Patrick DeCorla-Souza, Tolling and Pricing Program Manager, FHWA
Lee Munnich, Humphrey Institute, University of Minnesota
Kenneth Buckeye, Minnesota Department of Transportation
John Doan, SRF Consulting
Center for Innovative Finance Support
Federal Highway Administration
Seventh Part of a Webinar Series on Overcoming the Challenges of Congestion Pricing.
Moderator:
Jennifer Symoun
Presenters:
Jennifer Symoun
Good afternoon or good morning to those of you to the West. Welcome to the Overcoming the Challenges of Congestion Pricing webinar series. My name is Jennifer Symoun and I will moderate today's webinar, which will focus on Pay as you Drive Insurance. Please be advised that today's seminar is being recorded.
Before I go any further, I do want to let those of you who are calling into the teleconference for the audio know that you need to mute your computer speakers or else you will be hearing your audio over the computer as well.
Today we'll have three presenters - Allen Greenberg of the Federal Highway Administration, Robin Harbage of Towers Watson, and Dave Huber a Pay as you Drive Insurance Industry Consultant and Veteran.
Today's webinar will last 90 minutes. We will be primarily taking questions after all three presentations, but if a question arises that is important to answer after an individual presentation we will do so. If during the presentations you think of a question, you can type it into the chat area. Please make sure you send your question to "Everyone" and indicate which presenter your question is for. Presenters will be unable to answer your questions during their presentations, but I will start off the question and answer session with the questions typed into the chat box. If we are unable to get through all of the questions in the time allotted we will get written responses from the presenters and send them out with the follow up information.
The PowerPoint presentations used today are available for download from the file download box in the lower right corner of your screen. I would also like to remind you that this session is being recorded. The recording, presentations, and a transcript will be posted to the Tolling and Pricing web site within the next few weeks and I will send out a notice when they are available.
We'll now go ahead and get started. We're going to start off with Allen Greenberg of the Federal Highway Administration.
Allen Greenberg
Thank you and good afternoon and I appreciate everyone's attendance. I'm going to start by discussing just why the federal government has been interested and why state and local governments have been interested. It stems from the public policy implications of offering pay as you drive insurance. First, a little bit about the concept. What is it and what is pricing in general and what is it specifically? Once people own a car today and have insured the car, the cost of using the car is relatively small even considering today's gas prices. If you have invested that money, and most people have, we have more vehicles and licensed drivers in this country, there's very little incentive not to use it. What if we took some of the cost involved in auto ownership and registration and converted them into variable costs based on usage? Probably the largest of those costs is insurance but also there are others such as vehicle taxes and fees, parking costs and even the car itself through a concept that is fairly new but growing in this country called car sharing.
Why do we want to do this? The first reason I noted. The two key things that stem from that include that many households like the idea. People like to manage their budgets and they prefer to have variable costs to fix ones. From the public policy standpoint, there is a host of studies that project substantial driving reductions and found that a lot of public policy benefits. Also, consumer savings resulting in pay as you drive pricing.
This isn't a new concept. One can find extent extensive references going as far back as 1929. A quintessential academic work on the subject was published in 1968 by William Vickery and the title was "Automobile Accidents, Tort Law, Externalities, and Insurance." He goes into quite a bit of detail on not only where the risks are in terms of the roadway type and time of day, etc. but what would be appropriate proxies for the technology available at that time for pricing of these risks. Of course he didn't have the capabilities we have today in terms of instrumenting vehicles with GPS devices and other kinds of readers and even back then getting inspection data from the Department of Motor Vehicles was not commonly available. We have a lot more tools potentially to offer some form of pay as you drive insurance that was not available when researchers were contemplating the importance and possible ways of offering this. There is a lot of research recently that shows why pay as you drive pricing is an important thing. The key thing is it aligns the insurance company's interests with the public policy interests. If there wasn't an actual basis for charging in a usage-based fashion, there wouldn't be a mechanism for encouraging companies to do that. We have new research from Massachusetts where they got a huge amount of data from the states vehicle inspection. Insurance companies also have to report losses and they were able to get those databases and put them together and they found the statistic for showing how strong a relationship is, that was increased from 0.57 which was using just standard insurance variables, location, driver experience, and where the vehicle is registered. When you impose a new model which was mostly mileage-based pricing although it included variables into what was being charged. The r-squared value jumped from 0.57 to 0.72. That is a huge increase into the potential by doing pricing this way. We have seen a lot of small instrumented vehicle studies and they also point to the same direction. They show a strong linkage between certain driving habits and crashes and in addition to these small studies and the Massachusetts study, we know in number of insurance companies have done and are doing their own data collection efforts and the number of them have gone to his offering some sort of version of pay as you drive. We haven't learned what the companies learned directly, but we know after incentivizing drivers to instrument the vehicle so they can gather data that we do find these companies are pricing in the usage based fashion which means they have seen something in the data that justifies it.
I mentioned an instrumented vehicle study and I picked out a few of them. The key point is the most dangerous drivers are significantly more dangerous than the safe drivers in terms of crashes and I cited a study, the 100 car study in Northern Virginia and an Israeli study. If we let them know that drivers are risky and get them the opportunity to change that behavior with a financial consequence, the behavior does change fairly often. In a Swedish study speeding went down from 15% to 8% of driving time. This opens the door that it isn't just about charging drivers differently, it's about helping drivers reduce their risks, be rewarded for reducing risks, and the rewards go both to the driver and the insurance company because fewer claims would result.
I have done some research looking at hypothetical governmental incentive kinds of policies and I'm not going to go into the methodologies, but in one case I created a hypothetical government program for every dollar in pay as you drive pricing I would reward ever offered the pricing with $0.10 and I compared that to the cost of other government programs that would either reduce emissions or improve safety. I found quite substantial results that this was a competitive strategy financially in terms of value from an air quality standpoint. It exceeded the benefits of most air-quality investments we have made today and it was competitive meeting comparable to other safety investments in terms of reductions of injuries and fatalities that would result from this measure and the cost of getting this measure versus injuries and fatalities reductions. In a second study I looked at the cost-effectiveness of fatality reductions through the cost-effectiveness of fuel economy rules and this is to assess what cost would be appropriate for vehicle technology and this was done a couple of threads ago in the safety traffic at illustration and I took the numbers and said what is the benefit of pay as you drive insurance and what would be a comparable cost that would be justified based on that calculation? I'm not really going to focus on these two slides, I just want to note they are in the presentation and they are based on larger research papers and if for some reason you are interested in some of that I would look at these slides in the presentation and then you can send me a note and I can forward the research they came from. This last one, for every mile not driven based on the NHTSA approach, the total benefits is 21.1 cents if you assume an 8% reduction of driving for each mile not driven, the value of each pay as you drive insured mile is about 1.7 cents.
We know a lot of things about it and there are studies to suggest all these things on this particular slide. The vehicle miles traveled reduction is probably the best and most comprehensive and recent study from the Brookings Institute that shows up 8% reduction in driving from pay as you drive insurance. We have evidence crash claims reduce greater than the driving reductions. Reductions are much greater than a couple of years ago when vehicle miles traveled came down by 3% because of the economy. Congestion at major intersections according to INRIX went down by 30%. There's evidence of air pollution and carbon emissions reductions. Infrastructure cost savings are $0.03 to $0.05 of every mile not driven. There other benefits as well. The Brooklyn Institute shows the average of 63% households would save and it would be an average of $270 per vehicle. It is tremendous savings but savings that results from claims reductions. The insurance company profits can be maintained as well.
I throw these things out there, when we the federal government have been involved in pilots we try to maximize the benefits of all the pricing pilots and there's a few strategies we use and here are some on this slide that would be consistent with insurance pricing. You want the pricing to be direct, you want the displays of pricing related information to be transparent and you want reminders frequently, you want compared to alternatives to be free or discounted such as a transit pass were where the per usage cost of a single transit trip would be zero. You want to help people find alternatives and there are a lot of peer comparisons and competitions that get people to reduce their emissions further. In other sectors we have seen this like for utility usage and we could probably apply similar concepts to encourage people to reduce their driving with pay as you drive insurance. Okay, I will pass it on.
Jennifer Symoun
Robin, if you give me a second I will set up your presentation.
Robin Harbage
Hello, I am Robin Harbage and a director with Towers Watson. I have been working on pay as you drive in different roles, first with a private insurer who is one of the early adopters of pay as you drive and now with Towers Watson and helping some of our clients adopt the process. For today, I'm going to do a couple of things. First I'm going to talk about the market. Allen has given you a sense of why this makes sense from the federal government's public policy standpoint. Let me tell you about how fast the insurance industry is embracing this and then I'm going to turn it over to Dave for a little about why this makes sense from an insurance standpoint and then I will follow-up with a couple of other points with the data used for risk pricing and a bit of a forecast on where usage-based insurance which I will refer to pay as you drive, could be in the future.
This is a picture of people in the marketplace doing it. It's not a complete list for a couple of reasons. One, new insurers are failing quite frequently and secondly there are many insurers in a pilot phase and are not public with their filings yet. While I might be aware of other insurers, I can't publicly talk about them but this list gives you a sense when you look at the names of some of the insurers, especially in the US, the size and the prominence they have in the marketplace, it is quite significant. This graph gives a sense for it. In terms of private passenger auto share, right now six of the top ten personal insurers have implemented usage-based insurance programs in at least one state. If you look at the market share for all of the insurers writing private passenger auto, 60% is being written by insurers have already implemented the program in at least one state. Not to say 60% of the market uses this, but those insurers have that much of the market and have the capability to continue to roll it out to that kind of a market share and in addition to that another 16% of the market share is controlled by insurers exploring with internal pilots were preliminary studies of usage-based insurance so three quarters of the market is actively pursuing this at this point of time in terms of market share. That is quite significant in terms of the adoption rate. This chart shows the competitiveness of usage base insurance on a state-by-state basis where the darker the color the more insurers are actively pursuing this. At this point in time to our knowledge, Hawaii is the only state that doesn't have an active usage base insurance program and 17 have four or more insurers running usage base insurance programs in their state so it becomes a competitive market.
One of the leaders in this has already launched in 37 states plus the District of Columbia and it made several comments publicly about their approach to the product and their rollout of the product and how much they feel comfortable with continuing their rollout on a national basis. They have made quite a bit with their product in the marketplace. It is a market leader but not the only person rolling this out and other insurers have taken notice and understand it that program is successful to stay competitive, they need to follow suit. With that, I will turn it over to Dave and he will give you a few if you about the insurers viewpoint of the economics.
Dave Huber
Thanks, Robin. The question is why is it that all those insurance companies that Robin showed on those maps and charts have jumped into this thing called pay as you drive insurance? Before we can answer that, let's remind ourselves a few things about auto insurance. Generally as we know, insurance is about assessing risk and try to figure out the frequency and severity of losses that are going to occur in the future. The premium needs to cover costs. Some of those costs are well known but some of those costs are not very well known and in some cases insurers just have to make a guess. It is a tricky proposition. To add a little more complexity, the margins are pretty thin. Most insurers would be happy to make $0.04 on every premium dollar they collect and in some cases the profit the insurers make is on investment, but if they could make an underwriting profit, I think many insurers would be very happy. It's all about trying to understand the risk and to introduce a premium associated with an insurance product that would be attractive to consumers so you can gain market share.
Generally, all insurers across-the-board use the same kind of data and ask the same kinds of questions in order to assess the risk and determine a price. These are familiar to everyone. Insurers depend on age and gender and marital status and the type of vehicle you have and whether or not you had accidents and violations and limits and deductibles and events whether you own a home. Those all factor in to the way the insurance company determines the price. From the consumer's perspective it is not very obvious how that price gets determined. You tell them the things they want to know and they gave you an answer. Insurers have also figured out that the more sophisticated they can be in terms of assessing that risk and determining a price that can be a significant competitive advantage. Insurers are always on the lookout for introducing new types of data that would help them predict crashes and things like education, insurance history, and even financial responsibility have worked their way to the underwriting and pricing models and have proven to be very predictive. For those who have done it early it has given them a competitive advantage. They have gotten a lot more business because of their ability to introduce sophisticated pricing. All of this data is used to predict behaviors that might lead to an accident.
The behaviors about how, when and where you drive are at the heart of pay as you drive insurance. This has a lot of labels but all that information historically has tried to guess at or maybe even been a surrogate for the information about how, when and where you drive. Your gut says that must be predictive. The data is proving out that those kinds of data are more likely to be predictive than the other data the industry uses. Until the advent of telematics, whether they are embedded or aftermarket, that behavioral data hasn't been easy to collect. It has been elusive and in some cases the best you could do was have insurers self-report. You can imagine when you ask a bunch of customers how many miles they actually drive that it doesn't take long before most of them if not all of them are only driving 5000 miles a year. That makes it difficult if you are left to self reporting. The early adopters are taking advantage of knowing something more about a risk. Maybe information about how many miles they actually drive over the course of the year, or how often the car is driven after midnight on a Saturday, or even how often the cell phone is used while the car is being driven. That data allows those early adopters to enjoy segmentation and pricing advantage. In the segmentation and pricing advantage, if you can imagine a room full of risks and for all intents and purposes they look like the same risk and consequently they all have the same price. If you do something more about that room full of people than the rest of the competition, you are able to introduce a new predictive variable into the pricing calculation and some of those rates based on that new information might go up and some may go down. To those whose rates go down, your offer would look more attractive and perhaps to those whose rates may go up, they may choose to stay or they may move on and shop. If you're confident in your ability to accurately price, you're more interested in finding that new business that you can priced lower than the market and you are less concerned about the business that may bleed off because you are charging a little more. Over time, adverse selection can have a detrimental effect on one's business.
When insurers think about usage-based insurance or pay as you drive insurance, it is often the strategic investment. He can the offense it or defense it and I suspect there are folks playing in this now that have in their own minds gotten into it because they want to be offensive or may have gotten into it because they feel like they take a defensive position fear for the competitor of the road may take all of their best business and leaving them with business that might have increased frequency or severity but yet they don't know why because they don't have access to that additional information. Also, pay as you drive insurance that introduces telematics devices and fulfillment, there are a bunch of processes and deployed this kind of product that extends beyond the core competency of insurers. Also as they look at pay as you drive opportunities, the economics are not always crystal clear. They are trying to make $0.04 on the dollar. If you can imagine them trying to figure out how to cover the cost of maybe a $100 device and $5 a month data fees, those economics are not clear. I think as they get into if they become clearer and economics to work out so the benefits outweigh the costs, but you have to get into it to try it in order for the economics to play out. I think the good news is for the industry and even for us as consumers is that the hurdles are getting lower and so the opportunity for all of us to finally have control over how much we pay, have some say in what our premium will be, and in some cases prove to our insurer that we do deserve a better rate, those opportunities will be more available to us as those insurers and other insurers get added to the map and you have more than just three or four per state but you have 20 alternatives in the state and as the field against level, insurers will continue to innovate and look for other ways to segment the data and use it in a more granular fashion. So you go from miles and time of day to miles and time today on a particular road and traffic patterns. We are very much in the early stages of usage-based insurance today and I think it would be interesting to see how this evolves over time. I think that what Robin is going to talk next about.
Robin Harbage
Thank you, Dave. I'm going to give you a little insight into some of the data that insurers are using in order to price this usage base insurance. Let me give you a quick primer for those who haven't seen these or how they operate in the field on how this can work from the insurer and the consumer standpoint. First of all, the consumer goes into the agent's office or calls them up and normally they will be quoted a certain premium based on the information at the time of the quote, where that premium maybe $1200. In some cases the insurer would say that we have this new methodology where if you are willing to allow us to track some of your driving behaviors using this small device with plug into your car, we will give you a discount of let's say 10% for the first term. The insurer installs it, the data is collected, it's transmitted wirelessly from the vehicle to the server for the insurer who then takes a look at the data they gave in including the number of miles driven, a few events, maybe the time of day, they compare that to the models they have built on this and they come back and say that you had driven better than average so your premium drops not just the 10% we discussed but an additional percentage so maybe now you're only paying $1000. An additional step that can be beneficial to everyone in this process is the fact the insured for the first time have some control over the insurance premium because they can look at what they were charged, the elements into making up the rate and say I wasn't going to change my gender or marital status to get a lower rate, but I could choose to maybe drive less often, to carpool more, to use public transportation, to do a number of things to reduce the amount of time for driving I do. With enough feedback from the consumer they could change some of the consumer behaviors that may not be as obvious but result in lower lost costs. In some instances we have seen drivers that become safer drivers also reduced their per mile fuel consumption such as some fleets have reported decreases in fuel consumption of up to 10% just based on driving behavior without having to change the mileage they use their vehicles. The result may be the consumer controls the premium and they come back and make it a lower renewal rate in the next term and everybody is happy. They have control over this as before they weren't stuck paying based on other things.
If you look at these devices they are simple yet complex. There is a box on the car sending information to the server and they have a lot of different gizmos inside of them. Most of them are plugged in to the OBD port on the vehicle. That port is a direct link into the onboard computer so you can read anything off the computer that is known about the vehicle in terms of speed and operation. In addition, most of these chips installed in the device had a GPS chip so they can tell where the vehicle is being operated and other driving behaviors. They have accelerometers and will know things about speed changes. They will have a small memory and processor so they don't have to constantly communicating and they can hold the information and then eventually to make it easier for the consumer, they have a cellular service so they can upload the information without the consumer having to touch the device or do anything else so it becomes a convenient opportunity for the consumer. These devices as they start to reach larger volumes are beginning to come down in price as do most electronics once they are used by a lot of things and the cost for these things that used to be in excess of $200 are starting to come on the market place at or below $100.
This gives you a sense of the information. I mentioned they have a GPS on them so they know location and they know speed changes. They have forward and backward and side to side motion and that they are calibrated you can tell the changes going on. In this case I have listed the obvious things like the start time for the vehicle. You can know things like whether or not the seat belt is being used. You can know with information what kinds of roads are being used. This requires the GPS coordinates to a map of some kind and know what kind of roads are on that map but that is pretty common and available at this point in time. You can know things around braking behavior or accelerating behavior and you can bring in data and think about vehicle density and how many vehicles are on the road at any point of time when the vehicle is being operated. The wealth of information is astounding when you get down to it. As mentioned previously, this type of data is so much more correlated with the actual operation of the vehicle and not as a weak of a predictor of other variables. While mileage is an important component, it can be predictive of other elements and the prediction value of the models becomes much more significant.
I'm just going to give you a look at what this looks like. Here is a simple trip on the left-hand side from a device. You can see this vehicle has taken off and has travel on some road with his speed has gone up and down and at some point they hit a stable speed and then they came back down to a few stops. What's interesting about this is on the right-hand side we have taken the same vehicle and we have shown what data is being reported for five consecutive days at the same time of day. What you see is a pattern because this vehicle is obviously a commuting vehicle because you see it during rush hour over the same pattern of driving so you know this isn't a vehicle used in a pleasure situation where it may be going on simple errands but used consistently. This is simple data on a distance time and date of speed but there's a lot of underlying data available. Here is some information taken from three sample vehicles and what we have done is summarize for these vehicles some driving behavior information. First we have gotten average miles on a weekly basis and you see they have varied quite a bit. If you analyze the weekly miles, it goes from a low of 9,400 miles a year to a high of 20,100 miles. Vehicle number three reporting the average mileage for the US consumer, around 12,000 miles a year. The bottom one is the number of hard brakes per mile. This is measured in a fraction because a person doesn't break hard every single mile but it goes from a low of 0.04 to a high of 0.10. What's more fascinating is to know is I do put these people pay for a premium and to know that while vehicle two would probably be the most risky vehicle and one or three, depending on how you judge the mileage would be safer vehicles. For these three operators, vehicle two is currently paying the lowest premium just because of their gender, marital status and age so what looks like a reasonably accurate measure of one vehicle being much more risky than the other two is turned upside down in the marketplace because the operators are not sharing this information with your insurer.
We have taken these vehicles and added additional information. We looked at what percentage of the time was during rush hour versus non- rush-hour. We look at percentage of the time driven in the evening and late at night. You can see differences in the proportions. If you assume the time of day was a significant variable in terms of predicting mileage losses per mile, you would say this segment these three vehicles so again, one vehicle may be a lot more, assuming late night driving is riskier than something that is non- rush-hour during the day. One would look at this and say vehicle two is more risky and should pay a higher premium.
With that, I'm going to talk a little bit about where this is likely to go into future given what we know now. The great news from the insurer standpoint is this process appeals to consumers. One of the things we all get when we look at it is the dashboard of information, such as driving on the right hand side. It shows the driving behavior for the insured and with the feedback, there is driving behavior improvement that could occur. This is for an individual who is young and has been able to improve their driving behavior based on feedback. Consumers look at this information and say I like this because it is controllable. Most consumers think of this as being a valuable thing when they can control their premiums. In South Africa, one insurer using usage base insurance uses this as a cornerstone of their advertising. They say that this is the first time that you control your insurance premiums. You can reduce reliance with the proxies such as insurance credit scores, and changes in relatively rare accidents or convictions. You're able to substitute this behavior for other variables that have less influence. It appeals to consumers because the insurers who are fairly active in this have found they can add additional services because they know things about the vehicle operation and can provide safety services, they can do roadside assistance where they can direct the roadside assistance to the exact location of the vehicle even before the insured can describe where they are. There are a number of services. These do appeal to consumers. One of the problems with implementation is figuring out which device. We have already mentioned the prices are coming down. The number of telematics manufacturers out there making these devices available means it is a competitive market and the devices will get more reliable and the cost will come down. This is a small list of potential device manufacturers out there. As these devices proliferate, the cost will come down making this more attractive for the insurers to offer it in the marketplace.
In addition to the fact there are these telematics devices that plug-in, there are many other alternative models that may develop over time. First you can have things that are dedicated service such as the plug in devices where the insurance company pays for the device and has it installed. If it is self-installable it can appeal to a mass market. It can have reasonable cost and ease-of-use. As OEMs install these telematics devices installed at the point of manufacture so they are already there and nobody has to pay for the conditional device cost and to we have to do is transfer the data at the consumer's request from the OEM manufacturer over to the insurer so they can do the services for them. Finally, it is becoming clear that cellular phones and smart phones with GPS are one way to attract some of the consumer behavior and with some modification can be used as a way to communicate this such as the total cost of this can come down over time. It is clear this will proliferate. If you look at the host of companies I have listed, if you look at the speed with which some of these companies are moving this into the marketplace and you look at how the cost is coming down for the insurers to implement this, it's pretty clear the adoption rate is only going to accelerate as we move forward.
A couple of points about why I think this is game changing and I'm going to ask Dave to comment about the future as well. This type of device installed differentiates a product offering that has never been there before. You can increase the pricing significantly so as Dave describes clearly, the insurer gets ahead of this and has the ability to segment the market place and in addition that insurer knows information that no one else knows it is difficult for another insurer unless they are installing advices to actually get at the information. It can change the individual insurer and change the game for auto insurance going forward. It attracts lower risk so that best drivers are the early adopters and allows consumers to change their behavior. It could improve claims handling although it's not used commonly in the US, there are other countries were claims handling is becoming a pretty significant part of the value added proposition for the telematics being installed. Dave, I want to open it up if you had comments.
Jennifer Symoun
I have lost Dave from the phone line. If the operator is on the phone line, please drop Dave from the phone line as he is on hold trying to get back in. I think we will move on to Allen Greenberg and then come back to Dave for comments.
Allen Greenberg
Dave and Robin, those were very good presentations. I just wanted to conclude with a couple of comments with what DOTs might be able to do to help to promote pay as you drive insurance. The first thing is to educate yourself. You can be most helpful in advancing something if you learn the benefits and obstacles to advancing it. Most states require preapproval but some don't but all states have insurance regulations that have to be adhered to. It's useful I think for State DOTs interested in seeing pay as you drive insurance offered in their states to meet with the insurance commission and convey what they believe are the public policy benefits and the commission doing what it can do within the confines of the states laws to allow or encourage pay as you drive insurance to be offered. Now 14 states have action plans that include pay as you drive insurance and they reflect commitments to reduce greenhouse gas emissions within the state and 14 states have identified this strategy. That provides a good in road for the DOT staff to figure out how to collectively promote pay as you drive insurance.
State DOTs have connections to specific resources. The federal aid transportation bill has some programs, one is the congestion mitigation and air quality improvement program has been used to support a variety of pricing strategies and is an obvious fit for providing incentives for pay as you drive insurance. When I say incentives I mean things like helping to purchase necessary equipment and perhaps marketing and the kinds of things that would facilitate pay as you drive insurance to being offered. In addition to this particular program, there is eligibility under other federal aid highway and transit programs. Some states have taken upon themselves to provide state-level incentives. The state of Oregon offers a million dollars in tax credit for this purpose.
If you start down this particular path, there were other things going on that could be helpful as you move forward. This is an optimist critique, if you will. With unstable gas prices and poor economy, consumers are driving less and looking to save money. If you look at the last two years of shopper surveys, we see consumers not only are looking to save money or shopping more for insurance, they are viewing specifically pay as you drive insurance more favorably. Their knowledge about the concept is growing substantially year-to-year and their interest is growing quite a bit as well. There was one research project, the Strategic Highway Research Program and it has a study of about 2500 instrument of vehicles. They are beginning that now and it's going to provide important data about how drivers drive and how it affects crash risks, where they drive, characteristics of the road, location, time, etc. That data could be useful to insurance companies and it would be a public source of data. The federal government has a couple of ongoing pilots in Texas and Washington State looking at pay as you drive insurance pricing. The one thing we do as a condition of funding is we tend to make sure that we collect before data and that way we know to be able to attribute whatever behavior changes there are. That's the difference between what we do and point and insurance company might do on its own. After it instruments vehicle it would normally begin its pricing. They are more interested in prior pricing from an actual standpoint. There's a lot of modeling that suggests large reductions in driving the benefits but we want to confirm that in the studies we support. Next, regions collaborate and cooperate on a number of projects and this is one where it would be good for regions to get together and try to offer pay as you drive insurance. None of us know what will be in the final bill but there's likely to be new opportunities especially related to fuel tax alternatives and emphasis now trying to figure out what states can do because the gas tax is becoming less of a reliable source of revenue for highway and transit trust funds. To the degree that there is support for testing various strategies, and alternatives to the fuel tax, the same mechanisms that are tracking miles or allowing those kinds of alternatives to move forward could also be helpful in enabling or facilitating some form of pay as you drive insurance. That concludes my presentation.
Jennifer Symoun
Dave, we will go back to you and give you a chance to wrap up and comment before we move on to questions.
Dave Huber
I'm pretty confident Robin wrapped it up pretty well. I apologize; my phone dropped the conference call so I may have missed what was said or not said. I think generally speaking the industry is moving in this direction. The success of these programs will depend on it being easy and depend on it being transparent. I think it could be accessible by the customer. Customers need to know how the data will affect their rates.
Question and Answer
Jennifer Symoun
The first question I think is geared toward Allen. Has there been a study on how much of this data has been collected, aggregated, and by who to date?
Allen Greenberg
That is a tricky one. There are multiple studies and I'm not sure whether there has been a study or studies per se. In my own research I documented about a half a dozen public studies. Robin, I don't know if you have additional ones. You cited some data there, I don't know what is public or not.
Robin Harbage
There are several public studies. We have utilized those to give our clients a sense for the predictiveness of this and I'd be glad to follow-up if they want to contact me. I can point them to some of the studies that are available. I think the key is most of the insurers who have launched these programs see this as competitive advantages for them to have the data they collected in an aggregate and individual. It's a bit of a dilemma and we would like to encourage this is much is possible, but the data that is available publicly isn't always the most useful for doing a rating on this from an insurance standpoint. There are clearly a number of studies done that give us confidence this data is predictive and will be useful for the insurance rates.
Jennifer Symoun
Dave, would the insurers be interested in the number of occupants in the vehicle as a data point?
David Huber
I think they would be interested in getting their hands on any information but then would want to figure out if it is predictive. I have no evidence and have not collected data about occupants to know if it is predictive or not but I think it's generally true across usage-based insurance that the more data you can get your hands on the better because if you get enough of it, you can allow the math to figure out what is predictive. If you can include occupant data with how, when and where you drive, then over time you can figure out if that matters. It might be associated with other data. Occupant at different times of day with particular types of drivers might in fact be something insurers would be very interested in. You never say no to any additional data. Just as a side note, I know with very young drivers that are an issue but actually more of the risks relate to the age and gender of the other occupants as well as just the number of occupants. It would be worth pointing out there is at least one telematics device vendor out there who does provide information on the documents through an onboard camera so this information could be collected and utilized.
Jennifer Symoun
Are insurance companies interested in tying pay as you drive insurance with GPS providers such as TomTom that can provide significant information about drivers?
Robin Harbage
Insurers are open to the idea of partnering with anyone who can give them additional information. One of the problems about something like a TomTom or an aftermarket device installed in the vehicle is getting some kind of evidence if you will that the device is continuously and always operated in the car when it is in use. If you are only taking a sample of the driving behavior or you don't know what size of sample it is, it limits its value. At this point in time an aftermarket device like that, that can be installed and uninstalled without the insurer knowing that occurs and limits the usability of that device. With that said, some of the people who manufacture these GPS devices that are used by consumers are also some of the same players looking to operate in the telematics field in the way they can improve installation and use their technology and know how to pass that on and are impacted associations with insurers.
David Huber
This is Dave; I also want to add the same GPS providers have also found their way into the data side of things. I think companies like TomTom have access to speed limit data, posted speed limit data. Speed might be predictive if you can determine speed relative to the posted speed limit that may be even more predictive. I think there are a lot of players out there who can have a hand in this ecosystem as it relates to data or devices or the implementation of the telematics insurance based product.
Jennifer Symoun
I'd be interested in the relationship between pricing and behavior. For example, let's say person A drives at 5 mph over the speed limit but person B drives 10 mpg over. Is there a cutoff point allowing some fudging? Or a direct increase per mile over the speed limit? Or a penalty point, like averaging over 10 mph over?
David Huber
Good question. Somewhere in the answer is the secret sauce insurers are trying to figure out. I don't know that insurers are smart enough yet or have collected enough data yet to know the risk difference the 5-mile an hour over the speed limit compared to 10 miles an hour. Ten miles over on the 65 is a lot different than 10 miles over on the 25. I think speed relative to something probably matters. Certainly there are devices you can set to that based on GPS it can provide in vehicle alerts, audio or visual that says you are exceeding the speed limit. Those are effective when you think about team based telematics alerts. You can deliver alerts when speeding events have occurred. I think the answer is insurers very much want to answer that question but I don't know that they have answered it yet.
Robin Harbage
I would add based on my experience there is there interesting information with regards to speed and safety and the speed relative to the speed that meant and also the speed relative to other traffic following the same direction on the same road but mostly that is proprietary and probably not very available for public use.
Allen Greenberg
There have been some studies on speed and risk. There was a value pricing pilot pricing the Federal Highway Administration was working with in Atlanta with instrumented vehicles. The naturalistic driving study and mentioned, what they try to do is get at odds ratios so they look how prevalent a particular behavior is in general that they look at how prevalent that behavior is either preceding a crash or what they call a near crash and they have definitions of these things. Then there are combinations. For instance, speeding a bit over the limit may not be terribly risky in normal weather but very risky in wet weather and there may be some drivers who are experienced in speeding might be much more risky. These are the kinds of things this particular study is going to get at. There was a smaller version at Virginia Tech that only involved 100 drivers versus the 2000 to 2500 coming up and they were still able to get a lot of really good results. There is some stuff out there and I would anticipate there will be more coming up in the future.
Jennifer Symoun
How accurate is the telematics data in referencing a particular driver within the policy at the wheel? This is a huge factor in the rate calculation of a multiparty policy.
Robin Harbage
Most of the devices don't differentiate which driver is actually driving. It is possible, there are methodologies out there that can help you identify the driver, whether it is a self registration or whether there is a camera taking pictures of the occupants of the vehicle. There are ways of doing it. The important thing to note is most insurers are not fixated with who the driver is. They are fixated on the driving behavior. Insurers have spent a great deal of time assigning drivers to vehicles to a driver policy and are sort of a voluntary decision on how to make the assignments. With this information the insurers are moving away from whom the driver is and looking at the operation so if there are three drivers and one is much more risky to the extent their proportion of the driving is included in the behavior collected, it will be weighted into the total operation score. Most insurers will tell you what they are calculating when they look at the telematics information is the operation score based on the average of all drivers and much less worried about who the driver is.
David Huber
If you think about it, in some cases the arbitrary nature by which drivers are assigned to vehicles on the policy in order to determine the rate that description Robin just shared speaks to the fairness that is appealing to consumers. One of the reasons consumers are attracted to this kind of product, it is based on the actual driving as opposed to some arbitrary assignment that no one understands. In some ways, it may simplify the overall pricing.
Jennifer Symoun
The next question: how does this work with transit?
Allen Greenberg
Transit providers and transit advocates should love pay as you drive insurance. King County, Washington which encompasses the Seattle area, they first approach Federal Highways about funding and application after doing surveys of their van poolers and they have a large vanpool system, I think they are up to 900 or something like that, they asked a range of questions about how we help serve you better. The most popular response, and this includes questions about making the vans better in providing traffic diversion opportunities, the number one answer is to save money on car insurance. People pay per transit trip and they drive mostly for free minus gasoline charges. Would it be a lot easier if it was the reverse? One way to promote pay as you drive, first of all, commuters are on transit vehicles and many of those vehicles have advertisements so that would strike me as a targeted way of getting transit users to consider pay as you drive insurance to welcome them. You have for the gutter group making some decisions to drive less than they would otherwise drive. A way to lure people is perhaps offering a few miles of insurance that are included in the transit pass so people are getting something related to their insurance when they buy a monthly transit pass and then maybe they would be inclined to lean toward a pay as you drive. The opportunities for transit commuters are quite vast and I think they would be fans of pay as you drive insurance.
Jennifer Symoun
We have several related questions about the type of feedback drivers are getting. What type of information do the insurance companies get back to the customers? Do they provide information and what is the frequency to motivate drivers to improve driving behavior?
Robin Harbage
The information offered is quite varied according to the insurer and it is varied on the basis of what they have chosen to create their brand and services for customers and it varies to the information collected through the telematics device. Most of the insurers would give information such as daily trip summaries, the length and number of trips, the time of day of the trips, etc., so the insurer can look at those. Some of them go into a great deal of detail in that they will actually map those trips down so the insured can click on them and they can look at the actual location of the trip where the car was operated. They can see charts that show the speed over time during that trip so they can know that kind of information. At least one vendor gives an indicator as to safe driving habits and will show them when they have read events during the trip so they can see at what point in the trip those occurred. They can look at the kind of behavior and whether or not that affected their rate. If they say every time I do this commute I'm always stopping to rapidly at this stop sign and I'm causing an opportunity for the person behind me to crash into me so I can be a little more defensive. Whether they believe it is a red or bad event is indifferent as they understand that event is what is making them gets charged a higher rate and they can control it.
David Huber
I think also they show how their behaviors are affecting their rate. Whether it is miles or miles and other things, insurers are wanting to provide up-to-date information on how driving is affecting the rate and also it provides the what if modeling tools, if I did this, if I drove this many miles at this time a day at this speed, how might that affect my rates? I think it is in the insurers interest to promote, point out and do the best they can to advocate safe driving behaviors. Insurers don't want to pay for crashes. If it is cost effective, they will do lots of things to encourage safe behavior to prevent crashes. They do that with alerts provided in the car, outside the car, on the web, text message or e-mail. That speaks to the transparency of the data and allowing the consumer to have some control in terms of how they're driving affects their rate.
Robin Harbage
In regards to the frequency of data transmission and updates to the consumer, it is worth noting that varies as well. It is plausible to have data on a real-time basis but that is very expensive so except for exceptional services, most of the data is stored and transmitted at the end of the trip or at the end of the day.
Jennifer Symoun
I'm going to group the next two questions for Allen. Are there any more governmental programs like tax credits to encourage urban or suburban drivers to use this and are there any legislative barriers or patents to implementing this in some or all states?
Allen Greenberg
I mentioned the Oregon tax credits as one example of a government incentive. There have been a couple of other cases of governments going out and seeking partners for small pilots of their own. The North Central Texas Council of Governments has done that and has done some work with Progressive, for example. The 14 states that have pay as you drive insurance in their climate action plan, some of them have a reasonable amount of specificity as to what they want to see happen and what they are committing in writing publicly. A number are much less specific, they just identify the strategy is something they want to advance and don't say anything beyond that. In terms of barriers, my two other colleagues on the call could probably say more about how much a barrier state regulations are at this point. Some years ago there was a survey conducted by Georgia Tech and they asked insurance commissioners whether they would allow pay as you drive insurance and they found there was a reasonable amount of barriers to offering it, although not overwhelming. I think some of those have been coming down.
Finally the question about patents, I have written about this particular issue. Progressive does have business method patents on usage-based insurance. Other companies to my knowledge in the US have chosen not to license those patents. They have pursued pay as you drive insurance in ways I think they don't believe are in violation of those patents. It is a fairly litigated issue. I know Progressive has defended some of its patent claims. I think there is some ambiguity and risk on everyone's part when they pursue these strategies. I think there are some strategies related to pay as you drive insurance that are clearly not subjected to existing patent claims. I am not an attorney and I would encourage insurance companies to seek appropriate legal advice before offering pay as you drive products because of the patents out there.
Robin Harbage
Just a comment on the patents, while David I both worked for Progressive, we need to be careful about what we say about the patents. I can say from public record that while Progressive did sue two insurers over patent infringement, Allstate and Liberty Mutual, at least the case with Allstate has recently been settled. I don't know the details of the settlement or if they are public, but there was some progress on that front. I agree with Allen and I won't make legal statements but people need to be careful of this. These patents decided do not seem to be slowing down the adoption of usage base insurance programs nationally.
Jennifer Symoun
Is this being used in the trucking industry as well? I will put that out to anybody.
Robin Harbage
Telematics is very much a part of the trucking industry in that if you take surveys which we have of large long-haul trucking, probably more than 80% of the long-haul trucks have embedded telematics instrumentality in them that is used for things like creating logs like the drivers time at the wheel and miles driven and used for dispatch of commercial vehicles, etc. Telematics is quite commonly used in the marketplace. On the side of insurance and pay as you drive insurance for commercial vehicles, it is much less prevalent than it is for private passenger auto. That is for a couple of reasons. One is the insurers don't control those telematics devices and the standard of data is not the same as the ones installed by the fleet operators. It's been more difficult for insurers to get a foothold using the data. Not to say they haven't, several insurers do use programs for pay as you drive insurance; it's just not as widespread.
Jennifer Symoun
Okay, Robin, how much do these devices cost and how long do they last?
Robin Harbage
To the first question, the cost a couple years ago was in excess of $200. Today the cost is coming down. The device is being produced in high volumes for insurers, I'm sure they cost well under $100. We are starting to see telematics device manufacturers issue the devices and device prices in the range of at or below $100 so that is good news in the marketplace. Additional to that is the issue of data transfer which if you are using wireless which is the only reasonable way you can make this easy for the consumer to access the data, there is additional cost for wireless and that is a bit difficult. I think Dave mentioned $5 a month for transfer and that's not uncommon but there are ways to bring that cost down.
David Huber
I would add those devices we have been talking about, generally the self installable devices are able to plug into that onboard diagnostics port that are on 1996 and newer model cars sold in the US and that probably represents over 80% of the cars on the road. The prevalent telematics platforms don't work in all cars but work and the great majority of the cars. I think part of the other question is how long did the last? In theory, they should last forever. The data coming from the vehicle is pretty consistent. It may vary over makes and models but any particular vehicle, it is consistent. These devices get power from the vehicle so it's not like the battery would wear out. The chip in the device is just like the chip in your cell phone and those don't wear out. They will last as long as the customer continues to remain a customer of the insurance company.
Robin Harbage
Our experience is the failure rate is very low. The firmware or software on the devices can be updated wirelessly so they don't have to be uninstalled to upgrade and keep the software on current.
Jennifer Symoun
We have another question related to feedback. Driver behavior may be affected from a safety and consumption standpoint if the feedback you receive is ongoing, but the newest iteration of the Progressive data received is a short window of time. In other words, there was no ongoing feedback loop. This is probably more efficient for the insurer but of reduction in the positive impact on the discussion today. What are your thoughts on this?
Allen Greenberg
I think that is an astute observation and I agree with it. That's the difference between what an insurer might do on its own versus what one would expect an insurer to do in partnership with a government entity. The insurance company just wants to get a very accurate picture risk and they want to charge appropriately for it. I don't think they care that much whether the risk is high or low if they are charging appropriately for it. In this instance, I think this is where government has a different point of view. We are in the game to reduce the risk. We want to reduce risk exposure and want people to drive less and more safely. It's only through our presence and demands that in order for us to help support a program; we need the risk to be made transparent to the driver and the incentive to be made transparent to the driver. The monitoring has to continue for that to occur. To get the best results, somebody needs to be rewarded for changing their behavior and so I think that was a very astute comment.
Robin Harbage
I don't disagree with Allen in terms of segmentation that this creates. I think it's important to note that different insurers are choosing different ways to offer their service and create their brand. Some are saying I will keep it in the car because I can provide additional services. The breakpoint comes down to the insurers intent and it comes down is continuing to monitor the behavior, is that more beneficial from a profitability standpoint. It is a choice based on profit models.
David Huber
Clearly Progressive is less interested in changing behavior as attracting a certain behavior that exists today from an acquisition strategy.
Jennifer Symoun
Okay, we are actually out of time and I know we have been a more questions that have not been addressed so I will send them to presenters and ask for written responses and I will get those out to everybody who register today. By the number of questions, this is a hot topic and we want to keep the discussion going. If you think of others, feel free to type them in and I will make sure the presenters get them. We're about out of time for today. I want to thank all of our presenters and thank everyone for attending. The next webinar will be on September 22 and will be about Economics of Congestion Pricing and Impacts on Business. Please visit the web address shown on the slide to register for this webinar and view the schedule of other upcoming webinars. Â For those who have been participating with these for the last few months, please note the change of dates for the November webinar. Thank everyone for attending and have a great rest of the day.
