Basic corporate data is essential for understanding our world. The name of a company, its legal form, registration number, formation date, the identities of its directors, and the registry where this information is held are all fundamental to knowing who we are doing business with and who our employers are, as well as which entities should be taxed and in which jurisdiction. Access to that data over time allows us to assess the performance and structure of the economy as businesses form, merge, break apart, and fail. Another layer of analysis opens up if we move from simply identifying corporate entities to identifying their owners and those who ultimately control them, a concept referred to as beneficial ownership. The more jurisdictions that require corporate ownership data to be open, the easier it becomes to navigate through a myriad of shell companies, regardless of where they are located, to identify the actual owners.
Information provided as open corporate data is of interest to public, private, and civil society stakeholders, and has a universal geographical applicability. G20 leaders have discussed the need to use corporate data to improve financial stability and efficiency, to combat corruption, and to improve the exchange of tax information between jurisdictions.1These same goals are reflected in the Sustainable Development Goals (SDGs). SDGs 16 and 17 address the need to reduce illicit financial flows, ensure the return of stolen assets, reduce corruption, promote investment, and develop capacities for domestic tax collection, all of which are supported by improving the availability and use of corporate data.2
Progress, impacts, and challenges
Progress: Data availability and technical infrastructure
Despite the relevance of information on corporates, progress on releasing it as open data has been slow. In the fourth edition of the Open Data Barometer, corporate data, identified as company register data, had the third lowest score of all the datasets surveyed, and only 5% of all company register data was available as open data.3Nor has there been a significant shift over time to open these datasets. The first edition of the Open Data Barometer in 2013 found that only 3 of 74 company register datasets were available as open data; by the time of the fourth edition in 2017, only 6 of 109 datasets were open.4Corporate registers still lack a universal data standard that could be used to make opening these datasets easier in the future.
A more hopeful picture can be found from OpenCorporates, an aggregator that provides corporate data under an open licence. In 2018, OpenCorporates provided corporate data from 127 registries in 73 different countries; however, much of this comes from OpenCorporates’ own scraping work rather than from the release of open datasets at source. For the platform to move toward more comprehensive global coverage, more jurisdictions will need to open their company register data and remove the paywalls that limit access.
Although datasets containing basic corporate information were found in all but one of the jurisdictions assessed by the 2017 Open Data Barometer, albeit at many different levels of openness and machine-readability, data related to corporate ownership often did not exist at all.5In the latest Financial Action Task Force (FATF) consolidated assessment, only 19 of 69 countries listed are compliant with transparency requirements for beneficial ownership, and these requirements mandate only that information on beneficial ownership is obtainable by competent authorities, not by the public.6,7 Very little open data on corporate ownership exists at the present time. Open registers on beneficial ownership are available for Denmark (which also has a register of legal ownership), the UK, and the Ukraine, and a state contractors’ register is available in Slovakia. Furthermore, while the policy advances discussed below are likely to create more ownership registers in the future, these will not necessarily be open, free-to-access, or machine-readable.
However, important advances have been made. These advances might be described as “infrastructural”, providing the technical components that support the dynamic and international nature of corporate information, as well as the legislative and civil society support that have made these technical advances possible.
The most fundamental requirement for open corporate data is the availability of identifiers that are unique, stable, and interoperable across jurisdictions and that are openly licensed. In this respect, a great deal of progress has been made. Two systems are operational. Thomson Reuters’ PermID assigns identifiers and offers them, plus basic company and officer information, under an open license for over 3.4 million legal entities.8The Global Legal Entity Identifier Foundation (GLEIF) takes a different approach, requiring legal entities to sign up for an identifier through a local operating unit. The LEI data they provide also contains basic company information and will soon also contain “Level Two” data on an entity’s accounting parent. Over 1.3 million LEIs have been issued to date; however, despite their universal applicability, identifier coverage is greater across the wealthiest countries with almost 25% of the LEIs issued coming from the UK and USA alone (see Figure 2).9
Another supporting development has been the emergence of lists and services to point publishers and data users toward the right sources of corporate identifiers. One example of this is org-id.guide, which has evolved from the organisation identifiers registry list maintained by the International Aid Transparency Initiative to help encourage the use of stable identifiers from existing registers for all types of legal entities across the world.10Launched at the 2016 International Open Data Conference and developed to meet a commitment of the IODC Roadmap,11it has been fostered by a collaboration of open data standards providers, but has not yet seen wide adoption. GLEIF also maintains a list of corporate registries with unique identifiers known as the Registration Authorities List that is used in LEIs.12One danger here is that we face a proliferation of competing open standards to the point that tooling for crosswalks between identifiers will need to be part of the future landscape of open corporate data. GLEIF’s dataset linking LEIs to SWIFT’s Business Identifier Codes is a welcome step in this direction.13
The final technical component of progress on open corporate data is the emergence of universal data platforms. These services offer an advantage over national platforms in that corporate activity often crosses borders, so identifying corporates often requires searching multiple corporate registers maintained on national platforms. Reconciliation of company names and disambiguation of company officers is a significant value-added service that such platforms can provide. OpenCorporates is the most well-established example, offering access to open corporate data via search and an API. More recently, OpenOwnership, funded by the UK’s Department for International Development, has sponsored the development of the Beneficial Ownership Data Standard and brought together beneficial ownership information from several existing national registers on its own platform with plans to extend further and to allow self-submission by companies and individuals.14
An introduction to beneficial ownership
In the context of corporate ownership, “beneficial ownership” refers to the identification of the natural person or persons who benefit from, or control, legal entities, persons, or arrangements. Beneficial ownership can be achieved through such means as formal rights, like votes or dividend rights attached to shareholdings, or informal rights, like the ability to influence the direction of a company outside a formal ownership relationship. The identification of beneficial ownership involves looking through otherwise complex corporate ownership chains to find the “ultimate” beneficial owner regardless of how many shell companies or secrecy-based jurisdictions may stand in the way.
Beneficial ownership is also partly defined in the negative. While the “beneficiary” of an asset can be any legal person, natural person, or arrangement, a beneficial owner must be a natural person, because, regardless of how complex a corporate structure may be, control over it ultimately resolves to one or more natural persons. Beneficial ownership is also distinct from “legal ownership”, which refers only to ownership of legal title. For example, one natural person may legally own a company, while another natural person is the beneficial owner through a trust or nominee structure. Legal title is not necessary for beneficial ownership, because control may be exercised through informal means. 15
The concept of beneficial ownership has its origins in trust law, but, beginning in the 1970s, it has become part of the lexicon of international tax, anti-money laundering, and illicit financial flows. More recently, beneficial ownership has moved out of these specific fields into the broader policy debate on corruption and transparency. 16Of particular importance is the emergence of the OpenOwnership project, which is focused on enabling the publication of open beneficial ownership data and creating its own global register of this data. The project has a steering group comprised of civil society groups, such as the Open Contracting Partnership and OpenCorporates, and has built links with organisations like the Open Government Partnership and the Extractives Industry Transparency Initiative that have existing commitments to beneficial ownership transparency.
Progress: Policy and legislation
The foundational work for progress on opening data corporate ownership has involved winning the policy and legislative argument that corporate ownership data should be made available. At a high-level, these arguments are summarised in two communiqués from the G8 and G20 in 2013 around the themes of anti-corruption and open societies,17which, in October of that year, led to the first public commitment to an open register of beneficial owners by the UK at an Open Government Partnership (OGP) conference in London.18
The motivation for opening up corporate data has three interlocking themes. First, the financial crisis ushered in a historical period of reform centred around the dangers of uncertain information and unknown actors in financial markets. This has been particularly important in the G20 driving forward the Global Legal Identifier Foundation, which is based on their linking of time-consuming and uncoordinated practises for identifying counterparties to the potential for exposure to liabilities and consequent financial instability.19Second, the use of anonymous corporate vehicles in corruption cases and other illicit financial flows was highlighted in the World Bank’s influential 2011 Puppet Masters report.20Corporate anonymity has since been identified as a contributor to terrorist-financing, to corruption, to the expropriation of shareholders, and to impeding development goals.21Third, the technical requirements of tax-sharing and anti-money laundering requirements increased the demand for interchangeable data on natural persons and legal entities.22Together, these three themes have been critical drivers of the beneficial ownership agenda.
The concrete outcome of these policy initiatives has been national- and regional-level legislation to mandate open registers of business data and the systematic collection of beneficial ownership data (to which access may still be restricted). While early adopters were single European nations, new regional leaders like Indonesia are starting to emerge in the Global South.23In the European Union, the Anti-Money Laundering Directive (AMLD) obligates member states to create public central registers of beneficial ownership. In a major legislative advance in 2017, the AMLD was updated to include trusts and trust-like arrangements and to make that data accessible to those with a legitimate interest.24
Multilateral organisations have also drawn a wider range of entities into contact with the infrastructure for identifying and describing corporates and for publishing this information as open data. The Extractive Industry Transparency Initiative (EITI) 2016 Standard requires countries to publish roadmaps for beneficial ownership transparency in the extractives sector and a recommendation that beneficial ownership registers be public. As of March 2017, 21 countries had committed to a establishing a public register.25The Open Government Partnership also recommends robust registers of beneficial ownership as an intermediate commitment to open government, and recommends providing open access to machine-readable data from these registers as an advanced commitment.26
There is also a less visible layer of work related to corporate registers that is not yet yielding open datasets but still establishes concrete targets for advocacy work in this area. In many jurisdictions, such as Hong Kong, Singapore, Switzerland, Zambia, and others, governments have passed legislation to require closed beneficial ownership registers to comply with anti-money laundering standards.27Similarly, the implementation of the Markets in Financial Instruments Directive II (MiFID II) requires trusts wishing to trade in financial instruments to have an Legal Entity Identifier as of January 3, 2018,28so that prior to the implementation of the revised AMLD in the EU, we will get signals about which trusts in the region are economically active and another identifier that can be incorporated into central registers.
Case study: OpenCorporates
OpenCorporates was founded in 2010 and has received funding from the Alfred P. Sloan Foundation, ODINE, and EU Horizon 2020. OpenCorporates collects corporate data from public open data sources, official APIs, and data scrapings, and transforms it all into a standard form. Covering 127 registries, OpenCorporates offers a search engine, access to free and paid APIs, corporate data, company gazettes, and LEIs. The site also offers curated datasets on out-of-state corporations in the US. In 2017, OpenCorporates took the Quebec corporate registry to court after receiving a takedown notice for information originally accessed from that registry. The case indicates both how far the movement for open corporate data has come and the role that OpenCorporates has played and how much work is still left to be done.
Impacts: Markets for corporate data
Even at this relatively early stage, the opening up of corporate data has had significant impact. There is a significant market for data on corporate entities, and startups in this ecosystem have both built new business models based on available open data sources and challenged information providers. Significant businesses that have emerged in this space: OpenCorporates, providing standardised corporate and financial data; DueDil, offering credit checking and anti-money laundering checks; Arachnys, offering automated and manual tools for due diligence; and Calcbench, offering standardised accounts data for US companies.29Encouragingly, many of these businesses support the rhetoric around the expected impacts of open corporate data by serving to reduce the compliance costs for businesses, making decision-making about investment and risk more transparent, giving minority shareholders visibility on who controls legal entities, and allowing diagnostics for individual businesses and whole sectors of the economy through the use of detailed data. There is still an obvious absence of businesses from the Global South; however, the coverage of OpenCorporates and Arachnys, in particular, is also deliberately global and constrained only by data availability. This is an encouraging sign that businesses built on open corporate data can be supply-led rather than demand-led and that we will see a more diverse customer profile emerge as a result.
Impacts: Civil society users of open corporate data
Corporate data has been fertile ground for civil society with the origins of movement closely tied to transparency goals and now closely aligned with the Sustainable Development Goals. The NGO, Global Witness, has used corporate data to investigate corruption in Myanmar’s jade industry (see box) and money-laundering associated with Panamanian companies and the Trump Ocean Club.30Investigations that combine leaked and official data simultaneously illustrate both the promise of joining up data and the difficulties of investigating ownership when so many legal entities are registered in jurisdictions that support secret registration for corporations. As another example, Transparency International, when investigating foreign ownership of properties in London, combined information from the International Consortium of Investigative Journalists’ Offshore Leaks Database and official sources; however, despite combining sources, the report was unable to find information on 46% of the companies concerned.31When open data sources are available, the potential for civil society to investigate and exert pressure is clear. For example, court documents from a Brazilian case linked to the Odebrecht bribery scandal, enabled journalists in Scotland to both examine the role of Scottish Limited Partnerships (SLPs) as a possible money-laundering vehicle and highlight the lack of compliance with disclosure rules in the UK’s open register.32The UK has since signalled reforms on SLPs.33Similarly, campaigners have also demonstrated the potential impact of tying data standards together. In Malaysia, the Sinar Project’s Telus website, for example, will use the Open Contracting Data Standard, Popolo, and the Beneficial Ownership Data Standard to link legal entities and natural persons in public procurement disclosures as a way of finding corruption.34
Case Study: Global Witness and the Jade Industry
The Global Witness investigation of the jade industry in Myanmar involved unstructured source data that was turned into structured data by OpenCorporates. This process also meant that full access to the dataset was preserved even as director and shareholder details were being scrubbed from the original source.35Global Witness was then able to see links within the jade industry using official data and to disambiguate legal entities and natural persons with the support of on-the-ground interviews. Together, these techniques allowed for precise documentation on how important figures from government, the military, and the narcotics trade were heavily involved in the jade industry. As this example suggests, the investigative use of corporate data often requires anchoring in a particular external context. This might, as in Myanmar, provide information on the local significance of patterns of corporate ownership or, as with large-scale leaks like the Panama Papers, point investigators toward possible wrongdoing.
Challenges: Data quality
A significant challenge that is likely to become more acute over time is the quality assurance of data held in open registers, especially when legal entities self-submit and information is verified on a risk-assessment basis. In these situations, the information provided by honest actors is of variable quality, and the information provided by dishonest actors is often hard to disentangle. This problem is particularly acute when new compliance regimes are introduced and submitters’ understanding of their statutory duties is limited. A report from Global Witness has found that the UK’s person of significant control (PSC) register, which is unverified and has some unvalidated data fields, has serious data quality issues.36Verified data, or data submitted through a corporate service provider, is likely to be high quality, but imposes higher costs on corporates. The poor quality of open registries, and the high cost of verified alternatives, was used as an argument against open and verified corporate registries in a 2011 report by the Stolen Asset Recovery Initiative.37The same argument has resurfaced more recently in the context of registers of beneficial ownership.38Proponents of open corporate data will need to be wary of poor data creating negative feedback around corporate transparency. A possible solution is to adopt the LEI approach, where local operating units validate information for a small fee and accuracy remains high.39There will also need to be a push for verified data, using traceable processes by authorised persons to reduce the opportunity for plausible deniability when false information is entered into a register.40To guard against honest mistakes, registries may also be able to improve data quality through automated error detection, better guidance to submitters, and improved data validation.
Challenges: Data protection and useful data
Other than data quality, the other major challenge to improving open data on corporate identity and ownership may be the potential negative reaction from some corners to increases in transparency, primarily because arguments around privacy and data protection are currently unresolved. While the European court has ruled that there is no “right to be forgotten” for natural persons in company registers, legislation has not removed ambiguities around how long data can be stored for, who has the right to access it, and how far should it go to identify individuals.41While campaigners have made cogent arguments against privacy for beneficial owners, the argument is likely to re-emerge as ownership transparency is tied into legal reforms associated with corporations and to the fundamental rights of individuals. Given that extracting value from corporate ownership data involves making sometimes uncertain connections between datasets held in different jurisdictions, practitioners and civil society will need to balance the arguments for transparency with the need for coordination and harm prevention for individuals exempt from disclosure and for individuals who, while not exempt, may nonetheless be exposed to harm by the joining up of datasets.
One more challenge arising from success will be the need to encourage positive uses for ownership data, while discouraging detrimental or adversarial uses. There is a significant risk that some combination of poor data quality, complex or non-interoperable data, and a lack of capacity will lead to registers being created but not used. In this regard, it is useful to have sector-specific guidance (e.g. NRGI’s work on extractives) or targets for success or failure (e.g. increasing revenue collection in the Global South).42While positive uses of ownership data will need to be cultivated, adversarial uses of registers are likely to increase unsupervised. Dishonest actors will become familiar with the rules and find it easier to skirt disclosure requirements, and as long as secrecy-based jurisdictions still exist, they will also have the option to simply close legal entities and open new ones not subject to scrutiny.43This could itself be seen as a sign of success, but it will be crucial to be able to measure such behavioural changes.
Much progress on open data related to corporate identity and ownership has been made. While it might appear that the amount of open corporate data is still relatively limited, much of the infrastructure required for success, such as policy, legislation, and technical architecture is now in place or being developed. Moreover, progress to date has involved a broad base of stakeholders from government, multilateral institutions, civil society, and the private sector. Great advances have been made in creating a shared understanding that we need to be able to unambiguously identify legal entities using universal identifiers, which has led to an emerging ecosystem of links to complex corporate data. A more recent development is that the policy argument in support of beneficial ownership seems to have been won. We have seen early examples of registers providing ownership data as open data, as well as the legislative basis for many more to come in the near future mostly as the result of a civil society movement committed to opening up asset ownership as part of economic transparency. However, the withdrawal of the US from the EITI on the basis that it is a “burden for business” demonstrates the fragility of the achievements to date and the difficulty of maintaining corporate disclosure as a desirable goal.44For open corporate data to fulfil its promise for the SDGs, any momentum that has been created in Europe and North America will need to be recreated across the Global South.