Bank of Korea and Statistical Society Hold Joint Forum: Sharing Research on AI Applications in Economic Statistics

Kwon Minsoo: "Modern Statistics Have Advanced Steadily Through Trial and Error"

The Bank of Korea held a joint forum with the Korean Statistical Society on September 11 under the theme “AI, Data, and Economic Statistics: Changing Environments and New Approaches.”

Korea Development Bank Building

The event brought together statistical analysis experts and practitioners from academia, related organizations, and the Bank of Korea to present research findings and share relevant knowledge. During the forum, the two organizations planned to share research results on new statistical approaches and the use of AI in response to advances in AI technology and changes in the data environment, as well as to discuss future tasks from multiple perspectives.

In his welcoming remarks, Kwon Minsoo, Deputy Governor of the Bank of Korea, stated, “With the rapid development of AI technology and changes in the data environment, the production and analysis of statistics are also undergoing significant transformation. To quickly capture economic phenomena, it has become increasingly important to utilize new types of data and to extract meaningful information from vast data sets.” He added, “How to maintain statistical reliability in environments where the distribution and structure of data are changing is emerging as a key challenge.” Kwon explained that the forum’s theme was determined as a result of these considerations.

Kwon further commented, “Looking back, modern statistics have steadily advanced by adding new theories to existing methodologies and integrating real-world experience, sometimes through trial and error. I am confident that today’s forum, where theoretical perspectives from academia meet practical experiences from the field, will be an invaluable opportunity to learn from one another and elevate the level of our statistical practice.”

The forum, which continued through the afternoon, covered three sessions: keynote lectures, topic presentations, and discussions.

In Session 1, Minkyoo Park, Professor of Statistics at Korea University, examined the potential for machine learning techniques—based on high-frequency data such as card sales and search trends—to be applied as official statistics, and proposed ways to introduce these methods as complementary to the traditional statistical paradigm.

Session 2 addressed the issue of declining statistical model performance caused by discrepancies between microdata and macro statistics, and introduced methods to ensure both statistical precision and model reliability by combining statistical validation techniques with AI self-evolving algorithms. Seho Park, Professor of Industrial and Data Engineering at Hongik University, reinterpreted the differences between microdata and macro statistics from the perspective of distribution changes, and presented a cyclical framework that secures estimation stability from change detection through calibration. Kyunggu Song, Professor of Applied Statistics at Yonsei University, proposed directions for building “self-evolving AI agents” and introduced statistical methods to control hallucinations and uncertainty in AI-generated responses.

Session 3 featured the Bank of Korea’s latest statistical research achievements. Sojung Kim, Manager of the Bank’s Statistics Research Team, assessed the potential for utilizing AI agents in statistical development and analysis, and highlighted the need to identify and control agent bias. In addition, Jin Park, Manager of the Bank’s Household Distribution and National Income Team, presented improved methods for compiling the household distribution account in order to more precisely grasp distribution situations by household group.

This content was produced with the assistance of AI translation services.