The Civil Rights-ing of Risk

The Louisiana insurance commission recently had a hearing on the appropriateness of employing credit scores to set insurance prices. (Bestwire) The Commission found that appropriate actuarial technique support the use of the credit scores.  However, let’ just look at some of the comments of those fighting the use of credit scores:

An attorney representing policy holders states "The concern is that the credit-scoring mechanism used in setting rates is discriminatory and causes a disparate impact upon minorities who are poor," […] "This is akin to having separate but equal systems for Caucasians and minorities; a justifiable system that existed after slavery. Now, in insurance, the poor minorities are paying more for their insurance simply because they're poor."

This is a hoot.  First, the disparate impact notion is foreign to insurance price setting as it is only in the Civil Rights Act and the ADEA.  The general insurance pricing requirement is that rates accurately reflect risk factors.  Insurance pricing is an exercise in blatant discrimination.  High risks pay more than low risks!   Second, the policy holders’ attorney seems to think that all minorities are poor, or the law does not protect well-to-do minorities from discrimination. What hyperbole.   Third, This discrimination is all about having poor credit—so “Caucasians” with bad credit are also paying more.  Finally, even if one applies a new disparate impact test, there are still the possibility of defenses.  One under Title VII of the Civil Rights Act is the so-called business necessity defense.  If insurers couldn’t discriminate based on risk a market failure could arise.  We see this in auto markets and workers comp from time to time. If the failure was substantial enough, then insurers would make less money and some would exit the business.  For those that don’t leave -all customers then pay a higher price.  The whole rationale behind using credit scoring is for companies to be more competitive.  That sounds like a business necessity to me.

 

Update:  I had a lot of comments on this post and Doug Simpson sides with the RiskProf!  One thing that I noted is that I was not really talking about the propriety of using credit scores, but I was talking how people were attempting to attack its use because it had an alleged disparate impact.  Further, if its use in insurance was prohibited under some new disparate impact theory, wouldn’t it apply to bank loans and other types of credit?

 

 

Comments

[Joe Rotger]

“The general insurance pricing requirement is that rates accurately reflect risk factors. Insurance pricing is an exercise in blatant discrimination. High risks pay more than low risks! “

I strongly and deeply disagree. It’s up to all of us to decide how far.

I understand the position of the business, which is closely reflected in your presentation.

But, from the other side of the fence, the customers --not all them, I would agree, it seems to be a first step into an Orwellian world...

Let's take your reasoning a little further, into a possible future where insurance companies make extensive use of medical records to determine personal risks.

Why not? if the information is available.

In this world, it is clear that the unlucky will be --blatantly discriminated--, as you expressed.

And, I don't happen to like that. I have feelings and would rather be nicer to these poor unfortunate folks.

So, I'm part of that group of customers that would be willing to sacrifice and to pay higher premiums, --the prices we pay now, in order to avoid the associated --blatant discrimination.

As I see it, the insurance business is simply a probabilities game. If insurers do not have access to certain information, they have to do with actuarial tables built with the available information.

The probabilities, risks involved, determine policy pricing. And, rest assured insurance companies do make a nice profit with the actual limited actuarial tables.

True...the privileged insured would end up paying less.

But, I insist, I prefer to pay the extra charge.

It’s worth it!

[R.J. Lehmann]

Framing the issue as "either some people could enjoy discounts, or everyone could pay the same high rate" is usually offered as a reductio ad absurdum. The thought being that no one is so committed to the abstract notion of egalitarianism that they would prefer bad outcomes for everyone to good outcomes for a few.

Joe obviously proves that theory wrong. Such folks do, in fact, exist.

[Bob Detlefsen]

To be fair to Joe, he seems to be suggesting that some should pay a higher rate so that others may pay a lower rate, which is different from saying that everyone should pay "the same high rate." Whether the community rate is high or low depends on the rate one would pay under a regime that permitted risk discrimination.

My response to Joe is that while he may choose to pay higher rates to subsidize coverage for high-risk insureds, he does not have the right to impose that choice on others. Government price controls, as well as government restrictions on the use of risk-based underwriting criteria (which serve to impede the risk discrimination Joe abhors) do just that.

Joe should establish a charity to which he and like-minded souls could voluntarily donate funds that would be used to subsidize the insurance costs of high-risk individuals.

[Joe Rotger]

R.J., Bob,

Avoiding the obvious humanitarian reasons...

I could argue further that extreme free market scenarios foster enormous hidden costs to society.

Who is going to take care of the higher risk customers?

If cornered, they could even become violent.

On the other hand, if we only look at the numbers, it is quite evident to me that people have great economic potential, which if not catered, will be lost to society. Stephen King comes to mind...

Caring for other people has rewards...

Think about it. You're lucky, they’re not.

[R.J. Lehmann]

Inasmuch as credit scores go, I don't really think luck has all that much to do with it. Even granting that one's relative status in society hinges on "luck" -- which I think is, at the very least, a drastic oversimplification -- there are many poor people with good credit, and many rich people with poor credit (indeed, bankruptcy protection tends to be sought primarily by those with assets to protect.) You don't have to be "lucky" to live within your means and pay your bills on time. Doing so consistently -- whether you're rich or poor -- is a sign of responsibility. It's not much of a stretch to posit that those who are responsible in this way tend to be responsible in other ways as well, including in their driving habits. Whether that's definitively true or not, I won't pretend to know, but the statistical correlation is certainly there, and it seems as good an explanation for the phenomenon as any other.

[Joe Rotger]

I meant Stephen Hawking, the physics prof...

[Martin]

The way I think about health risk is much different than how I would think about credit risk. Credit risk pricing reduces the effect of adverse selection and moral hazard. People make decisions that have an impact on their credit. It is acceptable for society to charge people for making poor decisions or in R.J.’s words it is ok to charge people for being irresponsible. (We do it all the time, but perhaps not in every case). That way people can learn and can make better future decisions. In the long run everyone benefits as the costs of "controllable risks" decline. However, what if have a situation where the person is “high risk” through no fault of their own? This might be someone with a poor health outcome determined completely by genetics. Does charging them a risk adjusted price help society? The answer is probably no as the person with a poor health stock may not be able mitigate his cost of risk. In this case a risk premium has no effect except to reduce wealth to the high risk individual. This is an argument for social insurance (or a social subsidy) for certain types of poor health outcomes. This situation is completely different than that underlying credit risk and should be treated differently.

[kimberly townsend palmer]

dear martin f. grace,

but just how is credit risk correlated with auto accident risk? i mean, just because you're a financial disaster doesn't mean you're a bad driver, does it? where are those actuarial links? how does an actuary get out of a bottomless pit?? climbs on the backs of the actuarial subjects... take care,
kimberly townsend palmer

[Martin]

Hey there KTP:

There have been a number of actuarial studies (see http://www.tdi.state.tx.us/reports/credit3.html) for one by the State of Texas that shows a relationship between credit score and claims. Those with poor credit scores tend to have higher claims. A previous study (paid for by the industry) [at http://www.ask-epic.com/Publications/Relationship%20of%20Credit%20Scores_062003.pdf] showed that people with lower credit scores were more likely to make a claim than people with higher credit scores, but that the size of the claim was not much differnt between the groups.

One of the perceived problems with credit scoring is just that people do not see a link between credit scores and driving risk. However, just about every study has confirmed the link. What we have is study after study saying there is a link, so the use of credit scores satisfies the states’ fair pricing rules used for risk classification and pricing. This is essentially what the Louisiana commission did last week.

Just as a note to interested readers (other than my correlations), KTP swore me into the bar after I passed the bar exam.

[Joe Rotger]

Kimberly pointed out:

"but just how is credit risk correlated with auto accident risk?"

Does this mean insurance companies will collect higher premiums during recession periods? Plenty of unemployed, probably in a debt crunch, high credit risk...and also high insurance risk??

Does this mean the much needed entrepreneurs will have to add to their list of concerns, the higher insurance premiums they'll be paying for car insurance, if they go belly up in their new and risky venture??

I see a lot of innocent folks paying higher premiums for a questionable correlation...

Society too.

[Joe Rotger]

Martin wrote:
"What we have is study after study saying there is a link, so the use of credit scores satisfies the states’ fair pricing rules used for risk classification and pricing. This is essentially what the Louisiana commission did last week."

Martin,

I'm convinced the link is there. We could get into the nitty gritty of the math involved, some correlations are better than others.

But, accepting the link, the fishing net is going to entangle a lot of unintended innocent fish...

Too many things tell me the correlation is quite imperfect. It is far reaching into distinct private issues.

I don't like it.

[Joe Rotger]

Doug Simpson wrote:

"Risk Prof spurred a bit of a debate over the use of credit scores as a rating tool, as recently approved by the Louisiana Ins. Commission. Commentators repeat the usual objections, usually with a minimal background in the science, logic and economics of insurance pricing and risk selection. Risk Prof quotes and criticizes arguments from objectors that attempt to apply standards from civil rights cases to commercial insurance pricing.

One commentator, Joe Rotger, volunteers to "sacrifice and pay higher premiums" to avoid "blatant discrimination." All we need is a market filled with buyers who make decisions based on other than their economic best interest. Last I heard, those were hard to find. Hard enough to make it unprofitable to depend on it in a competitive world. And the various GEICO and Progressive ads offering lower rates would be a waste of money. My guess is they are more practical than Joe."

We pay more, all the time, for social security and medicare, for instance. How about taxes?

A competitive market is regulated. It is so, to safeguard citizens' rights.

It is also regulated because private companies do not care about the hidden costs to society. Waste disposal from chemical and mining industries come to mind.

Of course private companies, if left unregulated, will trash the Liberty Statue to dig for oil. I wouldn't call that practical, though.

As a matter of fact, the old ads mentioned show that insurance companies are still capable of lowering their costs under present actuary tables, without considering credit risk impact.

"Another is willing to let "Joe" pay extra, but opposed to forcing that choice on everyone."

I proposed status quo, leave the insurance actuary system as it is, avoid using this questionable new relation -- credit versus accident risk, to construct new actuarial tables.

"The key element is that credit scores do in fact correlate with claim ratios. Why? Hard to say, but its not important to the rating decision. "

Gee, it sure sounds like not much of an interest to get to the bottom of this.

"If a positive correlation can be shown, making the discrimination not unfair, the law is satisfied. "

Any positive correlation? The math says there are some very poor positive correlations, too. I'm willing to bet my underwear that the old accident report to accident risk correlation is better than the new one, credit risk to accident risk? Just plain common sense.

"If the legislature wants to outlaw a scientifically demonstrated predictor of claim costs, let it be their call, not the insurance commissioner's."

Why do I get the feeling that the legislature was taken for a ride by the insurance lobbying? There is something called bad science too. At the least, I've shown that this poor piece of legislation will have innocent people --not irresponsibles, paying higher premiums. To name a few, the thousands of unemployed suffering duress due to a recession, outsourcing, or loss of jobs due to Asian competition...

Finally, I pose a couple of questions:

* Why not use medical records to assess risk?

* Who will pay for the hidden costs to society?

* How could the insurance companies get away with murder?