Let’s be honest, the next $100 million blockbuster isn’t going to be titled The Actuary, starring Liam Neeson as an experienced FSA looking to make sure that pricing, valuation, and forecasting are all on the same page with their best estimate mortality assumption. We’re not going to hear that famous Hollywood voice-over saying “The fate of the quarterly dividend DEPENDS on it.”
It’s just not that interesting a subject for large audiences, but appropriate assumption management is a critically important topic for actuaries, the companies in which they work, their regulators, and their customers.
Assumption frameworks have been around for years. Many of these are driven by regulatory reporting requirements, whether necessary because of Statutory, GAAP, or, now, IFRS regimes.
Plus, different actuarial tasks may require various assumptions and various assumption bases, from best-estimates to margins for adverse deviation to values that are solved for in order to achieve management targets.
Not only do these bases and regulatory frameworks have some specified assumptions, they often also diverge from one another. And so we as actuaries may find ourselves using assumptions in a Cash Flow Testing model that differ from the same model and the same product we worked on two years before in a pricing rotation.
It’s a Bit Much, Isn’t It?
This proliferation of assumption sets, modeling requirements, and documentation can be overwhelming.
Many companies have, in response, developed their own model and assumption governance structures. Despite this start, challenges remain. In a presentation to the Society of Actuaries at the 2016 Annual Meeting, results of an interesting poll were presented. The plurality of respondents (43%) said that assumptions are still set in multiple departments [question #7]. That’s a bit of a problem, because such decentralization leads to many different priorities for assumption determination. However, it may not be the biggest challenge facing the actuary, because each department has specialized expertise in their product or function.
What’s more concerning, though, is question #9. “Where do official actuarial assumptions ‘live’ at your company?” had the greatest response (29%) for “Semi-centralized combination of above [database, Excel files, actuarial software]”. The second-highest (20%) was “Decentralized Excel files in many locations.”
From this, we conclude that fully half of the actuaries functioning at a high level are managing assumptions without much consistency, transparency, or accountability.
Let’s take a look at each one of these assumption management challenges in turn.
Actuarial Assumption Management Challenges:
#1 – Consistency
There is a clear need for consistency across many facets of an actuarial model. From the hierarchy of results reporting and storage to the table layouts themselves, having consistent structures means that anyone at the company can understand, and use, the model and its underlying assumptions without significant modification.
This might show up in the fact that various departments think differently about certain assumptions. Some always think of everything in terms of policy year, some always think in terms of calendar year. Passing inconsistent models between these two areas can lead to confusion, poor output, and workarounds.
Therefore, having a consistent structure (not just for tables, but also for how much of an assumption is a best estimate and how much is a margin, and what’s constrained by regulation) for assumption setting and management can only lead to smoother modeling process.
#2 – Transparency
A good way to create consistency is to first focus on transparency. That is, making sure that everyone can see and understand what assumptions are actually in use in the different departments and for which functions. Creating a centralized system for assumption management is a great way to ensure that everyone can observe all the other assumptions in use with regularity and that they’re not diverging from the rest of the business.
If divergence does start to show up, then the actuaries (and other business professionals involved) can have a discussion about which functions take priority and remain unchanged; which others are of lower status and should be modified; and which are best left to differ.
#3 – Accountability
This is a major problem for those decentralized, divergent “multiple Excel file” systems. Most of the time, these systems have no formal (or even informal) documentation about what assumptions used to be there and, if they’re now different, when they were changed, why, and by whom. Then, when results from the model are not what was expected (or desired), investigations into the values themselves ensue. “That lapse rate is 10%? Why? It was 5% last year. I know, because I changed it from 2% the year before! I remember that.”
Some systems allow for such change documentation. Few require it. However, when you’re dealing with decentralization and, more importantly, just using Excel files, it’s not likely that you’re going to have much luck ensuring compliance. You can hope that people remember to make notes about what they changed, but you will also do well to remember that “hope” is not a strategy.
A Viable Solution in the Cloud
Now, we may be a little biased, but at Slope we believe that cloud-based software is the coming revolution for actuarial models. Model code, assumption sets, and results stored on a cloud-based system offer a feasible solution for each of those problems. Consistency? Transparency? When every formula and assumption table can be made available to any actuary in any department, there absolutely will develop both consistency and transparency as everyone can see what everyone else is coding.
And as for accountability, cloud-based tools provide change logs for all versions, not only of code, but of tables: who updated it and when. Because of the cloud-based archive, you can also find out what it used to be, just to confirm that the 5% last year value really was 5%.
All of this leads to better assumption management, better model management, clearer results, and a better work product.
Can you imagine that voice-over? “Finally, The Actuary was able to explain EXACTLY what the IFRS margin for adverse deviation should be, and why it needed to differ from pricing’s best-estimate assumption. And everything was right with the world once more.”
Hey, it could happen.