Is targeting different loss ratios common?

[PSUActuary]

In my previous experiences with larger companies, generally there would be a target loss ratio for each underwriting company (preferred, standard, non-standard, etc). This target loss ratio by company was largely based on the differences in expenses for each of the companies. And then the loss ratios for business within each company would generally be similar across all risks.

With just one underwriting company and a book of business that varies considerably in terms of retention (and thus expenses), is it reasonable to target different loss ratios for different customers? For example, using a cancellation model and splitting the book into 5 segments, the best segment might retain at 90%, while the worst group only 30% - with the other groups in between. When marketing is spending a few hundred dollars per sale, customers that retain at only 30% need to be priced at a much lower loss ratio in order to have any chance at making money on those customers.

We are working on building a lifetime value model to help determine what the NB loss ratios should be given various retention levels. Assuming the results showed a group with 90% expected retention should be priced at 85% NB loss ratio and another group with only 30% expected retention priced at 50% loss ratio, would it be acceptable to target those different loss ratios when proposing to change factors in a rate filing? Or would that look unusual or discriminatory to a regulator?

When doing an overall rate level indication by coverage, the permissible loss ratio would be derived the more traditional way by using the overall expenses by coverage. What I am trying to figure out is the best way to price preferred business and non-standard business differently (once expenses are considered) within one underwriting company. Would showing differences in retention be enough support to justify targeting different loss ratios by credit score, number of vehicles, years with prior insurer, etc?

Greatly appreciate any comments or similar experiences! Thanks.

[MountainHawk]

I have never tried to file with a LV model as support, so take this all with a grain (or a heaping pile) of salt.

I would expect this to be a hard sell to some states (cough, New York, cough). You are essentially amortiziting the cost of selling the policy over the expected number of renewals, so they may take a hard line and say you can only recognize the policy you are writing, so the LV model is not a justification for lower rates on more-likely-to-renew business.

Lifetime value is absolutely the theoretically correct way to look at this in an economic sense, but I would expect significant pushback from the more hard line regulators.

[tometom]

I pretty much agree with MH about the LV model.

You didn't really mention a product line, nor how you were determining these segments, so just throwing an idea out there. Are the segments easily identifiable in the underwriting process? If so, I'd think most states (maybe not all, such as NY, but most) would accept an expense study showing that you have considerably higher expenses for one segment over another, thus allowing you to adjust the PLR by your segments. Your actual data would need to support what you are assuming in your model.

[Vorian Atreides]

What variables do you use for the LV model? How many of them involve some sort of control by the consumer?

Also, what's the potential for a risk to change/migrate (apart from tenure with your company) to a different category?

[Wmorrissey]

Originally Posted by PSUActuary

In my previous experiences with larger companies, generally there would be a target loss ratio for each underwriting company (preferred, standard, non-standard, etc). This target loss ratio by company was largely based on the differences in expenses for each of the companies. And then the loss ratios for business within each company would generally be similar across all risks.

With just one underwriting company and a book of business that varies considerably in terms of retention (and thus expenses), is it reasonable to target different loss ratios for different customers? For example, using a cancellation model and splitting the book into 5 segments, the best segment might retain at 90%, while the worst group only 30% - with the other groups in between. When marketing is spending a few hundred dollars per sale, customers that retain at only 30% need to be priced at a much lower loss ratio in order to have any chance at making money on those customers.

We are working on building a lifetime value model to help determine what the NB loss ratios should be given various retention levels. Assuming the results showed a group with 90% expected retention should be priced at 85% NB loss ratio and another group with only 30% expected retention priced at 50% loss ratio, would it be acceptable to target those different loss ratios when proposing to change factors in a rate filing? Or would that look unusual or discriminatory to a regulator?

When doing an overall rate level indication by coverage, the permissible loss ratio would be derived the more traditional way by using the overall expenses by coverage. What I am trying to figure out is the best way to price preferred business and non-standard business differently (once expenses are considered) within one underwriting company. Would showing differences in retention be enough support to justify targeting different loss ratios by credit score, number of vehicles, years with prior insurer, etc?

Greatly appreciate any comments or similar experiences! Thanks.

Exam 9 has a paper by Panning that considers the franchise value(value of future renewals) which you might find useful.

[Arlie_Proctor]

I've been out of the rate filing business for over a decade, so I really can't comment on what a regulator may or may not perceive as acceptable in the current environment. However, your approach seems to be based in sound actuarial theory, so my hope would be that many regulators would accept it. In general, however, it will sell better in most states if you build the rate structure using discounts as opposed to different target loss ratios, especially if insureds can do something to achieve those discounts. For example, insureds who have been with you for five years probably have a much better retention rate than those initiating coverage. If that translates into a 5% total cost savings, file it as a 5% discount, not as a difference in target loss ratios. Everybody likes discounts, nobody likes penalties, even if they get you to exactly the same prices.

[Mary Frances]

Well said, Arlie.

[great3981]

I also remember the Feldblum paper that discussed lifetime value and retention models. It was one of my favorite papers from the old Exam 5.

[tude]

Originally Posted by Wmorrissey

Exam 9 has a paper by Panning that considers the franchise value(value of future renewals) which you might find useful.

But the purpose of that is show that you can migitate IR risk through pricing. Moreover, it can be hid from rating agencies and regulators, unlike changing the composition of your asset portfolio. OP wants to actually file this with a regulator.

[PSUActuary]

Thanks all for the great feedback!

I guess my main concern over any other is that I could potentially get charged with discrimination. The last thing I want to be doing is defending the rates in court. It sounds like you all agree that the approach is at least sound actuarially. I fully expect some states to have a hard time accepting and may not allow different target loss ratios. I am pretty sure a few of the bigger states we are in will not have a problem initially. By the time we are big enough (assuming we are successful) to be on any consumer watchdog radar, hopefully we will have mutliple underwriting companies to place the different types of risks in.

The Feldblum and Panning papers came to mind - the Feldblum paper seemed the most relevant for this particular problem. That is the structure I followed with some subtle variations. We are working with the various depts to capture the detailed expense information. I think I will be able to support the difference in long-term costs and thus different targeted loss ratios necessary to produce similar lifetime profitability.

Ultimately, I am not nearly as concerned about having a filing rejected as I am about having the filing go through and then get questions on it sometime in the future and have to defend an approach that isn't necessarily common in the industry. I feel that it is absolutely necessary for our particular circumstance though. The CEO and mgmt all feel that this is the obvious approach (as MH said - makes sense economically). I feel that I would be damaging the reputation of the actuarial profession if I told mgmt we need to price the high freq / high cancellation customers at the same LR as those with low freq / high retention, when no one (including me) thinks that it is appropriate. But that doesn't mean regulators are going to feel the same way.

As Arlie said, using discounts is always good whenever possible too!

Thanks again for all the great feedback everyone. I feel better now. I will let you know if I need any of you to come to court with me