (n=200) of refugees will be allocated to a “Certified” group (C) and 75% (n=600) to a “Non-Certified” group (NC). The only difference between these two groups will be that some of them will be endowed with the Worker’sPAS and the others will receive a simple certificate of completion of the test (i.e. not the Worker’sPAS).
Step 2: Recruitment of firms, baseline, randomization in treated and control, and matching
We will recruit 800 firms to participate to a baseline survey. To do so, we will first conduct a census of firms active in the sectors we are interested from the main markets of Kampala. During this exercise, we will collect detailed information about the business (monthly profits and revenues, employees and assets value) and the firm owner (specifically, her attitudes towards refugee job seekers). We will select participants to the experiment with the following criteria: 1) have never hired a refugee and 2) have at least one employee which is not a family member (and are hence used to hiring workers whom they are not personally connected to). The baseline survey will last no more than 90 minutes and will collect detailed information on: general practices regarding hiring (new) workers, constraints to business growth, employees, revenues and business profitability, as well as beliefs about workers of different nationalities (especially asylum-seekers and refugees).
After the baseline survey, firms will randomly be allocated to two different groups: 200 employers will be matched to the C-group refugees, and 600 instead to the NC-group refugees. Hence, the only difference between the two groups is that the firms matched with the C-group refugees will be told and will be shown the skills certificate of the potential hire. We will randomly assign the treatment status stratifying by sector of the firm.
Step 3: Willingness-to-pay (WTP) to hire a refugee
Firms will be then shown the refugees they have been matched with. At this step we will elicit firms’ true willingness to hire a refugee for a short period of time (one week), using an incentive-compatible Becker-DeGroot-Marschak mechanism: a multiple price list. We will thus act as an intermediary firm aiming to match each firm with the randomly selected refugee. More specifically, firms will be shown 11 prices, wp, between 0 and 50,000UGX (approximately 15USD) in steps of 5,000UGX (approximately 1.5USD). They will be asked whether they would be willing to hire a refugee for a weekly wage of 50,000 receiving a subsidy equal to each wp until they say no. As soon as they say no, the recorded subsidy at which they will like to hire the refugee is (wp – 5,000UGX). Firms will be told that we will draw a random price, wr, and that if their proposed wp is higher than the wr, they can hire the refugee for a week. In order to be sure that the random assignment of the subsidy is quasi-exogenous to the firm’s preferences, we will randomly produce wr to have a large mass at 0 and 50,000UGX. Overall, 50% of firms will receive the subsidy to work with the selected refugee and 50% will not. Comparing the answers of firms exposed to the C-group refugees with those of firms exposed to the NC-group refugees will tell us about the value of the market certificate.
Step 4: Exposure to a refugee job seeker
The random price will be such that, conditional on the WTP, 50% of firms across groups will receive the offer to work with the refugee job seeker. In this way, both firms that have access to certificates and those who do not will be randomly exposed to a refugee worker for a week. This strategy will allow us to have four groups of firms:
1. No certificate, no exposure
2. With certificate, no exposure
3. No certificate, with exposure
4. With certificate, with exposure.
This way we can measure the impact of an employment subsidy and compare it to the effect of showing certificates only. We can also test eventual complementarities between the two policies. Importantly, we will ask the firm to hire the refugee worker for only one week. Together with the fact that the certification will just be an acknowledgement of skills already acquired by the job seeker, this will imply that we can rule out the concern that exposure leads to more productive workers, which could have an independent effect on the WTP.