So…what can you do during a rate revision?

So, apologies for naive questions, but then again that’s what this forum is for(and AO is a thing of the past).

Put yourself into the position of a ratemaking actuary. A revision for a line you’ve had in force for 5+ years is approaching, what can you do? I feel intimidated sometimes when making proposals, and am okay with getting feedback from internet strangers(though I’ve met one of you in person!)

• Base Rate only changes. This one is pretty straightforward. Change base rates with the goal of increasing or lowering prices equally for the entire line.
• One way analysis loss ratio supported changes. with credibility and a complement of no change, use relative loss ratio performance to increase or decrease relativities. This can be done to one segment or multiple, and you may, or may not adjust the premium for subsequent one way analysis. So if you make an adjustment to relativities in segment1, you may or may not use the proposed new premiums to support a one way analysis for segment 2. A one way analysis can include a base rate offset to rebalance or accomplish overall rate increase or decrease.
• limits/deductible analysis. This can be done through a loss ratio analysis, or using limited expected value/expected excess loss styles of analysis, and can follow with base rate changes to rebalance or accomplish overall rate increase or decrease.
• Modeling: This is a more complex process where multiple segments are revised through some modeling process(a GLM is not uncommon) to adjust rate in numerous segments while taking account the relationships between variables as defined by model assumptions.
• The CA DOI’s Sequential Analysis. I’ve not had to use this in practice so it’s not familiar to me.

And I’m concerned that’s about it. I realize there should be more in my toolkit, but I’m not sure what else there should be, and I like to be prepared for the situations where modeling isn’t something I have resources for, but want to be open to other changes in existing segments. Say you have a segment for which you have pure premiums for, would you implement just those pure premiums relativities? For example I wasn’t sure in a previous life if suggesting factors for vehicle age was actuarially sound. I want to not be limited in my thinking or awareness for what types of changes are typical or under the common scope of a rate revision(assuming regulatory approval and IT smooth implementation)

What else am I missing?



“What was done last time?”


You’d probably credibility weight the pure premiums or relativities against a larger cohort (the entire state, for example, or industry information). But honestly that kind of one-way analysis and modeling really are the toolbox (where modeling is basically a catchall for analyzing across more than one way). The loss ratio view is really the same concept but you’ve divided things by premium.


does it benefit you in any way to do more than a base rate change?

1 Like

Potentially, if there are segment level issues, but even asking this question and learning other fundamental questions to take ownership of a line are good reminders i think. I suppose you don’t know the answer to that question unless you either look yourself or reach out to claims staff or other business partners to see if they have concerns.

This is what i thought was the case, and am wondering if other experienced actuaries can confirm from their own experience too, that one way loss ratio analysis and “modeling” generally is the complete toolbox.

Regarding the pure premiums, they were countrywide so no company level data to be larger, but “industry data” is almost always a good consideration.

Thanks @NormalDan !

You have to do the basics to not get fired by management. This includes getting your relativities and ILFs on par with what competitors are doing, and if there are factors not currently in the rating plan, get them in there to not be selected against. The filing will be easy since you can use competitor filings as support.

If you are trying to beat the competition, then you need to do some in depth analysis (GLM, etc.) and will likely spend a lot of time answering objections.
I asked the question because, unless you’re going to get rewarded for doing extra work, it is definitely not worth your time.

All good. I’m not exactly under pressure or anything, it’s more of “under circumstances of your choosing, how would you go about choosing what to do in your revision.”

LOL @ not get fired by management. But yes, i would prefer my way to earning a positive rep, though your point is taken!

Note that this question wasn’t exactly a “tongue-in-cheek” response.

It is important to know how the current rates were determined; and a key part to evaluating any new method is to ascertain if it’s an actual improvement on the prior (i.e., the current) method.

1 Like

One thing I found helpful is to look at the policy profile of the book. Like policies by age, zip code, etc. If you have a lot of policies in one category, your rates may be too low. If hardly any, rates may be too high.

I find a good rule of thumb is to see whichever group the underwriters are really keen to write/protect. That one is probably rate deficient. :laughing:

if you’re not pissing off underwriters, you’re doing something wrong

I didn’t perceive your response as “tongue-in-cheek” and i appreciate the thought process of was the prior process good, and should i replicate the process, or is there an improvement.

An alternative view to “what was done last time” is asking if the past time the product was reviewed, was that review only one part of a larger, longer time scale product. Like this year we get XXX on countrywide, next year expand to rate on YYY, or some other longer term product. I’ve been in meetings where the discussion also includes to “look at things next time”