LIRNEasia has recently secured a grant from the Asian Development Bank to carry out research on the potential for using online job portals for labour markets in 12 countries of the Asia Pacific region over the coming months. This builds on our future of work portfolio and prior work, where we have developed a scalable computational mechanism to scan online Sri Lankan job boards, using natural language processing (NLP) techniques to extract skill sets from job descriptions and relay findings in explainable human-readable formats. This prior work was funded by the International Development Research Centre (IDRC) of Canada.

The traditional tools of understanding job markets include skills surveys, qualitative studies and the  manual analysis of job vacancies. However, these methods are expensive, time-consuming and often  incapable of capturing the complexity, the variability and the pace of change of labour markets at a  level of granularity that is useful for policy makers and other seekers of such information. It is in this  context that online job portals have emerged as a highly promising data source that can bridge this  gap of information. 

Online job portals offer near real-time information on the current skills demanded by employers,  while also enabling comparisons across time and a wide range of occupations, industries and  geographies. They also allow for the early detection of emerging labour market trends, providing job  seekers, employers, and policymakers with a forward-looking analytical tool. Further, job portals  with advanced functionality often contain data about the job seekers and their behaviour on the  platforms, which can provide additional insights into the supply of skills, preferences of job seekers  and job searching patterns on these job portals. 

However, there are several challenges that need to be overcome before online job portals can be  used to draw reliable conclusions on labour markets. The principal concern with using online job  portal data for economic analysis is that their data are not representative of the full job market. Not  all vacancies are advertised, and even among the advertised, there are differences in coverage from  one vendor to another based on their target market segment, language, approach used to collect  ads, etc. The second important challenge is data quality. Given that the data generated on online job  portals aren’t collected with research objectives in mind, there are no common standards for  vacancy formats, schemas and job classifications. Even within the same country, there can be  significant differences in the nature, the amount, and the quality of data captured by different job  portals. Thirdly, the legal and ethical frameworks for utilising job portal data are not always  clearly established. This concern is even more pressing when data about individuals and their  behaviour on job portals are used for analysis.  

Due to the recency of this field and differences in country contexts and  technical features of job portals, there are no standardised ways of conducting job portal analysis.  Researchers studying online job portals employ a variety of techniques drawing from different  disciplines including, statistics, econometrics and computer science. Furthermore, much of the existing work in this area is based on relatively more developed online job markets of the Global North, where digital access and literacy is at a higher level than the Global South, little is known about the latter in this context.

The current ADB-funded work will extend this work through a scoping study and a case study. The scoping study will seek to understand the role and the drivers of online job portals in the context of domestic labour markets, and their technical attributes. Further, different methodological approaches available for analysing online job portal data for a variety of policy and research questions will be explored; this will include different techniques available for overcoming the inherent limitations of this type of data. The literature review, conducted by LIRNEasia, which forms the basis of this work can be accessed here. 

The second part of the project will be a country case study for establishing a proof of concept, i.e., Sri Lanka. Specifically, for a selection of job categories covered by TopJobs Sri Lanka (topjobs.lk), the project will develop a data extraction pipeline which enables the translation of vacancies found in image form into a corpus of text ready for analysis, and subsequently conduct analysis around a major event/disaster of interest.  

The countries covered in this project are: Sri Lanka, India, Pakistan, Nepal, Bangladesh, Bhutan, Hong Kong, Singapore, Thailand, Philippines, Indonesia and Fiji. It is expected that the project will be completed by the end of August 2022.