The collaboration between LinkedIn, Eurostat and CBS started two years ago and has gone well, according to De Mooij. ‘We had many discussions, especially in the initial phase, about how the Dutch education system works. This allowed LinkedIn to formulate the right research questions. When the results became available, we looked at them together with Eurostat while CBS compared the results, where possible, with its own data.’ The collaboration with a private international organisation is unique to CBS. ‘But in terms of the content, we as researchers turned out to have many similarities. It is great that new ideas emerge from such a collaboration.’ Furthermore, CBS and LinkedIn did not exchange any data at the individual level, only aggregated data on groups of recent graduates. That said, there was intensive exchange of knowledge.
Divergent findings on labour participation
In assessing the representativeness of the LinkedIn population, CBS compared the results from LinkedIn’s study on whether graduates had found employment or not with its own publications, based on Dutch education and labour market registrations. What did CBS conclude on the basis of this analysis about the quality of LinkedIn data? De Mooij: ‘That depends on the level of detail chosen. The LinkedIn data accurately reflects overall developments. But if we compare the CBS data and LinkedIn data, we see a structural divergence in the graduates’ labour participation of 7 to 10 percent.’ There are all kinds of explanations for this: ‘People may only list jobs on LinkedIn that match their education. CBS has data about all the jobs held by an individual. We have also observed that LinkedIn is more popular in some industries than in others. That may produce a distorted image as well.’
New information about skills and emigration
CBS does not have specific information on the skills of recent graduates. De Mooij: ‘That is precisely why we found LinkedIn's research questions interesting. Skills data potentially offer many opportunities. You could, for example, compare the listed skills with those that you might expect after a particular study programme.’ As for the question of emigration among recent graduates, CBS is able to say something based on information from Dutch municipal databases. De Mooij: ‘However, LinkedIn’s information may be more complete, since people who only move abroad for a short time do not have to deregister at their place of residence. Here the LinkedIn data could be a valuable supplement to the existing data.’
Digital map
LinkedIn is building the so-called Economic Graph: a digital map of the global economy based on five main pillars: members, companies, jobs, skills and educational institutions. ‘We work in partnership with government agencies to chart and communicate insights on, for example, trending skills and emerging occupations,’ says Mirek Pospisil of LinkedIn. ‘With the aim of preparing employees worldwide for the skills and jobs of the future.’ LinkedIn noticed that there is only limited data available about how recent graduates progress in the labour market and about how their skills develop over time. Pospisil: ‘The New Skills Agenda of the European Commission, launched in 2016, introduces the idea of following recent graduates in order to fill this gap in knowledge. Our collaboration with CBS and Eurostat undergirds the New Skills Agenda.’
Ups and downs
The reason Eurostat attaches importance to LinkedIn's research is twofold. Albrecht Wirthmann from Eurostat explains: ‘Firstly, this research project helps us identify the potential contribution LinkedIn data can make towards statistics production. Secondly, we wanted to find out how using this new data source could generate valuable statistical information on career paths and skills of Dutch based graduates.’ The project had its ups and downs, he recalls: ‘We first had to align expectations and build a good working relationship within the project team. Then a lot of discussions were needed in order to develop the methods to assess the data and the findings while working under privacy constraints. The national statistical system is quite different from the environment in which internet platforms such as LinkedIn operate. The key to this project's success were the support provided by the management of CBS, LinkedIn and Eurostat, and the close collaboration within the project team.’
More research needed
The research report was presented to the European Commission in Brussels on 3 December 2019. Pospisil: ‘The report is the fruit of a unique public-private partnership between official statistics agencies and LinkedIn, in which the expertise of all participating parties is leveraged. With our research we are also addressing a knowledge gap that has been raised by the European Commission. This produced good results: first insights into the skills of graduates, their evolution over time and knowledge about international talent migration.’
Extend analysis
De Mooij emphasizes that it is too early to use LinkedIn data for official statistics. ‘As far as we are concerned, this is only the start,’ she says. ‘We would like to continue with the ongoing research to obtain more detailed figures. There are possible ways, for example by correcting for insufficient representativeness by applying statistical methods.’ Albrecht Wirthmann adds: ‘Eurostat would like to extend the analysis to other EU member states and continue working on the exhaustiveness, representativeness and comparability. Finally, we will discuss how the results may support the initiatives of the European Commission with the policy directorates involved in labour market and education.’
