Actuarial transformation to meet the needs of insurers
With the rise of big data and machine learning, some would ask where actuaries are positioning themselves in the next few decades. We came up with the term “actuarial scientist” to mean an actuary who possesses the actuarial skills and domain knowledge, coupled with the understanding of how to use today’s technology.
You may have heard other terms floating around such as “digital actuary”. In my view there is definitely space for both actuaries and data scientists to thrive in the insurance world for years to come. However, I do believe that there is a necessity in some areas for actuaries to fill more of a hybrid role as an actuary and a data scientist; that is, to be “actuarial scientists.”
At Barnett Waddingham, we have been upskilling our team for a few years and converting processes to R, SQL and Python, where appropriate. We still have some way to go, but we’re keen to share with you our vision and some of the things we’ve learned when helping clients.
Our process automation vision
We have been on a quest to automate our own processes for some time. Our ideal state would be a no-touch process; where everything from reserving to pricing is fully automated, preferably with voice activation. This may be a bit far-fetched right now but many actuarial teams are currently in pursuit of a one-touch reserving or pricing process.
Although we haven’t got to the no-touch state yet, we do have some processes that are run with a click of a button to obtain the first cut results. By optimising the time spent on processes, actuaries can spend time on value-added work and have more time to get involved in new areas across the organisation.
In this blog, we have used process transformation with R as an example of upskilling our actuaries in using new tools and exploring other ways to build models. These include building bespoke parts of internal capital models, for validation, reserving, pricing and data processing. However, the same will apply if you are upskilling actuaries with other software and tools. Some of our R case studies are on our website and can be found below.
Upskilling our actuarial team – key lessons learned
Based on our own journey and our clients, here are a few things that one should consider when upskilling one’s actuarial team.
In the past, we have tried asking our actuaries to learn R/Python by giving them a book or following online courses themselves. This is one approach but, in our experience, the amount of learning between individuals varied and the pace of transformation was slow. The pace of change in R adoption at Barnett Waddingham accelerated tremendously after we gave our actuaries a bespoke training course presented by Allan Engelhardt, a data analytics expert. This involved a few days of training in which a number of our staff were taught the basics of R and given the training they needed to be able to code a project that was relevant to their day to day work.
It is important to make time for automation projects. Testing new tools and building new models always take longer than expected. For example, we have sometimes found it necessary to allocate protected time for these projects, with a few consecutive days blocked out with no other interruptions.
Some of our earlier projects have taken a few attempts to get them right. It’s important to factor in time for the team to explore and fail. Failure is a learning experience, as even the best team will not get everything right the first time. Fail fast, learn quickly is the lesson!
For the actuaries who are learning how to code and are building new tools, it may be frustrating having to look up how to debug unfamiliar error codes in the R models. Or there may be no other team members who know the answers. Muscle memory plays a large part in coding and, with time and practice, you will become more proficient with R or other platforms you may choose.
There is no point in training the team without putting the skills into practice. It will get quicker and easier for the team once they get over their initial learning curve and there will be less key person risk. The more people who can pick up a task, the more flexible your team will be. They will be able to complete automation projects faster in future and the results will be to a higher standard. So the benefits are worth the time and investment!
Not everyone may buy in to your vision or some may not be sure if they want to learn R; it is very different to Excel. We have actuaries who had not coded anything prior to joining us but are now proficient in coding. We encourage our team to have a growth mindset, which fits very well into one of actuaries’ natural elements.
Many actuaries are curious to learn, so we instil a mindset that any skill can be learned as long as you put your mind to it. You probably couldn’t build a cashflow model in Excel straight out of school, but you learned. You might not be able to automate some data checking in R now, but if you put the time in you will learn. Upskilling takes time, effort and a willingness to continually learn.
Excel is an excellent tool, for some things. It is important to re-evaluate where Excel is and isn’t working for your team. If you are storing 100,000 rows of policy/claims data in an Excel spreadsheet, it’s time to try SQL. If your Excel model is so large you need a virtual machine to open it, it’s time to try R or Python.
Automation is not just about running the R code. To set up a good model it’s important to have a big picture view – not just hit run and job done. You should add a few more things to your to-do list such as the below.
- Check outputs match expectations
- Build in checks that flag if the outputs are not reasonable
- Ensure information is easy to read for others
- Add appropriate checks throughout the whole process
- Ensure this fits with the rest of the process
- Have adequate documentation
Plus . . .
- Ensure good version control. One of the advantages of R over Excel is that R can be used with Git and has very good version control capabilities which, in our opinion, are far superior to Microsoft Office.
The IT department may not be familiar with R and the environment required to run R in an optimal way. For example, getting R installed, using Git, hooking R up to existing databases and so on. Insurance companies often have legacy systems from which it may may be difficult to extract the data required, or there may be lots of levels of review to get programmes approved for use.
Having a good and flexible IT team behind you makes transitioning to R much easier.
Senior managers do not need to be R coders
It is possible to review the work done in R even if you do not know how to code. Senior actuaries can review assumptions and methodologies in R, together with analysts to check if the assumptions and results make sense. Assumptions can be updated instantaneously and the results can be signed off much quicker than previously possible. Graphs can be outputted as part of the process to allow senior managers to easily visualise and get comfortable with the results.
Actually, it encourages and forces us to do more on job training as we review the work done together. I am sharing more of my thought process with my team and, at the same time, I am improving their knowledge.
Be purposeful with your programme of change
In summary, to upskill your actuarial team, you need to be purposeful with your programme of change.
- Upskill your team with an expert who has the technology know-how, the skills to provide bespoke training for your business and the ability to set up best practices for your team to follow
- Be patient and give space for the team to grow
- Consider the IT infrastructure requirements and have an IT environment that optimises the tool and platform
- Be more than just an actuary
Many of our team members are as comfortable with R as they are with Excel. We routinely invest in automation projects ourselves and see the benefits of this work on a daily basis. We are also helping our clients design automated processes or enhance their existing ones.
We have shared some of the ways in which we have used R to deliver high quality outcomes for our clients. See the case studies below.
You can also see our client testimonial on how we have helped upskill their actuarial teams in this area too.
Get in touch below to speak to us today about your actuarial science needs.
Applications of R in insurance: demographic experience monitoring
Read our second case study in the series as we showcase how have we used R to improve demographic experience monitoring.Find out more
Stay up to date
Get the latest independent commentary and exclusive insights from a range of experts at the forefront of insurance, risk, pensions and investment – tailored to your preference.Subscribe today
Unlock your Employer DNA
Businesses share many similarities, but – as with human DNA – the precise way your business operates remains unique. Decoding this DNA can give you the insights and actions that will allow your business to evolve and thrive.Find out more