Modernising actuarial data processing - part one

Published in: Risk, Roundtable, IFRS 17

Actuarial data comes in huge quantities, different forms and is often handled using outdated systems. The IFRS 17 accounting standard is presenting additional challenges, while introducing innovations is not always straightforward, as experts discuss in this InsuranceERM / Reitigh roundtable


Abi Holloway, Head of Actuarial Services, Phoenix
Brian Walsh, CEO, Reitigh
Clara Hughes, Head of Automation and Insight, Pension Insurance Corporation
Darragh Pelly, Chief Strategy Officer, Reitigh
Kim Giddings, Business Information Manager, National Friendly
Lorraine Paterson, Senior Manager Actuarial Assumptions & Methodology, Lloyds Banking Group
William Diffey, Chief Actuarial Officer – Europe, Assurant

Chaired by: Christopher Cundy, Editor, InsuranceERM

Christopher Cundy: What are the main challenges with actuarial data processing?

Abi Holloway: We have a powerful actuarial valuation model, but one of the challenges is the processing of the data before it goes into the model, and the processing when it comes out. The model is in the cloud, and we have been doing transformations to get data systems into the cloud, but some elements are not there yet.

IFRS 17 has made the journey of actuarial data a little bit more complicated

IFRS 17 has added to the challenges, because we are now trying to connect our data with the accounting world, making sure we are tagging it right and feeding it through the actuarial and accounting systems. It has made the journey of actuarial data a little bit more complicated.

Clara Hughes: IFRS 17 has been a huge challenge for the industry as, while it fundamentally changes the basis of reporting, the requirements of the standard have not been fully stable until recently, making it difficult to set up the systems and data. We have a blended team of actuaries and accountants which has worked well for this project.

PIC has grown quickly and therefore the company's processes need to adapt in line with that growth. Future-proofing systems so they scale as the business grows is an area of focus for me.

Kim Giddings: We don't have large numbers of policies, so we don't have to deal with huge amounts of data. But we have such a breadth of policies – pensions, life insurance, private medical insurance, etc – that collating the data, making it uniform and ready for the actuarial models is an issue.

We have gone through several projects looking at the common fields among these product types and whether we should put them into one big data extract, or keep them distinct. At the moment, we are keeping the product sets distinct and looking at what is required for the models.

"Good data lineage, data cleansing, consistency and accuracy is important for actuaries"

William Diffey: The challenges on the non-life side are quite similar to those you describe. Having good data lineage, data cleansing, consistency and accuracy is important for actuaries doing reserving, capital modelling and pricing.

At the forefront of many of our minds is having multiple geographies, different data sets and legacy systems. Migrating to single systems is a key challenge.

Lorraine Paterson: Our organisation has had many acquisitions over the years each with its own legacy administration platform and data transformation code. Legacy code is often written in outdated programming languages which are effectively a 'black box'.

We are moving our business onto newer platforms that will give us much greater scope for one source of data, the ability to analyse data directly and decommission legacy code.

Darragh Pelly: You are not alone! Everyone is dealing with the same challenges and I would argue there has been underinvestment in technology and data in the insurance industry.

On the topic of data warehouses and dealing with multiple systems, we are seeing a few people look to hub-and-spoke models: the hub is the 'postman' that moves data between all the systems, so you know where each system is at and there is consistent data that can feed into systems.

Brian Walsh: When we see a big transformation project like IFRS 17 come along, there are two very different approaches. Some will take a macro view of the challenge, and try to use the regulatory spend as leverage to solve wider data issues.

Others have a more singular focus. They have their heads down trying to implement IFRS 17, and the decisions they make are sometimes not as good in terms of solving future regulatory data or reporting challenges. Meaning they potentially miss the opportunity to solve multiple problems in one go.

"The barrier to getting things done isn't always the budget"

Darragh Pelly: The barrier to getting things done isn't always the budget: sometimes it's the SMEs [subject matter experts]. There is a lot of corporate knowledge embedded in these people, and they only have so much time to do the day-to-day work and tackle the new initiatives. So it's a question of how to make the best use of SME time, so both IFRS 17 and data improvements can be done together.

Christopher Cundy: Do the drivers for change tend to come from internal developments (e.g. M&A, business growth) or external (e.g. regulatory change)?

Abi Holloway: For us, it has been driven by acquisitions and IFRS 17. That can pose challenges in terms of timing: there might be a solution you want to implement, but a new acquisition comes along so you have to park it and deal with the integration first.

William Diffey: A lot of process simplification is ultimately driven by shareholders or boards. We are getting into a high inflationary environment and often the only way of optimising businesses - in my experiences of the general insurance side – is through process simplification. The challenge over the next few years is can we automate faster than inflation?

Cultural shift

Clara Hughes: Another challenge with transformation projects is creating the cultural change needed when modernising the tools people use to work with data.

"Another challenge with transformation projects is creating the cultural change needed"

Actuaries are very comfortable with spreadsheets. For many years they have been the tool of choice as they are a quick flexible solution. If you put in a modern system, it can take time for it to be understood and accepted.

Additional governance and training can help smooth the transition from a system where individual SMEs are very knowledgeable about the specifics, to a more systematic approach, and manage any associated risk.

Abi Holloway: The people that have been with the organisation the longest often don't know about the different tools we can use. One of the ways we've approached innovation is to get someone straight out of university to come into the work group and say 'this is the way I would do it'.

Clara Hughes: We are an industry that stands on practices that have been in place for many years as requirements have tended to evolve slowly over time. I remember when I graduated and started working, there was an actuary who still had more confidence in the result coming from a calculator than a spreadsheet! The industry has moved on a lot since then, with people embracing and enjoying innovation, which is something that benefits all of us.

Darragh Pelly: I work on the Society of Actuaries in Ireland's Wider Fields Committee and I worry that students might not come into the actuarial profession because when they land in insurers and see they're all using Excel - while in college they have been using Python or R - they think 'this isn't for me'.

Hurdles to innovation

"When someone comes in and says 'we want to use Python', what is the blocker?" 

Brian Walsh: When someone comes in and says 'we want to use Python', what is the blocker? Is it IT or the actuarial department?

Abi Holloway: For us, it tends to be more IT. They want to look at different solutions, and then they need to take steps to make it production-ready. Sometimes the established review actuaries can be uneasy with technology they are not familiar with.

Lorraine Paterson: Part of our strategy is to become a data and insight-driven organisation. There is therefore a culture of innovation across all our colleagues.

We have improved our experience analysis over recent years using Scala, Python, R and Power BI with a connection across the different tools. This has enabled a decommission of over 100 spreadsheets and long processes of data manipulation. We have developed this in an agile manner using a building block approach. Building each block frees up time for further developments.

William Diffey: There is quite a lot of work going on in the UK actuarial community to show data science can be pursued within the actuarial profession. There is a challenge in how these tools can be used and fitted to the data, and obviously you need good quality data platforms in order to have a tool to fit.

Part two of this roundtable will be published next week.