13

I'm trying something in related to Auto Insurance and when I went through the State Farm's insurance portal, it asked me for a couple of questions like below.

  1. Vehicle ownership
  2. Car Age
  3. Person's Age
  4. Net worth
  5. Risk Tolerance
  6. Is the person owning a house.
  7. Does he require towing and labor coverage
  8. Does he require Rental coverage.

When I gave these details. It redirects me to another page where I give my zip code and personal details. And then it shows me the quote with the monthly premium that I'll need to pay.

Here I've 2 questions.

  1. Does the Insurance value differ from state to state (for example I've a car in Hawaii and there is another car in Illinois with same model, make and same features), does the Insurance vary for both?
  2. How is the insurance quote calculated? (can someone please help me with a small formula that does the job).

Thanks

4
  • 7
    It may well vary not just by state (where you may have different legal requirements), but by your zip code or even address. E.g. if you live in a congested urban area, your risk of accidents might be higher, so they'll charge more.
    – jamesqf
    Nov 13, 2017 at 18:49
  • 1
    It will also vary between insurance companies even if all factors are the same. You can't really answer this without a specific insurance company in mind and inside knowledge of how their pricing works. In which case you might as well get a quote from them
    – JohnFx
    Nov 13, 2017 at 22:32
  • Question 2 has incorrect assumptions in it. There is no formula. If there were it wouldn't be small. And each insurance company would have their own and they wouldn't share them with you.
    – stannius
    Nov 14, 2017 at 19:58
  • @JohnFx: It's more likely to vary by the prevailing race of people living in your zip code/on your street... Nov 15, 2017 at 0:41

5 Answers 5

28

First you should understand the basics of how insurance companies make money:

In a simple scenario, assume 1,000 have car insurance. Assume that on average, 100 people have accidents per year, and that each accident costs $10,000. So, we can expect total costs to be $1,000,000 per year. Some of those costs will be paid by the drivers, who have some sort of 'deductible'. That is - the insurance company will only cover costs after the driver has themself paid some initial amount [something like, the first $1,000 of repairs is paid by the driver]. So now the insurance company expects to have to pay out $900,000 in total claims this year. If they want to pay those claims (and also pay their administrative costs, and earn a profit), they might want to have $1,250,000 in revenue. Across 1,000 people, that would be $1,250 / year in insurance premiums.

Of course, the big question for the insurance company is: how much will they really need to pay out in insurance claims each year? The better they can predict that number, the more profitable they can be [because they can charge a much more accurate amount, which can earn them new customers and gives them insurance {pun} that each new customer is actually profitable to them on average].

So the insurance company spends a lot of time and money trying to predict your likelihood of a car accident. They use a lot of metrics to do this. Some might be statistical hogwash that they charge you because they feel they can [if every insurance company charges you extra for driving a 2-door instead of a 4-door, then they all will], and some might be based in reality.

So they attempt to correlate all of the items in your list, to see if any of those items indicates that you should be charged more (or less) for your insurance. This is equal parts art and science, and a lot of it comes down to how they market themselves. ie: if an insurance company gives a discount for being in college, is that because college drivers are better drivers, or is it because they want to increase the number of young customers they have, so they can keep those customers for life?

Therefore how each metric factors into your calculation will be based on the company using it. It would basically be impossible to 'come up with' the same answer as the insurance company by having the information you provided, because of how heavily dependant that answer is on statistics + marketing.

As for how your state matters - some states may have different accident rates, and different payout systems. For example - is Hawaii driving more dangerous because of all the tourists driving rented cars faster than they should? Is New York less expensive to insure because better public health care means less cost is borne by the insurance company in the event of an accident [I have no idea if either of these things are the case, they are purely for hypothetical discussion purposes].

In short, make sure you get quotes from multiple providers, and understand that it isn't just the cost that changes. Check changes in coverage and deductibles as well [ie: if one company charges you $100 / month when everyone else charges $200 / month - make sure that the cheaper company doesn't limit its coverage in ways that matter to you].

15
  • 3
    Equal parts art and science? That seems an unnecessary and inaccurate slight against actuarial science.
    – Hart CO
    Nov 13, 2017 at 15:44
  • 5
    @HartCO Last I asked around, auto insurance companies' quotes varied by a factor of 5 to 6. There's more than pure actuarial science at play. The ultimate goal is profits; science is one of several inputs. Nov 13, 2017 at 19:50
  • 13
    @HartCO as someone in the actuarial field, I would instead consider "equal parts art and science" a compliment.
    – glassy
    Nov 13, 2017 at 20:01
  • 2
    @Solomonoff'sSecret ahhh. That probably likely was why you got such a wide spread. You raised a big red flag in everyone's risk model; and then saw a lot of variance in how the other data countered it along with how much risk they were willing to take for a relative unknown's business. Nov 13, 2017 at 21:48
  • 4
    Actuaries and regulators give a mathematically supported and legally allowed rate "range" per category. Choosing whether to go high or low in the range is correctly an art, as it is a combination of commercial negotiation, marketing, customer acquisition costs, and not creating an illegal discriminatory pricing regime against protected demographics. +1 for the total answer
    – user662852
    Nov 14, 2017 at 2:52
9

Does the Insurance value differ from state to state (for example I've a car in Hawaii and there is another car in Illinois with same model, make and same features), does the Insurance vary for both?

Yes, quotes will vary based on where you live for various reasons, (propensity for accidents, value of cars, etc.), and state laws regarding required car insurance can vary.

How is the insurance quote calculated?

It's likely a proprietary formula that the insurance company will not disclose. If they did, they could be giving away a competitive advantage.

However, like all insurance, the goal is to determine the probability of the insured having an accident, and the projected cost of such an accident. That will be based on actuarial tables for each of the risk factors you mention.

4
  • Hi Stanley, thanks for the quick response, I want to know if there is any source/formula that gives me the average quote(not exactly what a company would charge the user). so that I can tell in this way, here is an approximate quote for you and the prices may vary in real time. Thanks Again! Nov 13, 2017 at 14:55
  • 1
    Not that I am aware of - some providers will quote you prices from other companies, but I tend to be skeptical and get several quotes on my own anyways.
    – D Stanley
    Nov 13, 2017 at 15:07
  • @user3872094 The best way to compare quotes is to actually compare quotes. Unfortunately it will take you time to gather the information yourself, but nothing will be anywhere near as accurate as a quote specifically provided to you by a company. If you can't act on the quote, then it isn't as valuable of a comparison, because it might be different from reality. Like D Stanley says, you might find "this is the competitor's quote" on one company's website, but unless you can click 'buy now' on that competitor quote, you'll never know for sure if it's real. Nov 13, 2017 at 15:15
  • In Massachusetts, the rates for the required parts of insurance are set by the state.
    – Barmar
    Nov 13, 2017 at 16:07
2

Former software developer at an insurance company here (not State Farm though). All of the above answers are accurate and address how the business analysts come up with factors on which to rate your quote.

I wanted to chime in on the software side here; specifically, what goes into actually crunching those numbers to produce an end result. In my experience, business analysts provide the site developers with a spreadsheet of base rates and factors, which get imported into a database. When you calculate a quote, the site starts by taking your data, and finding the appropriate base rate to start with (usually based on vehicle type, quote type (personal/commercial/etc.) and garaging zip code for the US). The appropriate factors are then also pulled, and are typically either multiplicative or additive relative to the base rate. The most 'creative' operation I've seen other than add/multiply was a linear interpolation to get some kind of gradient value, usually based on the amount of coverage you selected. At this point, you could have upwards of twenty rating factors affecting your base rate: marriage status, MVR reports, SR-22; basically, anything you might've filled into your application. In the case of MVR reports specifically, we'd usually verify your input against an MVR providing service to check that you didn't omit any violations, but we wouldn't penalize for lying about it...we didn't get that creative :) Then we'd apply any fees and discounts before spitting out the final number.

With all that said, these algorithms that companies apply to calculate quotes are confidential as far as I'm aware, insofar as they don't publish those steps anywhere for the public to access. The type of algorithm used could even vary based on the state you live in, or really just when the site code is arbitrarily updated to use a new rating system. Underwriters and agents might have access to company-specific rating tables, so they might have more insight at the company level.

In short, if there's an equation out there being used to calculate your rate, it's probably a huge string of multiplications with some base rate additions and linear interpolations peppered in, based on factors (and base rates) that aren't readily publicized. Your best bet is to not go through the site at all and talk to a State Farm agent about agency-specific practices if you're really curious about the numbers.

1
  • Exactly as I suspected. To the extent that there is any "formula" it's actually a piecewise function.
    – stannius
    Nov 15, 2017 at 18:30
1

Question 1: Yes

Question 2: There is no simple formula. Car insurance is mostly Statistics, because you have so many millions of cases that the variance is really low. This also means that, because the cost can be estimated so precisely, it is difficult to make an offer better than the competitors. For that reason every insurance company makes there own, arbitrary, segmentation of the data which leads them identify low risk groups they can offer a bonus to.

Common ones are type of car or and driving experience, but it could be anything that is not forbidden by anti-discrimination-laws.

Also additional perks like towing insurance etc. may give them an opportunity do differentiate themselves or to make easy profit.

In fact it is a common tactic to offer prices that make close to no profit to fill up your book, then raise tariffs in then following years an make you profit with those who are to lazy to switch.

0

On top of the given answers, the type of referral will also factor in.

When you're up for renewal and go to a comparison site (in the UK: CompareTheMarket, MoneySupermarket, Confused, GoCompare, ... ) and struggle accurately through all their lists of questions, you see that some of the data differs (e.g., not all the same jobs can be entered; if you have had an accident, not all ask whose fault it was and/or don't leave the option "not yet resolved" --possibly forcing you to guess which way it will be adjudicated,-- and/or what the total repair cost was).

So as these referrers feed slightly different data to roughly the same set of insurance providers, you will get slightly different quotes on the same providers. And expect your own provider to offer a slightly better quote than you'll get in reality for renewing: The referrer's (one-time) cut has to be still taken off, but they count it as a new client so somebody gets a bonus for that --- you they disregard as a captive client and give what boils down to a loyalty penalty. [Case in point: I had an unresolved car accident, resolved months later in my favour. With all honest data including unresolved claim and its cost and putting my 'accident-free years' factor at 0 instead of 7, my old provider quoted about 8% more than the previous year on comparison sites; but my renewal papers quoted me 290% more, upon telephone enquiry the promised to refund the difference if court found in my favour though they refused to give this in writing. So: No thanks!]

Then the other set of referrals they get is from you directly going to their website asking for a quote. They know what type of link you've followed (banner, or google result, etc), they may know some info from your browser's cookies (time spent where) or other tracking service, and from your data they may guess how tech-savvy and shop-wise you are, and scale your offer accordingly. [Comparison-site shoppers are lumped together at a relatively high savvy-level, of course!].

Companies breaking down your data and their own in a particular way can find advantages and hence offer you better terms, as said in the main answer (this is like Arbitration in stock exchanges, ensuring a certain amount of sanity: if there's something to exploit, somebody will, and everybody will follow). It may be that they find a certain group of people maybe more accident-prone but cheaper to deal with (more flexible in repair-times, or easy to bully in accepting shared-fault when they weren't at fault), or they want a certain client (for women, for civil servants, for sporty drivers, for homeowners --- often for cross-selling other insurance services). Or they claim to want pensioners because the company can offer them 'a familiar voice' (same account manager always contacting them) while they're easier to bamboozle and less likely to shop around when offered a rubbish deal.

Also, 100% straight comparison of competing offers isn't possible as the fine details of the T&Cs (terms & conditions) would differ, as well as various little pinpricks in the claims handling process.

And depreciation of a car, and various ways of dealing with it: You insure it for the buying prices, but two years later it's worth about 40% less on paper --- so in case of total loss, replacing like-for-like will cost you still at least 80% of the value for which you've been insuring it while they'll probably offer you the 100-40= 60%. Mostly because instead of your trusted car you have something unknown that may have hidden defects, or been mistreated and about to die. [Case in point: My 3-y-old dealer-bought car's gearbox died just outside the 6month warranty period, notwithstanding its "150-item inspection you can rely on". In the end the national brand agreed to refund the parts (15% of what I paid for the car) but not the labour (a few hours).]

And any car model's value differs (in descending order) from its "forecourt price", "private selling price", "part exchange price", and "auction price". Depending on your ompanies may happily insure you for forecourt price (=what you paid to dealer) but then point out that the value of that car is the theoretical P/X value, i.e., the car without anybody's profit, far less than you've been paying for. [Conversely, if you crash it after insuring below market value, they can pay you your stupidly low figure.]

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .