Do employers in Myanmar prefer workers who accumulated skills in more advanced countries? Evidence from a field experiment

Saw Htay Wah

doi:10.1186/s40176-017-0106-2

Original article
Open Access
Published: 11 January 2018

Do employers in Myanmar prefer workers who accumulated skills in more advanced countries? Evidence from a field experiment

Saw Htay Wah¹

IZA Journal of Development and Migration volume 8, Article number: 1 (2018) Cite this article

1780 Accesses
6 Altmetric
Metrics details

Abstract

I conducted an audit study in Yangon, Myanmar, to examine the effects of foreign labor market experience on employment outcomes in the Myanmar labor market focusing on three skilled occupation groups—software developer, accountant, and engineer—which combined account for more than 80% of all professional and technical job openings. I find heterogeneous effects by employer type (local, foreigner), occupation (software developer, accountant, engineer), and foreign country where such experience was accumulated (Singapore, Malaysia). In particular, I find a large and significant effect for software developers who accumulated experience in Singapore and no significant effect for the same experience accumulated in Malaysia; the Singapore effect is mainly due to foreign employers operating in Myanmar. In contrast, I find no significant effects of foreign experience for accountants and engineers. This is true for both the Singapore and Malaysia experiences, as well as for local and foreign employers.

JEL Classification: O15, J01, F22

Introduction

One prominent feature of international migration today is that a vast majority of migrants return to their country of origin after taking up employment or spending time in a host country for a certain period (Gibson and McKenzie 2011). Frequently cited economic reasons for returning include (1) higher returns in home countries to human capital accumulated in host countries (Dustmann, Fadlon and Weiss, 2007, 2011); (2) higher returns to self-employment activities in home countries relative to host countries (Dustmann and Kirchkamp 2002); (3) higher preferences for consumption in home countries (Dustmann and Weiss 2007); (4) a high purchasing power in home countries relative to host countries (Dustmann and Weiss 2007); (5) an increase in wage differentials between host and home countries (Dustmann 2003); (6) a fulfilment of a saving target that allows migrants to become entrepreneurs in home countries (McCormick and Wahba 2001; Wahba 2015; Mesnard 2004); and (7) worsening economic outlook in host countries relative to home countries. Non-financial factors such as personal preferences, social ties, and lifestyles are also found to be strong determinants of return migration (Gibson and McKenzie 2011).

Prior studies analyzing the effects of foreign experience find that return migrants are more likely than non-migrants to earn higher wages in home country labor markets (Wahba 2015; Reinhold and Thom 2013; Co et al. 2000; Barrett and Goggin 2010; Martin and Radu 2012). The effects are also found to differ by (1) gender (Co et al. 2000; Barrett and O’Connell 2001); (2) occupation (Reinhold and Thom 2013; Co et al. 2000); (3) host country or region where experience was accumulated (Co et al. 2000; Barrett and Goggin 2010; Carletto and Kilic 2011); and (4) education level of returnees (Barrett and Goggin 2010; Reinhold and Thom 2013; Barrett and O’Connell 2001; Barrett and Goggin 2010).

In this study, I examine the relationship between foreign labor market experience and employment outcomes that have received little attention in the extant literature.^{Footnote 1} It is important to study employment outcomes because, despite possessing valuable skills and overseas experience, returnees could face barriers to full participation in local labor markets. For instance, a mismatch between skills acquired abroad and skills demanded by local employers (Sun, 2013) and a lack of cultural awareness and sensitivity may make returnees less attractive than non-migrants to potential employers in home countries, which could worsen employment outcomes for returnees. Or a lack of social ties and connections in the home country context after years of absence could lengthen their employment search process and thus unemployment spell (Martin and Radu 2012).

Examining the employment consequences of migration experience is hampered by the fact that migration is a highly selective process. For instance, Ambrosini et al. (2011), Wahba (2015), De Coulon and Piracha (2005), and Dustmann and Weiss (2007) find that migrants are positively selected relative to non-migrants. Similarly, in return migration, individuals who choose to return may fundamentally differ on both observed and unobserved characteristics from those who do not return. Indeed, prior studies show that returnees are both negatively (Wahba 2015; De Coulon and Piracha 2005) and positively selected relative to migrants who do not return (Ambrosini et al. 2011; Martin and Radu 2012). In addition, when analyzing the effects of return migration, one also has to address the bias associated with returnees self-selecting into specific occupations in home countries. For instance, compared to non-migrants, return migrants are more likely to become entrepreneurs (McCormick and Wahba 2001; Wahba 2015; Martin and Radu 2012). As a result, differences in outcomes between returnees and non-migrants reflect unobserved factors such as personality and motivation that can influence both migration decision and outcomes of interest (McKenzie et al. 2010; McKenzie 2012).

Given the highly selective nature of migration and return migration, studies analyzing the effects of foreign experience using conventional survey approaches may fail to uncover the true explanation for differences in outcomes between returnees and non-migrants. In the outmigration literature, there are lottery programs (e.g., the US H-1B visa lottery and New Zealand visa lottery program) that enable researchers to causally estimate the return to migration (i.e., the impacts of people leaving their home country for other countries).^{Footnote 2} In contrast, in return migration, there are no precedents for policy programs that will serve as a natural experiment allowing for causal inference of return migration.

This study aims to address the shortcomings inherent in the use of conventional survey data approaches and provide causal estimates of the effects of foreign experience. In particular, this study employs a resume audit methodology that randomly assigns if a fictitious job applicant has foreign labor market experience or local experience. This methodology allows me to observe outcomes for returnees and non-migrants that are similar in both observable and non-observable characteristics, except for one important dimension—migration status. Then, any observed differences in outcomes between the two groups can be solely attributed to employers’ reaction to migration experience. The resume auditing method has proven useful for studying differences in labor outcomes by race (Bertrand and Mullainathan 2004), age (Lahey 2008; Neumark et al. 2015), length of unemployment (Kroft et al. 2013), college education (Martorell et al. 2014), and nationality (Oreopoulos 2011).

In this paper, I present results from a resume audit study conducted in Yangon, Myanmar, that seeks to examine two questions: (1) how much demand exists for return Burmese migrants who accumulated computer programming, engineering, and accounting skills in Singapore and Malaysia, relative to similar workers who accumulated the same skills in the Myanmar labor market? and (2) does the demand for these skills acquired in Singapore or Malaysia differ by occupation, employer type, and foreign country? The resume audit study involved (1) submitting hundreds of fictitious resumes online in response to postings for skilled occupations (software developer and programmer, accountant, and engineer) by two kinds of employers (foreign and local) in the Myanmar labor market and (2) measuring callbacks for each resume submitted. Each resume was randomly assigned whether an applicant accumulated work experience in Malaysia or Singapore or in the Myanmar labor market.

Overall, I do not find statistically significant evidence that foreign experience is favored overall by employers in Myanmar. This aggregate effect however masks a great deal of heterogeneity by occupation, employer type, and foreign country. Specifically, software developers and programmers with foreign experience are highly favored, particularly if the experience was accumulated in the Singapore software industry. When selecting on employer type, I find that the effect of Singapore experience is mainly driven by foreign employers in Myanmar. In contrast, I find no significant effects of foreign experience for accountants and engineers. This is true for both the Singapore and Malaysia experiences, as well as for both local and foreign employers.

This paper contributes to the existing literature on return migration in two ways. First, it addresses the endogeneity of return migration and provides causal estimates of foreign experience. Second, it helps better understand heterogeneous effects by occupation, employer, and foreign country where experience was accumulated. It also contributes to a growing list of labor market audit studies conducted in developing country settings (e.g., Galarza and Yamada 2014; Banerjee et al. 2009; Bóo et al. 2013).

The remainder of the paper is organized as follows. Section 2 presents the experimental design and describes the sample and data source used in this analysis. Section 3 details empirical analysis, followed by a presentation of the findings in Section 4. Section 5 concludes.

Experimental design

The experiment was conducted in Yangon, Myanmar. With a population of over 6 million, Yangon is the cultural, commercial, and financial capital of Myanmar, home to over 90% of all registered businesses, and thus accounts for a vast majority of private sector jobs. When it comes to examining how returning skilled migrants can potentially close the skills gap and strengthen a local labor market in a developing country, Myanmar offers an excellent case study due to its unique economic and political history. Since its independence from Great Britain in 1948, Myanmar was ruled by three consecutive military regimes until 2015 when the first democratically elected civilian government came to office in the country’s history. During decades of military rule, crackdowns on students, political oppression, and severe economic hardships triggered a departure of the skilled and talented to other countries in search of better economic opportunities. Meanwhile, absent meaningful educational reforms, the country’s higher education system was unable to produce graduates equipped with the skills employers needed. Although no official statistics are available, anecdotal evidence suggests that a large number of Burmese migrants work as engineers, skilled laborers, surveyors, IT analysts, and nurses in more advanced countries (relative to Myanmar) such as Singapore and Malaysia. Many of these migrants have worked for international firms and accumulated skills consistent with global standards. Thus, they could be a potential source of skills for employers in Myanmar if programs and policies designed to attract them to return home can be put in place.^{Footnote 3} The historical context and current political climate of Myanmar provide a unique opportunity to explore how return migrants fare in the labor market, which will generate policy implications specific to Myanmar but also applicable to other developing countries facing shortages of sufficiently skilled workers.

Preparation of resumes began with sampling job openings and collecting job requirements from job advertisements each week. I then prepared resumes meeting those requirements in terms of educational attainment, years of work experience, salary requirement, and relevant computer skills. For each job posting, I prepared two resumes, and for each resume, I randomized the following variables: (1) country where an applicant accumulated labor market experience (local/foreign) and (2) foreign country (Singapore/Malaysia) conditional on being assigned foreign experience. The choice of Singapore and Malaysia labor markets in this application is meant to reflect the actual situation that most skilled Burmese migrants work in these countries. In addition, the characteristics of the Singapore and Malaysia labor markets are fundamentally similar to those of Myanmar. For instance, although workers are allowed to form unions in these three Southeast Asian countries, their collective political clout is relatively limited and labor rights are significantly curtailed, compared with those found in liberal democracies in western countries; protests by workers are rarely tolerated and often suppressed immediately. Thus, we can rule out Myanmar employers’ concern that returnees who have been exposed to the Singapore and Malaysia labor markets might demand excessive political rights and stage protests.

If an applicant had foreign experience, this information was salient to employers in the subject line of the emails sent to employers by listing “[Position applying for]: Applicant with work experience in Singapore/Malaysia,” as well as in the content of the application letter by listing “I had relevant work experience in Singapore/Malaysia.” In addition, foreign experience is also salient in the resume by listing an applicant’s last three foreign employers and their foreign addresses. Each resume listed the job title, period of employment, and company name of an applicant’s recent three jobs in reverse chronological order. The total years of work experience listed in the resume were always the same for both applicants and met the job requirements. Each resume was randomly assigned three companies where the applicant previously worked from a list of companies that were carefully chosen from two company registries based on two criteria: they (1) are of small size (most employing less than 20 persons) and (2) lack reputation and recognition at the national and international levels. This effectively rules out the possibility that potential employers might infer labor productivity from previous employments and their prior belief about the quality of previous experience.^{Footnote 4} Both applicants were unemployed and looking for employment. Similarly, age of applicant, year of graduation, college major, university, computer skills, and other related skills were the same for both resumes. One of the resumes was randomly assigned a female name and the other a male name. For job openings with gender restriction, both resumes were assigned the same gender. Gender requirement for a job opening was always mentioned in the advertisement. Name of an applicant was always listed at the top of the resume with a large font size. Address and contact information were listed below the name. The resumes themselves were designed based on a large pool of actual resumes used by applicants in the Myanmar market and thus representative of actual job seekers. A pair of resume samples prepared for a site engineer position is provided in Additional file 1, with resume sample 1 belonging to an applicant with Malaysia experience and resume sample 2 belonging to a local experience. While each employer was sent two resumes, they were sent at least a day apart. Each resume was randomly assigned a template and a cover letter. The font and layout choices were also randomized. Because the number of features randomized by the experiment was large, it was unlikely that the resumes would have similarly listed characteristics to describe the applicants. Furthermore, different fonts and layouts were used for each resume sent to an employer. Thus, it is very likely that employers viewed each resume as coming from an independent and real applicant.

Finally, resumes were saved as word files and randomly submitted via emails (independent of any resume characteristics) to employers over a 2-day period (Wednesday to Thursday) in each of the 19 weeks between March 23, 2015, and August 7, 2015. Unsuccessful applications due to incorrect employers’ email addresses, which constitute less than 1% of the sample, were dropped from the sample.

The sample for job openings is from a popular magazine among job seekers in Myanmar (Opportunity; http://www.opportunity.com.mm/). The magazine carries only job advertisements, is published weekly, and accounts for approximately 90% of all job advertisements in the country. Job openings are classified by industrial sector: (1) production and manufacturing, (2) construction, (3) tourism, (4) IT, (5) trading, (6) service, and (7) non-government organizations (NGO) and international non-government organizations (INGO). I restrict the sample to (1) skills-intensive and professional job categories; (2) job openings that require applicants to submit resumes via email; and (3) non-NGO and non-INGO jobs. I also restrict the sample to one randomly selected job opening per employer although several employers list multiple job openings in a given week. Each company is uniquely identified by a combination of information available in the advertisements, including company name, address, phone number, and email; I use this information to ensure that each company is sampled only once in the event that a company’s job advertisements show up in several issues of the Opportunity magazine during the study period.

Only the occupations of software developer and programmer, accountant, and engineer are chosen for this study because they represent growing occupations as Myanmar has been trying to move from an agriculture-based economy toward a service-, manufacturing-, and production-based economy. An examination of the Opportunity magazine reveals they account for more than 80% of all professional and technical job openings in a typical week. A close investigation of all job advertisements also reveals no systematic differences in job openings between foreign and local companies in terms of qualification, job title, and advertisement content. In addition, job titles within each occupation group tend to be homogeneous. For instance, job titles for most accounting jobs are either accountant or senior accountant. Myanmar has seen a gradual increase in the inflow of foreign direct investment since 2011 when it opened up, resulting in an increased number of foreign companies and multinational corporations operating in Myanmar.^{Footnote 5}^, ^{Footnote 6} Further, as the Myanmar economy is projected to grow at an annual rate of approximately 8% in the coming years, according to the World Bank,^{Footnote 7} a shortage of workers with sufficient skills among these occupation groups will be an issue for both local and foreign employers in the coming years. The distribution of sample job openings, broken down by occupations, is provided in Appendix 1. Broadly, 46% of them can be grouped as accountant, 39% as engineer, and 14% as software developer or programmer.

Empirical analysis

The outcome used in prior resume audit studies (e.g., Bertrand and Mullainathan 2004; Kroft et al. 2013; Lahey 2008; Oreopoulos 2011) is if a resume submitted receives a callback for an interview. I study additional outcomes because of the unique challenges—relative to the literature—faced by conducting this method in Myanmar. First, employers in Myanmar do not rely as heavily on resumes in terms of screening applicants and inferring productivity, compared with employers in other developed countries, in particular the USA. Rather, personal interview combined with oral examination about knowledge on relevant subjects is the preferred method for screening applicants. Second, the marginal cost to interview an applicant is very low in Myanmar. As a result, a high percentage of applicants is interviewed for a job opening, making resumes relatively less effective as a screening device and thus weakening the effects of listing any resume characteristics (e.g., foreign experience). This study finds that the average callback rate is 22%, compared to 3–5% found in prior studies. Moreover, in prior resume audit studies, information regarding whether a resume submitted receives a callback is retrieved from content of the voice message left by the employer. In contrast, employers in Myanmar rarely leave voice message, and whenever an employer calls back, a trained auditor pretending to be the applicant has to answer the call and is instructed to mention to the employer over the phone that he/she is no longer interested in the employment opportunity. As a result, the employer is more likely to call back the second applicant for an interview.

To circumvent this issue and uncover an employer’s interest in an applicant, I focus on the following outcome: if a resume submitted is the first to receive a callback for an interview. The outcome produces variation even when employers call back both applicants, and some of this variation may be informative about employers’ preferences. As a robustness check, I also consider another outcome: if a resume submitted receives a callback for an interview conditional on employers calling back only one applicant out of two. This outcome eliminates employers that are calling back both applicants and, instead, focuses on employers that are more likely to be making active hiring decisions based on the resume characteristics.^{Footnote 8}

To assess the effect of foreign work experience on the outcome variables, I estimate the following model by OLS.

$$ {Y}_{ijt}={\beta}_0+{\beta}_1\ast {\mathrm{foreign}}_{ijt}+\omega {X}_{ijt}+{\gamma}_t+{\varepsilon}_{ijt} $$

where the indices have the following meaning: resume i gets submitted to employer j in week t. Y_ijt is the outcome variable of interest. The term ε_ijt is an error. “Foreign” is a treatment variable taking a value of 1 if a resume is associated with foreign experience and zero otherwise. I report all standard errors adjusted for clustering at the employer level (Bertrand and Mullainathan 2004; Oreopoulos 2011). X is a vector of randomly assigned resume characteristics that include resume template fixed effects, resume cover letter fixed effects, resume address fixed effects, and gender of an applicant. The regression model also controls for week fixed effects captured by γ_t. The parameter of interest, β₁, captures the average effect of listing foreign experience on the outcome variable. Given that resume characteristics are randomized, their inclusion should not matter but will help reduce unexplained error, thereby improving the statistical precision of estimates of the treatment effects. To investigate if the effect of foreign experience varies by foreign country where experience was accumulated, I disaggregate the total treatment effect by introducing dummies for Malaysia and Singapore separately. To assess if callback rates are different across employer subgroups (foreign, local) and occupation subgroups (accountant, software developer, engineer), I restrict the sample according to the subpopulation of interest and report results for each subpopulation group separately. For each specification and sample restriction, I report OLS results with and without controls and p values adjusted for multiple hypothesis tests using Holm’s method (Holm, 1979). Standard errors are clustered at the employer level.

Results

Table 1 presents the results for mean equality tests of the randomly assigned resume characteristics. The last column reports p values, with the null being that the means are equal between local and foreign applicants. The results suggest that randomization was successful. Table 2 lists callback statistics and shows that the average callback rate is 22%, which is significantly higher than 3–10% found in prior studies (e.g., Bertrand and Mullainathan 2004; Kroft et al. 2013; Lahey 2008; Oreopoulos 2011; Galarza and Yamada 2014). About 72% of employers did not call back any applicants, 13% call back one of the two applicants, and 15% call back both applicants. Descriptive statistics of the treatment variables and randomly assigned resume characteristics are also provided in Table 2. Of 1938 resumes submitted, 50% are assigned foreign work experience, and conditional on being assigned foreign experience, 50% of the resumes are associated with the Singapore labor market experience and 50% with the Malaysia labor market experience. Each of the four resume templates is equally represented. About 52% of resumes are associated with female names, and 50% are associated with an address in a rich neighborhood. The average age of applicants is 27 years, and half of the resumes are assigned application cover letter #1.

Table 1 Randomization tests

Do employers in Myanmar prefer workers who accumulated skills in more advanced countries? Evidence from a field experiment

Abstract

Introduction

Experimental design

Empirical analysis

Results

Conclusions

Notes

References

Acknowledgements

Funding

Competing interests

Ethics approval and consent to participate

Publisher’s Note

Author information

Affiliations

Corresponding author

Additional file

Additional file 1:

Appendices

Appendix 1

Appendix 2

Appendix 3

Rights and permissions

About this article

Cite this article

Keywords