Actuarial model conversion is perceived as difficult.
Ask any actuary who’s had to convert a model from one system to another, and they’ll tell you the same things:
It was confusing, because it was hard to understand just where we were starting from.
We didn’t really know what we were doing when we started, so it was tough to measure success.
We’re not sure we really got all we wanted.
It took too long.
And the best one:
We never want to go through that again.
Those experiences get shared and spread to colleagues and peers at the Annual Meeting or through local conversations. They convince others to avoid a “conversion” or “transformation” project, out of fear or insecurity. They allow inertia to restrain advancing an actuary’s practice, under the guise of “risk avoidance”.
All of those quotes are very fair statements. We’ve been there. We have multiple actuaries on the team who have said those exact same things.
But those are generally not good reasons to not improve your actuarial modeling systems. They are usually coming from challenges with the prior modeling system, not the new one.
Because, frankly, if actuaries were completely satisfied with their old systems, they wouldn’t be looking to convert.
So, yeah, they start a conversion (or “modernization”, or “transformation”) process, and, in the end, find themselves not much better off than they were before.
And therefore they say all those things above. Which, again, aren’t great reasons not to make the next advancement in your actuarial practice.
They are good reasons to have a great process in place.
It doesn’t have to be so hard.
It’s no secret that actuaries are not generally seen as the most cutting-edge of professionals. We joke all the time about introversion (“She’s looking at your shoes!”), lack of personality, and the tendency to fetishize precision over accuracy.
There are some good reasons for this: actuaries spend a lot of time considering risk, they have highly specialized knowledge about complex insurance products and processes, and they’re often so dedicated to their craft that it takes a decade or more to become fully qualified.
All that said, it should be no surprise that many actuaries are still using actuarial forecast systems from five, ten, or twenty years ago. It may be a series of Excel worksheets that were set up once for a test model, and just became the standard. It may be an old commercial system that has been in place for a long time, and since it does what you need, why bother worrying about it?
For all the good that such structures have allowed actuaries to do over the years, like any tool (or appliance, or item of clothing), they eventually wear out.
Those macros might not run very fast, but you just don’t even want to think about the downtime to rebuild them. Your assumption sets, well, they’re not laid out very well or transparent, but you know how to work around all those flaws to get what you need in a “reasonable” time frame.
Which you’re constantly justifying because that’s just how long it takes.
If those are challenges you’re dealing with, it might be time to upgrade your actuarial systems.
You need the right structure and mindset.
Over a series of forthcoming blog posts, we’re going to provide some structure to how an actuary may go about implementing a new system in a very structured way. From deciding that it’s time to do so, to evaluating options, to actual implementation, to completing the relationship with a former vendor, we’ll provide some guidance so you can properly arrange your thoughts.
That way, you don’t have to think through everything on your own. And maybe you won’t end up saying any of those quotes above.
Now, that doesn’t mean we’ve thought of all the little nuances you’re going to encounter. We can’t be that prescient. [If we were, maybe we’d be buying lottery tickets instead of selling software.]
But we have considered most of the steps you’re likely to encounter, and will be offering you a framework for how to make the majority of those decisions you come across. You will be able to take these outlines and adapt them as you need for your company, your product portfolio, your risk profile, and the way you wish to do business in the future.
Because, remember, just because something exists now, isn’t a great reason for it to do so in the future.
If your systems don’t allow you to actually be an actuary (you know, that thing you may have earned a degree in, and certainly studied hundreds of hours to pass exams over), and instead find yourself managing the system itself, it might be time for a change.
Here’s a quick outline of the articles in this series:
1. When Should You Consider a New Actuarial Software System?
There are many instances in which it makes sense to consider buying a new system to replace what you currently have. Or, you could look to add on some functionality for a specific product line or process. What are some good reasons to consider a new system? Are there any bad ones?
Read part 1 here. And part 2 here.
2. How To Make a Business Case For Conversion
Actuarial model conversion processes often need buy-in from many levels: the front-line actuaries, their supervisors, and those who are funding the project. What are the aspects you need to consider when suggesting this kind of project? What might be the budget elements to include? How much is your time worth? Can you find evidence of others (peers or competitors) who have gone through a similar process?
6 Key Components of the Case for Actuarial Modeling System Conversion offers an option to access an Excel spreadsheet you can use to quantify potential costs and advantages of switching systems.
3. How To Evaluate Your Options
What dimensions are going to be important to you as you consider differing systems? Is user flexibility versus standardization important? What about cloud-based versus housed on your local server? This article provides a large number of categories for inclusion in your decision process.
9 Criteria for Evaluating Modern Actuarial Software Systems.
4. Good (and Bad) Reasons To Choose One System Over Another
Following-up on the How To Evaluate Your Options, this article offers advice on how to consider some of those intangible elements of the decision-making process. We’ll offer our thoughts on good, bad, and ugly decision-making paradigms.
Read Good (and Bad) Decision-Making Criteria.
5. How To Convert Models Into a New System
Finally! We get to the good stuff, the actual model conversions and comparisons of all the little decimal places.
This article is a welcome resource for those actuaries who may be already marching down the path of conversion, and looking for some guidance on what to think about as they actually do the work. Error tolerances? Prioritizing processes versus prioritizing products? Planning around standard work responsibilities? We’ll provide our perspective on how you can think through many of these issues.
How to Convert Your Actuarial Models
6. How To Break Up With Your Old System
Now that you’ve gotten everything converted, what do you do with your old models and procedures? What about results and documentation? How long might you need to keep all these records around?
How to break up with your prior vendor.
7. How To Maintain Momentum and Validate Your Decision
Once the dust settles and you’ve achieved your “new normal”, what does that look like? What should you be considering as you go forward? How do you keep your models in line so they don’t get to the point (as in article #1) of being so unruly and burdensome that you feel like you need another conversion process? How do you ensure that you aren’t wasting time with value-detracting processes? How do you set your team up for even more successful advancements in the future?
So, there you have it – a 7-part structure for planning and executing your actuarial model conversions. Keep watching this space for updates. We’ll be producing content very regularly throughout this series, such that you can follow along and be ready for your own successful transformation. Be well, and do good work!