Eneli Kindsiko and Andres Kütt: How €360 discovered the education gap

The state could significantly save money by focusing on the use of existing data instead of commissioning expensive and repetitive studies, write Eneli Kindsiko and Andres Kütt.
Estonia's e-government is often considered a global model, but the harsh reality is that the public sector's weak institutional memory and inability to utilize existing data hinder our progress. Instead of relying on already collected information, new studies are frequently commissioned from scratch, wasting time, resources and money.
This issue is most pronounced in the field of education, where connecting different registries could provide valuable insights into educational disparities and help identify solutions. Unfortunately, the potential of the e-government remains underutilized. How long can we afford the underuse of data and the duplicative expenses on studies?
Weak institutional memory holding back public sector
We are data hoarders. We collect as much as possible, yet we fail to apply it effectively. We underutilize our country's greatest advantage and when questions arise about the state of a specific field, the response is often to commission new studies from scratch.
This tendency likely stems from the weak institutional memory of public sector organizations. When a person leaves, the knowledge of what has already been done in the field often disappears with them and their replacement immediately begins commissioning new studies. As a result, we find ourselves in situations where four studies have already been conducted on the same topic, but no solutions have been reached.
It's not uncommon for research processes to reveal that similar studies on the same subject are simultaneously underway in three different public sector organizations. Even more striking are cases where parallel studies are being conducted within the same organization.
Commissioning studies — especially those based on surveys and interviews — is not inherently bad. However, the state could save significant time and money by first using registry data to map the situation and then focusing subsequent steps on investigating identified problem areas in depth through targeted surveys, interviews or other methods.
As EBS professor and education economist Kaire Põder has also pointed out, registry-based data is more reliable for obtaining trustworthy results than data collected through survey studies.
How was the educational gap found?
The clearest example of the damage caused by failing to integrate registries is what could be called "PISA blindness." In 2023, one of the authors of this article approached Statistics Estonia with a request to link the income of parents to all students in Estonia's general education schools. Importantly, this was not about publishing personal data, as the output data reflected parents' incomes at the level of school averages (distinguishing between mothers' and fathers' incomes).
Since such data integration was new in Estonia, it took a month to agree on data protection requirements, particularly for very small rural schools. The primary restriction was that a school had to have at least 20 students with either a mother or father, which excluded schools with fewer than 20 students from the analysis.
To this layer of information, data from the Ministry of Education and Research on average results from the basic school mathematics exams (from the EHIS database) were added. The results showed a positive correlation between mothers' (and families') median income and students' mathematics exam results.
The educational gap has not silently crept upon us. While PISA press releases over the years have largely confirmed that students' academic outcomes in Estonia are not dependent on their socioeconomic background, Estonia's own data tells a different story. In recent years, nearly a quarter of basic school students have failed their mathematics exams and in some schools, as many as half do.
It is particularly paradoxical that PISA tests are taken by 15-year-olds, the same age group that is required to take Estonia's basic school mathematics exam. Globally, these approximately 6,000 Estonian students rank among the world's best in mathematics, yet domestically, a quarter of their peers fail the subject.
In conclusion, while international surveys appear to provide positive results and portray Estonia in a favorable light, our local (registry-based) data should be the foundation for drawing substantive conclusions about the educational gap.
PISA covers about 6,000 students, but Estonia's registries can provide insights for over 100,000 students.
Rather than commissioning studies in Estonia based on surveys or interviews to map the situation, which can cost €20,000-€50,000 and take six to nine months to complete, integrating registries could provide answers for the entire population in just one day.
In the case of linking parental income, it took about a month to resolve details related to data protection and similar concerns, but the data linking itself was completed in a single workday. Thus, if ministries know how to ask, they could receive answers through registry integration within a day or two. This is the potential of e-government that we should embrace far more boldly.
Cost: €360 (approximately six work hours), which Statistics Estonia spent linking data from different registries.
Data analysis must become simpler
Cross-registry data analytics must become simpler from a technical, organizational and legal standpoint. For critical questions affecting the entire country — not just individual agencies — it must be possible to use existing registry data to find answers.
Changes are necessary across all three dimensions (technical, organizational and legal), as even the best technological solution will remain unused if there is no established workflow to implement it or if navigating a maze of legal requirements proves too complicated.
Our current situation has a clear root cause: we have built our state with a focus on the unity of the state and its information systems, as well as data protection. The first principle ensures that data is not stored in one massive repository but is instead managed and maintained by specific agencies. The second ensures that these agencies are surrounded by relatively high walls of data protection.
Both approaches have served us well, but how do we move forward? How can we adapt our current system to maintain data security and legal certainty while enabling more efficient cross-registry collaboration?
On the technical side, there is no need to invent something entirely new; it is enough to effectively implement the solutions already available. Why hasn't this been done? Because no one has taken responsibility for it. Each agency is responsible for its own data and even Statistics Estonia defines its role as the "reliable home of data." Each organization focuses on the data under its control, but there is no entity responsible for the analytics infrastructure that connects these registries.
In addition to clearly defining responsibility, institutional memory could also be strengthened by providing training, creating standardized processes and documentation and sharing best practices.
From a legal perspective, this is likely the most complex challenge, as compliance with GDPR requirements must be ensured. At the same time, we must ask whether we have sometimes built "walls" around data protection as a precaution. As Kaire Põder aptly points out regarding inconsistencies in data protection: "If we can download student-level data from the OECD website for the PISA study, why should data protection issues be more stringent in Estonia?"
Still, much can be achieved here, for example, by establishing cross-agency rules that allow for the anonymous secondary use of collected data for service provision purposes. At present, even within individual agencies, data protection rules are sometimes interpreted with varying levels of strictness.
By addressing these technical, organizational and legal gaps, Estonia could unlock the full potential of its registry data while maintaining the high standards of security and trustworthiness that have defined its e-government success so far.
In summary
The state could save significant amounts of money by focusing on the use of existing data instead of commissioning expensive and repetitive studies. The potential benefits could likely be even greater.
A recent European Commission report on evidence-based policymaking highlights the need for a more systematic approach to scientific advisory processes. On one hand, the report shares many similarities with the issues outlined above: from cross-sectoral collaboration to a centrally managed ecosystem. However, improving the ability to manage reliable knowledge about the state as a whole could, in itself, lead to a significant leap forward. This would also enhance the practical functioning of the e-government system. Such progress would undoubtedly be highly beneficial for Estonia.
P.S. The 2026 Estonian Human Development Report will focus on education and its broad impact on societal development.
--
Follow ERR News on Facebook and Twitter and never miss an update!
Editor: Marcus Turovski