Credit: Better Up
This blog shows the statistics and empirical facts behind the likelihood of re-entering medium to low skilled jobs based on how long an individual has been employed or unemployed.
The process of entering and re-entering the workforce is characterised by complicated dynamics and is not well understood by researchers. There exists a substantial body of empirical literature that extensively explore the relationship between unemployment duration and reemployment probabilities, however, their ability to reflect whether results reflect unobserved heterogeneity or a causal relationship is limited because of the use of administrative data. To solve this problem Kroft et al. (2013) conducted resume audits and submitted carefully constructed 12000 fake resumes to roughly 3000 real job postings in each of the 100 metropolitan areas in the United States to investigate whether the duration of unemployment affects the chances to receive a call-back when applying for a job and how it varies with market tightness which is new to the literature.
The primary advantage of using field experiments is that it allows us to estimate the causal relationship between unemployment duration and call-backs, separating the effect of other resume characteristics on callbacks (Ghayad,2013). Secondly, the empirical approach can be tied to the theoretical prediction that employers use observable information available on a resume to form expectations about the worker’s productivity (Fraber et al., 2016). Thirdly, firms don’t know they are being tested, and since IRB does not allow firms to be harmed in any way, they have little incentive to change their behaviour. Finally, even if firms change their behaviour, these audit studies will still be able to reflect market conditions correctly.
It is important to highlight the focus on medium to low-skill jobs: customer service, sales, administrative and clerical. The study has been criticised by Erikkson and Rooth (2014) for not considering high-skill jobs. However, it is important to note since the job requirements for these medium to low-skill jobs are the same across firms, the chances of call-back are higher which can help in identification. Moreover, the skills in these professions will not diminish significantly during the early unemployment spell which will help to understand whether recruiters review the unemployment spell as a loss of skills even under these circumstances (Oberholzer-Gee, 2008).
They sent four resumes to each job posting: two males and two females for customer service and sales jobs, while 4 females resume for administrative/clerical jobs. Moreover, there were two types of resumes low quality and high quality. Low quality resume is assigned the minimum qualifications required while high quality resume exceeded the minimum qualifications. The crucial point to consider in designing resumes individuals were randomly assigned employment status, and the duration of the unemployment ranging from 1 to 36 months (if the individual is unemployed) and it assumes that employment status and unemployment duration is important to employers and that was confirmed by a web survey they conducted. The unemployment spell they considered is larger than previous studies so allowing researchers to estimate the relationship flexibly. The age composition of their sample is mainly young (mid-twenties).
After applying for jobs, they track call-backs from employers, and they didn’t observe the employer hiring stage where models of employer discrimination often feature discrimination. Firms interview multiple candidates but hire one with the lowest unemployment duration endogenously (Fernandez-Blanco and Preugschat. 2018). However, Riach and Rich (2002) suggest that the initial stage accounts for as much as 90% of the discrimination levels measure, implying later steps are unimportant. This point though, may not be as relevant as this is an older paper where changes in views upon discrimination have changed insinuating that these later steps do still pay a key role.
The dependent dummy variable is a callback indicator, and the explanatory variable provides estimates of the mean impact of employed assigned randomly with a probability of employment equal to 0.25 or unemployed. Another explanatory variable provides the mean impact of changes is the log of unemployment duration of 1-36 months if unemployed. A variable accounting for efficiency gains and individual and area characteristics is included.
To examine how duration dependence varies with local labour market conditions they restrict the sample to the unemployed only and a new explanatory variable, proxies for market tightness is included and also interacted with the log duration variable. Proxies for labour market tightness are the unemployment rate, vacancy unemployment ratio and the difference in unemployment rates between 2008 and 2011. Secondly, fixed effects models include MSA fixed effect and another explanatory variable MSA-specific estimate of the unemployment duration effect on call-backs. This is motivated by that there is a relationship between the call-back rate for a newly employed and level of market tightness.
They find a slight increase in unemployment duration (1 log point) reduces the probability of call back by 1.1%, from a mean of 4.7 percentage points. These results are statistically significant. The steepest decline in call-back rates occurs in the first 8 months of the unemployment spell. Secondly, they find that newly unemployed applicants are more likely to receive call-back relative to currently employed this can be interpreted as they preferred because they will be available for immediate start. Moreover, they find that the unemployment duration effect is stronger in relatively tight local labour markets, implying that employers view unemployment more negatively when it is easier to find jobs. This is also consistent with ranking models as an applicant in tight markets is less likely to face competition with a shorter duration. Interestingly, they find after 8-10 months of unemployment call-back rates are the same in weak and strong labour markets. Duration dependence in women is larger than in men (statistically not significant). These results are also consistent with job search behaviour, unemployed worker should lower their reservation wage due to human capital erosion or initial overconfidence and change in search behaviour as individuals get disheartened.
Overall, this study is highly regarded for its ability to account for the effect unemployment has on call-back duration and how it varies with local labour market conditions.
Bibliography
Eriksson, S. and Rooth, D.-O. (2014) ‘Do Employers Use Unemployment as a Sorting Criterion When Hiring? Evidence from a Field Experiment’, The American economic review, 104(3), pp. 1014–1039. doi: 10.1257/aer.104.3.1014.
Farber, H. S., Silverman, D. and von Wachter, T. (2016) ‘Determinants of Callbacks to Job Applications: An Audit Study’, The American economic review, 106(5), pp. 314–318. doi: 10.1257/aer.p20161010.
Fernández-Blanco, J. and Preugschat, E. (2018) ‘On the effects of ranking by unemployment duration’, European economic review, 104, pp. 92–110. doi: 10.1016/j.euroecorev.2018.02.003.
Ghayad, R. (2013). The Jobless Trap Job Market Paper.
Kroft, K., Lange, F. and Notowidigdo, M. J. (2013) ‘DURATION DEPENDENCE AND LABOR MARKET CONDITIONS’, The Quarterly journal of economics, 128(3), pp. 1123–1167. doi: 10.1093/qje/qjt015.
Neumark, D. and Button, P. (2014) ‘Did Age Discrimination Protections Help Older Workers Weather the Great Recession?’, Journal of policy analysis and management, 33(3), pp. 566–601. doi: 10.1002/pam.21762.
Oberholzer-Gee, F. (2008) ‘Nonemployment stigma as rational herding: A field experiment’, Journal of economic behavior & organization, 65(1), pp. 30–40. doi: 10.1016/j.jebo.2004.05.008.
Riach, P. A. and Rich, J. (2002) ‘Field Experiments of Discrimination in the Market Place’, The Economic journal (London), 112(483), pp. F480–F518. doi: 10.1111/1468-0297.00080.
By Momin Rizwan
Guest Writer
Edited by Samuel Golder
Comments