Matchings in the Job Market
One example of a matching market that particularly affects college students is the internship/job hunt. With recruiting season in full force, some of the largest tech, finance, and consulting companies will receive thousands upon thousands of applications, but they can only accept so many people. This makes for an interesting matching problem with multiple levels, which will be described below.
We can construct this matching market with students on one side, and firms on the other. Because each firm can hire multiple students, we add n nodes for a company, where n equals the number of students the company can hire. There are edges between a student and all of the firm’s nodes if and only if a firm has extended an offer to the student. If this has happened, it means both that the student submitted an application to the company and is interested in working there, and also that the firm is interested in hiring the student, which they express by offering the student a job. We can then imagine this matching market as “unbalanced” for the students: it is likely that some students will receive many offers, while other students may receive very little, or maybe no offers, in which case those nodes will have no edges coming out of it. If this is the case, then it is already impossible for a perfect matching to exist, as there are simply no edges coming out of some of the nodes. The lack of a perfect matching is also reflected in the job market today, with many companies having open positions year round, as well as a non trivial unemployment rate that varies from month to month.
This matching market analogy to the job market can be extended further. Students who receive multiple offers will undoubtedly have a preference as to which company they would like to work at. In this case, we can add the preferences to the market. We can assume that employers have no preference (or the same preference) for the students they extended offers to, as they have little to no control over who accepts/rejects their offers. Adding in preferences, we can then run the Gale Shapley algorithm on this imaginary graph to prefer students, in order to give students their best possible choice while still creating a matching, and since the companies’ preferences are all the same.
The problem of job matching has existed for many years, and in recent times companies such as LinkedIn and Indeed have sought to help relieve the problem. The article below presents an interesting statistic: 40 percent of companies polled “reported lack of skills as the primary reason for job vacancies”, while “37 percent of job-seekers in a LinkedIn study said they were underutilizing their skills in their current positions”. This implies that the matchings created by the job search process today have inefficiencies and are not close to optimal. However, these companies’ on the job market has had mixed consequences: on one hand, they allow applicants to find positions that fit their skills and interests more easily, but also make it easier to apply so that each position receives many more applications, making it harder to find the best potential employees. In recent years, companies such as Google and Indeed have tried to use machine learning and natural language processing to help applicants find jobs that better suit them and reduce the rate of labor mismatch. According to McKinsey, these technological advances could increase the global GDP by 2% within the next 10 years. Whether or not these new technologies will be effective remains to be seen, but they are exciting new prospects that could help more people get jobs that fit them.
Article: https://equitablegrowth.org/how-job-matching-technologies-can-build-a-fairer-and-more-efficient-u-s-labor-market/