Measuring Saudi Arabia’s Job Market: The Untapped Potential of Using Job Postings Data to Bridge the Labor Market Information Gap
Across most Arab countries, labor force surveys (LFSs) are published on an annual or semi-annual basis, and even with the few exceptions such as Saudi Arabia that release quarterly labor force surveys, there is often a delay of 5 to 7 months between the time period measured in a survey versus the actual publication time. For instance, in the case of Saudi Arabia, the latest LFS publication was for Q2, 2020, and thus from July 2020 onwards, policymakers have little to no information on the job market’s status.
This contrasts with other countries such as the United States, Japan, and Germany where LFSs are published on monthly-basis and where decision-makers and researchers can track the latest month to month changes in the labor market, and thus the health of the economy overall. While in normal years the need to have frequent LFSs may be still crucial, in a year like 2020 the urgency to access timely information on the labor market is more pertinent than ever.
As a result, in the past few months, I have thought extensively about developing practical methods to address the labor market information gap across Arab countries, which can also satisfy two important conditions. First, I wanted to propose a solution for this issue that goes beyond recommending governments to collect and publish survey data more frequently. This is because asking governments to increase spending and reallocate their limited resources on survey-collection seems to be an unrealistic policy proposal given the state of the world. Second, I wanted to develop a policy solution that if scaled-up, has a relatively low data collection cost compared to conventional methods such as phone or in-person surveys.
A valuable source of high-frequency labor market information that remains unutilized in the Arab world is using web-scrapping techniques to gather and organize job postings from the various sites that constantly update this information. Therefore, I specifically wrote a code that collects data through web-scrapping, on daily job postings in Saudi Arabia as a proof of concept. More specifically, the proposed solution here does not address the information gap on the supply of labor as conventional LFSs are still required for this task, however, I argue that it fully addresses the lack of data on the demand for labor. Using the data collected on job postings in Riyadh, Saudi Arabia, I show how such data can not only be supplemental to survey-data but also provide a more granular data level that is often unmatched through LFSs.
However, when scrapping a job posting site like Bayt, which is one of the most commonly used sites for job postings in the Arab world, one can easily be overwhelmed by the amount of unnecessary information that a policymakers may not need to understand the labor market’s status. Therefore, I will further focus here on the labor demand information that a decision-maker or researcher may be specifically interested in learning.
First, job postings data inform us about the geographic location of the job at the city-level and district-level, which is more granular than LFSs data across the Arab world, which are only published at the national or regional-level. Second, the type of sector and industry where the job was posted. More specifically, for the data I collected on Riyadh between November 8th to November 20th, 2020, there were job postings from 33 different industries and 19 sectors, respectively (See figures 1 and 2 below for more details). For instance, in Figure 1, a decision-maker may easily notice that most of the hiring demand in Riyadh during this time period was driven by firms in tech-related industries, such as Computer and Hardware or IT services companies.
In Figure 2, a policymaker can once again observe that firms in the IT sector have driven most of the labor demand and possibly hiring in the time period of interest, followed by companies in the “Business Services” sector. Furthermore, a researcher can also see in sectors that were more adversely impacted and vulnerable to Covid-19 restrictions such as Construction or Retail, hiring remains sluggish compared to other sectors.
Third, the data I am using here further tells us how many and which job titles were demanded the most by employers in each time period, which informs us about both the skill-level and educational qualifications that employers are searching for. For instance, the top three job titles that were demanded by employers in Riyadh between November 8th to November 20th, 2020 are as follows:
· 5 job postings for “Project Manager.”
· 4 job postings for “Account Manager.”
· 3 job postings for “Business Development Manager.”
Fourth, another piece of useful information that this data provides to policymakers is on the estimated revenue size which again not only offer insights into what type of jobs and skills are demanded in the labor market but also on the characteristics of firms that are driving the labor demand. In Figure 3, one can quickly notice that the demand for labor has been driven mostly by firms with an annual estimated revenue of $10 billion or greater, thus suggesting that such firms are not only the most resilient to this year’s ongoing economic slowdown, but have also been the fastest to recover since the Covid-induced recession began in March, 2020.
Finally, I present in Figure 4 information on job postings per firm size, organized according to the range of number of employees in each firm. Specifically, Figure 4 shows us again that most of the labor demand is being driven large companies (i.e. those with 1000 workers or greater). Interestingly, however, one can also see that small firms (i.e. those with 1 to 50 workers) have also posted more jobs than medium size companies (i.e. those with 51 to 500 workers). This drastic variation in terms of labor demand per firm size, is likely reflective of the fact that economic aid by Saudi Arabian Monetary Fund (SAMA) was sufficient for small-sized firms and that large firms were generally resilient to the recession, however, for medium-sized firms with 51 to 500 workers the economic damage caused by the pandemic remains in place and substantial.
In conclusion, I think that in a developing country’s context where reliable labor market information has a long time lag, there is an even greater case to employ data from job posting sites as a proxy measure for the economy’s health overall, and the job market’s in particular. Additionally, while the insights generated from job postings are by no means an alternative for conventional LFSs, I strongly believe that mobilizing such high-frequency information as a supplement to survey-data will provide a unique and more granular view into local labor markets across Arab countries.
