From unstructured to structured product data

I have always been intrigued by the word “disclosure.” Any standard dictionary defines “disclosure” as the act of making something known. Interestingly, that definition carries a subtle connotation: It implies something that was obscured has become visible.

The asset management industry has fully embraced the act of disclosure for a long time. The European Commission has even included the word in its flagship regulation to prevent customers being misled around the ESG subject[1], ensuring that all that should be disclosed will be disclosed.

The fund prospectus is without doubt the most exhaustive document issued by asset managers to describe their product offering. Companies seeking investments from retail or institutional clients are expected to disclose this information in an offering document. There are myriad regulations that mandate the content of the prospectus depending on the nature of the financial instrument.

Take a UCITS fund: The Board of the Fund is expected to describe the product features of each investment strategy in detail in the prospectus. This is not a “one-and-only” act but needs to be repeated each time new products are added, existing product features change, or a new category of content is prescribed through regulation, e.g., SFDR.

This requires continuous diligence on behalf of the promoter to update the prospectus and reflect the evolving change of its product offering. Different internal and external stakeholders need to be coordinated until the document can enter the approval process at the host regulator. The approved prospectus is then shared with the custodian, fund administrator, transfer agency and other dealing platforms allowing them to update their core systems. This should surely happen automatically—one thinks. But no, far from it. Diligent back-office staff comb scrupulously through this voluminous fund bible to locate the relevant changes, frequently taking significant time to triangulate data in order to find that one golden source. Yes, this is common practice.

Imagine the benefits emerging from a digitalized, data-driven model—not only for operational departments but for the asset management industry in general.

But what does this mean, and more importantly, where should we start?

A logical beginning would be to consider the prospectus for what it is. It is true that the genesis of the document is the direct consequence of regulation, however, the overarching purpose is to explain the investment strategy, associated risks, and commercials. Meaning is made explicit, and relationships are being disclosed. Who is the investment manager? Who is the custodian? Who is the paying agent? What is the nature of the company? That combination of meaning and relationships should allow the reader to understand the product. If we reduce it to its bare essence the purpose of the prospectus is to make the acquisition of knowledge possible albeit in a very specific context.

Of course, the information contained in the prospectus is not exhaustive, but from small beginnings come great things—so the proverb goes. SFDR for instance clearly contains a product angle and the specific disclosure is at fund level. Once more, we will be witnessing the creation and distribution of additional disclosure in an unstructured form.

When I refer to the structured to unstructured dichotomy, I see two worlds in opposition. Firmly standing in the unstructured corner are documents or Excel files. Humans can read and understand them, but computers struggle to make sense of the way humans have described investment risks differently even in the same document. Here we are operating in the realm of conceptual inconsistencies. I am touching again on a familiar subject of mine: the lack of a common fund ontology inside and across asset management firms.

The structured world, by contrast, is the world where, for example, fund product data is clearly defined in terms of meaning and relationships, and where systems take the lifecycle and temporal nature of that data into account. That is where technology can help.

Think of any prospectus’s paragraph on their dividend policies, in most cases containing a plethora of different data nominators all of which will be assigned value two to three times a year. These values are often recorded in separate Excel files which rarely record or show the temporal evolution of that information. In a perfect world data would be shared between the rather static prospectus and a data management system which would allow for the automatic adjustment of e.g. the threshold of dividend payments based on the performance, historical pay-outs above threshold limit and re-investment into distribution shares, in which currency and why. This not only gives the investor a much deeper understanding of the fund but adds valuable insight into the relationships that exist between current thresholds and changes over time—ultimately helping to answer the ominous “what’s next” based on the known “what was.”

The challenge, however, is bigger than this. Automatically extracting such static data from a prospectus into another database to ensure usability throughout an organization and beyond is an onerous task heavily dependent on further advances within AI and machine learning. Some companies have developed solutions to do just that, but still it seems that this is approaching the problem in the wrong way. The goal should be to possess one golden data source of ultimate “truth” throughout an organization, which itself would feed into and drive the prospectus.

That in essence means using the prospectus not as the golden data—as that only holds true for a given point in time—but change the paradigm of the entire industry (and even beyond) allowing professionals to wrestle back control of the infamous prospectus and reclaim their rightful prerogative of control over the life of this document and the gold mine of product information that lives in and on the pages.

Internal and external counsel often control that container of product information, largely oblivious to the fact that this data is crucial for the adequate operational support required for the global distribution of investment products. Allowing product information to be accessible throughout the organization not only brings meaning beyond the static at-one-point-in-time prospectus, but also brings value and operational efficiency to all stakeholders.

In summary, then: Why do not we tackle the requirements of disclosure at the more fundamental level of the data? If we take a more structured approach to data management, where the “ultimate golden” truth lies not in a Word document but in bespoke data-focused software, we will ultimately show the meaning and the relationships of the fund product data.

For stakeholders, what was previously hidden and obscured will now be made visible.

Disclosure will have come full circle.

[1] SFDR - Sustainable Finance Disclosure Regulation