The context​

Toward the end of 2021, more than 28.2 million Vietnamese people aged 15 years and older lost their jobs or otherwise experienced loss of income or changes in work hours as a result of the Covid-19 pandemic and measures to contain it.

Impacts were most significant in labour-intensive industries such as tourism and export-oriented manufacturing industries such as textiles (ILO, 2022).

As was typical in the region, the hardest hit were disadvantaged groups living in remoter provinces or working in the informal labour market. These groups have reduced economic visibility which hindered their ability to receive governmental assistance. As women are overrepresented in these groups, and were subject to additional demands for unpaid care work during the pandemic, female labour force participation declined particularly sharply (ILO, 2021).

In order to quickly restore the country’s economy and labour market to pre-epidemic levels and improve long term resilience, the Government issued Resolution No 06/NQ-CP which called for the development of a flexible, modern, effective, sustainable and integrated labour market. Speed and granularity of workforce data was identified as a core problem. The DoE was tasked with “building an information network and database on the labor market and future skills needs; invest in research and construction, regularly analyze and publish labor market forecasts by industry, occupation and region in the short, medium and long term, serving as a basis for timely and effective construction and implementation of policies on labor - employment, social security, [and] providing labor market information for employees, employers, training and research institutions”.

How it was implemented​

Data ecosystem mapping and policy questions

One of the project's main goals was not just to provide a one-off technical solution for this one instance, but to explore the potential of different data sources to further the Ministry's overall digital transformation. Hence, the partners conducted a data ecosystem mapping to gain a better understanding of their baseline situation and to address data gaps.

At the same time, expert interviews were conducted to identify issues where more granular information for decision-making was needed. This was to ensure that that the project concentrated on data that end-users find relevant and helpful.

After completion of these parallel activities, the results were presented to a workshop of all relevant policy stakeholders. One of the key things revealed in this process was that some ministerial units are already generating high-quality data that colleagues in other units were not aware of. Furthermore, some data processes were duplicated while in other places gaps existed. For example, policymakers highlighted their need for more information on which professions are over- and under-represented in order to better advise vocational schools and universities on training needs.

The LMIS Policy Task Force

Based on these results, the Vietnamese partners identified three priority workstreams. Firstly, the communication between various political institutions should be improved to enhance intra-governmental data sharing. Secondly, the identified data gap on occupational profiles and competency needs should be addressed through a collaboration with private data providers and this approach should serve as a blueprint to tackle other information gaps in the future. Thirdly, MOLISA’s capacities for using evidence in decision-making needed to be increased. This can be done through peer-to-peer exchanges with policymakers from countries with similar challenges and with the advice of national and international experts.

A cross-institutional LMIS Policy Task Force was established in August 2022. The task force consists of representatives of DoE, GSO and the think tank ILSSA. The task force members coordinate the implementation of the workstreams, communicate learnings and decisions internally and provide a platform for exchange with partners from academia, private sector and civil society.

Cross-institutional policy task forces can be an effective method for institutionalising data-informed decision-making.

1. Collaboration between experts from different institutions can lead to the development of common standards for data collection, analysis and sharing, enabling better data utilisation.
‍
2. Combining data from various sources can lead to more comprehensive analyses.

3. Policy task forces can share best practices and enable all stakeholders to benefit from proven methods and approaches.

4. Establishing a task force can increase transparency and accountability in political decision-making by ensuring that decisions are based on reliable and granular information. This can help to strengthen public trust.

Online Job Vacancy Data

The LMIS Task Force explored the potential of various new data sources to fill the identified information gap related to occupational profiles. Inspired by successful approaches in Indonesia, Malaysia, Canada and Germany, the partners decided to explore the potential of online job platform (OJV) data for Vietnam.

OJV data can provide detailed information on what types of jobs are available, what qualifications and experience are required for certain roles, and where exactly these jobs are being offered. In addition, the analysis of anonymised users provides information about which skill profiles are currently available, how many applications have been submitted for certain adverts and what salary expectations candidates have.

However, making OJV data accessible in Vietnam comes with some challenges:

1. Representativeness: The informative value of OJV data is influenced by many factors, such as general job search practices, the size of the informal labour market, population's internet access and the hiring policies of national companies. In Vietnam, there are several online job platforms, e.g., TopCV, CareerBuilder and VietnamWorks, but only about 30% of formal job seekers use these platforms and this 30% are not typical of the whole. However, this number is increasing by about 15% annually and will become more significant and representative over time.
2. Availability: OJV data services such as Lightcast or Textkernel use sophisticated technology such as Natural Language Processing to extract granular and timely labour market information from multiple job platforms and convert these into easy-to-analyze datasets. These services, however, are only available for the most commonly spoken languages globally and cannot yet be used to analyze Vietnamese job sites. Stakeholders had to organise the collection, processing and analysis of data themselves.  
3. Accessibility: The partners identified two approaches: either developing an algorithm that reads open information from multiple platforms and websites simultaneously or entering into a direct data-sharing agreement with individual job platforms to access their data through an API (an Application Programming Interface – a mechanism for two pieces of software to communicate). Both approaches have advantages and disadvantages in terms of representativeness, sustainability of collaboration and legal requirements.

To explore these issues further, a pilot based on an initial data donation from a large job platform in Vietnam was conducted with the collaboration of ILSSA. Through this, technical feasibility was proven to policymakers in DoE.

How can better data contribute to better policy?​

The results of the mapping of the data ecosystem show that many conventional data sources already exist within MOLISA and neighbouring governmental institutions, which can be used in a more systematic and targeted way in decision-making processes. The mapping also revealed where there are gaps and how those gaps could be filled.

While OJV data is not yet fully representative of the Vietnamese labour market and cannot capture the dynamics in the huge informal sector, it already contains relevant information for policymakers. For example, it can help identify which industries are gaining in importance and in which fields employers are struggling to find applicants with the needed skills. Based on this information, the government can take appropriate measures to support growth in these areas, for example by providing the necessary support to universities and vocational schools to provide relevant re-skilling and training.

OJV data can also help monitor the impact of recently adopted policies, something which was shown to be lacking during the pandemic.

By integrating OJV within the broader Labour Market Information System it is possible to cross-check the information with other labour market data. This ensures that the limitations and blind spots of OJV data are well understood, and policymakers develop a clear overall picture of Vietnam’s labour market.

Where do we go from here?

The project team is currently reviewing technical and legal terms to make OJV data accessible. An important part of this process is to better understand the needs of the various beneficiaries of the prototype, both to ensure that the data generated is user-friendly and to design training and capacity building programmes accordingly. At the same time, the prototype looks to ensure future scalability by creating the relevant conditions for a potential integration of other data sources or the opening of the system for new user groups.