A recent working paper on the practicalities of establishing a legal entity identifier database highlights the tough challenges ahead and rather than promoting an extensive reference data collection exercise, it suggests that a ‘less is more’ approach is the only sensible way to proceed. The paper is a continuation of the research that was initially detailed in the Linchpin report, which was published back in December, but, importantly, “Legal Entity Identifier: What Else Do You Need to Know?” signals a change in tack from the regulatory community in that it indicates the high costs and lack of a business case for a detailed reference data store.
The paper, which has been written by Linda Powell, chief of the Fed’s Economic Data Management and Analysis Section of its Division of Research and Statistics, Mark Montoya, chief of the Data Strategy Section of the Division of Insurance and Research at the Federal Deposit Insurance Corporation (FDIC) and Elena Shuvalov, director of the Financial Industry Regulatory Authority (Finra), highlights the potentially high costs of maintaining “extensive reference data” and instead suggests that a “minimum” amount of this data should be captured and maintained. It highlights that systemic risk monitoring does not necessarily require the collection of vast quantities of detailed legal entity data at the outset: “It is not an attempt to capture all information that could be useful to researchers or important to regulators. In fact, some regulators currently capture more detailed information about the subset of the financial industry they regulate.”
This is a somewhat more pragmatic approach from the regulatory community and demonstrates the progress that has been made since the start of the year in terms of understanding the legal entity ID challenge. No doubt, this is in some degree attributable to the industry feedback received during the Office of Financial Research’s (OFR) comment period on the subject.
In terms of the main implementation challenges, the report concedes that some corporate actions will entail a change of LEI, but says that this data must all be retained as operational metadata (more on this later). Privacy is also noted as a serious issue and a “hurdle that must be discussed and addressed,” especially with regards to LEI data’s potential impact on competition within a market, as well as legal issues.
It suggests that the establishment of a legal entity database should be tackled in a phased approach and that the OFR should therefore walk before it can run. “For example, recording hierarchy data may not be possible until a critical mass of LEIs has been assigned because the parents of entities may not have LEIs during the initial population of the reference data,” it explains.
The paper sticks to reviewing publicly available data rather than vendor provided information in order to avoid the licensing related arguments that will likely occur when that can of worms is opened. It breaks entity data elements down into three categories: reference data that describe the entity; operational metadata that describe and document the data set; and economic or financial data related to the entity.
To this end, the most important information as identified by the report is: “legal name of the entity; status of the LEI (such as active, inactive); country of incorporation or registration (referred to as formation); state, province, or region of formation, if applicable; address of the headquarters; industrial classification; organisation type; direct parent (more than 50% ownership); and ultimate parent (also more than 50% ownership).”
There are two other data items that could, in some cases, also be useful, according to the report: alias (any names that the company may be known by, other than its legal name) and secondary industrial classification (for those firms that straddle industries).
However, the name of the entity is highlighted by the report as the most important item out of all of these because it is the one most commonly used for identification purposes.
The country or region of formation is likely to be a contentious issue (taking into account registration in multiple jurisdictions), but the report states: “Knowing the country of formation may be necessary to ensure accurate identification of an entity as well as to assist in analysing macroprudential trends. Similarly, in countries that are broken into distinct legal regions, such as states, provinces, or other administrative divisions, knowing the region of formation may be equally important. Within a country, important regional distinctions could have macroprudential implications that make the state or province, as well as the country, of incorporation important.” If a firm is registered in multiple locations, which should be chosen?
There is a similar issue with regards to dealing with address data, which can vary “significantly” in the way it is stored and the way it is determined (could be the HQ, operational offices, marketing and communications offices or another), accept the report’s authors.
Industrial classification and secondary classifications are determined to be important due to the need to monitor systemically important participants in the shadow banking system, for example. The report says: “Both knowing what industries are participating in various financial transactions and observing changes in the fundamental structure of market participants will likely be important in systemic risk monitoring.” In terms of which standard of classification should be used, it suggests the United Nations’ International Standard Industrial Classification of All Economic Activities (ISIC).
The report also notes that classification could be tricky, given that there are different levels of specificity (general finance versus central or investment banking, for example) and the activities of subsidiaries may not necessarily be captured at the outset. To this end, it suggests: “As much specificity as possible is desirable, but there is value in being able to aggregate firms even at the highest level of the nomenclature. The necessary precision should probably be discussed, taking into account international differences, as part of the implementation process.”
On the subject of organisation type and hierarchical data, the report suggests this data could provide important information with regards to risk in the case of bankruptcy and interconnectedness. However, these are the trickier data items to standardise, given that local law determines organisation type (plc in the UK versus GmbH in Germany, for example) and hierarchical data can exceptionally complex (see the troubles faced when unwinding Lehman Brothers, for example).
The report suggests that either a standard set of general organisation types should be determined, or organisations should be defined by their ownership: publicly traded, privately held, government owned, or the generic catch all of ‘other’, which would be noted in a specific field in the ID itself. It adds: “This field may also be helpful in solving some particularly thorny problems associated with the LEI. Comments on the OFR’s LEI policy statement noted that there are market participants that may need identifiers but are not legal entities, such as pension funds. If regulatory rules result in a need for this type of market participant, funds can be flagged and identified by having a specific organisation type.”
In terms of hierarchical data, the report notes that it will not be an easy task to meet all regulatory requirements for this data, but it is a valuable endeavour to collect it for systemic risk monitoring: “Given that regulators and market participants have different needs and directives, it is unlikely that a consensus of detailed ownership data will be achieved. It is also unlikely that the detailed regulatory needs can be efficiently maintained through an open standard.”
The reference to “dates” within the data sets that could be collected relates to important dates regarding the entity (such as date of establishment and corporate actions) and operational metadata (for the maintenance of the LEI itself). On the former, the report suggests that the time, effort and expense of collecting this data outweigh the benefits. The latter, it argues, are much more important in order to monitor systemic risk.
As for what this metadata would include, it suggests: start date of the LEI, revision dates, retirement date and information on why it was retired, corporate actions that affect the LEI (M&A, ownership changes or spin offs, for example) and a privacy indicator (to show how much of the data should be made public). Not all corporate actions data should be included in the database, however, according to the report: “There do not appear to be compelling arguments that most corporate action data would be helpful in identifying entities or in monitoring systemic risk. Therefore, other than in the cases in which a corporate action results in a change to the LEI, maintaining these data as part of the reference data or operational metadata would likely be burdensome and provide limited benefit for these purposes.”