International Swaps and Derivatives Association (ISDA)
Training / additional staff (skills)
Changes in processes
Changes needed in the context of other counterparties / third-party providers
Time required to find new solutions
Other (please specify)
Data Dictionary - Semantic level
Data Dictionary - Syntactic level
Data Dictionary - Infrastructure level
Data collection - Semantic level
Data collection - Syntactic level
Data collection - Infrastructure level
Data transformation - Semantic level
Data transformation - Syntactic level
Data transformation - Infrastructure level
Data exploration - Semantic level
Data exploration - Syntactic level
Data exploration - Infrastructure level
In recent years, the industry has embraced the ISDA CRIF as the standard to hold and report risk sensitivity data. Initially developed as part of the ISDA Standard Initial Margin Model (ISDA SIMM) to enable the exchange of risk data between firms and facilitate initial margin reconciliations, the CRIF has evolved to cover market risk (the Fundamental Review of the Trading Book – FRTB) standardized approach and credit valuation adjustment risk (CVA) standardized approach capital models.
Using the CRIF standards and the CDM (see the answer to Question 12 for further details) has several potential advantages, including:
• It leverages existing standards that market participants are familiar with, avoiding the need to adopt new standards for a similar purpose.
• It establishes a robust and comprehensive framework for risk data representations, which are directly applicable as inputs into standardized capital models and the ISDA SIMM.
• It allows for a granular analysis of the drivers of capital and margin calculations and their relative importance, at asset class and risk factor levels.
• Applying the CDM will help to partially automate the creation of CRIF files and ensure their validity, as well as enable automation of risk data reporting to regulators.
ISDA estimates that more than 200 firms are using the CRIF standards on a regular basis, including 40 globally active firms using the FRTB-SA CRIF standards.
Widespread adoption of the CRIF also benefits regulators, allowing for the simple identification of risk drivers, easy computation of capital and margin amounts, and interoperability and comparability of CRIF files. ISDA has already licensed the CRIF to several regulators interested in using the standards.
Any new or revised data dictionary or standards should, wherever possible, cross-reference to (and have interoperability with) existing standards and terms which are in use within other regulation(s) / legislation(s), and recognise current industry practices. Where possible, a new or revised data dictionary or standards should leverage upon existing widely used standards, such as the ISDA CRIF when it comes to the collection and exchange of risk data. As the financial markets are global, a data dictionary should also be adopted globally to deliver the maximum effect. Fragmentation of reporting requirements between jurisdictions increases the cost and complexity for market participants, and even where reporting processes are to be streamlined for the benefit of both firms and regulators, to the extent that this becomes an EU only project, the benefits will be limited. However, while the establishment of, or improvements to, a data dictionary for reporting and collection of data would ideally be solved at a global level, it is perhaps unrealistic to expect coordinated changes to be agreed and applied simultaneously across jurisdictions. Therefore, a more realistic approach may be for the EU to start locally and encourage other jurisdictions and regulators to follow suit. It is also helpful to avoid (at least initially) any extra-territoriality impacts, as well as avoiding the adoption or alignment with regimes that may impose restrictions on choice, competition, transparency, or undue cost (unintentionally or otherwise).
A good data dictionary can help with establishing a data model to fully represent specific financial markets, (with some models having the potential to represent multiple financial markets), and present the data as machine readable and machine executable. There are some industry initiatives currently underway which regulators should consider leveraging, such as the ISDA CDM, (see the answer to Question 12 for further details) where possible, potentially with the option of participating in the further development and endorsement of the models and data dictionaries. This should allow such data dictionaries and models to be ready for use in the more immediate future and gain greater traction from market participants. The next stage would be to find where standards uses may intersect, or where there is need for interoperability to enable communication between parties, or to aggregate information, such as for regulatory reporting or risk management.
An integral step towards achieving longer term goals of establishing a robust data dictionary is to develop common data inputs. The way that ISDA have approached this, through the ISDA CDM, is in the development of industry data standards. We would caution against deriving common data inputs from reporting needs – good reporting should be reflective of industry standards at the outset. It should also be reflective of industry data standards that are already used practically for performing financial tasks (such as the settlement of a trade). Such practical standards will generally be understood by the financial industry and give the relevant level of granularity needed to regulators.
This work would help to facilitate the representation of a data dictionary, and of reporting instructions, as code.
Understanding reporting regulation
Extracting data from internal system
Processing data (including data reconciliation before reporting)
Exchanging data and monitoring regulators’ feedback
Exploring regulatory data
Preparing regulatory disclosure compliance.
Other processes of institutions
Processing data (including data reconciliation before reporting) - Use of the CRIF ensures risk sensitivity data is consistent with reported capital.
A well developed data dictionary and model for common data inputs that can be applied across financial products will lead to faster and more efficient industry implementation of new regulatory requirements. This is where a data set can be used to demonstrate implementation of regulatory requirements which, in the pre-publication/consultation phase of the legislative process, can be built transparently with market participants and regulators involved. This allows the results of different approaches and draft requirements to be analysed by all parties, to hone the implementation, in an iterative way, so that eventually a specification can be reached that fully achieves the regulators’ goals.
And moreover, this specification can and should be made available to market participants as open source code for them to implement consistently in their own systems or with their service providers on different technology platforms as necessary.
As an example of a data dictionary helping to extract data from firms’ internal systems, the ISDA CRIF has significant value when it comes to collecting and exchange risk data.
Initially developed as part of the ISDA SIMM initiative, the CRIF was designed to standardize the risk data that is inputted into the ISDA SIMM to compute initial margin calculations. It was important to ensure all ISDA SIMM users adopted a common format to exchange risk data so they could easily identify and investigate the sources of potential bilateral initial margin mismatches and reconcile as appropriate. The CRIF is a simple column-based and machine-readable set of flat files, governed by a set of rules documented in the ISDA risk data standards, available for each CRIF version.
In recent years, the ISDA standardized approach benchmarking initiative (https://www.isda.org/2020/06/17/isda-sa-benchmarking/) has triggered the need for further standardization of input risk data for standard capital models, such as FRTB-SA and SA-CVA. The aim is to allow participating firms to exchange risk data with ISDA to validate capital calculations and analyse the drivers of capital results at the risk factor level. In support of this benchmarking initiative, ISDA has developed additional CRIF standards for FRTB-SA and CVA risk capital models, each governed by their own risk data standards.
ISDA collects data from member firms to assist with its prudential advocacy work, such as Basel reform. When carrying out such data collection exercises, ISDA have encountered many of the same issues experienced by regulators, such as data being submitted in an incorrect or inconsistent format, data sent late, or data not being sent at all. The absence of a non-standard format for submitting this data means that ISDA need to organise the data before it can be used. This is both costly and time consuming to ISDA. Alternatively, ISDA can push back to the submitting firm(s) for the data to be re-sent in the correct format, which is costly and time consuming to the firms.
As an example of how ISDA have addressed some of these challenges, the “Common Risk Interchange Format” (CRIF) was developed, which is the ISDA standard format for risk sensitivities and trade details, and hence standardises the inputs to margin calculations. This format ensures that firms exchange trade information in a standardized way and reduces the need for interpretation, manual intervention, and minimises room for error.
ISDA also found that using FpML to describe trades and portfolios helps firms consistently book these trades and portfolios, ensuring consistency of the exercises.
Based on these experiences, establishing standardised templates and formats that are accepted industry wide results in data collection being performed at a fraction of the cost and with minimal human interaction.
Today, the licensed ISDA CRIF standards support multiple purposes and are being adopted by an increasing number of market participants, including:
• Vendors, aiming to provide SIMM margin, FRTB and CVA capital reconciliations, or capital management services;
• ISDA member firms using it for model back-testing, benchmarking and other validation processes;
• Regulators collecting portfolio risk data from supervised firms to support standardized approach benchmarking exercises;
• New firms may benefit from the synergies offered by the CRIF standards in their interactions with regulators, vendors and ISDA.
The CRIF is a simple, robust and comprehensive risk data format that organizes and stores any risk data input needed to compute SIMM margin, FRTB-SA and CVA capital amounts. CRIF files can be semi-automatically generated by leveraging the ISDA CDM and are also sufficiently intuitive to allow for visual inspection. The CRIF is specified for the set of relevant risk factors associated with each standard model, and can easily be extended to handle new risk factors in the future. Each CRIF file can be automatically processed by the relevant model calculation engines to compute margin and capital amounts.
The CRIF standards are typically used for SIMM margin reconciliations and standard capital benchmarking analyses. The risk factor information contained in a CRIF file allows users to easily identify the key drivers behind different SIMM margin amounts calculated for the same bilateral portfolio, or different capital amounts calculated for the same hypothetical benchmarking portfolio. The CRIF can be implemented as an excel file or in any other column-based file extensions like “.csv” or “.txt”, with defined data fields arranged in columns. The elements of a CRIF file can be either textual (string), date or numerical, and their types can be context dependent.
More broadly, any move to a common data model or data dictionary will require market participants to implement changes to how they represent and manage their data. The impact will vary for each market participant depending on the scale of change, but while there will be additional costs in the near term, it will lead to more efficient data reporting and cost savings in the longer term. These cost savings can be maximised by relying on existing and widely used industry standards such as the ISDA CRIF.
A standard data dictionary and data model is essential for regulatory data to be represented consistently across reporting requirements, allowing for reduced costs to market participants, interoperability between regulations and more transparency of the financial markets.
To this end, ISDA have developed, and continue to enhance, a solution to the problem of inconsistencies in the ways of representing and processing derivative transactions via the ISDA Common Domain Model (CDM). The CDM establishes a common set of representations for derivative products and their components, events and processes – a common language – designed to solve the problem of fragmentation in the interpretation and implementation of various processes, such as meeting regulatory reporting requirements. Therefore, firms using the CDM would be able to represent data in the same standard format which could then be cut in different ways to meet different regulatory reporting requirements. For example, trade data could be reported once in CDM format to a single pool of data, and this single set of data can be used to meet the requirements of multiple EU and other global jurisdictions with similar trade reporting requirements. This in turn would allow for much more meaningful aggregation of data at an international level, with regulatory authorities in a position to look at this data in the knowledge that what they were seeing was consistent across the firms who supplied it. Such an approach to representing data would align with a move to a ‘pull’ model for data collection.
Although the initial development of the CDM has been focused on OTC derivative products, the scope of the model is much broader. ISDA are working with partners and other trade associations to expand the range of financial products the CDM supports, such as incorporating exchange traded derivatives and securities products into the model. This would enable a single model to have a greater reach across the financial industry and regulatory regimes.
ISDAs report on ‘Regulatory Driven Market Fragmentation’ identifies reporting as one of the key examples where firms are forced to develop and implement different systems and solutions in different jurisdictions because of varying regulatory requirements – even though those requirements are being implemented to meet a global standard (https://www.isda.org/a/wpgME/Regulatory-Driven-Market-Fragmentation-January-2019-1.pdf):
‘Data and reporting is an obvious example [where firms are forced to develop and implement different systems and solutions in different jurisdictions]. If all jurisdictions require market participants to report generally the same information to trade repositories, but each requires different data forms and formats in which such information should be reported as part of its rule set, then firms will incur significant expense in complying with myriad rules.
Discrepancies such as those related to data standards will also impact the ability of regulators to monitor risk on a global basis’.
More recently, the CDM was used in the G-20 TechSprint of 2020 to develop a digital regulatory reporting pilot for the derivatives reporting regulations set by the Monetary Authority of Singapore (MAS). The solution enables reporting firms to access an executable code version of the regulatory requirements and to run trades through a reporting engine to validate the business logic, which can then be used consistently across the market. This pilot was submitted by ISDA and fintech firm REGnosys, winning the regulatory reporting category in the G-20 TechSprint (https://www.isda.org/2020/10/06/isda-and-regnosys-win-g-20-techsprint-for-regulatory-reporting/).
Additionally, the Expert Group on Regulatory Obstacles to Financial Innovation (ROFIEG) paper ‘30 Recommendations On Regulation, Innovation And Finance’ (https://ec.europa.eu/info/publications/191113-report-expert-group-regulatory-obstacles-financial-innovation_en) puts forward several proposals that lend themselves to the aims of ISDA CDM and for common data inputs. For example, Recommendation 9 – ‘RegTech and SupTech’ – refers to machine readable legislation, and Recommendations 10 and 11 promote the use of standardised legal terminology and producing legal and regulatory language that is both human and machine readable.
The challenge of different jurisdictions implementing data and reporting requirements in various different ways, is compounded by market participants each needing to implement the reporting requirements separately and potentially arriving at differing ways to represent various trading scenarios. Essentially, if every industry participant describes trade events in different ways, it will not be possible for dealers and vendors to speak to each other in a standard manner. By adopting the CDM however, there will only ever be one single way in which to define and manage trade events.
In the area of regulatory compliance and reporting, the CDM will have a transformative impact. Using this open source model, market participants and regulators can come together to tackle new regulatory requirements and build prototype solutions as code through open industry initiatives. These solutions can be tested in the open with regulators involved, and if they so wish, regulators can be given the opportunity to view the test results and offer their guidance for changes that should be made to prototypes, the final implementation code will be then made available to the whole market for consistent implementation of the regulatory requirements, without risk of thousands of different interpretations of rules being implemented across the market as happens today.
But effective automation can only be built on standardization, which is why the ISDA CRIF standards and the CDM can work together to achieve the goals of standardizing risk data representation and enhancing CRIF process efficiencies. The CDM is able to take in combinations of firms’ CRIF input data, CDM position data, FpML data, and trade reference data, using the CDM staging algorithm, which organizes, enriches, and validates the relevant data inputs. The CDM staging algorithm then generates a set of valid CRIF files, readable by standard models such as FRTB-SA, SIMM and CVA. These models in turn generate the relevant capital and margin amounts based on the CRIF input data, and the CDM can digitize and tag these amounts, which can ultimately be consumed by connected regulators or third parties.
The CDM capabilities can provide regulators with access to an accurate and up-to-date instance of transaction risk data at any time, as well as the associated capital and margin amounts computed by each firm. This would help to reduce some of the current burdens of regulatory reporting. Systems based on the CDM and distributed ledger technology typically offer a structure that solves for these aims.
Wider adoption of the CRIF and CDM would contribute towards standardizing the market as a foundation for structural change. Steps are already being taken towards this. For example, ISDA has been supporting the work of the Bank of England in developing common data standards , and a number of industry initiatives already leverage the CDM. ISDA also plans to conduct a real world pilot of the proposed CRIF and CDM framework in collaboration with regulators to further demonstrate the value proposition.
As the Basel III requirements are implemented, the industry and regulators need to take stock and closely collaborate to adopt CRIF as the industry standard and leverage the CDM to form a standard approach to risk data.
More details can be founds in the ISDA White Paper “The Future of Risk, Capital and Margin Reporting”
High cost reductions
Different regulations will have various aims and purposes, such as to detect market abuse or to identify systemic risk. Despite the differing aims, the same basic data may be required by multiple regulations, and only the format or representation of the data may vary.
As an example, this is seen with the trade data reported under the EMIR and MiFIR regulations. When the same trade is in scope for both regimes, many of the same data points within a trade booking are to be reported, however the way in which EMIR and MiFIR require that data to be represented within their reports can differ, resulting in firms implementing two ways of representing and reporting the same set of trade data. Where a definition of data elements can be standardised within a single data dictionary and data model, the same value could be reported to multiple reporting regimes without losing any visibility to the overall trade data.
Therefore, where there are overlapping requirements within regulations, there is an opportunity to reduce costs by aligning the definitions of those requirements by use of a standard data dictionary and data model.
There are also significant costs incurred during the implementation of new or updated regulatory requirements. This requires a coordinated effort between multiple divisions within a firm including (but not limited to) legal, compliance, government relations, business specialists, technology and trade associations. Translating regulations into data requirements and reporting logic creates a high demand on the resources of firms in terms of cost, time and personnel. Currently, every market participant implements its own solutions, but greater efficiency can be gained through a mutualised industry effort to create standardised rules and reporting output that can be applied consistently between regulations.
No costs (4)
Collection/compilation of the granular data
Additional aggregate calculations due to feedback loops and anchor values
Costs of setting up a common set of transformations*
Costs of executing the common set of transformations**
Costs of maintaining a common set of transformations
Complexity of the regulatory reporting requirements
Other: please specify
No benefits (4)
Reducing the number of resubmissions
Less additional national reporting requests
Further cross-country harmonisation and standardisation
Level playing field in the application of the requirements
Simplification of the internal reporting process
Reduce data duplications
Complexity of the reporting requirements
Other: please specify
not valuable at all
valuable to a degree
Data definition – Involvement
Data definition – Cost contribution
Date collection – Involvement
Date collection – Cost contribution
Data transformation – Involvement
Data transformation – Cost contribution
Data exploration – Involvement
Data exploration – Cost contribution
Data dictionary – Involvement
Data dictionary – Cost contribution
Granularity – Involvement
Granularity – Cost contribution
Architectures – Involvement
Architectures – Cost contribution
Governance – Involvement
Governance – Cost contribution
Other – Involvement
Other – Cost contribution
Early engagement with the industry and the involvement of firms with any changes to data collection processes will help to ensure all the relevant data for a given regulation or financial section will be made available to regulators. Direct engagement between regulators and market participants in the development of standards, data dictionaries and data models will help to avoid any potential gaps that may otherwise be created and pave the way for regulatory endorsement of the resulting data dictionary / data model.
A well developed model for common data inputs that can be applied across financial products will lead to faster and more efficient industry implementation of new regulatory requirements. This is where a data set can be used to demonstrate implementation of regulatory requirements which, in the pre-publication/consultation phase of the legislative process, can be built transparently with market participants and regulators involved. This allows the results of different approaches and draft requirements to be analysed by all parties, to hone the implementation, in an iterative way, so that eventually a specification can be reached that fully achieves the regulators’ goals.
And moreover, this specification can and should be made available to market participants as open-source code for them to implement consistently in their own systems or with their service providers on different technology platforms as necessary.
In order for regulators to collect data using a “pull” model, the data itself must first be represented and managed across the industry using a standard data model. Furthermore, regulations would need to be written in line with this same data model in order to be compatible. Provided such a data model is in place and regulatory rules are written to be consistent with the model, it should be possible for market participants to submit a pre-determined set of data to a central repository on a regular basis, or even from regulators to extract the data directly from market participants, thereby adhering to several of their regulatory requirements at once, without the need to submit similar and overlapping sets of data to multiple repositories / locations.
The pull model provides regulators with the potential to modify the data they collect for a given regulation, provided that data is already available within the central repository. However, even if such changes do not impact the data a market participant reports, any changes to data used for a regulation should still be communicated to the industry.
It should be noted though that the pull model may not work for all types of reporting, (for example it may not be appropriate for capital reporting), and consideration must be given to issues around transparency and a firms control over the data they make available.
In order to adhere to various financial regulations, firms are required to report similar, or even the same, data to a number of data repositories and service providers. Establishing a single central service provider will reduce the burden on market participants to submit the same data to multiple locations. Additionally, regulators will have access to a greater range of data in a single location which would otherwise be disseminated amongst multiple regulations. This would provide regulators with faster and easier access to data which they may require now or at a future date.
The ideal central service provider model would span multiple regulators. However, before implementation, either globally or within the EU only, the scope, role and use cases for a central service provider needs careful consideration to ensure the needs are fulfilled sufficiently. For example, the full scope of the role; whether it will be multi-jurisdictional or can be expanded to be multi-jurisdictions over time; whether there be a single provider or multiple providers; and if there is to be a single provider, who would operate it. A single monopoly provider, even one that is well governed and regulated, may result in a ‘common denominator’ service and struggle to provide for specific needs. Outsourcing to third parties can add an additional layer to the process with potentially increased complication, service level considerations and possible increased cost for the process. Some of these concerns could be addressed if there were a ‘mesh network’ of centralised service providers, or perhaps the focus should be on a central data model (in line with the common data inputs) that can be serviced by multiple providers.
While a reporting model that utilises a central service provider has many benefits and would be welcomed as a means of reducing the reporting burden to both regulators and market participants alike, such a model needs to be fully considered prior to being implemented.
Developments in RegTech provides opportunities to digitise and standardise regulatory data models which allows for more efficient and cost-effective regulatory reporting solutions that have not been previously available to the industry.
The CDM has proven that reporting rules can be represented digitally as machine executable code in the G20 TechSprint (see answer to Question 12). An industry initiative is currently underway to develop this further and digitally represent the full scope of the EMIR Refit reporting rules. This initiative is bringing together SMEs across buy-side and sell-side firms, vendors, trade associations and RegTech companies. Organisations for all sectors of the financial markets will be required to work collaboratively together to oversee a successful data integration across regulations, both within the EU and globally.
In order for RegTech to help achieve a more integrated and joined up reporting ecosystem, it is essential to establish robust and non-ambiguous data standards and common data inputs. As previously mentioned under Question 10, these standards should be reflective of industry data standards already used for performing financial tasks and not developed only for the purpose of reporting.
International Swaps and Derivatives Association (ISDA)