Good morning, ladies and gentlemen. Welcome to the third quarter results teleconference for Travelers. We ask that you hold all questions until the completion of the formal remarks, at which time you will be given instructions for the question and answer session. As a reminder, this conference is being recorded on October 18th, 2018. At this time, I would like to turn your conference over to Ms. Abbe Goldstein, Senior Vice President, Investor Relations. Ms. Goldstein, you may begin.
Thank you. Good morning and welcome to Travelers' discussion of our third quarter 2018 results. Hopefully, all of you have seen our press release, financial supplement, and webcast presentation released earlier this morning. All of these materials can be found on our website at travelers.com under the Investors section. Speaking today will be Alan Schnitzer, Chairman and CEO, Dan Frey, Chief Financial Officer, and our three segment presidents, Greg Toczydlowski of Business Insurance, Tom Kunkel of Bond & Specialty Insurance, and Michael Klein of Personal Insurance. They will discuss the financial results of our business and the current market environment. Then we will take your questions. Before I turn the call over to Alan, I would like to draw your attention to the explanatory note included at the end of the webcast.
Our presentation today includes forward-looking statements. The company cautions investors that any forward-looking statement involves risks and uncertainties and is not a guarantee of future performance. Actual results may differ materially from those expressed or implied in the forward-looking statements due to a variety of factors. These factors are described under the forward-looking statements in our earnings press release and in our most recent 10-Q and 10-K filed with the SEC. We do not undertake any obligation to update forward-looking statements. Also, in our remarks or responses to questions, we may mention some non-GAAP financial measures. Reconciliations are included in our recent earnings press release, financial supplement, and other materials available in our Investors section on our website. Now I'd like to turn the call over to Alan Schnitzer.
Thank you, Abbe. Good morning, everyone. Thank you for joining us today. By now, you've seen our numbers. We're pleased with the results. This morning, we reported third quarter net income of $709 million, or $2.62 per diluted share, generating return on equity of 12.6%. Core income was $687 million, or $2.54 per diluted share. Our core return on equity of 12% was strong, particularly considering that while catastrophe and non-CAT weather losses were significantly lower than last year's quarter, in the aggregate, they were nonetheless elevated relative to our expectations. Our performance this quarter is the result of excellent underwriting execution across a diverse portfolio of businesses, a successful and consistent investment strategy, and the continuation of our strategy of returning excess capital to our shareholders.
Notably, our results this quarter benefited from record high net earned premium and a sub 30% consolidated expense ratio, reflecting progress that we're making on our strategic agenda. About a year ago, we explained that our strategic plans were largely geared toward creating top-line opportunities and that improving productivity and efficiency was an important component of our strategy. Each of our business segments has contributed on both counts. They've done so by successfully implementing technology workflow and product enhancements and by making sure that we're delivering compelling value to our customers and distribution partners. Before I get further into the results, I want to acknowledge the devastation caused by Hurricane Florence and Hurricane Michael. Our thoughts and prayers are with all those who have been impacted.
Following events like these, our claim organization works hard to make sure that our customers see the value of the promise they bought from us and our agent and brokers see the value of the Travelers promise they sold. Case in point, as of today, we've closed more than 90% of our homeowners' claims arising out of Hurricane Florence, and we're bringing the same level of urgency for our customers impacted by Hurricane Michael. We're thankful for the extraordinary effort by our claim professionals. Turning to production, we were pleased with the success of our marketplace execution. Growth in each segment contributed to consolidated net written premium growth of 6%. In domestic Business Insurance, retentions remain at historical highs, which is a reflection of the fact that we have a high-quality book of business, and a significant percentage of it is meeting our return expectations.
Renewal premium change, including pure rate and exposure, is strong at around 5%, which is about two points higher than in the prior year quarter and stable sequentially. Inside that, pure rate is up about a point quarter-over-quarter and down a little sequentially. Economic activity is contributing to a healthy level of exposure growth. At this level, pricing is more or less offsetting loss trend, but as you know, we don't manage the headline number. There's a continuum of price change with workers' comp at the low end and commercial auto at the high end. In between, there are accounts, products, and geographies where we need more price and others where we don't and will execute accordingly. In terms of pricing and profitability, there are a number of factors that impact our returns beyond rate and loss trend.
First, we actively manage all the available levers, such as risk selection, mix of business, claim and expense initiatives, volume, reinsurance, and so on. Beyond that, there are a few environmental factors at play. A decade after the financial crisis, interest rates and, as a result, net investment income are moving higher. As I mentioned, economic activity is generating exposure growth. Thanks to tax reform, we're benefiting from a level playing field and a lower tax rate. We take all that into account in our pricing strategy to achieve our target returns. We follow the same approach across all of our businesses. Turning to Bond & Specialty Insurance, net written premiums increased by 5%, with strong production in both our management liability and surety businesses. In Personal Insurance, net written premiums increased by 6%, benefiting from continued successful execution in Agency Auto and PIF growth in Agency Homeowners.
You'll hear more shortly from Greg, Tom, and Michael about the segment results. Before I pass the microphone, I'd like to share a few thoughts about some of the investments we've made and are making. We've grown our top line year-to-date by over $2 billion or 11% as compared to the same period in 2016. Over that same two-year period, we've managed G&A expenses to about flat. You've seen the impact on our expense ratio. In addition to a lot of hard work by the best field organization in the business, our ability to have done that has been enabled by significant investments in things like core technology and workflow. We've been making these investments over some time while delivering leading returns, and we continue to invest thoughtfully and strategically.
As we shared with you before, in terms of innovation, we have three clear priorities: to extend our lead in risk expertise, provide great experiences for our customers, agents, and brokers, and improve productivity and efficiency. We're encouraged by the progress we're making. As you've probably heard, we recently announced that we've teamed up with Amazon to launch an industry-first digital storefront. This initiative is designed to create new opportunities for us and our agents to attract and serve customers and help them take a more proactive approach to home safety in an increasingly digital world. It's obviously very early days, but over time, we are hopeful that this and similar initiatives will lead to new business opportunities, improve loss costs, and provide us with valuable insights. This past summer, we made a majority investment in Zensurance, a Toronto-based digital provider of online insurance solutions for small businesses.
It's a natural fit with our Simply Business team, further adding to our digital capabilities and medium-term strategic positioning to serve the micro and small commercial customer. We also recently announced an investment in Kittyhawk, the market leader in enterprise drone operations software, further enhancing our position as an industry leader in the use of drone technology. In short, we believe the winners in our industry will be those with deep domain expertise who can continue to deliver industry-leading results while innovating successfully on top of a foundation of excellence. Internally, we refer to it as our perform and transform mandate. From a position of strength, we believe that we have the talent, resources, expertise, and focus to do just that. For the past 66 quarters, this is the point in the call that we pass the microphone to Jay Benet.
Today, I'm pleased to be passing the microphone to Dan Frey. Before I do that, a word about Jay. It's hard to overstate the impact that Jay has had on Travelers. He's a big talent, and virtually every milestone in the modern life of this company bears his mark. Having said that, this day was foreseeable, and we're well prepared. Thankfully, we'll continue to benefit from Jay's experience and wisdom as he'll continue to serve on the senior leadership team as Vice Chairman. One of Jay's biggest achievements is having developed a world-class bench of finance talent, including Dan Frey. Dan brings more than three decades of finance and accounting experience to the role. He served in a number of strategic financial management roles across Travelers over the past 15 years, most recently as the CFO of Personal Insurance.
I couldn't be more pleased to have Dan as a partner. His deep technical expertise, strategic mindset, and collaborative style make him the perfect fit to help lead Travelers into the future. With that, I'm pleased to turn the call over to Dan.
Thank you, Alan. Core income for the third quarter was $687 million, up $434 million from the prior year quarter, and core ROE was 12%, up from 4.5%. The significant improvement in both measures from last year's third quarter resulted primarily from a lower level of weather-related losses and very strong net investment income. Our third quarter results include $264 million of pre-tax CAT losses, a significant amount, although well below the $700 million in the prior-year quarter, accounting for 6.9 points of improvement in the combined ratio. Prior year reserve development was slightly favorable and comparable quarter-over-quarter. The consolidated underlying combined ratio of 93%, which excludes the impacts of CATs and PYD, remains steady as a higher level of non-CAT weather losses was mostly offset by the improvement in our expense ratio.
Our pre-tax underlying underwriting gain of $448 million remained steady as increases in Business Insurance and Bond and Specialty Insurance were largely offset by results in Personal Insurance. Pre-tax net investment income increased by $58 million from the prior year quarter to $646 million, while after-tax NII increased by $90 million to $547 million. Both measures benefited from the more favorable interest rate environment, an increase in average invested assets resulting from continued growth in net written premiums, and favorable results in our non-fixed income portfolio. On an after-tax basis, NII also benefited from the lower U.S. corporate income tax rate. As we had expected, after-tax fixed income NII increased by $60 million, and we expect after-tax fixed income NII in the fourth quarter will increase by $60 million-$65 million compared to the fourth quarter of 2017.
As net written premiums continue to grow and reinvestment rates are expected to exceed existing rates on upcoming maturities. Net favorable prior year reserve development of $14 million pre-tax, or $10 million after tax, was comparable with the prior year quarter. On a year-to-date basis, net favorable prior year reserve development was $350 million. Both Bond and Specialty Insurance and Personal Insurance experienced net favorable PYD this quarter. Business Insurance experienced net unfavorable PYD, driven by a $225 million increase to asbestos reserves, along with a $57 million increase to commercial auto reserves that resulted from elevated severity in recent accident years, partially offset by favorability in the workers' compensation line. Excluding the asbestos charge, Business Insurance had $169 million of net favorable prior year reserve development. The asbestos reserve increase resulted from the completion of our annual asbestos review during the third quarter.
The increase was driven by higher estimates of projected settlement and defense costs for mesothelioma claims compared to what we had previously expected. As has been the case in previous years, while there was some slight improvement in several of our asbestos indicators, our expectation had been for more of an improvement. This phenomenon has continued to result in periodic reserve strengthening for us and for others in the industry. The number of deaths from mesothelioma, which we believe is a leading indicator of future asbestos payments, has trended downward somewhat during this decade. At the same time, the mix of mesothelioma deaths by age has shifted to older cohorts.
That may be the result of medical advances over time successfully treating a variety of other deadly diseases, allowing more people who worked with asbestos when it was prevalent in the workplace prior to the late 1970s, to live long enough to develop mesothelioma in their 70s and 80s. Ultimately, this high-risk cohort will become smaller over time. We continue to expect that the current claim environment will improve as time goes on. Bond & Specialty Insurance's net favorable prior year reserve development of $53 million pre-tax was driven by domestic management liability, while in Personal Insurance, the net favorable PYD of $17 million pre-tax was driven by domestic personal auto. Turning to capital management. Operating cash flows for the quarter of $1.7 billion were again very strong.
All our capital ratios were at or better than target levels. We ended the quarter with holding company liquidity of approximately $1.4 billion. While it was already overfunded, we made a discretionary contribution of $200 million to our qualified pension plan before we filed our 2017 tax return in September, which provided a 35% rather than a 21% tax benefit, resulting in a $28 million economic tax benefit in the third quarter. Recent increases in interest rates, which have benefited our fixed income NII, have resulted in a modest net unrealized investment loss impacting shareholders' equity. As of September 30th, we had an unrealized investment loss of $447 million after tax. Remember that changes in unrealized investment gains and losses do not impact how we manage our investment portfolio, nor our business.
We generally hold fixed income investments to maturity, the quality of our fixed income portfolio remains very high. Changes in unrealized gains and losses have had little or no impact on our statutory surplus or regulatory capital requirements. Adjusted book value per share, which excludes unrealized investment gains and losses, is now $86.51, 4% higher than at the beginning of the year. We continue to generate excess capital. Accordingly, returned $607 million of capital to our shareholders this quarter, comprising share repurchases of $400 million and dividends of $207 million. On a year-to-date basis, we have returned $1.76 billion of capital to shareholders through dividends and share repurchases. While it obviously did not impact our third quarter results, I'd like to provide some commentary regarding Hurricane Michael.
While much of the media focus on this storm has been on the devastation of homes in Florida, and our thoughts are with all those impacted, it's worth noting that this was a multi-day event impacting both personal lines and commercial lines in a number of states. In addition to Florida, the PCS footprint also includes Alabama, Georgia, Maryland, North Carolina, South Carolina, and Virginia. Given that it's just a week or so since the storm hit, we are still assessing our view of losses from the event. At this point, we expect the losses from this storm to be significant in terms of meeting the threshold for inclusion in the schedule of catastrophes in our quarterly SEC filings, but manageable relative to the fourth quarter overall, not something that would preclude us from resuming stock buybacks next week. I'll turn the microphone over to Greg.
Thanks, Dan. Business Insurance produced segment income of $410 million and a combined ratio of 100.6% for the quarter, both significantly better than the prior year quarter, due primarily to a significantly lower level of catastrophes this year. The unfavorable prior year reserve development that Dan mentioned had a point and a half impact on the combined ratio. The underlying combined ratio of 95.4% was one point better than the prior year quarter as non-CAT property fire losses, which were elevated in the prior year, returned to more normal levels. Although the underlying loss ratio improved from the prior year, it was impacted by about a point from higher than expected non-CAT weather. For perspective, the third quarter of 2018 included 18 PCS in equivalent events outside the U.S. that did not meet our CAT threshold, while 2017 had only six.
While losses from non-CAT weather were unfavorable to our expectations, CAT losses were favorable, and the aggregate of the CAT and non-CAT weather was generally in line with our expectations for the quarter. Our expense ratio improved by around a half a point as we remain focused on managing expenses thoughtfully while making ongoing strategic investments and prudently growing the business. Turning to the top line, net written premiums were strong for the quarter at $3.6 billion, up 6% over the prior year quarter, driven by strong production results. Regarding domestic production, we achieved strong renewal premium change of 5.1% with renewal rate change of 1.8%, while retention was exceptional at 86%. New business of $467 million was up 7% from a year ago. Pure rate is up more than a point from the third quarter of 2017 and down a bit sequentially.
As you know, we execute at a very granular level. We continued to achieve renewal rate gains broadly across our middle market accounts with Auto, CMP, GL, property, and umbrella all having positive rate increases. Auto continues to be the line with the highest level of rate, consistent with our view of performance of the line. We're pleased with these production results, and consistent with Alan's comments, we continue to execute our marketplace strategy to meet our return objectives with thoughtful balance towards retaining our best business, improving pricing where it's needed, and pursuing attractive new business opportunities. Turning to the individual businesses. In select, renewal premium change and renewal rate were stable, while retention remained strong at 83%. New business was 5% over the prior year quarter as we continue to leverage our investments in technology and workflow initiatives.
In middle market, renewal premium change was 4.7%, with renewal rate change of 1.6%, while retention remained historically high at 87%. New business premiums of $269 million were flat to the prior year quarter in aggregate. In our core commercial accounts business, where our business initiatives are most advanced, new business was up 5% as we continue to be more active in the market with increased quote activity. In closing, Alan laid out our innovation agenda across the company. As we shared with you at last year's Investor Day, for Business Insurance, that means furthering our capabilities around customer solutions, convenient processes, and competitive cost structure. Our strong marketplace execution, including the impact from the progress we've made on these strategic initiatives, have contributed to our ability to deliver nearly $600 million of additional net written premium year-to-date in 2018, while keeping insurance G&A expenses generally flat.
With that, I'll turn it over to Tom to talk about Bond & Specialty Insurance.
Thanks, Greg. Bond & Specialty delivered strong returns and growth across the business. Segment income of $196 million was up $60 million from the prior year quarter, due primarily to a higher level of favorable prior year development. The underlying combined ratio was an exceptional 78.3%. Net written premiums for the quarter were up 5%, driven by broad growth across our businesses. These results reflect the impacts of strategic products, marketing, and distribution initiatives to grow these profitable lines, including workflow enhancements to drive improved efficiency and productivity internally, as well as for our agents and brokers. Advanced analytics leveraging our valuable proprietary and third-party data to refine our marketing and underwriting strategies, and improvements in value-add capabilities to further distinguish our product and service offerings in the marketplace. Turning to production in our domestic management liability business.
Given the level of returns we are achieving, we continue to execute our strategy to retain a substantial percentage of our high-quality portfolio while pursuing attractive new business. We are pleased that the retention came in at a very strong 89% for the quarter, with a renewal premium change of three points. New business was at a record level, up 12% from the prior year quarter. Bond & Specialty results were excellent, and we continue to feel great about our execution in the marketplace, our growth in returns, and the opportunities that our strong market positions and competitive advantages present for the future. Now I'll turn it over to Michael to discuss Personal Insurance.
Thanks, Tom, and good morning, everyone. In Personal Insurance this quarter, we continued to deliver on our objectives of balancing growth and profitability in Agency Auto and sustaining momentum in homeowners. Segment income of $153 million increased $76 million over the prior year, while our combined ratio of 97.2% improved two and a half points. These improvements were primarily driven by lower catastrophe losses and a modest amount of favorable prior year reserve development, partially offset by higher underlying loss experience, which I will discuss in more detail shortly. Agency Auto delivered a strong quarter with a combined ratio of 91.3%, down 14.7 points from the prior year quarter. While the quarter benefited from 1.8 points in favorable reserve development as well as 6.7 points from lower catastrophe losses, the main story is the continued improvement in underlying auto profitability.
The underlying combined ratio improved over six points to 92.6%, primarily as a result of the continued successful execution of our underwriting and pricing strategies, as well as frequency that was better than we expected. Taking our year-to-date underlying combined ratio for Agency Auto of 94.8% and factoring in our normal expectation for a seasonally higher fourth quarter combined ratio, we're on track to be within our targeted combined ratio range for the full year, which is sooner than we originally expected. In Agency Homeowners and Other, the third quarter combined ratio of 100.3% was impacted by a significant number of catastrophe and non-catastrophe events. Catastrophes, including the Carr Fire and Hurricane Florence, added 11 points to the combined ratio this quarter, a slight improvement relative to the third quarter of 2017.
The high number of other PCS events that Greg mentioned, along with a variety of additional non-catastrophe weather events, accounted for roughly six points of the 10-point increase in the underlying combined ratio for the quarter. The remainder of the increase was largely attributable to non-weather loss activity, including elevated water and fire losses. As we indicated last quarter, we're actively factoring this more recent loss experience into our granular pricing and underwriting decisions. Renewal premium changes have increased year-over-year. We expect an acceleration in homeowners pricing in future quarters. Turning to the top line, Agency Auto premiums grew 6%, driven primarily by price increases, which continue to moderate consistent with our plans and in line with improving profitability. In addition, we're pleased with the modest improvements in retention and a 2% increase in new business, the first year-over-year increase since the first quarter of 2017.
In Agency Homeowners and Other, we remain pleased with our momentum, delivering 6% growth in both net written premiums as well as policies in force while achieving renewal premium change of 3.8%. All in for the segment, results continue to improve, and we're pleased with our progress and our plans to generate profitable growth going forward. As Alan mentioned, we're excited about our new relationship with Amazon, along with the other strategic initiatives we have underway, including the ongoing rollout of Quantum Home 2.0, further expansion of our IntelliDrive Auto Telematics program, and continuing investments in digital capabilities that improve ease of doing business for our customers and agents while making us more efficient and effective. I'll turn the call back over to Abbe.
Thank you. We're ready to begin question and answers, please.
Ladies and gentlemen, at this time, we will now conduct a question and answer session. If you would like to ask a question at this time, simply press star followed by the number 1 on your telephone keypad. Your first question today comes from the line of Kai Pan with Morgan Stanley. Please go ahead.
Thank you, good morning. First, best wishes for Jay for his retirement and congrats to Dan on the new position. My first question is on non-CAT losses. You have elevated non-CAT losses for several quarters now, yet in your outlook, you expect them to normalize in 2019. What gives you confidence that recent experience will now be the new normal?
I guess, Kai, the outlook section, Business Insurance, that underlying margin, I think as we reflect on those words, is a reflection of large losses returning to more normal levels by historical standards.
At what time, at what point, I'm wondering, those recent experience will factor into your outlook or normalize non-CAT losses?
Is that a PI comment?
Is that a PI question, Kai?
I think probably for both, if you can comment on both PI as well as BI.
Kai, if we're talking about weather, this is the insurance business and there's going to be volatility in weather. We acknowledge that we've had a run of quarters where we've had some variability in both cat and non-cat weathers. If you go back and take the longer view, looking at cats and non-cat weather together for the whole place, 2009 was better than we expected. We had a run from, I think it was 2010 to 2012 that was worse than we expected. At the end of 2012, we were thinking, gee, is this a new normal? We went into a period of 2013 to 2015 that was favorable. 2016 was sort of spot on. You got 2017, so far 2018, that's been elevated. Our objective is to get it right.
Missing it too high is no better than missing it too low. Our objective is to get it right and make sure that we're pricing our product over time for what we expect on an over time basis. We think that we're effectively doing that. In terms of if this is a question about modeling and you're trying to figure out how to think about your model for next year, we give you all the help we can in the outlook section of the 10-K and 10-Q. Thinking about this business on a longer term, we're taking all the data we have into account. We're thinking about everything we know about the future and making sure we're pricing our product for the long term.
You don't think it's sort of like, is there a secular change in term weather pattern that could change your expectations?
Kai, I don't know whether you measure weather cycles by the year, by the decade, or by the millennia. We know it's in our data, and we can see that. Part of this isn't just weather, it's population migration into more wildfire and weather-prone areas. We take all of that and all the history into account when we think about both our underwriting, our risk selection, geographically where we want to be, what kind of risks we want to write, terms and conditions, what our risk control people look for when they're going out looking at the property to underwrite it, and we take it into account in pricing. We're very interested in the question you ask, which is why off the top of my head, I knew what weather had done going back almost a decade.
We follow it very carefully and take all that into account when we're thinking about the future.
Thank you very much for that. My second question on workers' compensation. Some of your peers indicated there's a rising frequency trend in workers' comp claims. I just wonder if you could discuss your experience as well as the pricing is going down, will you be able to maintain the underlying margin in this line of business?
Yeah, Kai. We've obviously followed all the chatter in workers' comp with great interest, and that's a business that we've historically been very good in, and we've got big stated data, and we like that business a lot. Getting back to your question, Loss trend, frequency, and severity, those are selections, right? Those are our management estimates based on actuarial analysis of history and a view of expected trends. That is a view that goes into both pricing and reserving. You set up those expectations, and then you look at the data quarter by quarter as it comes in. There are always ups and downs in the data every single quarter in virtually every product line.
You look at those ups and downs, and you take a step back and you say, "Okay, what do we think caused those things to go up and down?" Sometimes we know and sometimes we don't. Is that normal variability, normal volatility in that measure? Or is there something more fundamental going on that we expect to persist into the future? Particularly in a long-tail line, you'd be pretty careful about making a dramatic change based on one quarter of data. I'll tell you, we're watching it very closely. We do see ups and downs in these measures from time to time. We haven't seen anything year to date that's caused us to move off our loss picks. Kai, back to your other question. You asked, given pricing, we'll be able to maintain returns in the product.
If pricing is going down and everything else is equal, pricing is for sure an input into returns. If pricing is going down and everything else is equal, then the margins in the product will go down. That's just arithmetic. Pricing isn't the only factor. There is loss trend, as we just discussed. There's risk selection, there's claims handling, there's all the other levers that go into the return of a product. Your question, will the decline in rate cause the return to go down? If that's the only thing it changes, yes. That's not the only lever that there is. We're not going to get into our forecast of margins on a product-by-product basis, though.
Great. I really appreciate all the answers.
Thanks, Kai.
Your next question comes from the line of Ryan Tunis with Autonomous Research. Please go ahead.
Hey. Thanks. I just had, I guess in Business Insurance, the $160 million or so of total favorable development, I think that's excluding the commercial auto and excluding the asbestos. What's contributing to that? Is that basically entirely workers' comp, or are there other liability lines that are also showing some redundancy?
Hey, Ryan, it's Dan. One clarification. The $169 million was excluding asbestos only. It is reduced by the commercial auto strengthening. Most of the driver of the remainder of the favorability in Business Insurance is coming from the workers' comp line, although there is some favorability in other general liability, once you exclude asbestos. Workers' comp, that's the big driver.
Great. I think probably for Michael, just on the home results, I think you said elevated water and fire, I think it was 4 or 5 points of loss ratio deterioration. Just trying to understand exactly what's going on there, why that's a trend, why that's something you would gate a price for. What's driving that? What could be driving that? Just trying to understand what the loss trend is in the non-CAT.
Sure, Ryan. Thanks for the question. I'll just clarify. The 10-point deterioration in the quarter, again, roughly 6 points, roughly two-thirds of it is coming from non-CAT weather, which includes PCS events that don't meet our threshold for a major CAT, as well as the other weather that I talked about. Your point about the other water losses and the fire, again, the remainder of the 10 points is largely due to those. That's not 4 or 5 points. It's-
It's a smaller number than that, call it roughly a third. It's about 50/50 non-weather, water, and fire inside that. I would say the fire experience is very much sort of normal period to period volatility. I think the non-weather water is something we've been watching. I think you've heard others probably talk about non-weather water losses. We're digging into it. We haven't found a smoking gun underneath it. We have seen a deterioration in that experience, and that is one of the underlying drivers that's causing us to raise price, and expect an acceleration in Homeowners RPC on a go-forward basis. Tying back to Alan's comments earlier on non-CAT weather and underlying results, again, we look at all of that and are baking a portion of that into our outlook and into our underwriting and pricing strategies on a go-forward basis.
We are factoring in the elevated loss levels that we have been seeing recently into our pricing and underwriting decisions and our pricing and underwriting strategy, which is again, why you see our outlook for pricing to be higher.
Thank you.
Your next question will come from the line of Brian Meredith with UBS. Please go ahead.
Yeah, thanks. A couple quick questions here. First, I'm just curious, higher interest rates, does that have any impact on how your pricing models are coming out at this point? Any impact on it?
Well, I guess a couple of things on that. One, it's certainly an input into our pricing model. There's no question about that. It's hard to isolate how any one factor is going to impact pricing, right? You start with the return on the product when the policy renews, you put rate loss trend, interest rate assumptions, et cetera, into it, and you come up with a price. Is that your question, Brian?
Yeah, exactly. I'm just wondering if some of the impact we're seeing maybe on your renewal rate has anything to do with just the higher interest rates going into your kind of calculations.
Oh, I don't think so. First of all, because of the duration of the assets, it's going to take a while for that to really ramp up and work its way in. As the risk-free rate goes up, our cost of equity goes up, that's also an input into our pricing model. I wouldn't say that the increase in interest rates is a dampener on price.
Great. Just quickly on the Amazon and the, I guess, connected homes. I wonder if you could dive a little more into that. What do you think the kind of benefit you'll see from a loss trend perspective, or is there some kind of metrics you can give us that when you've got a connected home, loss experience is supposed to be 20%, 15%, 30% better?
Sure, Brian. It's Michael. Thanks for the question. Again, I would say it's a long game we're playing here, and it's certainly something that we think will create opportunities for both us and our agents. We do believe that connected home devices should have a favorable impact on loss experience. It's very early days. We've not only been doing the work with Amazon, we've been conducting other experiments with other smart home device providers to try to quantify the benefits. What I can tell you is that we're filing to expand the discount for connected home devices in conjunction with this and the other work we've been doing, and moving that discount to about 5% as we roll the new Quantum Home 2.0 product out across the country. That's probably the best number I can give you to put a point on it.
Again, it's early days, and we're very much looking forward to the teaming up with Amazon and to some of the other experiments to gain insights into the impact of these devices on loss experience.
Great. Very helpful. Appreciate it.
Your next question will come from the line of Paul Newsome with Sandler O'Neill. Please go ahead.
Hi. Good morning. I was hoping you'd give us a little bit more thoughts on the ultimate trend for the expense levels and expense ratios in both Business Insurance and the Personal Insurance business, and just how far you think you can push that number around.
I think we're going to avoid the temptation to try to forecast the expense ratio. We give you all the outlook on margins in the outlook section that I think we want to give. There's a really important strategic point here. Productivity and efficiency are very important initiatives around here. Everybody is focused on it, but it's not necessarily for the purpose of just lowering the expense ratio. It's for the purpose of giving us flexibility from a financial perspective. We're trying to create flexibility to take that benefit and either let it fall to the bottom line, we can reinvest it in important strategic initiatives, or we can make a decision to put it into price without compromising a return threshold. While it's on everybody's mind here, and it's a big focus for us, it's not just about managing an expense ratio.
Do you have a target or a goal in terms of the level of investments that you want to put in some of these sort of new InsurTech type businesses that you're focused on this quarter?
Yeah. There's not a bogey that we would
Share from a just competitive sensitivity perspective. We certainly think about the opportunity and what it's worth when we invest in these things. We've made a deliberate decision so far not to set up a fund or allocate a certain amount of dollars to an initiative like that. We may change our mind, but at least our perspective now is you do something like that, and then you end up spending the money, whether you find value in it or not. The fact of the matter is we generate a lot of excess capital, and our objective is to reinvest that money anytime we can when we think we can create shareholder value with it.
Having said that, we're very disciplined about every dollar we spend, and we put it up against a lot of metrics and measures to make sure we're getting that value out of it. I wouldn't put a number or scope the opportunity.
Thanks for the call.
Thank you.
Your next question comes from the line of Amit Kumar with Buckingham Research. Please go ahead.
Thanks, good morning. Two quick, I guess, follow-up questions. The first question goes back to Kai's question on workers' compensation. Would it be possible maybe just to talk a bit more about how should we be thinking about the pricing? Because if you look at some of the pricing surveys, comp pricing is still in the negative territory, but coming back. What would be really helpful is, just broadly, if we exclude workers' comp, how will the pricing look like?
Well, on workers' comp, I don't know if you're looking for a forecast on pricing, where that's going. Obviously, we're not going to get down to.
No
forecasting the pricing in a line. That's not much more competitively sensitive than that. But what I will say about pricing in workers' comp, it's a reflection of a line of business that's been very profitable for us. So there's nothing about the pricing dynamics so far that's a surprise to us.
Okay. The other question is a follow-up on the Amazon relationship. I was a bit surprised that it wasn't somewhat of an exclusive relationship, and in the fine print, it shows that it's in four or five states, and it's initially targeting, I guess, 15,000 customers who will get the free Echo. Can you just talk about the trajectory, and did you try to be an exclusive relationship or not? Just give some background on that.
Yeah. Sure, Amit. The 15,000 you referenced is a specific target of devices for the state of California only.
That's just a unique feature of the way we set the arrangement up in that state. To your point, just to give background for the rest of the group. The first four states, and this is really driven by regulatory filings and regulatory requirements. There's actually an endorsement to the policy that's required for us to be able to make the offer. So we're essentially linking the Echo Dot offer and the discounts on the connected devices with that filing as we make it and get it approved across the country. The first four states are just the initial four states. We plan to extend it to additional states as we go forward. And again, the 15,000 is just the number of devices specifically for California, subject to the terms of the agreement with Amazon.
As respects your question about an exclusive, we're not going to comment sort of specifically about the details and the terms of the agreement beyond that. Again, I bring you back to we're excited about this opportunity. We think it's a great opportunity for us. We're excited about the relationship with Amazon, and we're excited about the opportunity it presents for our agents and brokers to promote the program and help sell these home safety solutions to our customers.
Thanks for that color. I will stop here. Thanks.
Your next question comes to the line of Yaron Kinar with Goldman Sachs. Please go ahead.
Good morning, everybody. First question, going back to the expense ratio. It just seems like a real substantial improvement this quarter. I know you've highlighted the expense management initiatives for a while, and we've seen the earned premium growth as well. I was just curious if there was anything else specific this quarter that was driving this improvement, whether it was timing or other one-offs that helped the ratio here.
No, there's nothing unusual underneath it. There's no timing or sort of one-time things that are artificially improving it. It's a lot of hard work over a long period of time, enabled by years of investment in technology and workflow that's enabled it.
Okay, it just all came together this quarter, I guess.
You've seen an expense number that's been flat over a pretty long period of time. You've seen an expense ratio that's been on a decline. It's sub 30. It's as good a number as we've printed in a long time. As we see it, the continuation of a trend.
Okay, fair enough. Then going back a second to non-CAT weather, specifically in BI. I noticed that you exited the, I think the open market property in Lloyd's this quarter. Would that or should that have some positive impact on non-CAT weather going forward, you think?
Yaron, this is Greg Toczydlowski. That's a real small portion of our overall Lloyd's book, and hence, is a real small portion of our overall BI portfolio. Really wouldn't have a material impact on the loads of the business going forward.
Okay. Would you be willing to tell us what the year-over-year change in non-cat weather was in BI?
Yeah. I don't think that's detail that we've given. We did tell you that non-cat weather and BI, I think what Greg said was about a point relative to what we would've expected.
Right.
Hopefully, that's useful color for you.
Okay, I appreciate it.
Your next question will come from the line of Josh Shanker with Deutsche Bank. Please go ahead.
Yeah. Thank you for taking my question, and good morning. Following up on Yaron's question, I might have it wrong in my notes, but I think in 3Q 2017, you said you had 200-250 basis points of manmade fire losses that were elevated in 3Q 2017, and now we have 100 basis points of elevated non-cat weather. Maybe my numbers are wrong there, but if I strip those two items out, am I wrong to come to the conclusion that the underlying true underlying loss ratio in BI has deteriorated over the last 12 months?
Well, it's certainly worse by, I don't know exactly what your math was, but I don't know, there's another 50 basis points or something like that we would just consider the normal variability in the insurance business. I wouldn't consider it a structural. The word deterioration suggests a systemic structural deterioration. We wouldn't say that. We'd say there's another 50 basis points of just variability, and this is the insurance business.
Well, if I go back to like the three Q17 10-Q, the projection was that margins would improve as, I guess these items. Are my numbers right? There were 200 to 250 basis points of fire losses a year ago embedded in the numbers. That's still correct?
That is correct.
Okay. Then, I think we've been through it in the past, but I just want to walk through. You had a very good quarter on alternative investments results. In terms of when you record them, the limited partnerships and whatnot, are they on a lag? Are they in real time? Can we just talk about how those are recorded and how we should think about that going forward?
Sure. It's Bill Heyman. Actually, it's a mix. If a private equity fund sells a portfolio company and distributes the proceeds within a quarter, that comes into NII for this quarter. If at the end of the quarter, those proceeds are still in the fund, that doesn't come into NII until we receive the financial statements from the fund. If you carve out distributions, and this year they've been responsible for the minority of private equity fund NII, there is a lag in terms of written-up values. I know people ask us, maybe you, to try to project, but with 75 sponsors, 200 funds, and 3,000 portfolio companies, we probably couldn't come up with something that was even meaningful to us, much less something we'd venture to share. We probably can't offer much guidance on what the next quarter will look like.
Hey, Josh, let me just go back to your question on BI underlying. It seems implied in your question that you're looking at our outlook or explanation of prior periods and thinking about that as a forecast for a particular quarter. We're not giving the outlook information to try to give you guidance on any particular quarter. We're just trying to give as much help as we can on the general trends, as best as management can see it and interpret it, and what those trends are and how they're influencing margins over the forward year. We identified the large losses last year because that was an outlier, and in this quarter, it just so happens that those return to a more normal level. There are other factors that are going to impact this business from time to time. There's going to be non-CAT weather.
We had that this quarter. We told you that was about a point. There are base year changes. There are premium adjustments. There are all sorts of things that in any particular quarter could impact the result. I would just caution you in terms of the way you use that outlook in terms of an expectation that it's guidance.
I want to get off someone else's question, do you feel that margins are about stable on where they were in your mind a year ago, generally trying to adjust for all these items?
I don't think backwards, I think forwards. We are where we are. We think there's an elevation of 100 basis points of non-CAT weather, we've given you in the outlook as best we can where we think it's going to go.
All right. Well, good luck with the future. Thank you very much.
Thanks, Josh.
Your next question comes from the line of Larry Greenberg with Janney Montgomery Scott. Please go ahead.
Hi. Thank you. Actually, my questions have been answered, but just one other thought. Why not just use the PCS definition or allocation of catastrophe losses in your own CAT definition, which, I don't know, might not do anything for you guys internally, but probably just would help with communication with the investment community, and probably doesn't eliminate the non-CAT discussion completely, but probably reduces a bit. Any thoughts on that?
Yeah, sure. It's a topic that we certainly debate among ourselves all the time. We created the convention that we had, I don't know, probably in 2002 or something like that, or maybe even before that. The theory behind it was, there's going to be some level of weather losses that ought to be stable and relatively predictable. Then you have things that are more severe that aren't. Obviously, particularly more recently, we've seen a lot more volatility in that, we call it working layer of weather. I don't know if we'll do it, if we won't do it, but we appreciate the comment, and we have thought about it and continue to think about it from time to time.
Great. Thank you.
Thank you.
Your next question comes from the line of Meyer Shields with KBW. Please go ahead.
Great, thanks. Two quick questions. First, I guess last quarter you talked about pricing within domestic BI, excluding workers' compensation at 3.6 points. I was wondering whether you could give us an update for the third quarter.
Yeah. We gave that last quarter. It was relevant. We hadn't intended to give it going forward. There's some competitive sensitivity to giving that on an ongoing basis. Having said that, and having given it to you last quarter, I don't think we would intend to give it going forward, but I'll tell you, that also ticked down a few tenths of a point. Again, I'll point you back to the RPC of five, which, if you think about that five in a historical context, that's a very strong number. While I'm not going to quantify it, but the RPC ex workers comp is even higher than that. Hopefully a couple of helpful data points.
No, that is helpful. Second question, I was hoping you could dig a little bit into what observations drove the commercial auto reserve charge.
As we indicated, it was severity in recent accident years, and we could speculate on some of what the causal factors might be and whether that's inexperienced drivers or higher density on the roads given the economy. We can give you some factors that we could all read about. The fact of the matter is, it's hard to isolate it to one or two things that we can pull out of our data and say that it's this. It's a line that's been challenged from a profitability perspective for a while. We've known about that. We've been pricing for it. We've gotten, I don't know, over a couple of years, cumulatively 20 points of rate or something like that. We're trying to get ahead of it. It's hard to get to the causal factor.
If we can analogize to PI auto, we've seen some fluctuations in frequency and severity there, and sometimes those are explainable and sometimes they're not. We can speculate about some things, but I wouldn't know exactly what to point to tell you what it is.
Okay, great. Thanks so much.
Thank you.
Your next question will come from the line of Jay Cohen with Bank of America, Merrill Lynch. Please go ahead.
Yes, thank you. Most of my questions were answered as well. I guess just on the Amazon relationship, this is such a massive platform, and I know what you're talking about now is relatively modest for Travelers, but was there any discussion around a broader product offering on their platform? Was auto thought about at all as far as being on that platform as well?
Yeah, Jay, I think, again, I would stick with we're very excited about what we've done with Amazon, and not get into speculating on what other opportunities might exist there. We're really focused on making the most of the relationship we've established here, again, for us, for our customers, and importantly, and I keep emphasizing this because I think it's gotten lost a little bit in the rhetoric, but importantly, creating opportunities for our agents and brokers to take advantage of it with their customers as well.
Yeah, Jay, we love the question, and we are very enthusiastic about it. I think we're a little bit hesitant to give any further insight into the way we're thinking about it strategically beyond what we've said.
Certainly shows you guys are being first movers in this, so it is pretty exciting.
Thank you.
Your final question today comes from the line of Michael Zaremski with Credit Suisse. Please go ahead.
Hey, thanks for taking my question. I wanted to focus on Personal Auto for a minute. It feels like the year-to-date accident year combined ratio is kind of back where you want it to be. I don't know if you agree with that. I know you mentioned frequency was better than you expected. Maybe you could provide some color on whether it's kind of maybe back to its long-term negative trend and you're maybe being conservative. Along the same lines, I don't get the sense you guys are going to start growing that line, but maybe I'm wrong. You've been growing Homeowners at a nice clip, but Auto has kind of been in maintenance mode.
Sure, Mike. Thanks for the question. I think as we've been talking about the Auto strategy at the top level has shifted from reduce growth and manage profitability to balance growth and profitability. We have shifted a bit in our mindset there. We are very pleased with the results year-to-date. As I indicated, expect we'll be within our target combined ratio range for the year, which is ahead of schedule. Most of that is driven by the pricing and underwriting actions that we've taken. Some of it is driven by better-than-expected frequency that we've seen. I think for all the reasons that Alan just outlined in the Commercial Auto discussion, sort of we're cautious about our outlook on frequency, but one of the drivers of the improvement has been frequency's been a bit better than we expected for a little while now.
We're cautiously factoring that experience into our outlook. I would say cautiously optimistic. Again, you can see our outlook for pricing and for margins in the line in the outlook section as we spell it out there for you. Really pleased with where we are with Auto and looking forward to taking it forward from here.
Okay, great. One last follow-up on the fixed investment income guidance for 2019 and the outlook. It seems like it's about a 5% increase in the run rate. I know there's a lot of moving parts here, but premiums are growing mid-single digits. Interest rates have moved higher. Just maybe there's some moving parts I'm not appreciating. It seems like it's conservative.
I think it's Dan, I think if you look at the run rate that we're on and the fact that the interest rate environment should continue to improve modestly and we're modestly growing the level of premiums, I think that trajectory makes a pretty good amount of sense. The fixed income portfolio is a fairly predictable number. We've been giving you that type of guidance for the last several quarters, I think we've been coming in pretty close to that guidance.
Yeah, in the outlook section, I think we give you the driving factors, and we can take it offline.
Okay. Thank you very much.
Right. Happy to help you with that. Thank you.
We have no further questions at this time. I would now like to turn the call back to the presenters for closing remarks.
Thank you very much for joining us this morning, and investor relations is available if you have any follow-ups.
Thank you to everyone for attending today. Ladies and gentlemen, this will conclude today's call, and you may now disconnect.