Digital Platforms, Data and Artificial Intelligence: New Tools for Developing the Secondary Market for Distressed Assets
As part of the 10th International Conference of the International Public Asset Management Companies Forum (IPAF) in Astana, organized by the Fund jointly with the Asian Development Bank, the third session, “From Transparency to Liquidity: Secondary NPL Markets, Digital Platforms and Data,” was held. The session focused on the development of secondary markets for distressed assets, digital trading infrastructure, and improving data quality.
One of the central topics of the session was creating conditions under which the secondary market for non-performing loans could move from simple availability of information about assets to actual liquidity and sustainable investor participation. Particular attention was paid to the role of digital platforms, data standardization, and new technological solutions in reducing barriers arising in the preparation and execution of NPL transactions.
The keynote presentation of the session was delivered by Mr. Burkhard Heppe, Co-Founder of Accuria (note: Accuria is a platform for credit portfolio management and the application of artificial intelligence technologies, specializing in the valuation, monitoring, reporting, and trading of credit assets). In his presentation, he demonstrated the practical capabilities of modern technologies for developing a transparent and scalable market for distressed assets. Accuria works with data on more than 350 million loans, including performing and non-performing assets, while its solutions have been used to upload and assess approximately 50 million NPLs across 3,500 portfolios. The company has also cooperated with the International Finance Corporation (IFC), the World Bank, and the Asian Development Bank on projects and research related, among other things, to the development of NPL markets and platforms.
The presentation noted that digital infrastructure can address several systemic challenges of the secondary market simultaneously: reducing the information gap between sellers and potential investors, improving the quality of due diligence, strengthening participants’ confidence, and ultimately increasing the efficiency and liquidity of transactions. At the same time, simply publishing information about assets on an electronic platform does not in itself guarantee the emergence of demand or the successful closing of a transaction.
To illustrate this issue, Burkhard Heppe presented the Five Gates concept – the 5 stages, that an investor must pass through sequentially before capital can actually be deployed to acquire a distressed asset. The first stage concerns the visibility of the offering and access to information about a potential transaction. The investor must then meet the requirements for acquiring a particular class of assets, have sufficient data and market benchmarks to assess them, be able to execute the transaction itself effectively, and, finally, understand the mechanisms for subsequent servicing of the asset and exiting the investment.
Therefore, the effectiveness of a digital trading platform should not be assessed solely by the number of assets listed on it. More indicative metrics include the number of qualified investors, their participation at different stages of the transaction, the number of non-binding and binding offers submitted, the number of participants in the bidding process, the time from asset listing to receipt of an offer and closing of the transaction, as well as the share of successfully completed transactions and the repeated participation of sellers and buyers. This approach makes it possible to assess the extent to which the infrastructure actually facilitates the movement of capital rather than merely providing visibility of the offering.
Particular emphasis was placed on data quality and standardization. For effective portfolio assessment, investors need not only basic information about loans but also detailed loan-level data, information on collateral, documentation, recovery history, and actual recoveries. Standardization of such information makes it possible to segment portfolios more quickly, compare them with market benchmarks, and reduce the time required to analyze a potential investment.
A significant part of the presentation was devoted to the application of artificial intelligence and AI agents at various stages of working with distressed assets. In particular, tools were demonstrated for the automated uploading and transformation of the seller’s source data into standardized structures, analysis of legal and financial documentation, matching documents with the Loan Data Tape, preparation of materials for investment and credit committees, and conducting due diligence.
AI agents can also be used in portfolio assessment: to analyze the sufficiency of available data, determine the need for additional reference information, conduct statistical analysis of representativeness, assist with model calibration and interpretation of results, and prepare drafts of valuation memos. At the same time, the presentation specifically emphasized that cash-flow calculations themselves should remain deterministic and reproducible, and that the language model should not perform the financial calculations themselves.
Another area of technology application was work with complex restructuring mechanisms. The approach presented envisages the creation of a unified process in which AI tools help process fragmented financial and legal documents, extract and verify the required data, conduct research, prepare financial models and draft credit opinions. At the same time, specialist involvement is retained at critical stages: the process includes review of the analysis plan, manual adjustment of the financial model, and subsequent incorporation of changes into the prepared materials.
Smart Virtual Data Rooms (VDRs) and AI Document Analysis tools are of particular importance in this process. They make it possible to extract information from documents while preserving the structure of financial tables, classify and match materials, and create more comprehensive datasets for subsequent valuation and due diligence. As a result, technologies can reduce the amount of manual document processing and accelerate the preparation of a portfolio for sale.
One of the key conclusions of the presentation was that developing a liquid secondary NPL market requires a combination of several elements: transparent digital infrastructure, detailed and standardized data, reliable valuation tools, and efficient transaction processes. New AI solutions can significantly accelerate work with large volumes of information and automate individual stages of analysis; however, their effectiveness depends directly on the quality of the underlying data.
As the speaker emphasized, artificial intelligence cannot create facts that are not available; it can only structure and use the information available to the seller. Therefore, digitalization does not replace high-quality preparation of assets for sale but becomes a tool that makes it possible to transform available data into a basis for investment decisions more quickly and efficiently. Ultimately, visibility of assets does not in itself mean that a transaction can be completed: secondary-market liquidity increases when a greater number of investors are able to go through the entire process: from obtaining information about and valuing an asset to acquiring it, servicing it, and subsequently exiting the investment.