Nortal HQ, July 5, 2021
Although one of the main fears about Artificial Intelligence is its possible negative effect on employment, this story is about AI helping people get back to work faster.
When the COVID pandemic started in 2020, it was soon apparent to governments worldwide that the economy and employment were under serious fire by minimizing the hits on healthcare. That year, 96 thousand people in Estonia registered as unemployed, the highest unemployment rate (6.8%) since recovery from the 2008 global economic crisis.
Imagine the 96 thousand unemployed, each with different backgrounds, obstacles, and strengths in finding a new job. The 350 consultants at the Estonian Unemployment Insurance Fund (EUIF) must inspect and analyze countless information to create a suitable action plan for each individual for a successful return to the labor market. This process is time-consuming, but it is crucial to act fast for a person to rejoin the workforce. Because in addition to the unemployed person’s arising financial difficulties, the longer a person is unemployed, the less likely they are to return to work. And for full disclosure, long-term unemployment is also expensive for the state.
And so, Nortal, in cooperation with the Center for IT Impact Studies (CITIS) and Resta, developed a decision-support tool raising the quality of the service provided to the unemployed and increasing the efficiency in organizational processes.
Two years ago, the Estonian government felt that AI adoption in the public sector needed a boost, so potential problems that could be solved through AI solutions were mapped out. EUIF, as a data-driven organization, started its own AI project coincidentally at the same time, and in October 2020, the EUIF decision-support tool for consultants, OTT, went live.
The decision-support system OTT (also a boys’ name in Estonia, think rally world champion Ott Tänak) is a tool that helps the Unemployment Insurance Fund’s consultants quickly and effectively collect and analyze all the data about an unemployed person (i.e., their client), takes into consideration the temperature of the economy, and predicts the chances of them getting a new job.
OTT applies AI, specifically the random forest machine learning model, trained and tested based on the last five years’ unemployment data. Using the trained model and 60 different attributes and indicators, each unemployed person is evaluated, and their chances of finding a new job is calculated. The attributes are both about the person, e.g., their education, previous job experience, right to benefits, health restrictions, and about the labor market, e.g., the number and type of available positions in different regions and the number of newly unemployed people.
The statistical engine of the decision-support tool is based on the R freeware. It incorporates data collection from the data warehouse, preprocessing of the data, evaluating the models, calculating the risk scores based on this with the help of machine learning and importing the updated results back to the data warehouse from where the results are made available to the consultants.
The model calculates the prognosis of moving into employment for every newly registered person unemployed for 35 days. It also calculates the probability of becoming unemployed again within a year, and the factors affecting these probabilities are presented. The consultants can offer suitable labor market services based upon the factors. The consultant’s interactive dashboard displays the results.
In this way, OTT shows the consultant where to put their effort and priorities – if a person is at low risk, they can most likely find a new job independently and need little assistance. However, a person with a higher risk of long-time unemployment needs more attention, help, and a comprehensive plan to get back to the labor market. So, it is possible to provide a better service for the unemployed without raising the costs for the organization.
Because the department heads and managers also see the overview of the consultants’ client bases, it enables them to distribute the workload between consultants more evenly and support the consultants who have more difficult client bases, thus making the organizational processes more efficient.
Consultants who act as mentors for the newly recruited appreciate OTT as a tool helping the inexperienced consultants assess their clients’ situation and the extent of help needed, notice their clients’ obstacles of returning into employment and act upon it.
“Thanks to the decision-support tool, our consultants can instantly decide which actions to take next, for example, whether to help the person improve their computer skills or re-evaluate their work capacity,” said Karina Leinuste from EUIF’s department for job seeker and employer services. “OTT has been in use for less than a year, but we can already say that we are very satisfied with the tool. Now we can make further developments based on practical user experience.”
The decision-support tool also helps the Unemployment Insurance Fund better understand the labor market’s ongoing processes and develop better labor policies.
Preventing long-term unemployment is a crucial challenge for the government, in part due to its costliness. Center of IT Impact Studies (CITIS) has calculated that, even if the decision-support tool reduces average unemployment by one day only, Estonia already saves up to about 3.8% of its annual expenditures each year.
Gartner’s list of top 10 government technology trends for 2021 highlights operationalized analytics, which is the strategic and systematic adoption of data-driven technologies – such as artificial intelligence (AI), machine learning, and advanced analytics – at each stage of government activity to improve the efficiency, effectiveness, and consistency of decision-making. Gartner predicts that by 2024, 60% of government AI and data analytics investments aim to impact real-time operational decisions and outcomes directly.
However, EY’s survey Artificial Intelligence in the Public Sector: European Outlook for 2020 and Beyond showed that while many local, regional, and national governments recognize the potential of AI, only 4% of surveyed public organizations have scaled it to transform their organization. As a result, just 10% of respondent organizations are using AI to solve complex problems, and 9% are using it to significantly change ways of working. And only 12% were using it to create significant value for external stakeholders such as citizens and businesses.
The OTT project proved that modern machine-learning models could be successfully implemented in public sector work processes and IT solutions to improve the organization’s service and efficiency, make work more pleasant, and save money.
The decision-support tool project also won a special prize at Estonia’s government digital services competition for best use of data. The jury noted that OTT is an excellent example of how e-Estonia’s infrastructure (i.e., the X-Road, electronic identity, and registries) creates an opportunity to use the data so that it generates added value in everyday decision-making, be it on a personal, organizational, or even country level.
Nortal has been a strategic partner for the Estonian Unemployment Insurance Fund (EUIF) for more than 10 years. If you would like to know more, contact the EUIF Project Manager Vitali Korniltsev.
Read more about how Nortal helps governments provide seamless services for its citizens, which increases social inclusion and justice by eliminating red tape and making resources readily available to those entitled to them.