Dumbest Rule in Insurance...

MrGolf

Super Genius
100+ Post Club
150
has to be the combining of Experience Mods for common majority ownership. It never fails to work against you, but very rarely helps you.
 
has to be the combining of Experience Mods for common majority ownership. It never fails to work against you, but very rarely helps you.


It can be viewed as a necessary evil. About 20 years ago, there was an epidemic of businesses with horrible loss experience simply creating new corporations and getting new mods of 1.00 even though nothing had changed as far as the ownerhip's lack of loss control.

----------

Credit scoring model for pricing.

I've probably shared my personal experience with this here. My homeowners renewed with a $1,000 premium increase based on a bad insurance score. Turned out they had the wrong information...they used the score for a couple with the same last name that lived 1,500 miles from us.

Two years later, we got a $700 increase, again citing a deficient insurance score. Our credit is impeccable. When I inquired, they cited three "reason codes" that were very easy to refute. I went all the way to the carrier's VP of underwriting to get it fixed and his response was the insurance score reasoning was incorrect and the increase was still warranted based on a statewide rate increase granted by the state DOI. Having contacts at the DOI, I found out they had not had a rate filing approved in at least 3 years, so he lied. I moved my account at that point.

There is a lot of credible evidence that insurance scoring is often based on erroneous or misapplied credit information (credit reports are notoriously inaccurate), is too rigid and sometimes unfairly applied, and allegedly too often based on correlation and not cause-and-effect. But it's cheap and easy. Credit reports don't have salaries, benefits, or take vacation or sick leave like underwriters.

Expect it to get much, much worse with the growing reliance on "big data." Just recently an agent friend had an account with a large premium increase and the underwriter could not tell him why or how to fix it. They said that their rating system had dozens of variables and they had no practical way to determine how these factors combined to produce the final premium. These types of "black box" rating algorithms make it impossible for insureds to manage their risks and insurance, not to mention making it impossible for regulators to determine if rates/premiums meet statutory requirements for being adequate but not excessive nor unfairly discriminatory.
 
It can be viewed as a necessary evil. About 20 years ago, there was an epidemic of businesses with horrible loss experience simply creating new corporations and getting new mods of 1.00 even though nothing had changed as far as the ownerhip's lack of loss control.

----------



I've probably shared my personal experience with this here. My homeowners renewed with a $1,000 premium increase based on a bad insurance score. Turned out they had the wrong information...they used the score for a couple with the same last name that lived 1,500 miles from us.

Two years later, we got a $700 increase, again citing a deficient insurance score. Our credit is impeccable. When I inquired, they cited three "reason codes" that were very easy to refute. I went all the way to the carrier's VP of underwriting to get it fixed and his response was the insurance score reasoning was incorrect and the increase was still warranted based on a statewide rate increase granted by the state DOI. Having contacts at the DOI, I found out they had not had a rate filing approved in at least 3 years, so he lied. I moved my account at that point.

There is a lot of credible evidence that insurance scoring is often based on erroneous or misapplied credit information (credit reports are notoriously inaccurate), is too rigid and sometimes unfairly applied, and allegedly too often based on correlation and not cause-and-effect. But it's cheap and easy. Credit reports don't have salaries, benefits, or take vacation or sick leave like underwriters.

Expect it to get much, much worse with the growing reliance on "big data." Just recently an agent friend had an account with a large premium increase and the underwriter could not tell him why or how to fix it. They said that their rating system had dozens of variables and they had no practical way to determine how these factors combined to produce the final premium. These types of "black box" rating algorithms make it impossible for insureds to manage their risks and insurance, not to mention making it impossible for regulators to determine if rates/premiums meet statutory requirements for being adequate but not excessive nor unfairly discriminatory.

I am all for using any piece of information for underwriting it if can be shown to affect claims. And I have heard a number of times that insurance score is a great predictor of claims, however I have never seen anyone show evidence to support this.
 
"I am all for using any piece of information for underwriting it if can be shown to affect claims. And I have heard a number of times that insurance score is a great predictor of claims, however I have never seen anyone show evidence to support this."

Someone once said something like "It's better to risk saving 10 guilty men than condemning one innocent man." Even if it was predictive and causal, there are so many opportunities for error, there should be concern whether it's appropriate. We should also question WHY/HOW it works as a predictive factor...as you say, no one really knows and all I've ever seen are suppositions. No question it serves the interests of insurers, but does it justly serve the interests of insureds?
 
"I am all for using any piece of information for underwriting it if can be shown to affect claims. And I have heard a number of times that insurance score is a great predictor of claims, however I have never seen anyone show evidence to support this."

Someone once said something like "It's better to risk saving 10 guilty men than condemning one innocent man." Even if it was predictive and causal, there are so many opportunities for error, there should be concern whether it's appropriate. We should also question WHY/HOW it works as a predictive factor...as you say, no one really knows and all I've ever seen are suppositions. No question it serves the interests of insurers, but does it justly serve the interests of insureds?

I disagree.

Insurance is statistics, there will always be outliers. I have no problem using credit if it really is a predictor of claims. With any type of underwriting, there will always be people the underwriting model says are high risk, but they aren't and vice versa.

Also, I disagree that it serves the interests of insurers if it is not of predictive value. If insurance scoring causes someone to be viewed as a higher risk than other underwriting evidence would suggest, then they will be charged a higher premium. So they may go with a company that isn't as credit driven in their underwriting. So ultimately, it is costing a company premium if they are using underwriting criteria that is inaccurate.
 
I don't think they're outliers. I believe the data sources are very often inaccurate or there are mitigating circumstances that aren't reflected in the model. I've only been adversely impacted (as far as I know) by credit scoring twice but both times it was in error. I believe the error rate is very, very high. In my early bureau days, we audited policies. In a good month, the error ratio was 40%. You would think that would be better today, but anecdotal information tells me different.

----------

There have been a number of studies about the equity of credit scores. The ones funded by the insurance industry have generally said everything's fine. The ones funded by those opposed to the industry claim that it's discriminatory. It's hard to find objective studies. Here's one for the federal government:

http://wweb.uta.edu/economics/workshop/Credit Scoring Paper 1 for UTA.pdf

First, using the two commercially-available generic credit history scores available to this study, we find that the credit scores of blacks are consistently below those of white nonHispanic whites or Hispanics whites, with the scores of Hispanic whites falling between those of blacks and non-Hispanic whites. These differences can only be partially explained by differences in age, marital status, income, and neighborhood (census tract). Indeed, the within-tract difference in creditworthiness between blacks and non-Hispanic whites is more than one-half of the overall difference, suggesting that blacks and non-Hispanic whites living in the same neighborhood can be expected to have substantially different measures of creditworthiness.
 
I have written thousands of personal lines accounts.

Bad credit is 100% indicate of claims. I've seen it every single day for the last 12 years. It's burned in my head through repetition.

bad credit ------> lapses -----> claims -------> lower limits
 
I have written thousands of personal lines accounts.

Bad credit is 100% indicate of claims. I've seen it every single day for the last 12 years. It's burned in my head through repetition.

bad credit ------> lapses -----> claims -------> lower limits

And I am willing to accept that. I just want something more than anecdotal evidence.

And to InsCommentary, I would say be careful of what you cite and in taking only one section. This was in the very next paragraph.

Second, differences between blacks and non-Hispanic whites in credit performance appear to be consistent with differences in measures of creditworthiness. That is, all else equal, a black borrower would be expected to perform worse (have a higher level of loan default) than a non-Hispanic white borrower with the same credit score. Everything else the same, this result suggests that lenders might have an economic incentives to discriminate by charging black borrowers higher interest rates (or having higher deny rates for black applicants) than white borrowers with comparable assessments of creditworthiness (as measured by credit scores). However, examination of loan denial patterns and interest rates on closed-end credit, suggests that fears of such actions may, for the most part, be unfounded. Conditioned on credit score, interest rates charged to black borrowers are only marginally greater than those charged to comparable non-Hispanic white borrowers, with differences significantly less than might be expected if differences in loan performance were taken into account.

I won't interpret as I find it rather clear, and certainly advocates a position I would rather not. Finally, InsCommentary also failed to mention that the very first page of this report says draft. So is it a final report, and if not what changed in the final report?
 
Back
Top