Labor Markets as Complex Networks

I am attempting, with classmates Grant Lovellette and Guillaume Liegey, to characterize labor markets as complex networks where nodes are different occupations and links are the sets of skills and characteristics required to be employed in each of them. The idea is a parallel to the work of Hausmann, Hidalgo and Klinger in developing the "Product Space", where they develop a "bipartite network in which countries are connected to the products they export". In this case, we are building a bipartite network that connects workers or categories of workers to the occupations they are employed in.

Applications

Characterizing the labor market as a network could allow for a better understanding of the skills and competencies a certain workforce possesses and how those skills and competencies match those demanded by employers in the market. Conceptually, this idea could be applied to different categorizations of workforce, from the graduates of a particular school to the economically active population of an entire country. Imagine, for example if the Dean of the Harvard Kennedy School asked himself whether the curriculum being offered by the MPP program is equipping its graduates to compete for the jobs that the labor market has to offer. A representation of the labor market as a network could help him understand what regions of the "Job Space" his graduates are occupying and what skills they would need to be equipped with in order to populate better connected or more central regions of the network. On the other hand, a World Bank team advising a national government could use the characterization of the labor market as a network to better understand whether the education system or a job training program is providing the labor force with the skills that will allow them to insert themselves in the formal labor market. Furthermore, an analyst could derive important lessons from the evolution of the job space over time, such as the effect of trade or financial policy changes on informality or the types of employment trajectories that different groups of people follow over their lives.

Scope and Data

The scope of the project described above is big enough to occupy an entire research agenda for several years, but our aim is to begin preliminary work on the topic. Our data comes from quarterly employment surveys conducted by the Mexican National Statistics Office. In the most basic form of the Job Space, we will create a bipartite network connecting the occupations codes with the career codes reported for each worker in the survey. We will then project this bipartite network to the space of occupations and therefore each occupation code in the Mexican catalog of occupations (CMO) will become a node in the Job Space. Occupations will be then connected to each other if the people who reported being employed in them share the same career code. A further iteration will be to use the North American Classification of Industries, also reported in the survey to explore what region of the North American Job Space Mexican workers occupy. In the future, researchers interested in exploring this further could create web crawlers to mine data from sites such as Monster.com or Devex.com to create more disaggregated versions of the Job Space.