Usable beneficial ownership data

Overview

Introduction

For beneficial ownership data to be useful it must be usable. That is, it must be easy enough for people to access, interpret, connect, and rely on. Only then can data be used to create impact, whether that means promoting effective taxation, improving public procurement integrity, or tackling financial crime.

Open Ownership recently undertook a major piece of research to understand the experiences of those who use beneficial ownership data. We found that most people have fairly similar needs, whether they work in government, financial institutions, commercial data services, or journalism. We also found that, often, these needs are not being adequately met by existing beneficial ownership registers.

Our research aimed to better understand people’s work with data in order to inform more user-centred, impactful reforms for beneficial ownership transparency. Embedded within our findings were a number of clear functional requirements for those building, assessing, and seeking to improve beneficial ownership registers.

We have revisited the research with this in mind, identifying 22 features that we believe make beneficial ownership data more usable, especially when it comes to interpreting it and connecting it to other data. This short working paper outlines 1) how we reached this list of features, and 2) the features themselves.

Open Ownership has started using this framework to review and offer advice on implementation of beneficial ownership transparency reforms to various governments. Appendix I summarises all the features in a ready-to-use table for anyone reviewing or designing digital systems.

This is a first draft of the framework, and it is likely to evolve over time. Based on our experiences and those of our partners and other interested parties, we will consider whether there is value in building out this framework by adding further information and recommendations.

We welcome any feedback as part of this process. In particular, we are interested in responses to following questions:

• Do these features reflect user needs uncovered in other user research?
• Are there other features which are necessary to deliver high-quality beneficial ownership data?
• Is this list of features useful for reviewing system requirements, supporting register design conversations, user research, or other aspects of implementation work?
• Should we build out this overview of features into a dedicated resource? If so, what additional information about each feature would be useful? (Appendix II presents some ideas about what a more complete treatment of each feature could look like).

Please share any feedback on this working paper with Miranda Evans ([email protected]).

Successful user-centred policy and system developments should consider the specificities of each national context. This resource should complement rather than replace direct user research on the design and implementation of a particular beneficial ownership register.

Unpacking “usable” beneficial ownership data

In order to develop a better understanding of what usable beneficial ownership data looks like, we interviewed 37 people from law enforcement agencies, financial intelligence units, tax authorities, procurement agencies, anti-corruption bodies, financial institutions, data services, the media, academic institutions, and non-profit organisations.

Key findings from our data use research include the following:

• There is little useful distinction to be made according to people’s particular roles or organisations. Whether a person was a tax investigator or a private sector data provider, what they needed from beneficial ownership data sources was mostly the same, because they were addressing the same type of questions.
• People’s needs change according to the form of the question they are posing. For instance, understanding how diverse or centralised beneficial ownership is in a particular sector of an economy requires access to data in bulk; learning the beneficial owners of a single company does not.
• The questions a user is trying to answer will likely evolve over the course of their investigation, going from one line of inquiry to another. Therefore, the features people need will evolve as well.

In going back through the research to identify particular functional requirements for beneficial ownership registers, we identified sets of related questions, tasks, and features.

1. People use data and information to answer questions.
2. In order to answer those questions, people need to undertake certain tasks.
3. In order to conduct those tasks, people require certain features to be present in the data.

Here is an example of how the framework might help clarify the features required by a data user:

Questions

• Which companies within the data currently have French nationals as beneficial owners?

Tasks

• Find all beneficial owners that meet specific criteria (attributes)
• Find all companies beneficially owned by a list of individuals

Features

• Search by beneficial owner name and ID
• Search by other details
• Up-to-date information
• Structured data in a well-used format
• Relevant person information
• Clear field and object descriptions

Even for this relatively simple question, which only breaks down into a couple of tasks, a large range of features is necessary.

First, it is necessary to find beneficial owners who have French nationality. This requires that “Relevant person information” (nationality) is a feature of the data, and that nationality is identifiable in the data (“Structured data in a well-used format”). It also requires that the data can be “Search[ed] by other details” – that is, by nationality.

Once the set of French beneficial owners is identified, it is necessary to find all the companies related to each individual. This requires that the data can be “Search[ed] by beneficial owner name and ID”.

Across both tasks, the feature “Clear field and object descriptions” is necessary, as it is for most analysis. In this example, imagine a field named “country” is the only field with jurisdiction information for beneficial owners within the data. An analyst will need to check further documentation to see whether that is nationality information, residential information, or something else.

Similarly, the feature “Up-to-date information” is needed here (and for many analysis tasks). Without it, no insight into the current situation can be gained.

An initial outline of the features can be found below in the Features of better data section. As far as possible, they have been defined and presented so that they are discrete from each other. For example, these two features are related but distinct:

• Up-to-date information: Included in the data about companies, people, and their relationships is the latest known information.
• Basic disclosure information: Basic details about how the information was disclosed is included in the data. This includes, as a minimum: who is responsible for disclosing the information, and when they disclosed the information.

A beneficial ownership dataset may contain the latest known information while not showing exactly when each item of data was last updated. Conversely, a dataset may show exactly who disclosed the information and when, but a search of the online register shows that the data is six months old. Often, the basic disclosure information within a dataset will confirm that the rest of the information is up to date. Thus, these features are related but distinct.

Why does this matter?

This framework and the list of features could be useful for agencies developing beneficial ownership registers (and people assessing them) in a number of ways. It provides:

• A framework for discussions between data users, beneficial ownership policy teams, and digital teams developing registers: An agency developing a beneficial ownership register for a specific policy aim (for example, reducing collusion in public procurement) can use this framework to aid consultation with data users. Typical analysis questions might be elicited (e.g. “What proportion of bidders to this public contract have the same beneficial owner?”), then broken down into related tasks and data features. This would then inform the functional requirements of the system to be built by software developers.
• A guide for developing requirements for a new register: Including all of these features in a register system will lead to more usable beneficial ownership data. Therefore, an agency building a new register could use these as a starting point for a set of functional requirements.
• A tool for assessing the usability of data from existing beneficial ownership registers and identifying potential improvements: This could be used as an assessment methodology for understanding the quality of implementation of a beneficial ownership register.

What is not included?

The features presented in this framework belong to systems which deliver data and information to people (and to other systems). They are necessary but not sufficient to ensure the data is usable. They are not characteristics of data collection and data management systems, although they deliver data that is collected and processed by those systems. The characteristics of data collection and management systems (and their legal underpinnings) are also essential to ensuring that data is usable. Some useful questions related to these systems include:

• Is the concept of beneficial ownership well defined in law?
• Are all relevant types of company (or legal vehicle) covered by the scope of the beneficial ownership disclosure laws?
• Is all the beneficial ownership information centralised?
• Are appropriate mechanisms in place to verify that the information declared by companies is accurate, complete, and up to date?
• Have companies and individual users complied with their obligations to disclose information?
• Do the appropriate users have access to the information?

The Open Ownership Principles for effective beneficial ownership disclosure give greater detail on a number of these questions.

Next page: Features of better beneficial ownership data