Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Thomas Saliba on the Price Writer Podcast Ep 21
9 Plays23 days ago
Tom's Journey and the Future of Insurance Pricing Join us in this episode as Jeremy Keating interviews Thomas Saliba delving into his career journey and expertise in insurance pricing. Tom recounts his early fascination with maths, leading him to a successful career at Norwich Union and eventually becoming an independent contractor. He discusses his work with Lloyd's syndicates and MGAs, emphasizing the integration of data science in pricing. Explore debates on off-the-shelf vs. custom-built pricing software and the influence of AI on coding and pricing practices. Ideal for pricing professionals, this episode offers invaluable insights into the complexities, innovations, and the evolving future of insurance pricing.
Recommended
Transcript
Tom's Path to Actuarial Career
00:00:00
Speaker
Hello Tom, welcome to the show. Cool, hi Jeremy, good to see you. Excellent, it's great to see you as well. We've not caught up for quite a long time. So the first question I want to ask you is, how did you get where you are today? Unlike a lot of people, it wasn't by accident. I actually had an idea that I wanted to go into insurance when I was in sixth form college. So yeah, I was enjoying maths at sixth form.
00:00:22
Speaker
Wanted to do a career that was maths related. My career's guidance counsellor happened to know what an actuary was. So she suggested the actuarial profession. So I had not an eye for doing that.
00:00:32
Speaker
Went to uni, did a maths degree, focused on stats as my specialist area. Found I enjoyed and understood the stats a bit more than the number theory and the complex numbers and things like that.
00:00:44
Speaker
um And then really started thinking about insurance and actuarial when during my advanced stats course we were doing regression modelling, multiple regression and then generalised linear modelling.
00:00:56
Speaker
And I asked the question of, could you switch the explanatory variables for like questions in a car insurance quote? And the lecturer kind of smiled and said, you might want to consider a career in like non-life insurance.
00:01:11
Speaker
Unfortunately, grew up in Norfolk. Ended up starting my career in insurance at the Norwich Union.
Transition to Commercial Lines
00:01:18
Speaker
First job in insurance was lots of yeah lots of data, so it was personal lines, data-driven stuff. Then moved into something a bit more actuarial, so still tech still technical pricing, still technical modeling.
00:01:31
Speaker
I've been doing some more commercial line stuff where the risks are a bit less homogenous and there's a bit more judgement that needs to be needs to be laid on top of what your model says. found that quite interesting.
00:01:43
Speaker
I like the contrast between you churn out the model and it goes into the process and someone else takes that and then they and then they do something with it. I've started to work on some niche lines of business as well. so where you would be the actuarial person.
00:01:58
Speaker
So even if you're in a big team, if you're working on lines of business like yeah like your creditor business, possibly yeah your travel business, things like that, um you probably won't be able to justify having a whole pricing team on it unless you are unless you're a large insurer. And then, yeah, in future jobs, did a bit more risk modeling.
00:02:17
Speaker
So you're you using things like Emblem, lots of programming in SaaS. Then after that, decided... No, I want to start moving down that.
Consulting and Predictive Modeling
00:02:27
Speaker
let's be yeah Let's be a specialist. Let's move into an area where pricing isn't a solved problem, where we don't have, here's our data.
00:02:34
Speaker
Here's the data we need. It's all processed. These people will build the models. These people will come up with the market these people come up the market price. Let's go somewhere where it's not a solved problem, or where there's a lot more collaboration needed.
00:02:47
Speaker
yeah, that's what took me into now that's what took me into um commercial lines and your more complex commercial lines where you do some modeling, but an algorithm does not push out the price.
00:02:57
Speaker
There'll be an underwriter in the middle who's taking what the model says and adding in yeah and adding in their judgment or even just using it as a guide for them to come up with their price. So that's where that's what brought me to moving into kind moving into contracting and consulting.
00:03:13
Speaker
So now starting to work in areas like the Lloyd's market working for MGAs who may work in lines of business that I've looked at before, but they don't have as much data. So that's the problem of build a GLM for every payroll that has 15 factors in it was never was never going to work.
00:03:31
Speaker
Yeah, so that's what brought me into yeah to what yeah to what I'm doing now. So tend to pick up a lot of work yeah so pick have a lot of work as an independent contractor for the likes of Lloyd syndicates, ah ng llod syndicate MGAs, doing a mix of and predictive modeling and ah data work.
00:03:48
Speaker
worked on a couple of projects where I'm doing the first round of predictive modeling that a company has done. So there'll inevitably be some, you need to make the data look the way you look the way you want it to. So Tom, what do you see then as your mission now for insurance pricing?
00:04:06
Speaker
And so deliver from my clients. When you're a consultant or your contractor, Your main mission is deliver for your clients. So for me, that means to delivering what they need what they need or what they've asked for.
00:04:18
Speaker
So very rarely will a client ask me four ask me to go away and investigate their data and come up with all the interesting things that I find interesting and go and do a whole bunch of stuff that interests me.
00:04:30
Speaker
they'll be wanting to improve their pricing models, or they'll be wanting to update their pricing tool. So drawing that line. So my mission is to draw the line between doing the analysis that is useful for my client and knowing where to draw the line and say, I've now found something that can add value.
00:04:47
Speaker
So now's the time to move out of the theoretical mode, move out of the analysis mode, and now it's now let's actually get it into the business and add and add some value.
Diverse Roles in Actuarial and Insurtech
00:04:57
Speaker
Yes, deployment is what really matters, getting stuff out into the world away from just being theoretical into reality.
00:05:05
Speaker
So what kind of work is it you're doing and who for? So it's a good mix of... and Your traditional actuarial pricing data analysis. So testing out new rating factors, validating existing models. So does your existing rating structure match up to what your emerging claims are telling you?
00:05:25
Speaker
So both at a high level, like individual rating factors. um So there's a fair bit of that. I'll do a bit of strategy stuff as well. So I tend to work in the specialty market, but I'll pick up the odd bit of... High powered, high speed, yeah aggregator type business.
00:05:40
Speaker
Building data pipelines, working with clients to get their data doing what they doing what they wanted to do. and And quite a bit of than technology.
00:05:52
Speaker
So the building bespoke building bespoke pricing tools or advising on which off the shelf technology they should go for. I find
Future of Software in Pricing
00:06:02
Speaker
that really it's a really interesting really interesting situation that some insurtechs will find themselves in.
00:06:07
Speaker
So the ones of work the people I've worked with tend to have been built with a core team of data scientists who've been yeah building the front end, building the tool, focusing on user experience, and then getting the data done.
00:06:19
Speaker
And then when it comes time to build a robust pricing tool, to get a robust pricing tool, there'll be this idea between, did we buy it off the shelf? yeah Or these data scientists are telling us, no, it's okay.
00:06:32
Speaker
Yeah, it's a bit of, we can do we can do it. Advising them on, what the right, yeah, advising them on. what the right way to do it is. and And also a bit of um operating model stuff as well.
00:06:45
Speaker
So again, if you're an insurtech, you have this team of data scientists, you know you have some data that you need to analyze, you know that pricing has got something to do with analyzing your data. So is that what the pricing team does?
00:06:56
Speaker
or did the data scientists analyze it? Like what are the data scientists, what should they be doing and what should the pricing team be doing? Like, do they just deploy the stuff that the data scientists are building? Like know that what's that line?
00:07:08
Speaker
I found that really interesting as well. And also acting as a subject matter expert for an insurtech who's moving into ah moving out of their niche and starting to go more mass market.
00:07:19
Speaker
Yeah. um So explaining, so being the SME to say, if you want to trade in this space, this is how this is how you should be doing it. This is the data you need. These are the models you should build.
00:07:29
Speaker
Yeah. I think it's, I mean, you know, I always prefer off the shelf because ultimately someone who's an x expert has built all of this stuff for you. And it's relatively inexpensive um if you're a startup.
00:07:44
Speaker
And unless your literal reason for existing is something to do with pricing innovation, you're going startup. You should buy it off the shelf because building is just a distraction away from whatever the startup actually needs to be doing and learning about and and finding a market in.
00:08:02
Speaker
So, Tom, what do you see as your vision for the future of insurance pricing? The thing that's got me most really interested at the moment is... almost the explosion in options for pricing software.
00:08:17
Speaker
yeah So um yeah, I can see the benefits off off the shelf. I can see the benefits of building certain parts in-house as well. um There are so many options now four and for buying but proprietary software built by a third party.
00:08:34
Speaker
um So the argument of saying The off-the-shelf software, it's amazing, but it's far too expensive. Maybe we'll consider it when we're you know when we're writing enough business and and we'll just code it ourselves for now.
00:08:49
Speaker
There are so many options and so many price points. yeah and So I can see, yeah, so my vision would be consolidation, reset resettling in that in that marketplace. So you've got youtube
Specialization and AI in Pricing Teams
00:09:02
Speaker
you've got your two big incumbents.
00:09:04
Speaker
When I started out, it was one big incumbent. So yeah you have your two big incumbents. You have so many other and competitors now at much lower price points.
00:09:15
Speaker
yeah um I can see there being that there being room for at least one or two at least one or two more in the digital personal line space, particularly in in your niche areas. I think it's one thing. If you're one of the big you know hundreds of millions, billion billions in premiums,
00:09:34
Speaker
you can justify spending as much as you need to on pricing software because you need to be able to switch things really quickly. and But if you're doing things like yeah high yeah high net worth, travel, you're writing a particular niche within motor, something like that,
00:09:51
Speaker
and you've now got much cheaper, you've got now much cheaper options. You don't have to say, we can't afford the big stuff. We're to have to code it ourselves for now. So I can see, yeah, some consolidation there.
00:10:04
Speaker
In the specialty market, the consolidation of specialty pricing software is going on. That's where that's really interesting. I think everyone has their own horse that they that they're backing in that area.
00:10:15
Speaker
So I can see off-the-shelf pricing software becoming the norm in that part in that part of the insurance market as well. and The other thing, vision for the future, is pricing teams doing what pricing teams do best.
00:10:31
Speaker
So I was thinking about what I've done as a pricing analyst or as a pricing actuary throughout my career, the different things I've been asked for. And I remember a particular time early on in my career when an underwriter, Commercial Lines underwriter, came over to the actuarial desk and said, we need an actuary's help.
00:10:49
Speaker
And the head of the team said, Tom can help you. And I was very excited. was like, what am i doing? Am I pricing a new product? Yeah. on Part of the book that needs analyzing. And I was, went over to their screen. They had an access database open and they said, we need to filter out all the policies with this postcode. And my heart sank little bit. I was still happy. I thought at first I thought, well, each good, I can solve their problem.
00:11:12
Speaker
So yeah they're going to be impressed with the answer when I do this in about five seconds. So that's great. and But then I thought, there's probably someone else who can who can like who who can do but who can do this. So and I've seen over my career, pricing analysts, pricing actually be able to focus on what they do best as other and other parts of the business have started to take over the deployment of the rates or the preparation of the rates.
00:11:40
Speaker
Yeah, so in terms of that, the vision would be the pricing skill set. I've seen it kind of in a state of flux as data science and data science AI become more useful. So the questions in the business are who should be doing the data science?
00:11:54
Speaker
So um there's a tendency to say, well, The pricing team have been doing data for a long time, so the pricing team can do it. And I've seen some so i'm really experienced pricing actuaries who are used to work with. They're now principal data scientists. and our data They're now directors of data scientists. yeah They're directors of data science because the skill set was so similar. The problems we were solving have been so similar to the problems that data that data scientists are solving.
00:12:22
Speaker
um So think I've seen a lot more of pricing teams bringing in that data science knowledge, so bringing in those tools and techniques into their work. I think there may be some settling down once businesses have decided this is how we're going to embed these skills
Merging Pricing Practices
00:12:37
Speaker
within the business.
00:12:38
Speaker
So if they're going to have a central analytics team that everyone will and interact with on a client um client consultant basis, then the pricing teams will know, great, that's what the business are going to provide.
00:12:50
Speaker
Where's the gap? What do we need particularly to keep within the front to keep within the pricing team? And then we'll start then businesses will start to understand, well, who needs to know how to build an LLM? like That's not the pricing team.
00:13:01
Speaker
and like That will be like out out that will be a central team or that will be ah that will be a third party. yeah me Who needs to be able to process tabular data? Well, it would be really hard if that was in the pricing team, because that's all they work with. So yeah um that's the most efficient. yeah Who needs to be doing really complex like and IT, t yeah ETL processes straight from core system?
00:13:23
Speaker
Probably not. We'll have someone or have someone else do that. Slightly different skill set being used. I can see more collaboration. So if we're doing less of it ourselves, then there will be a need to collaborate with the people who are now taking over, who are now taking over that part of the process.
00:13:41
Speaker
as It's really interesting you're saying about the software options because there has been an explosion. And rather than just having one big one that's not really very good for anyone, you've got really good options for different types of providers. So if you're a small one in a niche market, you can get software that it doesn't have all the bells and whistles, but it takes you through the process. so You can run it with one or two people.
00:14:08
Speaker
Or if you're a big provider with many multiple lines, you can build out a big bespoke thing in someone else's software. So you've got different options for where you are in the market. And it's not just one linear line. There's a good tapestry of options.
00:14:21
Speaker
I think that there's a lot of options. I think it means... there's a challenge for pricing experts to keep up to date with the options that are available. Yeah, definitely. Yeah, definitely. So for the for the big pieces of software that have been out there for a long time, I can call on tips and tricks I learned in 2008 that would still be useful now in how to use it. Yeah, that's fair. Yeah. So I could trade on my knowledge of that software but along for a long time, whereas now,
00:14:51
Speaker
Or in the future, I may go to work for a client and it could be any one of four or five different different pieces of software that they're using. So I've been making some time to keep up to date with all the new offerings that are out there and so that I understand what skills are needed to use them and so that I can advise clients on this might be the right thing for you.
00:15:11
Speaker
Well, that's brilliant, Tom. Is there anything you would like to add? If I was going to give some advice to people who are starting who are starting out in and insurance and in insurance pricing and domain knowledge goes a long way.
00:15:25
Speaker
Yeah. of particular Particularly in the specialty market where the data will never give you all the answer. Yeah. Having some domain knowledge and respecting the domain knowledge that your underwriters will have.
00:15:36
Speaker
Yeah. Really, and really important. um I'm getting a lot of value from being able to code. So I've been coding like a lot of actuaries for the last 20 years or so, various programming languages.
00:15:48
Speaker
even when I'm using AI regularly to write code for me, being able to tell the AI what I want it to do yeah and being able to review it and tweak it when it doesn't quite do what you want.
00:16:01
Speaker
Yeah, absolutely. I find a lot of the time Coding with AI, it's a bit like playing golf and you only have a driver. So it will get you on the green pretty quickly. It will get you. Yeah. Yeah.
00:16:14
Speaker
But then you're on the, you're on the green and you just have a driver. You can't, it takes you ages to actually finish the last bit. So there may be a temptation to say, well, I don't need to learn to code. because the AI will code for me, but being able to tell it what you want it to do and understanding the understanding programming paradigms well enough, I found is still is is still serving me well. That's probably everything other than when I started out, I saw pricing as being two separate disciplines. I saw kind of stuff that i started out in, personal lines pricing.
00:16:49
Speaker
yeah Lots of data, as we said, lots of data, sold almost a solved problem of you build your process, you build your models, you have lots of data. And then you had this thing over here, specialty pricing.
00:17:01
Speaker
Yeah. Didn't really do any rating factor analysis. And you were effectively doing statistical analysis, constructing these complex insurance arrangements. So both those two are still valid, but I'm starting to see them merge a lot more.
00:17:16
Speaker
Yeah, definitely. So that last point on bit on Vision for the Future, I don't see those two being separate anymore.
Adapting to Changes in Insurance Pricing
00:17:23
Speaker
Yeah, we'll see a bit more individualization. It's interesting as well, because in some ways, I think the personal line side has gone too too like manufacturing mindset to the price and not enough analysis. And I think that tripped people up with the inflation issue. They had such rigid processes to keep deploying the prices.
00:17:44
Speaker
And they'd lost sight of that exposure management side, the after doing the speciality and that feedback loop as well. Yeah, so that you build a really clever model, but if you don't have the exposure management to tell you that you've ended up writing every single Ford Focus in Birmingham or something like that, yeah but exposure management, that is the bread and butter of the specialty market. Yeah, exactly.
00:18:09
Speaker
Brilliant, Tom. It's fantastic to catch up. Really good talking to you. You have a good rest of your day. Same to you, Jeremy. Take care.