Welcome to the Progressive Corporation's Investor Relations Conference Call. This conference call is also available via an audio webcast. [Operator Instructions] In addition, this conference is being recorded at the request of Progressive. If you have any objections, you may disconnect at this time. The company will not make detailed comments in addition to those provided in its quarterly report on Form 10-Q and letter to shareholders, which have been posted to the company's website, and will use this conference call to respond to questions. Acting as moderator for the call will be Julia Hornack. At this time, I will turn the call over to Ms. Hornack.

Good morning. Welcome to Progressive's conference call. Participating on today's call are Glenn Renwick, our CEO; John Sauerland, our CFO; and Bill Cody, our Chief Investment Officer. The call is scheduled to last about an hour. As always, our discussions on this call may include forward-looking statements. These forward-looking statements are based on management's current expectations and are subject to many risks and uncertainties that could cause actual events and results to differ materially from those discussed during this call. Additional information concerning those risks and uncertainties is available in our 2014 annual report on Form 10-K and our first quarter 2015 quarterly report on Form 10-Q, where you will find discussions of the risk factors affecting our businesses, safe harbor statements relating to forward-looking statements and other discussions of the risks, uncertainties and other challenges we face. These documents can be found via the Investors page of our website, progressive.com. Tory, we are now ready to take our first question.

[Operator Instructions] Our first question is from Vinay Misquith from Sterne Agee Center.

So the hot topic of discussion, I guess, is frequency. So I'd be curious to hear from you, what changed this quarter? Was it the past for the higher frequency? Was it individual states? And what do you think is going to happen for the future?

Yes, sure. And I can easily understand why that'd be a topic. So let's try to give as much color as possible. Tell me if I'm being responsive to your questions. If I look at frequency, the one thing I would tell you, at least I would encourage you, look at the time sequence of frequency because individual data points can be sort of interesting, but all over the map. So if we look at sort of through the second quarter, we were right in line with the industry. And so nothing particularly notable there on frequency. And we did have, and I'm sure it's been a story, obviously, a couple of companies, big companies, very credible companies reported frequency that was considerably higher than the industry. This quarter, we're reporting frequency up and not dramatically so. So I'm going to break that down a little bit and try to give you a sense of how with -- or our thinking around that is. So BI and PD are sort of in the 4 range. And frankly, from my perspective, if you remember, this time a year ago, we're at negative. So recognize percentage change over a prior data point is important as well. But 3 to 5, I would tell you, my expectations of any reasonable period of time, maybe not a single data point or a quarter, 3 to 5 on bodily injury tells me, sort of, we're in reasonable normal range. So we're not seeing anything on frequency that is shocking or different than or inconsistent with our expectations for the future. And I say expectations for the future because that's what it all comes down to is how are you priced for June of next year, not necessarily tomorrow. So the one place on frequency that…

Yes, it does. In fact, just to follow up on that, so you hit a good point that severity for Progressive is about minus 1. We've seen for the industry more in the 3% to 5% range. So if you put it all together, frequency, as you said, 3% to 5%, and let's say that severity caused that to normalize levels. Progressive, I believe, is raising rates around 4%. Do you anticipate Progressive having to raise rates more near term? Number one. And number two is would you be willing to sacrifice near-term profitability since everybody's hurting? And so would you be willing to let your combineds go up near term so that you can gain share? And then sort of price it back up in 2017?

Two questions in there. First of all, our current pricing actually is in very good shape. I'm not going to give you our indications, but I -- probably more than I've ever done before. We'd tell you instantaneously, our indications are not positive. So that's a good place to be. The real issue is taking a little bit of rate to stay on the crest of the breaking wave and not go over it. That is the challenge, and that's the challenge I outlined. I don't think, a, I am willing and nor do I think it is necessary for us to sacrifice our #1 commitment to shareholders, and that is to produce an aggregate 96 for our products. For a close observer, you will notice that Direct, which is growing very, very well, will have, at least on an accident year basis, a little bit of pressure on it relative to that 96. It's a complicated message to say we'll never sacrifice the 96. But can I expect that with a higher and very satisfactory inflow of new business in Direct, that I would be more comfortable operating an accident year aggregate for that part of the business, higher than a 96? You bet you. Because we know how that business runs up. We have high confidence. We're not betting on the future. That's a simple bet. And if we can take on the new business where we think that's appropriate, we're going to do that. So don't expect us to be moving in directions of cutting back our advertising. We think we've got the momentum. We're going to keep that momentum. We're just going to keep a little bit of rate coming along so that our rates stay and keep us on the crest of that wave. And frankly, these are market conditions that are extraordinary for us. And it's up to us to make sure that we ride them out as best we possibly can.

If I could just add to that, this is John, if we allow that calendar or accident year period for Direct to drift up to the 97, we're still very confident that business is running out on a lifetime basis at or materially below that 96 combined ratio. And the other business segments we run, we don't necessarily target a 96, so we think that our different businesses having different target margins, depending upon the marketplace, their appetite. But again, the aggregate we are always very conscientious around running that below the calendar year 96.

Our next question is from Ryan Tunis with Crédit Suisse.

I guess, my first question, Glenn, I was a little bit surprised to hear you talk about BI frequency, and a 3% to 5% range seemed normal. Because I think back in Investor Day, you talked about contemplating a longer-term downtick secularly in frequency. So I'm just wondering if that's normal for now, or if there's a change there, now you're thinking about frequency longer term.

Yes, that's fair to call out now, Ryan. My career has been based on something that might not be the next 30 years as well, so 3% to 5% is how I've always felt. I think that's how we feel is the norm for now, to the extent there are pressures, macro pressures in terms of vehicle technology and so on and so forth, but that will change. But the likely influence of that in the short term is going to be so small that it won't change my 3% to 5% for some time. But do I believe that 30 years from now, we'll be sitting with different measures? Probably.

Just add to that, so Glenn started out saying we should be looking at a trend series as opposed to a data point. And absolutely, if you look at the very long-term trend, a positive number on frequency and a negative number on severity is not what we expect. And as we showed you at Investor Day, has not been the case for the past 4 decades or so. So important to think of the longer term. Again, the most important question is are we priced for tomorrow and for next year. And as Glenn said, we feel pretty confident where we're sitting, and we continue to have robust appetite for growth.

Okay, that's helpful. And then just shifting gears, I guess, to thinking about capital management, growth this year has been robust. Obviously, you've grown in Home, which is a little bit more of a capital-intensive product as well. And I think in the past, obviously, you've been able to play some capital at year-end via special dividend. I'm just wondering how we should think about balancing organic growth and capital return heading into the end of the year relative to maybe what you did like 2012 and 2013.

I think we get this question -- I'm actually smiling a little bit, I'm hopeful -- in my response, if we get this question about this time each year, and let's just -- let the year play out. The special dividend, you have the means to determine where we stand on that. So that's not a secret in any way, shape or form and any other capital management would have said the same thing so many times that we care a great deal about effective capital management, good capital husbandry. And if we think there is something more to do, we'll do it if there are other options. But we've always said, we want to invest in the business. We've got some nice growth going. We can fund that growth. That's great. But if there are other opportunities, that would be used. And if not, we'll return on the leveraged capital if we think that's the right thing to do. But we just don't make forward-looking statements on that. And frankly, there's enough volatility in the investment marketplaces to just say, "Let's just -- let's count it on the day that we're supposed to count it."

That makes sense. And then I just want to sneak one more in on Agency. I think, Glenn, you highlighted in your letter that renewal -- that the renewal -- or the retention measures in Agency still are lagging a little bit on the new issued apps. I'm just wondering if you had to take more rate in Agency, that would seemingly put more pressure on renewals. I mean, is it reasonable to think that if you had to take more rate in the Agency channel, you could still generate pith [ph] growth there?

Yes, it is. And I want to sort of try to measure this. This is -- comes back to the same statement I just said, 3 1s better than 1 3. It's a question of how you take rate. We have certainly been on these calls and told you that we've taken rate in larger quantities. That was absolutely the right thing to do. We needed to do it. But it has an equal and opposite offset on affecting consumers and putting consumers into the shopping marketplace. Clearly, there's no perfect number, but we do a lot of work on elasticity of the consumer relative to the size of rate change that they can absorb. And also, you have to sort of factor in the competitive environment. So right now, we're seeing a lot of rate action, greater rate action than we ever needed to take this year in all channels, but certainly, the Agency channel. And as a result, we're competitive in that channel and it's shown by our new apps. Can we sustain a gradual rate increase to match our future expectations of trend? Yes, I think we absolutely can and shouldn't be assumed that it's quite as dramatic as a binary kind of take rate, therefore, growth comes up. We're really trying to signal. And my comment, while designed or not, about the paranoia that I have is exactly that. It's how can you just keep creeping your rate in. And for some close observers of Progressive, we actually have some things built in structurally to our rate order in many states where we actually have ways of matching continuous inflation or trend. And those are ways to keep all changes relatively low, so we don't shock the consumer, and we don't want to do that, but at the same time, keep up with trends. So the answer to your question is, we almost certainly will have to take some rate in all channels over a reasonable period of time. This is not an industry where rates yet are systemically going down. So we'll do that. The real question and the only question I think is on the table for the continued momentum of Progressive, can we match that in small increments and stay right with inflation, and we have a perfect opportunity to be able to do that.

Our next question is from Josh Shanker with Deutsche Bank.

A question about the variable dividend. I saw the note in the 10-Q. I'm just wondering, is that at your discretion? If underwriting income were below comprehensive income, could the -- the board can do anything, could you choose to pay the dividend anyways?

Let me answer that a little bit -- I'll answer it, but the board has final discretion. The board also has discretion to do a special dividend. So they have a lot of discretion. So my comment about managing capital, there is clear discretion built in. With regard to the trigger, it'll either trigger or it won't trigger. If we gave you an analysis as of right this moment, it would be different than the analysis at the end of the quarter. So this -- the variable dividend is designed to give to the board a very clear set of understanding. If it triggers, expect it to be paid. If it doesn't trigger, yes, there is some discretion.

And can I assume that the circumstances that have led to this condition right now are not the circumstances that this trigger mechanism was designed for in the first place?

No, I wouldn't say that. No. Think about it from a shareholder perspective. We're having a great banner year in terms of operations. I fully respect that. But even if it's on a total mark-to-market basis, if your capital is not doing what you want it to do, then you might question the logic of giving it back at that point in time. And that's what the trigger was designed to, at least, be in place to do. It worked exactly as designed in 2008, and I admit that was a much more dramatic set of circumstances where the operating income was positive. But clearly, the mark-to-market on our portfolio suggested it would not have been the right move to do. This year, you know the market as well as I do. It's interesting that we were close. And when I say close, well, I mean close. So we'll just let that one play out. But no, it was in place to do exactly this situation. And relevant to your first question, if you ever had a situation, and please do not try to read more into my statement than just what I'm saying, if you ever had a situation that was simply -- that doesn't make sense, then that's why a discretion was put into place as part of the whole dividend structure.

Next question is from Meyer Shields with KBW.

So one question on the severity, and I'm trying to think the best way to ask this, was the moderation in BI severity a function of the frequency increases? But if you look at the same sort of claims, you're seeing the same rate of increasing costs. Or is there something else going on?

That's a good observation. Certainly, you could say there are a higher frequency of smaller claims coming in, and that will clearly change your severity. And that probably has a contributing piece, but I would tell you the greater contribution has been a very orchestrated initiative in that claims organization to take a very close look at specials, managing the specials appropriately, and there are literally hundreds of ways that -- to do that and to do it accurately and correctly, making sure that we have our liability assessments. As we grow, we always have newer people. So it's a never-ending job to keep going back and making sure liability determinations and contributory damages and the like are absolutely taken into account in the right way. So those initiatives are absolutely in place, and we're seeing some very positive results from them. I would tell you our specials to general ratio is going down. These are all active managements. So I would read in most of the change in severity to be active management as opposed to mix-driven.

And just for clarifications, when we say specials, we mean the medical cost portion of the bodily injury claim and the generals were the pain and sufferings. So that's managing the medical cost portion of it to, we think, more accurate outcomes.

And that's helpful. If I could switch gears to Commercial Auto. You noted in the Q something of a mix shift with writing larger policies, and I'm guessing that the competitive situation is getting better there. Is that the reason that you're seeing some year-over-year compression in the underwriting margin in Commercial Auto? Or is that just the anticipated erosion from the nominal possible levels?

Why don't you take that?

So we are very happy with our Commercial Lines margins. So any erosion we're seeing is simply a -- migration is -- slow migration back to what we think might be our targets. So we are growing a lot in the truck end of the Commercial Lines spectrum. We report our new business average premium to you as well as our renewal, and we're seeing a lot of business coming in on the trucking end. We took a lot of rate there. Probably 2 years ago now, 2 to 3 years ago, instituted some material changes in our underwriting to tighten up. So we think we got ahead of that market. And when we look at our combined ratios across the segments, meaning truck down to what we call business auto, we're very comfortable with what we're seeing. And there again, are very interested in continuing to grow at a rapid pace. We're seeing great trends there. So any margin erosion you're seeing, we're not at all concerned with, and we are still well below our target combined ratios in all of the segments we write in Commercial Lines.

And just to add one other comment, it wasn't specific in your question, but maybe it's a good reminder, if you didn't know. We often say we're focused on high-frequency, low-severity claims but adjudicate relatively quickly. Some of the zones of Commercial Auto take you into higher-limit-type business. Just to make a mental note that anything over $1 million is reinsured, so we're not really subject to that sort of very low-frequency, high-severity hit. We've decided to run our business in a way where we're much more comfortable with the underlying limits. And therefore, we don't have quite the same volatility at higher MBI claims.

And as you note in our Q, perhaps a little different place to some other companies that have been reporting lately around reserve development in that segment. We have been taking reserves down in our Commercial Lines group. So we're pretty confident with our reserves there as well as our combined ratios.

Next question is from Paul Newsome with Sandler O'Neill.

I want to ask a capital question and a long-term capital question related to how you think about capital management as well -- and the diversification benefits of -- anyway, we've had a lot of interesting press on AIG talking about a large diversification benefit for being a diversified financial. Historically, you have not been, but you've added Home. And I'd love to hear how you think of the interaction of capital management as the Home business and as the commercial business increases in size.

Okay. I think -- well, maybe I said most of that before, but it's worth reinforcing. So I'm going to let Bill come on the call. Is he there?

Yes, I'm here. Sure. As far as the diversification benefit from a portfolio point of view, we do think about the assets in that part of the business being a longer duration, the claims a little bit longer tail risk and managed from a more conservative perspective than our auto book. [Technical Difficulty]

We may have lost Bill. Maybe I can pick that up. So the Commercial Lines is certainly growing and is now getting to a point of sort of 15% of premium outside of the property, which has slightly higher capital requirements. For sure is we're writing higher-limit business there but not materially higher than the Personal Auto. The property side is still, while new to us for sure, it's still very small. And we continue to follow the approach that ASI had when we took a majority interest in the company, which is to only have a maximum of 10% of surplus at risk in any given store. So you can see from the statutory filings, they used reinsurance extensively and protect capital pretty, pretty well. And again, relative to the total Progressive capital, it's pretty small. So there may be some diversification benefit there. We, obviously, consider the premium-to-surplus ratios the regulators require by line of coverage, maybe require is a strong word, but generally look to. Obviously, the rating agencies as well. So there's greater capital requirements in property. Where we're very comfortable with auto at a 3:1, property, you can think of probably half that, and we consider that into our capital models and ensure that we continue to have plenty of contingency capital. And to the other question around dividends and capital deployment, we are very much comfortable at call it our balanced range. So we're doing modestly different things in terms of capital management, given ASI. But again, it's a pretty small piece of our business now, pretty conservatively managed with reinsurance, so not a big deal at the total balance sheet level.

Our next question is from Brian Meredith from UBS.

I just go back to the frequency situation, sorry to belabor this, but the 3% to 5%, when you talk about 3% to 5%, is that different from what we would see ISO reporting? Because if I look at long-term frequency trends, they've been actually flat to down for more than a decade. And even if I look at when you're reporting on frequency since the beginning of '09, it's been flat.

Gary, you probably have a better perspective on long term ISO. Clearly, I recognize the 3% to 5%, I'm just saying it's sort of not unrealistic and not sort of unusual. We've also shown you, as John talked about in the IR meetings, the long-term frequency trends 30 years, and frankly, we're forecasting the next 30 will not be significantly different, maybe even steeper. Frequency is going down. But at any moment in time, for pricing purposes, especially when you have a macroeconomic change in the economy and miles being driven, 3% to 5% does not sort of take me out of a range that I'm uncomfortable with or have seen in general. Maybe Gary, if you can comment on the long-term 12-month ISO term.

Sure. This is Gary Traicoff, Chief Actuary. I think to your point and Glenn's, the long term both with Fast Track industry data and Progressive data, frequency is down. When we look at just particular quarter, so if we take second quarter of this year to second quarter of last year, from fast-track data, it's a little over 2% and we were about 1.5%. So we're a little bit below that. If I look at Fast Track over the last year, they're about 1% as well as Progressive. We don't have third quarter Fast Track data yet. But when I am comparing our data versus industry data, at least, over the last 3, 4 quarters, it is up overall but tracking very closely to each other.

Got you. That's very helpful. And then Glenn, going on to the Agency businesses, looking at your renewal ratios, continuing to be negative here, I'm just curious when do you think that could potentially stabilize or turn positive. And then also, in that question, the Platinum product, when do you think that could potentially have an impact on those retention ratios in your Agency business?

Yes. Retention on the Direct side is actually behaving predictably and well. And what I try to do, and I admit I'm a little bit in the cautious camp, I hope I'll always stay there, that I like to see things before I report them happening. And on the Agency side, it's just -- frankly, it's just a lot slower to sort of show the pickup that I think we should reasonably expect, given our market competitiveness and rate competitiveness. But I think it's coming, and I would just say, give me another quarter to confirm that. And I'll confirm it or tell you exactly what's going on. I had thought that we probably would be able to do a more positive statement about Agency retention in the third quarter. When you can't do it, don't do it. With regard to Platinum, recognize Platinum for now is going to be a very small part of the business. So that's going to be hard to see blend in to the entire retention trend. That probably will be more something we'll start to call out for you more deliberately because you just simply won't be able to pull it out of the macro numbers. And that's fine. We'll be happy to sort of give you, at least, color as to our expectation on Platinum. But we can start with a very clear assumption that I have given. Part of the reason why ASI and Progressive are working the way they're working together is that we had no doubt from our first work in PHA that when customers bundle, they have highly retentive behavior. And there is nothing that has suggested to us that, that is not going to play out in the scenarios that we have in place on the Direct side and now on the Platinum side. So Platinum's pretty exciting for us and for the Agency. So we look at it as a source of growth in the Agency. We look at it as a source of new customer mix, and we certainly look at it as a source of customer mix that will show different retention behaviors longer than we have typically experienced, so all positives on that one.

And while Platinum is going to be a fairly small part of our independent agent production here in the near term, we're starting slower there. And absolutely, that bundle is far more likely to stay with us. The precursor of the Platinum program is our newest product model. And we have worked hard to continue to get more and more competitive on preferred business with iterative product models we're rolling out. And what we're seeing with our latest product model is more mixed to multicar, which inherently retains longer, more mix towards full coverage, towards customers who have their insurance in place for a longer period. So that's new business mix. That takes a while to move into the book to affect overall retention levels. But while we're seeing our current book getting a little more stable and improving in terms of retention, we're getting new business in that we think will retain longer as well. So we think we've got a number of forces there that will push that retention number into the positive territory soon.

Right. Just for those following along, we often talk about the product as 8.3, the new product going into the marketplace. But we also talked at our IR meeting about the next one, 8.4. I know they're not glamorous titles, but 8.3 is in about 12 states right now and is tightly coupled with Platinum, even though Platinum is in 8 states right now with about 4 more to go. Those are the same states. There's just a slight bit of delay. And then you've also heard us talk about the new model for Snapshot. That's really all part of 8.3 as well. So the same states will have each of those. And I've reported previously both disappointment with uptake in Snapshot by the Agency channel and the new product model offering a discount upfront has actually moved the needle on that. I would be far away from suggesting it's gotten to the Direct levels. It hasn't, but it has gone in the range of double what it was. So a new Snapshot model also leads to what we fully expect is longer retention, simply because those who get the right rates are likely to stay in the marketplace and they won't be able to necessarily shop for a better rate than that. Having said all of that, we've also told you that there will be the opportunity for a smaller percentage of the book to get rate increases, though some of those people will be more inclined to leave. So we'll try over the next year to give you a very clear picture of the moving parts that involve in retention, but we would -- sitting here today, we would be disappointed if we didn't see the Agency retention start to track on a nice upward path. That's a good question, if not answered in our other documents for the fourth quarter.

Our next question is from Mike Nannizzi from Goldman Sachs.

Just back to the frequency conversation just for -- just a second here, if we could. Can you talk just about what you saw through the quarter and into October in terms of the trend that it wasn't stable, did it deteriorate? And then when you think about this call again next quarter, what would you have to see from frequency to start to be concerned about the potential for adverse selection, if you feel like that's a risk at all?

Let's say I quaked at the adverse selection piece on frequency, so I'll start there.

If you're just writing new business, you're pulling new business in and the frequency trends deteriorate...

Oh, mispricing.

Yes.

Okay. Actually, your point on adverse selection is not a bad one either. Obviously, Snapshot work is trying to find people with what we believe, as a group, will be lower frequency and bring them into the book, welcome them in at a rate level that is good. And for those that have what we consider to be high frequency likely behaviors, they may not join us as rapidly. So there is some adverse selection. We hope it's actually not adverse to us. We will -- it's almost impossible to answer your question specifically. We have to do our indications based on forward-looking estimates of frequency and severity. We combine them. We call it pure premium. And that's going on, on a daily basis, and I mean daily. We will make estimates, and we have estimates. They are probably a little bit tamer than the single data point at the last quarter. And if that data point continues to start to suggest it's a little bit more of a time sequence, we will adjust very quickly for that. The point that I've been trying -- I, hopefully, have been making is we will take what we see now in small increments. So even if we get higher frequency or higher severity in the future, we can -- we're already on path to have placed that in. We just don't want to get behind the curve on doing that. And if we saw something -- it's not a single rate revision. We don't take rates across the country on a single day. We're taking rates all the time. It's a very dynamic structure. We've spent a lot of time and money making sure our systems can allow us to take rate changes relatively quickly, and that's the key. If you've got…

And through the -- you mentioned through the quarter and into October that frequency trend stabilized, that it wasn't rising. Was it reverting? Can you just briefly touch on that?

The good news is I really can't give you enough detail to be meaningful on October numbers, so you'll get to see those very quickly.

Yes. And throughout the quarter, I would say, again, frequency is important to look at on a trend basis. Looking at it a month-to-month is probably not what we want to start communicating and not what we react to. So you see the combined ratios throughout the quarter on a monthly basis, that should give you a pretty good indication of where pure premium, which is again, as Glenn mentioned, the product of frequency and severity is going. And I don't think it makes sense to comment on monthly frequency changes.

Okay. And I just -- if I could squeeze in one last one, just on Snapshot and the Agency channel. Just trying to understand one thing. I mean, it seemed like the idea of Snapshot upfront was sort of good drivers selecting in because they believe that they would be able to get a better discount, a better price based on their own driving as opposed to these other demographic factors. And then it sounds like you introduced an iteration of Snapshot in the Agency channel, where it's sort of a discount upfront. But it seems like just psychologically that, that's a different value proposition, is -- the consumer may not feel like they're a great driver. Maybe they'll get a discount during that first year. And then if they're not a good driver, they'll just not renew once you guys figure that out. Have you seen -- has any of that manifested? That may be way off base, so I don't really know. But have you been doing this long enough to see in the Agency channel to see whether retention trends are different there than in the Direct channel?

In that way, I base it all human behavior is highly predictable, that, at least, some will take a discount at any year at any opportunity. We don't have a ton of data. This has not been in the marketplace a great deal of time. So we are seeing the uptake at the Agency point of distribution to be considerably improved. And I have little doubt that there are some customers that take a discount and maybe aren't exactly the right people, and we'll determine that in the first 6 months and their renewal rate will go up. That absolutely is a scenario that can and will happen. It is, however, the smaller part of the population. And agents, I think, are now starting to see the benefit of Snapshot for the larger group of customers that they have. We're seeing, and these are early numbers, we're seeing midterm retention, i.e. in the first period actually increased and not much yet notable, which wouldn't be surprising because we haven't been in the marketplace with this product that long for an update for what I'll call that first point of truth of the 7 month over the 6 month, that point of retention, which is clearly very important. And that's where people are getting quoted either the rate that they got as an initial discount, maybe even better, maybe slightly worse than the initial discount, but it will be the rate that they can expect going forward. And there will be a number, 30% or less, that will actually get the discount taken away. And those are, obviously, people who are going to react differently on renewal. But for the most part, this is attracting more new customers of the right type, and that certainly is an indication we can give earlier than getting retention estimates. We're seeing the kind of right credit mix, the right age and vehicle mix. So the mix of customers that are opting for Snapshot is actually meeting expectations in that regard, maybe even better. And just for one slight correction to your question, this product exists both in the Agency and the Direct channel, so it's available to both. So we really do have a lot of experience at much higher levels of acceptance on the Direct channel. About 1/3 of our sales are being made with Snapshot, so we have a really lot of data. And while there will almost certainly, as there is in every part of our business, be some aberrant behavior, some fraud, whatever it might be, the greater good here is far outweighing any concerns that we have about the individuals.

Our next question is from David Small with JPMorgan.

I just had a -- my question was, could you put your comments of this quarter in the context of what you said last quarter when the frequency trends were obviously different? Last quarter, you said you were comfortable with your pricing trends when you were seeing frequency lower. Now obviously, you're seeing frequency higher, and it seems like you're still saying you're comfortable with your pricing trends. I'm just trying to put those 2 together. Are you comfortable with your -- are you -- do you feel like -- is that meaning that you're going to wait to see if this trend persists, which it sounds like you think it will, so you're going to have to adjust your pricing to that? I'm just -- it seems like there's a different message last quarter than this quarter?

Yes, that's -- again, it's a time series. So last quarter, we didn't see ourselves being particularly different than the industry. The time series that we think is more reflective of future trends. So we take 2 things into consideration, historic trend and future trend. We're always making that estimate. And as we see individual data points that change that trend, either instantaneously or long term, we will adjust that future trend. But the future trend is not going to be nearly as volatile as a single data point or a single quarter. As I said, if we go back a year from now, we were reporting negative frequency, and so we're reporting a positive frequency over a negative frequency last year. The longer-term trend might be quite different, if you sort do 2 or 3, and don't do my math on that one. I'm just giving you an example. So we do feel comfortable with our rates because we're never stable. It's not like rate level was, a, at the beginning of the year hasn't changed. If you take a look at the market in general from a premium weighted perspective and there's going to be a standard deviation that is higher than normal or what I would consider normal. But you're going to look at rate that has changed for most competitors somewhere between 2 to 3.5, maybe get a little closer, clustering 3 -- 2.5, 3.5 of more recent times. We would be at the lower end but in that range, so we're continuing to take our rates up. And that's the point I'm making with comp. Our rate trend has approximated our pure premium trend. My comment that says I'm always, always concerned and we as an organization are always concerned, if that future trends change,…

At the risk of making that a longer question, I would tell you that it's important to look at what we're getting in the marketplace versus what we're taking. So we can take rates up. We can take them down. What we ultimately get in the marketplace is obviously a function of where our competitors sit. So if I were in your shoes, I would pay close attention to the average premium changes that we report. And for Personal Auto, that average premium change, albeit influenced by mix changes, is 4%. It was up 4% this quarter last year as well. Last year-to-date, it was up 3%. This year-to-date, up 4%. So average premiums are at least -- assuming a common mix, are certainly keeping up with pure premium trends.

And so if I could just follow up on that last comment, was actually my second question was, if I think about price versus loss trend and how that relates to margin, if I understand your earlier comments correctly where you're saying -- you talked about that long-term frequency and to look at everything over the long term. If I take frequency and severity together, I kind of have 5% in trends. And so if I'm getting 4% in premium, does that mean, kind of just simple math, there's a little bit more potential for margin compression. But obviously, things could bump around. So is that the right way to think about it?

We talked about pure premium, as Glenn mentioned previously, I think that does well. So we try to price to pure premium plus our cost structure. Pure premium is the product of frequency and severity. And yes, we try to project that out. So future trend is what we talk about. We're trying to predict the product frequency and severity out through the course of the period in which our rates will be in effect, plus our cost structure. Cost structure obviously is very competitive. Both expense ratio and LAE, they're not the lowest in the marketplace. They are sort of top 3-ish in both categories. So we project those out to create the rates. And yes, we're trying to keep up with that pure premium trend. So the math you're doing, while it's -- we give you, obviously, pretty high-level numbers there, is the math we do to set the rates. And again, that average premium, if you look over time, as reflected in the combined ratios, kept up with trend.

Our next question is from Adam Klauber with William Blair.

It seems like your paid loss ratio and paid to incurred for so far this year compared to last year is doing better. But again, frequency is ticking up. So could you help us with that?

So I think at the -- the core of your question is reserved changes, and are we priced adequately for next year, given where our accident year is. You can correct me if I'm wrong on that. So our accident year, loss and LAE is around 74. You add in our expense ratio, and you come to a number that is comfortably below 96. So we're approaching that 96, and we're cognizant of that, and appropriately taking rate as needed. That bodily injury trend that we discussed in terms of severity is what is driving the change in reserves and the positive development. So we think that is a systemic change, meaning we see why bodily injury severity is down. We think it's as a result of actions we take in our claims department to be more accurate in our settlements. We have more initiatives in the works to ensure that we continue to be as accurate as possible in injury settlements. So while we can't project where that severity line is going to go with tremendous accuracy, obviously, that's what we're charged with doing in our pricing group and our reserving group. We feel pretty good about where our reserves sit today and, again, can see the source of those changes. And in aggregate accident year basis, we are below that 96 and thinking of that well managed going forward.

Okay. And then also frequency has been an issue for the whole industry, and everyone's been saying loss has ticked up. It seems like you've managed it better than a lot of the other competitors in the industry. Did having data from Snapshot and miles driven -- did that play a part in that?

I would -- that would be a nice complement to take, but let's -- frequency is always so tricky. Frequency is what it is. We're not living on a isolated island from our competitors. So while there will always be statistically different numbers reported, they all come from the same ecosystem of roads, drivers and vehicles. However, that's not true when we all have different mixes in different states and different types of drivers, rural, urban. It really does get quite tricky. I tried to put a small note in my cover letter that really suggests that we do a lot of work to try to make sure that we understand frequency at a level that we can at least reconcile with all the data points in the marketplace. But I would tell you that managing frequency better than anyone else is something I'm not sure that we're going to put our hand in the air and say, "We can take credit for that." It is what it is. If everybody measures it the same way, it should be a representative, for example, of the overall industry, especially for larger players. Having said that, the point I'd make, and I will just use -- don't bring it up and use a competitor, but Geico's mix in New York is very different than our mix in New York. So as frequency in New York is different than in other state, that will show up, and you need to be very careful to draw conclusions about whether frequency is up or frequency in a specific place is up. I did mention before that Snapshot has a long-term benefit, we believe, of attracting lower-frequency drivers and, in some sense, making sure attracting them simply because we offer them the rate that we believe…

That concludes our call today. Tory, I will turn it back over to you for closing remarks.

Thank you, and that concludes the Progressive Corporation's Investor Relations Conference Call. An instant replay of the call will be available through Friday, November 20, by calling 1 (866) 346-7116 or can be accessed via the Investor Relations section of Progressive's website for the next year. Thank you for your participation today. We will now disconnect.