Skip to main content

Evaluation 2 of "The Benefits and Costs of Guest Worker Programs: Experimental Evidence from the India-UAE Migration Corridor"

Evaluation 2 of "The Benefits and Costs of Guest Worker Programs: Experimental Evidence from the India-UAE Migration Corridor" for The Unjournal. Evaluator: Anonymous

Published onMay 16, 2024
Evaluation 2 of "The Benefits and Costs of Guest Worker Programs: Experimental Evidence from the India-UAE Migration Corridor"
key-enterThis Pub is a Supplement to


This is an evaluation of “The Benefits and Costs of Guest Worker Programs: Experimental Evidence from the India-UAE Migration Corridor”[1]. The evaluator summarizes this as follows: The key contributions of this paper are its analysis of intermediary costs of migration, migrant well-being and diversity in the workplace, expectations of job-seekers and joint comparisons of various pecuniary and non-pecuniary measures of migration. A major limitation is the lack of a welfare justification for randomisation and little to no discussion on equipoise. My suggestions largely revolve around improvements to the discussion of methodology, representativeness, attrition and representation of treatment effects.

Summary Measures

We asked evaluators to give some overall assessments, in addition to ratings across a range of criteria. See the evaluation summary “metrics” for a more detailed breakdown of this. See these ratings in the context of all Unjournal ratings, with some analysis, in our data presentation here.1


90% Credible Interval

Overall assessment


60 - 90

Journal rank tier, normative rating


3.0 - 4.0

Overall assessment: We asked evaluators to rank this paper “heuristically” as a percentile “relative to all serious research in the same area that you have encountered in the last three years.” We requested they “consider all aspects of quality, credibility, importance to knowledge production, and importance to practice.”

Journal rank tier, normative rating (0-5): “On a ‘scale of journals’, what ‘quality of journal’ should this be published in? (See ranking tiers discussed here)” Note: 0= lowest/none, 5= highest/best”.

See here for the full evaluator guidelines, including further explanation of the requested ratings.

Written report


In this paper, the authors describe their findings from a randomized experiment on jobseekers in India. In particular, the paper studies applicants to employment opportunities at two construction firms in the UAE. Data is collected from approximately 4-5 thousand workers at recruitment centers in six Indian states of a Singapore-based recruitment firm. Five out of every seven candidates who successfully clear the recruitment firm's selection process are randomly selected and offered positions. Through a baseline survey, two tracking surveys, one follow-up survey, and multiple alternative administrative data sources, the authors observe and analyze the effects of being offered an opportunity to migrate on several outcomes including actual migration, labor market indicators, indicators of well-being, indicators of work-satisfaction, religious and linguistic diversity in networks and attitudes toward democracy, religious diversity, and inequality.

The primary findings of this paper are as follows:

  1. There is a high rate of non-compliance among workers who are offered positions, only 58% migrate to the UAE.

  2. Treatment groups experience significant increases in earnings and hours of work and significant decreases in physical well-being at the workplace. There are no significant effects on other dimensions of well-being.

  3. 10% of gains from international migration are captured by intermediaries or migration agents.

  4. Among those who migrate to the UAE, prior expectations of earnings compare well to actual earnings.

  5. Treatment is also associated with increased exposure to individuals of different religions and linguistic groups and favorable attitudes.

  6. There is little noted heterogeneity in returns to migration, except for the group of workers who have experienced migration before.

  7. A theoretical model predicts that doubling of total offered compensation would induce full compliance among the non-complying workers (who are offered the job but do not migrate).


1. Methodology

A. Ethics of migration

Did the authors of this study randomly offer employment and migration opportunities to some eligible candidates and consequently randomly prevent employment and migration opportunities to other eligible candidates? Or did the authors of this study exploit an existing institutional randomization in the job market?

The authors make a rather weak argument for the latter, by describing the randomization process as "a natural extension of an existing system in which firms request visas from the MOL and sometimes are granted all of them and sometimes fewer than they request" (Page 10). However, footnote 21 clarifies that the two construction firms in question did indeed have a guaranteed number of visas. This appears to be a contradiction, in that in the absence of this study, presumably, the workers would not have experienced any randomization through the visa selection process. Footnote 21 further clarifies that the firms "did have to agree to screen more applicants than they usually would need to for every position they wanted to fill". Did the recruitment firms increase the number of screened applications by inviting more applications or by loosening their criteria for the screenings?

Without more information, it is difficult to assess the welfare costs of this study on its participants. In its current form, it appears that in the absence of this study, at least some of these control group participants would have received an opportunity to be employed and migrate to UAE and others in the treatment group may not have successfully passed screening. Both these possibilities have important implications, most importantly ethical implications but also empirical implications (See 2A1). Furthermore, the authors do not mention any IRB, nor do they describe the ethical oversight or involvement of local and other stakeholders.

Knowing what is shared in the paper in the form of previous literature on the returns to migration, it is hard to believe that authors did not have any priors about the welfare implications of employment and migration on study participants, and if that is the case, how do they justify equipoise? Taking this argument further, it is also likely that authors might have had credible priors on which groups of participants benefit most from employment/migration based on the baseline survey.

B. On selection and representativeness

  • The literature review discusses the contribution of this paper as better understanding selection into migration. This deserves further qualification: the paper is able to qualitatively describe general selection into migration, but the empirical contribution is rather addressing the selection into migration conditional on applying for an international job, i.e. conditional on having intent to migrate.

  • The experimental design mentions that the recruitment sites are a sample. If so, how were recruitment sites selected? How many recruitment sites do the partner firms have? Do they only recruit from the 6 listed states? The recruitment sites are mentioned as being close to construction training schools. Are workers mostly fresh graduates of these schools?

  • One issue of selection and/or representativeness that the paper does not tackle is whether the 6 states are representative of migrant workers from India to the UAE (in construction or general). It is well known that southern Indian states, particularly Kerala, have historically contributed most in numbers toward migration to UAE. The paper would benefit from further comments and comparisons on this front, potentially using data from the Kerala Migration Surveys.

  • The exclusion of southern states also has an implication for the analysis of friendship and linguistic and religious exposure. Specifically, given the historic pattern of migration from south India to UAE, it is likely that Indian migrants from historically non-contributing states are more likely to have increased exposure to new linguistic and religious groups through migration than others from historically contributing states. This paper has an empirical advantage on this front, and this deserves some discussion.

  • At the same time, one potential disadvantage of this exclusion of southern states is that it limits the generalizability of the analysis of marginal treatment effects, especially if the authors expect that heterogeneity (either in baseline characteristics or in returns) [will] be more pronounced among historical migrant groups. Just as how authors discuss that estimates may be a lower bound on the non-pecuniary costs (Page 41), because the sample is drawn from candidate interviewees, it would be helpful to discuss similar implications because of sample states.

  • For similar reasons, it would be useful to discuss state-based heterogeneity in the analysis of heterogeneous returns.

2. Empirics

A. Always takers and Never takers

A more detailed description of never-takers and always-takers which would help contextualize the ITT results better;

  • On never-takers, what is the historical percentage of applicants to refuse to take up the offer? Given that the authors mention that the recruitment firms screened more candidates for the purposes of this study, is it possible that the percentage of never-takers is more than usual?

  • On always-takers, is it the case that applicants that visited the survey recruitment centers also visited other recruitment centers during this recruitment period? What are the channels by which the applicants who failed the screening entered the UAE job market? Were they recruited by similar recruitment companies? Did they end up working for similar firms?

B. Attrition

An important part of this paper is how authors deal with attrition. Attrition is impressively contained and [this] is testament to what must have been painstaking work for the research assistants and authors.

In describing the observed attrition, authors could include more details. For example, what are the dimensions in which there is imbalance? While this is self-evident in Appendix Table A.1., this information should also be discussed on Page 16. Is there any proposed explanation for the observed selection into attrition? Another discussion worth having is whether attrition is observed differently for the treatment and control group, as the authors seem to assume this in bounding their results in Section 4.4. If so, what is the difference, and what are the authors' priors on how this would bias the results? A qualitative discussion of this would help [convince readers] about the usefulness of the bounding exercise carried out later in the paper.

C. Treatment effects

  • It is unclear why authors do not control for baseline wealth (as measured by net assets) in the specifications, given that there is imbalance in this dimension between the treatment and control group. Do other fixed effects take care of this? What is the implication of this baseline difference on the follow-up finding of no significant differences in net assets between the treatment and control group (Table 7)?

  • While most treatment effects reported in the paper are conservatively ITT, and authors do share with us the scaling factor for TOT earlier in the paper, it would be useful to also have a TOT/IV represented in the main results tables, especially in the context of the impact on well-being and work satisfaction.

  • Is it possible that workers who have received offers have fairly accurate expectations of their earnings because the partner firms/recruitment firms have communicated this information well? If so, is this an unusual characteristic of the employer firm?

  • Is the no-offer density of well-being bimodal in Figure 4. because of the sub-group within the control group that did migrate to the UAE?

3. Policy Implications

In addition to the national policies mentioned on Page 9, it would be good to learn about … state policies and bilateral agreements with the UAE [if there are any]. What are the specific rules and regulations governing migration from the 2022 MOU that have little enforcement? This is especially important to [help us] understand whether there is a current policy audience for the findings of this paper.

4. Editorial Comments

  • Recommend not using the terminology "lower caste".

  • Recommend moving Appendix A2 to main text.

  • On page 10, it would be helpful to clarify whether offered wages have little variation within the firm or across the two surveyed firms?

  • On Page 24, the authors mention that "They are earning less in India than they expected to earn if they migrated, consistent with the idea that they only migrate for higher earnings." This claim is more credible without the use of the word "only".

  • The term "Indian men" is often used to refer to the sample, but it would be more accurate to say, "male construction workers from north India".

  • Figure 5 legends mention "Less" of certain indicators, which is unnecessary and confusing as it appears as a double negative (for example, negative coefficient on less Fighting).

  • On Page 41, perhaps it should be given the lack of heterogeneity in returns on observables.

  • Calculations in Section 4.8 is the most interesting contribution of this paper, and it would be helpful to have more subsections in this section, delineating the model, estimates from literature, assumptions/calibrations made and final calculation.

  • On Page 22, the uthors mention that this study provides evidence against the idea that migrant workers in the GCC are forced to work excessive hours. However, in footnote 15, the authors describe that, in this context, the two firms that they are partnered with are large firms that are more likely to follow labor regulations. Perhaps they can qualify this statement more.

  • In page 3, the authors describe that never-takers have somewhat lower UAE earnings. They should clarify that these are calculated potential outcomes (given that it is only shortly before that the authors define never-takers as those who do not migrate).

Evaluator details

  1. How long have you been in this field?

    • Mid-career development professional, currently pursuing a PhD in economics

  2. How many proposals, papers, and projects have you evaluated/reviewed (for journals, grants, or other peer-review)?

    • Journals/Grants: 0

    • Other peer-review: ±5

1 of 1
No comments here
Why not start the discussion?