Search costs and the determinants of job search

Highlights

• I look at how qualifications and perceptions predict search at a job fair.

• A randomly assigned job-search subsidy attracts less-qualified workers.

• The subsidy does not directly lead to an increase in offers.

• The subsidy leads those who are uncertain about their prospects to apply.

• The perceived likelihood of job finding is an important predictor of search.

Abstract

This paper examines how individuals select into job search in terms of their individual qualifications and perceptions and measures how recruiting additional applicants with a modest job-search subsidy affects selection. I use experimental evidence to examine individuals’ decisions to attend and participate in a job fair. Thirteen percent of invited but unsubsidized respondents attend the job fair, and they are positively selected from the overall distribution of respondents. While the subsidy attracts those who are less qualified and less confident in their ability to find work abroad, the least qualified do not search intensively. Although the subsidy does not lead to any additional offers, it induces individuals with a high degree of uncertainty about their likelihood of job-finding to apply with recruitment agencies. These results demonstrate the importance of imperfect information about the returns to search and highlight how reducing search costs can increase search effort among those most uncertain about their prospects.

Introduction

Labor market frictions generated by high search costs can lead to higher rates of non-participation and worse-quality matches between firms and workers (MacMinn, 1980, McCall, 1970). These costs, and their impacts, are often heterogeneous across populations, affecting those who are geographically remote or credit constrained in particular (Abebe et al., 2020). However, search costs are not necessarily harmful; they could help screen out unqualified or uninterested candidates, as Kuhn and Skuterud (2004) and Hadass (2004) find in internet job search and as Alatas et al. (2016) find in the realm of social protection programs. Understanding the empirical correlates of job-search decisions is important in order to understand the potentially heterogeneous nature of search costs and labor market frictions. Additionally, understanding who responds to reduced search costs, and how that affects program effectiveness more broadly, has important implications for determining optimal recruitment strategies by firms, as well as for labor market program design (Heckman and Smith, 2004)

This paper examines the determinants of job-search decisions and the role of search costs on selection into search. I ask, how do individuals’ backgrounds and perceptions affect their job-search decisions, and how does incentivizing search affect these determinants? I consider the decision to look for work abroad in the rural Philippines, where the overseas labor market is large but job search is costly. I combine survey data from 860 individuals ages 20–35 with a novel dataset of overseas labor demand to measure how the perceived benefits and costs to search, based on individuals’ qualifications, labor-market perceptions, and past exposure to the overseas labor market, affect their decisions to attend and apply for work at a nearby job fair for overseas work. Job and career fairs are commonly used by employers across both developed and developing countries (Chang, 2009, Abebe et al., 2018), and they are a popular means of recruitment in the Philippines for both domestic and overseas positions (Esguerra et al., 2001).

I experimentally reduce search costs by offering individuals from randomly selected neighborhoods a modest incentive to attend this nearby job fair. Specifically, they receive a gift certificate to a local fast-food restaurant conditional on job-fair attendance. The randomized incentive to attend a job fair reduces search costs relative to respondents’ outside option by offsetting the time and transportation costs of attending. I show that the cash value of the voucher is greater than the estimated time costs of traveling to the fair for most participants.¹ Although its value is not perfectly fungible, it will, at a minimum, offset the cost of one meal for approximately four people. Because all respondents are invited to the fair, it should not play an informational role. The incentive is similar in spirit to door prizes or giveaways that employers or coordinating organizations commonly used to entice job-seekers to attend job and career fairs.

The survey team hosted this fair in partnership with the municipality, and it attracted more than 750 participants, roughly 30 percent of whom were survey respondents. Enumerators tracked individuals’ attendance and participation at the job fair, including interviews with employers and their outcomes, and I matched this administrative data with survey responses to compare the characteristics of those who search with and without the subsidy and to compare how those who stay and participate in the fair differ from those who leave immediately after retrieving their incentive.

Under standard job-search models, individuals will look for work if the expected benefits exceed the expected costs. Those with higher expected benefits from working abroad are more likely to attend the fair, which could include individuals who are more qualified for overseas work or who are more likely to successfully deploy overseas conditional on receiving a job offer. However, more-qualified workers may also have higher opportunity costs of searching and migrating, inducing negative selection on qualifications and education levels. Ultimately, the distribution of underlying relative returns to search, which reflect qualification levels, individuals’ outside options, and the costs of search, will determine the overall pattern of selection (Borjas, 1987, Roy, 1951). The presence of imperfect information further complicates selection into job search if individuals’ perceived returns to searching do not reflect their actual returns. Individuals may over- or underestimate their own qualifications, or they may be highly uncertain about their prospects (Diagne, 2010, Falk, Huffman, Sunde, 2006a, Falk, Huffman, Sunde, 2006b). Consequently, labor market perceptions may directly play an important role in the decision to search for work.

I expect that incentivizing attendance should bring more people to the job fair, but heterogeneous search costs mean that incentives also may affect selection into search, with the net impact ultimately depending on the relationship between search costs and ability (Abebe et al., 2020). Additionally, the impact on actual applications is potentially ambiguous, as it also will depend on the costs and benefits of applying conditional on attendance.

Among my study sample, 13 percent attend a job fair for overseas work. Of those, 79 percent apply and roughly 50 percent are invited for a final interview. These individuals are positively selected not only in terms of their qualifications for work, but also in their perceived likelihood of job-finding. However, perceived wages are not an important predictor of attendance or participation. A randomly assigned incentive dramatically increases attendance regardless of underlying qualifications, but the least qualified leave immediately after receiving their incentive, preventing potential crowd-out during the fair. Only those who are highly uncertain about their overseas job prospects stay to apply for work, indicating that the marginal applicant may use job search as a tool to learn about his or her own labor market prospects. In aggregate, however, the incentive does not increase the number of final interview offers.

This paper uses data from the same experiment described in Beam (2016), although the two papers differ in their temporal and topical focus. Beam (2016) examines the impact of job-fair attendance on employment outcomes and labor-market perceptions in the months following the fair, using randomized voucher assignment to instrument for attendance. That paper finds that while attending a job-fair does not lead to overseas migration, it does affect how individuals look for work. Specifically, attendance increases the likelihood of looking for work outside the region and being employed in the formal sector in the months following the fair, indicating that job fairs may help job-seekers learn about their labor market prospects. It also discusses the impact of providing information about overseas wages and about qualifications for overseas work, which affect individuals’ labor market perceptions but not their labor market outcomes.

In contrast, this paper unpacks that initial decision of whether to attend a job fair. Specifically, I focus on examining selection into attendance and participation and how reducing search costs with an incentive affects that selection into search. I also expand the set of predictors beyond standard characteristics like age and education by including labor-market perceptions, estimated search costs, and a proxy measure of qualifications for overseas work.

This paper joins a growing literature on the causal impact of reducing job-search frictions. Transportation subsidies increase the number of job matches (Phillips, 2014, Abebe et al., 2018), and Abebe et al. (2020) find that incentivizing job applications leads to higher employment, particularly from those more likely to be credit constrained or face high search costs. Conversely, subsidizing job search in South Africa led to no change in employment or wages (Banerjee and Sequeira, 2020), and a job-matching program in Jordan effectively yielded no matches (Groh et al., 2015). In the more general realm of program participation, Duflo and Saez (2003) find that incentivizing employee attendance at a retirement savings plan enrollment fair yields large increases in attendance, but those induced to attend are no more likely to enroll.

Additionally, this paper contributes to the literature on subjective expectations about labor-market prospects. These expectations play an important role in job-search and migration decisions (Diamond, 1982, Harris and Todaro, 1970, Mortensen, Pissarides, 1994, Sjaastad, 1962). Biases in expectations have been documented across multiple contexts and tend to be particularly strong in developing countries, where labor market frictions can be high (Arcidiacono, Hotz, Kang, 2012, Attanasio and Kaufmann, 2014, Jensen, 2010, Nguyen, 2008, Banerjee, Sequeira). I find that the perceived likelihood of job-finding, as opposed to wage expectations, is particularly important in the decision to look for work. This complements McKenzie et al. (2013), who find that (male) potential Tongan migrants underestimate the likelihood of employment and wages they would earn, and Baseler (2020), who finds that families underestimate domestic migrant earnings. However, unlike McKenzie et al. (2013), who find that expected earnings are associated with higher migration application rates, I find that only the perceived likelihood of job-finding matters for the decision to search.²