8-Hour Monetary Policy Analysis Now Takes 10 Minutes... NPS Also Speeds Up AI Transformation

First AI Agent-Based Analysis System at a Public Institution
Plans for Gradual Expansion, Including Automated Corporate Earnings Analysis

Fund managers at the National Pension Service (NPS) will be able to use artificial intelligence (AI) to complete analyses of major economies' monetary policies in about 10 minutes, a task that previously took around eight hours. Analysis of corporate earnings, which required several fund managers to spend about 180 hours each quarter, will also be automated.

On October 13, the NPS announced that it had become the first public institution to establish a data analysis system using AI agents. The system can automatically analyze and summarize major economies' monetary policies and corporate earnings, substantially reducing the time required for related work.

Over approximately four months, from June through September, the NPS conducted a proof of concept (PoC) for a "cloud-based research AI agent." A proof of concept is a stage in which technology is tested to determine whether it is suitable for practical use before full-scale adoption. Under the new system, AI agents handle everything from gathering and summarizing data to making comparisons and preparing reports.

On October 16, the NPS will hold a results-sharing event with Amazon Web Services (AWS) in Seoul to discuss the findings and ways to apply the technology to fund management.

AI to Write Monetary Policy Reports and Automate 180 Hours of Quarterly Earnings Analysis

The NPS pursued three projects: analyzing the monetary policies of major central banks, analyzing corporate earnings commentary, and conducting integrated research on the stock market. It has completed proof-of-concept testing for the first two projects.

For monetary policy analysis, AI automatically gathers and summarizes monetary policy decisions and press conference materials from the U.S. Federal Reserve (Fed), the European Central Bank (ECB), the Bank of Japan (BOJ), and other central banks. It analyzes how policy stances and key wording have changed compared with previous announcements, then prepares a report.

For corporate earnings analysis, AI agents automatically collect earnings commentary for individual companies. Earnings commentary consists of explanations and outlooks provided by company management when reporting results. The AI compares quarterly changes and differences in outlook among companies, then summarizes expected future results and projections.

The NPS said, "We can substantially reduce the time spent on repetitive analysis that previously required several fund managers to devote about 180 hours each quarter."

The proof of concept also used only public data. An NPS official said, "By conducting the project without transmitting internal investment management data or nonpublic information externally, we minimized security risks."

Expanding AI Use to Stock Market Analysis and Tax Reviews

Building on this project, the NPS will continue work on integrated stock market research and analysis. It also plans to identify and gradually expand areas where AI can be used, including tax and accounting reviews of overseas alternative investment assets, such as real estate and infrastructure, apart from stocks and bonds.

Kim Sung-joo, chairman of the National Pension Service, said, "This proof of concept confirmed the potential for using AI to improve the efficiency of research work in fund management. We will continue to gradually expand the areas where AI is used, creating an environment in which fund managers can focus on their core investment responsibilities and enhancing the expertise and efficiency of fund management."

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