Tuesday, August 18, 2015

Consumer Reports on auto insurance

The September issue of Consumer Reports (CR) magazine features a "Special Investigation" called "The Truth About Car Insurance." In their investigation, CR discovers a number of "hidden truths" they're concerned about.

CR's biggest beef is that auto insurance companies practice "unfair discrimination" by basing premiums on other factors than just driving record -- credit history being the one they deem most inequitable. They demand that this practice be stopped, by legislative action if necessary. They ask readers to sign a petition to insurance companies and state regulators: "price me by how I drive, not by who you think I am!"

Reasonable people can disagree on what risk factors are "fair" and which are "unfair," so if Consumer Reports laid out a decent argument, you could treat it as a productive part of an ongoing debate. But ... well, not only does CR seem to have no idea how insurance actually works and how prices are actually set, but they wander all over the place, with scattershot attacks on insurance companies for completely unrelated practices. 

The article reads as if, having already decided that insurance companies are the bad guys, CR just decided to blame them for every aspect of auto insurance that makes them feel unhappy. As a result, the article is pretty much a total mess.

Here are some of the more obvious problems; I'll get to the meat of CR's arguments in a future post.


At one point in the article, CR asks, "Which Insurers Charge More Or Less?"  It turns out that, by their calculation, Allstate charges $1,570, while State Farm charges $1,147, and USAA charges only $817. 

Well, that's impossible. Auto insurance margins are tiny. According to this article, which corresponds to what I've read elsewhere, less then 10% of premiums are kept as profit -- the remaining amount goes towards paying claims and covering expenses. 

So, with only single digit margins, at best, how can one company possibly sell insurance 40% cheaper than another?

My first reaction was, well, maybe CR didn't adjust for the type of policyholder. It was probably that Allstate had lots of teenagers, while USAA had lots of Grannies who only drove to church on Sundays.

But, no: CR actually did control for state (and, I think, also ZIP code), age, credit store, and driving record. 

So what's going on? I have an idea, which I'll get to in a future post. For now, my point is only that CR *does* believe some companies charge almost twice as much as other companies, on average, for the exact same risk.

I know that because, in light of the wide range of prices, the CR writers very wisely "recommend you check prices from at least a dozen companies in your state, [to] help you assess whether you have a good deal or it might come up with an even better one." And they recommend that when shopping, you start with the lower-priced companies, like USAA.

OK, fair enough. But, then, they turn around and contradict themselves in the very same article, by endorsing the idea that prices actually *don't* vary.

"'The advertising creates the impression of price competition when there actually isn't any,' says Doug Heller, a consumer advocate and an insurance consultant in California."

Do they not see how this is the exact opposite of everything else they say in the article? With a straight face, they tell us that 

(a) some companies are substantially cheaper than others, but
(b) there is no price competition

How can they possibly believe both of these are true? 


CR is also critical of the insurance companies' advertised claims. Geico, famously, says that "15 minutes could save you 15 percent or more on car insurance." But CR's "special investigation" reveals:

"But did you know that the word "could" could also mean "could not" just as easily? When we checked, Geico's state-average premiums [were actually higher than two of their four big competitors]."

Well, yes, CR, now that you mention it, I *did* know that the phrase "could save you" is not the same as "is guaranteed to save you" or "is likely to save you."

Similarly, I also know that "can save" is also not a guarantee, even when uttered by Flo, the pitchwoman for Progressive. But CR warns me anyway:

"... Flo likens the number of bouncy balls -- 500 -- to the number of dollars you can save by switching to Progressive. In fact, she says, 'you could save even more.'"

But that's not fair, CR complains, because of selective sampling. Progressive's calculation was based only on those who actually DID save over the other companies. So,

"... whether a typical shopper would save with Progressive is still an open question."

Well, duh. *Some* people save with Allstate, and *some* people save with Progressive -- exactly the same way that, for instance, some items are cheaper at Target, and others are cheaper at K-Mart. What's the problem? 

What's really funny is how Consumer Reports is shocked -- shocked!! -- that the two companies aren't both cheapest across the board. Presumably, before their "special investigation," CR was willing to believe that, on average, Geico was 15% cheaper than Progressive, while, simultaneously, Progressive was $500 cheaper than Geico!

Reading between the outraged lines, Consumer Reports' argument really goes something like this:

-- the insurance companies advertised that you *could* save.
-- a naive shopper, or careless eight-year old, might interpret that as that "overall, we are cheaper than the competition."
-- if the insurance companies said that, they'd be lying.
-- therefore, the insurance companies are almost liars!


Oh, and by the way ... on the inside cover of the very same issue, CR advertises its car buying service by saying that "Buyers have saved an average of $2,990 off MSRP."

And ... yup, in a footnote, they confirm: that figure is based only on those customers who actually bought their vehicle after receiving a quote from CR.

So, it's exactly the same practice for which they rake Progressive over the coals! In fact, it's WORSE. Progressive actually compares the savings to the competition, to the policyholder's current offer. CR compares the savings to MSRP, which is higher than almost anyone would pay after even the briefest negotiation.


One of the "hidden truths" headlines in the article is that "promised discounts might not materialize." For instance,

"Some of the discounts that are advertised the most, such as ... installing anti-theft equipment, save very little: just ... $2 per year ..."

Well, OK, fair enough, $2 is indeed "very little." But how much was CR expecting? In five minutes on Google, I found the following statistics:

-- There are 253 million cars and trucks on the road in the USA.
-- There were 721,000 thefts in 2012.
-- The average loss per theft was $6,019.

721K divided by 253MM multiplied by $6K is ... $17. Take into account that not all the stolen vehicles were cars -- some were buses and snowmobiles -- and maybe you're down to $15.  (Probably less, because losses from buses and transport trucks are probably substantially more than the $6019 average. But, never mind.)

So, you save $2 from $15 by installing anti-theft equipment. That corresponds to the idea that an anti-theft device reduces the risk by about 15%. That doesn't sound too unreasonable, does it? (Especially when you consider that most new vehicles are *already* built with anti-theft technology.)

So, if $2 is too low, what percentage does CR think it should be? My guess: they haven't thought about it at all.


CR also objects that insurers give discounts to students who can show proof of good academic performance:

"It's nice that Johnny does his homework, but ... the good-student discount doesn't emphasize factors relating to actual driving behavior. ... 'According to our research, young drivers are inexperienced no matter how good a student they are, and that is their primary risk,' says Ruth Shults, senior epidemiologist at the National Center for Injury Prevention and Control at the Centers for Disease Control and Prevention."

Well, that sounds reasonable, that inexperience is the primary source of teenage driving risk. But ... well, it's also the primary source of teenage insurance costs! Because, the discount for academic performance is relatively small -- only "up to 14 percent," as CR told us just one paragraph earlier.

The article already told us how adding a teenage driver bumps premiums by an average 90 percent. So, even with a good student, premiums still rise by a minimum 63 percent (86% of 1.9, assuming the discount applies to the entire family, not just the student's portion).

That certainly sounds like the premium DOES still emphasize "factors relating to actual driving behavior," exactly as CR demands.


Oh, and another problem with the good student discount:

" ... it might reward families with high incomes at the expense of lower-income ones."

Notice that CR is assuming, without argument or acknowledgement, that students in high-income families do better in academics. Why would CR assume that? Are poor people dumb? Maybe we need a petition to force magazines to judge students by their actual academic record, not by "who CR thinks they are."

OK, that was just easy sarcasm. Taking CR's argument more seriously: 

First, a lower price is not a "reward," if the insurance companies are accurately evaluating the risk, and better students actually have fewer accidents than worse students. It's like, pork is cheaper than beef, but that doesn't mean supermarkets are "rewarding" Christians and atheists with lower prices for their meat "at the expense" of Muslims and Jews.

Second: if certain students learn math better, for whatever reason, is it not reasonable to assume they also learn *driving* better, for those same reasons? Shouldn't this have at least occurred to CR?


That one example, I think, encapsulates the problems in CR's thinking. They simply choose not to confront the reality of how risk works. They choose to ignore the idea that you can tell, statistically, who's high risk and who's low risk by factors other than actual accident history. 

I don't know whether they don't understand how insurance works, or whether they just choose to ignore it. For what it's worth, the article says nothing about matching premiums to risk, and does not once quote an expert or actuary on the subject. 

By CR's logic, pitchers should be paid only on their W-L record, even if they had ERAs above 5.00. "Judge them by how they win games, not by who you think they are!"

(to be continued.)

Labels: , , ,


At Wednesday, August 19, 2015 2:33:00 AM, Anonymous Anonymous said...

I haven't read the CR article, but I don't see why there is anything wrong with insurance companies pricing according to risk even if the attribute/risk correlation is not entirely intuitive to us. It does put high risk people in a bit of a bind (and perhaps the government can regulate this a little as they did with health insurance and the ACA), especially if that risk is through no fault of their own, as with teenage drivers, but the upside is that it encourages people to behave more optimally. For example, if I have a DUI but I can pay the same as you for car insurance, I have no incentive to curtail my dangerous behavior, at least as far as my insurance premiums are concerned. However, if you give discounts for say, taking a safe driving test, or installing a tracking device and then driving safely, that gives me an incentive to do those things, which is probably a good thing for everyone.

In any case, why is CR calling being a good actuary unethical? Is that because it creates too wide a range in premiums? If that is the reason, they may have a point. For example, isn't it illegal to charge some classes of people, like African Americans, more money for insurance even though that class is in a higher risk group?


At Wednesday, August 19, 2015 2:35:00 AM, Anonymous Anonymous said...

I realize that I contradicted myself with my first sentence and the rest of my comments. I also think that the regulation I was referring to might have been with respect to life insurance, but I am not sure.


At Wednesday, August 19, 2015 11:40:00 AM, Blogger Zach said...

I mean if we can't use insurance companies to entrench white privilege we'll only have everything else left to keep doing it with.

At Thursday, August 20, 2015 10:01:00 AM, Blogger Phil Birnbaum said...


My impression isn't that CR is concerned about the wide range of premiums, but that people are being charged in "discriminatory" fashion according to their group rather than their individual risk. That is, the African-American case applied to other groups too, like people with poor credit.

They seem to have no problem accepting higher-risk assumptions for driving-related criteria, like DUI, but not for non-driving criteria. That seems to be their dividing line, as far as I can tell. Well, with exceptions ... they don't complain about men paying more.

At Thursday, August 20, 2015 10:51:00 AM, Anonymous Anonymous said...

MGL - yes, of course it is illegal to rate based on race, religion, ethnicity, etc. Besides that, it's mostly fair game, although some states have restrictions on credit. All insurance rates for personal auto are filed with the departments of insurance, and some states post all filings online, if you want to see the exact algorithms used.

The problem with the whole "only rate me based on my driving history" argument is that most states limit the history insurers can use to the past 3 years. Throw in delays for renewal timeliness and reporting of claims and it's probably closer to 2.5 years. That's an incredibly small sample to use for something with a low frequency like car accidents. So a pretty small percentage actually get charged more for their driving history. For the rest of people with no incidents in the last 2-3 years, something is needed to further measure the potential risk.

Credit is highly predictive and it has nothing to do with race. It's about responsibility. It's not a reach at all to assume, given limited other information, that someone who is responsible in their finances will be responsible in their driving. Of course it doesn't apply to everyone, but as a general rule is very predictive.

At Saturday, October 03, 2015 5:10:00 AM, Anonymous Anonymous said...

In today’s competitive world education is essential for man after food, clothing, and shelter. Education is the solution of any difficulty; it is the only education which promotes good habits, values and alertness towards anything like terrorism, corruption and much more.
essay writing service reviews

At Monday, October 05, 2015 3:20:00 PM, Blogger nancy john said...

The insurance industry is filled with fraudsters and scammers. It is important to research well and ask for pieces of advice from trusted friends and colleagues before buying a plan. 

Forex Trading software

At Friday, February 05, 2016 3:02:00 AM, Blogger Unknown said...

I don't see how car insurance companies can be that discriminatory when it come to offering a person insurance! At the end of the day credit score information shouldn't even be part of the applications process!

At Thursday, February 11, 2016 2:37:00 AM, Blogger Justin L. Brown said...

A demonstrated and time-tried methodology to enhance your business is to purchase your rivals. I trust the data in this article will help you in your objective to assemble a more strong, more quickly developing and more gainful organization. Progressive competitor

At Thursday, April 07, 2016 10:20:00 AM, Anonymous April Nelson said...

Consumer Reports' two-year investigation found that many major insurance companies base prices in part on a hidden credit score that can take into account things like what type of credit card you have or whether you've applied for credit recently.


Post a Comment

<< Home