It goes without saying that a risk engine must get the basics right – functions such as instrument valuations, scenario simulations and stress testing. But in today’s environment, that is just the start. New regulations, especially Solvency II, plus strategic business goals mean insurers are now looking for considerably more from their risk engine provider. While UBS has been steadily enhancing the technical functions of its Delta system to meet evolving requirements, in many ways it is the services and support around the system that explain its continuing success among insurers.
First among these services is quality data – the critical raw material of all risk analysis. “We’ve been helping clients with the Solvency II Pillar 1 and Pillar 3 calculations, leveraging our data and the analytics in our system. Our big differentiator is the clean and normalised data we provide that can be used consistently across multiple-use cases,” says Ian Lumb, UBS Delta head of Northern Europe based in London.
Key to ensuring data quality is the fact UBS Delta is a hosted platform with a single centralised data warehouse. Although the Delta team takes responsibility for gathering, cleaning and maintaining all the data, in practice users quickly become willing collaborators. “If you have one dataset that everyone is looking at it effectively cleans itself in that clients will soon tell you when there is something they think is wrong or should be amended,” Lumb says.
The issue of data cleanliness and consistency is likely to come even more to the fore as Solvency II beds down and regulators begin to give feedback on the reports they receive. Already the Prudential Regulation Authority in the UK has questioned the quality of data it has seen in interim submissions. “We expect next year there will be a move from firms focusing on just achieving compliance toward more consistency and efficiency,” Lumb says.
One challenge is the consistent classification of assets. Many entities in which insurers invest have multiple or ambiguous identities. Take the UK’s Network Rail, which can be considered an infrastructure provider, a rail company or a sovereign entity. “There is no right or wrong answer, but firms must have a process for ensuring that it is consistently classified and aggregated across asset holdings, especially when look-through into third-party funds is involved,” says Lumb. And this is where another key feature of UBS Delta becomes evident. Although the bank normalises all asset data into standard classifications, it does not insist firms adopt these. UBS Delta has the flexibility to allow users to upload their own classifications or rules and apply them to their asset holdings.
Solvency II’s focus on asset data raises a further complication, but again one where UBS has particular strengths. In practice, many insurers use third-parties to manage their assets. As well as its many insurance clients, UBS Delta is used widely by asset managers. “Although there are initiatives such as the tripartite template to help asset managers share data with insurers, we are finding ourselves in the role of mediator and translator between the actuarial language of insurers and the investment language of asset managers,” Lumb says.
Going a step further, UBS is using the Delta platform and its expertise to help asset managers understand the importance of capital for insurers in today’s environment and how they can take this into consideration in their investment decisions. And among some of the more advanced insurers, it is helping firms use UBS Delta not just for risk and capital calculations, but also for performance attribution – a concept usually associated with asset management, but employed in this case to explain balance sheet change between reporting periods. “Being able to click a button and see why your balance sheet has changed either in market value or capital terms from one month or quarter to the next and whether it is because of interest rate moves or inflation moves or equity markets puts lot of power in the hands of insurers,” says Lumb.
If you have one dataset that everyone is looking at it effectively cleans itself