What makes someone an actuary?

I work on a data science team at an insurance company. As such, some of the work we do is similar to what actuaries might do. We file some of the models we create. Some of the people on the team have worked in teams that were considered actuarial, but more than half have not. I think only one or two people on the team have any actuarial exams passed.

When filing models, some DOIs ask questions that seem to assume they’re dealing with actuaries, such as when they specifically mention actuary or ask how guidance from certain ASOPs was followed. This is somewhat awkward. I wonder if ASOPs even apply to us.

ASOP 1 says:

The Actuarial Standards Board (ASB) promulgates actuarial standards of practice (ASOPs) for use by actuaries when rendering actuarial services in the United States.

But, it never defines actuary, as far as I can see. I think it’s just sort of assumed that only actuaries would use ASOPs. I’m not sure if the definition of actuarial services is helpful, since it might be somewhat circular logic:

Professional services provided to a principal by an individual acting in the capacity of an actuary. Such services include the rendering of advice, recommendations, findings, or opinions based on actuarial considerations.

What do you think? Would this team be considered a group of actuaries? If not, how do you think it would go if we responded to a DOI by saying ASOPs don’t apply to us?

Well, since you’re filing rates in the US some credentialed actuary would have had to sign it. So most technically they’re probably directing the question at said credentialed actuary.

2 Likes

Assume DOI questions ask for details about the model and modeling process that only our team can answer and are directed at those who did the modeling.

I think the onus would be on the actuary to ensure ASOPs were followed through his discussions/due diligence with the modeling/data science team.

1 Like

“actuaries” work in the design, building, and maintenance of insurance systems. so if you’re working in that capacity, you’re doing “actuarial” work.

then I think you’re an “actuary” (small-a), rather than an “Actuary” (designated by a professional body and accountable for following regulations).

IMO, you’re modelers doing one aspect of actuarial work. You are most definitely not an actuary, even in the little-a sense. ASOPs don’t apply to you, but you should probably understand their impact on the creation and maintenance of rates where necessary because actuaries (little-a or Big-A) are responsible for knowing it, and it would be helpful for you to understand what you can do to help them fulfill their responsibility. [The discussion of how to do that is way beyond the scope of a post. Plus, I’m spending my last 13 hours pre-surgery doing other stuff.]

You’re able to answer questions about how the model works and all of that. That’s important. You probably have zero clue how your work satisfies various ASOPs. That’s OK, because as noted above, you’re not responsible for that. An actuary - someone credentialed by a professional actuarial organization - would be.

I’m not a DOI, so I have no idea how they’ll respond when you say you’re not an actuary and so ASOPs don’t apply to you. If it’s a nit-picky state, my guess is that (1) the DOI will ask if an (credentialed) actuary has reviewed the appropriateness of your models in developing rates; if one has, then it’s on that actuary to answer questions about the model’s appropriateness vis a vis the applicable ASOPs, and (2) if an actuary has not reviewed the appropriateness of the model, expect the DOI to either farm it out to an actuary to pick apart your model’s flaws or tell you “the filing is disapproved; go have an actuary review it in light of the applicable ASOPs, then come back to us.”

If it’s not a nit-picky state, it may be a finger-wag about it or they’ll just check a box and move on.

4 Likes

You should not be doing the response work if you’re not an actuary. An actuary would take your response and review it, understand all the modeling processes and assumptions, and respond to the DOI on your behalf. And that actuary would make sure all the ASOPs are followed.

Also, what Ted said. If you respond saying you’re not an actuary, your response will most likely get shot down until you can find a third party consulting actuary to review on your behalf.

An actuary is someone who is credentialed by the appropriate organization. We’ve had this discussion on my team before since we are split between actuaries, actuarial students, and non-actuaries. As a non-actuary, I am not obligated to follow any of ASOPs but my credentialed teammates are. It does allow me to be a bit more cavalier in some aspects of my work, such as documentation, but the ASOPs are generally good practice to follow. If one of my models was filed, it would be up to the signing actuary to respond to the ASOP questions like Ted said.

We have actuaries, actuarial students, and non-actuaries on the team. We read/review/discuss/answer questions on the applicable ASOPs with every new actuarial student in the first couple months, and every couple years we do a refresher for the whole team on them. If there is something that I need modeled by someone else, and it needs to follow ASOPs, I will specifically note what additional things I expect to need reviewed up front, and then also review for any potential miscommunication, or things I didn’t think of at the beginning. If I’m putting my name on it, I’m making sure it’s done in accordance with ASOPs to the best of my ability/knowledge. I don’t think I’m odd in that respect.

1 Like

You’re not odd. This is required by the Actuarial Communications ASOP.

3 Likes

I’m not American, but I’d expect that it depends on what you’re doing. Id expect regulators have some things that require a credentialed actuary (credentialed by one of the actuarial societies) approve or sign off on it. If it’s not covered by the regulators, then doesn’t matter if they’re ana ctuary by any definition. But if you’re dealing with regulators, I’d expect that they are assuming you have a credentialed actuary in board. You’re best bet? Ping the chief actuary at your company and ask them.

In Canada, the term actuary is reserved for people credentialed by the Canadian society, so there’s no confusion over the term.

Had a data scientist on my team who I tried my best to explain this to. I don’t care how much lift your model output produces, if your entire process for determining is not documented, it’s useless to me.

He never really got it. Only person I’ve ever worked alongside in my career who got fired. It was deserved.

1 Like

I see a lot of cowboy coding from data scientists. How do you expect people to be able to review your work and understand your assumptions if you don’t document? But most of the time when I’m looking at a data scientist’s model I get, “Trust me, this works.” or “It’s really complicated, you won’t understand.”

1 Like

I take it they’re also into bitcoin?

1 Like

One of the DS I work with definitely is full of himself. I make sure to put him in his place whenever he pulls that that kind tone with me.

“It’s really complicated, you won’t understand.”

Lol, dude did the same thing to me all the time. I took that as code for “I don’t understand how this works, please don’t dig too deeply or my ego will be hurt.”

Literally never worked with someone so condescending.

I think Bro is narrowing in on the issue. In the US, for the American Academy of Actuaries, the Code of Conduct defines actuary (uncapitalized) and Actuary (capitalized).

It seems to me that a non-credentialed actuary may perform actuarial services, but only a credentialed Actuary is subject to the Code of Conduct. I don’t see much of the Code as applying to actuaries. ASOPs however are binding to Actuaries, but they are addressed to actuaries.

So, I don’t think an actuary can get in trouble with the American Academy for not following an ASOP. But I can see an insurance department questioning whether an ASOP was followed, and if not why not.

Any comments?

1 Like

I wonder if he used premium as a predictor and losses as a response variable, and then concluded that policies with more premium have more losses and are thus less desirable.

I also wonder if he used any of the following as predictors in a pricing model for one line of business:

  • ratio of elementary students to number of heavy trucks licensed, per ZIP code
  • ratio of apartments to non-fast-food restaurants, per CBG
  • ratio of plumbers to construction companies, per ZIP code
  • number of new construction single-family houses in last 10 years, per ZIP code