Episode 88: Climate Scenario Analysis

My guest for Episode 88 is Lloyd Richards, FIA, CERA, CStat, Director and Head of Actuarial at Crowe UK and Teaching Fellow at the Queen Mary University of London School of Mathematical Sciences.

The theme for the episode is 𝗖𝗹𝗶𝗺𝗮𝘁𝗲 𝗦𝗰𝗲𝗻𝗮𝗿𝗶𝗼 𝗔𝗻𝗮𝗹𝘆𝘀𝗶𝘀.

Lloyd and I explore the following topics:⁣

✅ Sustainability and its relevance for risk management
✅ The current sustainability landscape 
✅ Bank of England’s CBES directive 
✅ Physical, transition, and legal risks
✅ Climate scenario analysis and stress testing
✅ How actuaries get involved with climate scenario analysis
✅ Addressing key gaps in climate scenario analysis
✅ Data visualization and its connection to climate scenario analysis

If you are interested in learning about how actuaries are getting involved in Climate and ESG initiatives, you want to watch this.

Thank you for being on the show, Lloyd.

If you are interested in learning about how actuaries are getting involved in Climate and ESG initiatives, you want to listen to this.

My Website: maverickactuary.com

Episode Transcript

(00:00):
Welcome to episode eighty eight of Livewith a Maverick. My name is Dominic
Lee, founder of Maverick Actuary.We are content community. Our mission is
to maximize the impact and value ofQUN professionals on a global scale. The
goal of this series is to educateour community on the most relevant themes in

(00:21):
actuarial science, risk management and analytics. The theme of today's discussion is climate
scenario analysis and we are very excitedto have with us our guests for today's
episode, Lloyd Richards. Lloyd isDirector and Head of Actuarial at Crow UK,
a consulting, auditing and technology firmwith offices across the globe. He

(00:45):
is also a teaching fellow at QueenMary University of London in the School of
Mathematical Sciences. Welcome Lloyd, Thankyou very much for having me. Nice
to be here, excellent. I'dlove to give you an opportunity to introduce
yourself. Yeah, absolutely so,As you said, I mean, I
think you've done a better job thanI would have done. I'm Managing director

(01:06):
at Crow and I head up ouractual services in the UK, and really
I've spent my entire career in consultingCrows my third employer, and I've touched
on lots of different areas over thattime, pensions, life, GI risk,
and now I largely work across climateand sustainability, and it's quite an
interesting exciting time to be there becausewe're seeing lots of actories getting more and

(01:30):
more involved in in climate, insustainability and in those areas you also mentioned.
I'm a teaching fellow in the Schoolof Methemical Sciences at QMUL. We
probably have a lot of my studentslistening to this episode because I know they're
amongst your top listeners. So helloto all of my first years and previous
year students who are listening today.I hope this is interesting, as my

(01:51):
lectures probably weren't. So a littleabout myself. That's not my job.
I'm very involved in the actual professionin numerous ways, usually around education,
so I've been a career ambassador sortof since my early days. A couple
of years after joining the profession.I've also represented a number of working parties.
Currently I co share the Climate Reportingand Disclosure Working Party, which is

(02:15):
very relevant for today. We've recentlybeen published in the British Actual Journal,
which was a great milestone for usas a working party, and we're potentially
going to be at some conferences laterthis year. I'm also a liveryman with
the Worship Company of Actories, whichis one of the City of London's livery
companies involved in the governance and runningof the city and also has charitable and

(02:35):
community aims as well. And reallyI like to bring the education of mathemetics
into a lot of what I do, so you'll often see me on LinkedIn
trying to explain mathematical concepts in adifferent or interesting way, or using the
skills I've developed as an actory andan unusual context. Finally, a little
bit about my life outside of work. I'm a keen trail runner. I
find it's a really interesting way tosee different parts of the world by running

(02:59):
through the countryside as fast as youcan in different places, so I try
to find the times do the occasionalUltra marathon. Later this year I'm going
to be running my first ever backyardUltra, which I'm very excited for.
Oh that's exciting Ultra Wow. Yeah. I've done one marathon and I've done
some distance in the past, butonly simple you know, more shorter distances

(03:22):
for the most part, so youyeah, so showed out first of all
the Queen Mary University of London students, you know, I love thank you
so much for all the support.And one a question I want to askwer
what courses do you teach there?Of course or courses? Well, I
teach actual science. I teach thefirst year module called actual Professional Development,
which is essentially what actories do inthe real world. And I try to

(03:43):
bring sort of as many different guestlecture as I can to talk to my
students because I found I was atuniversity I wanted to be an actory.
I didn't speak to an actory untilI was going to interviews with companies.
All of my students have spoken toprobably at least ten actories in their first
year and hopefully get another ten intheir second year as well, so they
really understand what the profession involves.And I've heard of lots of different areas.

(04:05):
They know it's not just pensions,not just insurance. They speak to
risk actories, health factories. Iknow you had Bart's on a recent episode.
Yeah yeah, one of Bart's teamwas talking about football analytics to my
students a few months back. SoI try and give them as broad a
set of lessons as I can.Yeah. Yeah, that's one thing I

(04:26):
noticed with schools no Iday, notjust in the UK, in the US
as well. There's more that's atremendous growth in the actual profession in terms
of bringing the practical side into theclassroom, which I think is great.
I think when I was there,we definitely had a practical court. Well,
we actually had a couple of peoplecome in, but it was still
in this infancy at the time,but it seems to be much more mainstream
now. So so you know,you mentioned that you're alright, did I

(04:49):
mentioned you mentioned your crow Crow UK, your director, managing director, direct
you know, director head of backto Ourial. I'd love to explore that
more, like I'd love to learnmore about the company and you're a current
role there. Yeah. Sure,it's an absolutely fascinating business to work for.
That There'll be a lot of actorieseven here in the UK who've never
heard of us, or may knowus only as pensions auditors and mostly it

(05:11):
is an audit practice. But Iwork for a really small part of the
business, that's the consulting practice.It's about twenty five people now and we
offer services to our clients across buildingresilience, delivering transformation, mastering data,
integrating sustainability which is where our climatework comes in, and risk and governance.
And it's because it's so small.It's a lot like working for a

(05:31):
boutique consultancy, but then we're partof that broader audit firm, which means
we have really well established support teams, you know, IT and HR and
marketing that you don't often get insmaller consultancies. The company that I joined
eight years ago that eventually got boughtby Crow called Baxter Bruce. Fantastic place
to work. Still a lot ofthe old Backster Bruce people are still here

(05:51):
at Crow, but we didn't havean IT team, we didn't have a
HR team, we didn't have amarketing team. You had to do all
that yourself whilst also being consult foryour clients. So it's very nice to
have those support functions. I've alwaysreally enjoyed how FRO works as an organization.
So I've worked in a variety ofdifferent roles throughout my career, and

(06:12):
I think lots of actuarial roles thatyou can do a jobs where the client
has to spend money on you asan actuary. They have to get a
nactury to do pensions valuations or transfervalues. They have to get a nacturary
to sign off on reserves. Theprojects that I do now are far more
discretionary and value add nobody has tospend money on me. They choose to
spend money on me, and theyrecognize that the value and the skill set

(06:33):
that an actually can bring, andthat's what creates a really different relationship from
the outset when a client is actuallychoosing to spend their limited budget on you
and recognize as your expertise. AndI think a large part of why clients
use us is they recognize we're reallygood at tailoring solutions to their needs.
So we don't come with pre existingnotions of what the solution is going to

(06:54):
be. We actually try and sortof learn what our client is facing,
learn what makes them unique, andtry and upscal all our clients and leave
them empowered so ideally they no longerneed us. That sounds great, and
it sounds like a relationship, veryrelationship driven, very niche, very interesting
work of broad range of projects.That sounds very neat. Now I'm you

(07:15):
know the I didn't meant. Sotoday we're going to talk about climate scenario
analysis. But and the episode,of course is related is very close to
the climate. But before we getinto the details of that, I want
to take a step back to understandthe big picture. So let's talk a
bit about sustainability. So when wehear the term sustainability, what does that
mean or how is that relevant inwithin the context of risk management. Yeah,

(07:39):
I think that's a really good question. It's actually a really smart place
to start, because what you findwhen talking about sustainability is there's no commonly
agreed definition between organizations, and evensometimes within a single organization, people mean
different things when they say the wordsustainability. So a lot of organizations equate
climate change sustainability and the only sustainabilitymatters. They worry about the climate and

(08:01):
global warming. A lot go abit further and have started thinking about nature
and biodiversity in water usage. Somethen go even further and think about the
financial sustainability of their own organization orbroader still sustainability of the economy as a
whole. And then when you startto think about those things, you'll often
capture other diversity metrics. So howcan you be a sustainable business if you
aren't fostering a diverse workforce and youstart to bring in the DEI areas that

(08:24):
get captured. I think as youstart to bring all those together, you
start to get at how the moremature organizations define sustainability, which is how
you bring all these various strands ofenvironmental impacts together with a clear social purpose
that your organization has, and howyou try and make decisions that foster a
culture that lives up to all thoseelements. So it sounds a little bit

(08:45):
even though I know within a termEESG, I think that S is social.
I think it's correct. So eventhough the S is social, I
think it sounds like it kind ofties to the S in the ESG.
Would that be a fair assessment.Yeah. Absolutely, They're all part of
the whole, and I think companiesthat do the best job of this are
recognizing the interconnectedness and drawing things togetherunder a single strategy. Okay, great,
Now, as you sit in theUK, you're very close to this

(09:09):
work. You word, you doa lot of projects regarding climate and you
sit in the UK, so yousee what's happening locally and you look across
the global how would you describe thecurrent sustainability landscape. Yeah, that's also
a really smart place to start,because there's a joke going around at the
moment that if you take any fourletters and mash them together, you'll find
they represent the sustainability frame. SoI'm going to list some and we'll see

(09:31):
if listeners recognize any of these.Tcfd T nfd C s R d E
s R s s f d Rtpt pecaf c d p I s O
g r I n g f sn zami g fans p r I unt
u n g r c SDGs.I could go on, isn't there I
S SB two SB S one thistwo? Yeah, thank you on missing

(09:52):
as absolutely loads. Pretty much anyletters you bring together will be one.
So I think it's it's really challengingto keep up if you've got a busy
job, even if you're very focusedon climate reporting. How can you keep
up with which of these apply toyou and what do you have to do
under each of them? And Iactually have a huge fear at the moment
that there's a lot of very dedicatedpeople who really care about climate sustainability and

(10:16):
they're working in sustainability functions or organizationsthat can have a real impact, and
they're just stuck on this reporting treadmill. They're not getting enough time to actually
enact real change, which are thingsthat they want to do, because they're
so busy reporting what their organization's doing, and that's taking ten eleven months of
the year, leaving one month toactually do stuff. And it's a really
difficult typewrote because reporting is really important. It forces companies to make commitments,

(10:41):
and it allows people to hold thosecompanies to account. And there are some
voluntary initiatives out there that I thinkare really valuable that try and sort of
draw a pathway together through all ofthese initiatives and try and make people understand
how to understand the interconnectedness between allthese initiatives and how you can give one
answer to many different sets. ButI think we need to help help organizations

(11:03):
find a way to focus on thebits of reporting that matter to them,
so ideally in a principles based wayrather than ticking boxes, which is the
bit that takes a long time andleads people very disheartened. And in the
UK, a lot of what you'reseeing really in terms of climate reporting and
some of the detailed stuff on scenariomodeling which we're going to talk about,
dates back to the Sebez exercise,which is yet another acronym. That's the

(11:26):
Climate Biennial Exploratory Scenario. That wasan exercise conducted by the Bank of England
about three years ago now, andit used the NNGFS scenarios and ask organizations
to essentially evaluate their investments on aline by line basis under those scenarios.
So, if you think the worldis going to keep warming to within one
point five degrees, what does thatmean to your investment? If you think
we're going to have a four degreeworld, what does that mean for eature

(11:48):
investments? And I think it's areal indication of maturity how different organizations approach
that exercise. So some organizations tookthe exam question, which was runney scenarios,
tell us the result, They didthe maths, they reported the Bank
of England, they congratulated themselves ona job completed and left it to one
side. And I don't think thatwas what the Bank of England wanted.

(12:11):
The better organizations there is more matureand understanding sort of realize that what the
bank was really asking it's fundamentally theydon't care what the numbers are. What
they were doing is saying, weneed you to start investing in scenario analysis.
We need you to start building outthese models, start triggering out what
the challenge is, start seeing whereyou have data gaps, and start using

(12:33):
that investment because it's going to bea really important tool to help you make
decisions going forwards. And I feelthe organizations that I've worked with that use
the sea bears as almost as aleap frog to start investing in scenario analysis,
they're now at a real advantage andyou can see the real step up
in terms of quality, in termsof what they're putting out greed. So

(12:54):
actually I follow up on that.You talked about the sea best exercise and
I think it came from the Bankof England. Now, when we talk
about climate, different people have beentalking about this topic for different lens of
time. The earliest I kind ofremember this is before your student's time.
This is when I was fairly youngas well. Is I remember I think
al Gore had this film. Iforget the name of the film, but

(13:16):
somebody it was on climate. Soyou know people that people there have been
people at the forefront of that movement. I think Bill Gates might have been
talking about it as well. Hehad a really good Ted talk on it
as well. But what was acatalyst or was there a catalyst for the
Bank of England's directive our own SeaBest? Is something specific happened or was
it just more kind of a snowballeffect. It's an inconvenient truth. I

(13:37):
think it's the film that you're you'rereferencing that it's quite quite interesting. I
think in terms of where things areat in the UK, it's almost that
the time was right. Globally nowthese things are being pushed in different parts
of the globy bit in different ways, but generally there's a consensus that you
know, we get it now,we get that climate change really is happening,

(13:58):
really is having huge impacts. There'sno more debating whether it's a thing
or not. It's now about debatinghow bad is it going to be and
how do we mitigate the risk.So I think a lot of the cop
events that have happened to have ledinto that and have pushed some real action
there, even though they've got abit of a slating in the media for
not really enacting change. Actually,some of the discussions that happen behind the
scenes there have led to some ofthe regulator initiatives that are now coming out.

(14:24):
Okay, agreed, Now, climatescenario analysis, we're going to get
the promise. We're going to getto some of the details on that.
But one more thing I want tomention first. So climate a climate scenario
analysis is used to measure climate relatedfinancial risks. So what are the key
categories of risks being measured in thiscapacity. Yeah, so climate risk isn't

(14:46):
just one risk. I think ifyou read the media and look at stuff
about climate change, the risk thatthey focus on is physical risk. And
that's kind of the obvious one.That's quite simply as the climate change is
extreme weather events increase in both frequencyand severity. That increases the likelihood of
insurance claims. If you're looking atyour underwriting portfolio, asset values declining,
if you're talking about your investment portfolio. Potentially also operational risk impacts. You

(15:09):
know, if a climate event knocksout an office or a data center and
you can't use that office or datacenter for a day, then that's a
real physical impact of climate change onyour business, causing an impact. But
probably the bigger risk, and theone that you won't see from the media
is transition risk. And that's themacroeconomic risk that represents the really real challenges

(15:30):
of transitioning to a low carbon economyand everything that means for organizations throughout the
supply chain. So as we reduceour reliance on carbon, what does that
mean in terms of more expensive transportof goods, more expensive sourcing of power
and different types of power, potentiallyintroducing carbon budgets and being charged for your
emissions. But also then there's theopportunity side of things, so increase investment

(15:52):
in green energy and solutions that areneeded to reduce those emissions and what those
can mean as opportunities. And thenfinally we can think about litigation risk,
which is sort of the third pillarof climate risk. It's less relevant to
scenario modeling, but it's important tothink about nonetheless, and that's kind of
the risk of climate related lawsuits andthe associated costs. And generally I think
the media has you thinking about,you know, class action lawsuits by indigenous

(16:14):
people against shell as a classic exampleof mitigation risk, but it can happen
on a much smaller scale. Yousee a lot of climate protesters around the
city of London, sometimes targeting banks, investment managers, even insurance companies.
Now they've realized that the underwriting ofcoal plants is an essential part of that
financing structure, and then actually targetingthem is maybe a way to put pressure

(16:34):
on them. So I think nobodyin the financial services industry is immune from
potential lawsuits in the future. Soas we think of those risks of physical
risks of transition or is to legalrisk, you know, is there any
like as you think of it froma financial perspective, how it impacts the
balance sheet for companies? Is there, I don't know, a hierarchy,

(16:56):
a prior todation, or is itjust kind of like a cross the board
that everyone is looking at them intandem. Yeah, it's so different for
different organizations. In financial services,really you tend to not have a massive
physical presence in the world, butthen your investments do. If you're investing
in coal, oil, gas andstuff that has physical presences in the world

(17:17):
and long supply chains, then that'swhere the physical risk comes in. But
if you're thinking just about pure investmentsin retail whatever else, then that's that's
much more of a transition focus thing, so different companies are looking at different
things. There's one insurer that I'veworked in the past which has a very
large real estate book in the UK, and they're doing loads of flood risk
modeling. They don't care about sortof more extreme weather events that you'll be

(17:40):
getting like hurricanes and tidal waves andall that stuff, because that's not going
to hit the UK, but we'regoing to have increased flooding, so that's
what they're modeling over here. No, I would imagine that well, like
you said, depending on the natureof the organization, there could be some
resistance, so especially in the energysector if you're talking about none. But

(18:00):
one of the things I think isinteresting and hopefully I'm not going too far
off scope pair is when it comesto energy, I think diversification is important.
I think sometimes a conversation is simplifiedto the point where we just go
from non renewables to one hundred percentrenewables, where I think the truth is
somewhere in between. So I guessmaybe we're in that spectrum. I know
that like carbon caption and cup sorry, carbon capture and stories, there are

(18:23):
certain things you can do mitigative measures. You can do our own non renewables.
You know, like, where doesthat kind of fit into the broader
conversation of Yeah, I mean,technological solutions are an interesting thing. I
mean that they're not a silver bullet. I think you find people are split
that There are those sort of whoare very evangelical about technological solutions, say,
look, we can carry on asmissing as much as we want,

(18:45):
because in twenty years time we'll beable to safe you remove all of the
carbon from the atmosphere and it willbe fine. That's not true. But
there are also people who say youshouldn't allow for any technological solutions in any
of your modeling, any of yourprojections. You should be reducing carbon to
zero. And that's quite a harshview of the world as well. That
doesn't really recognize the real issues thatcompanies have to face and that global economies

(19:07):
have to face as part of thetransition and the answers somewhere in the middle.
You know, you have to youhave to recognize you can't switch off
every coal plant tomorrow and replace inthe solar panters. It takes time to
transition, and the amount of timeit takes depends on what your economy is
like, what your energy mix isalready like. You know, some countries
are already quite far along that transition, are able to provide most of their

(19:27):
power through renewables. Some countries,even in the YEW, are very behind.
If you look at Poland, forinstance, it's heavily reliant on coal
power, even though the rest ofthe year is quite far ahead globally in
terms of renewables. Yeah, Ithink it's a really interesting discussion when you
especially when we talk about transition oris one of the things I had noticed
even taking a step back during thepandemic. I just think from a broader

(19:49):
societal perspectives. You know, riskmanagement is tricky. Let's be real.
You know we're in that space.Then you have to look at many things
happening at the same time. Sowhen I think of someone whose organization,
and first thing that comes to mindis what happens to the workers. You
know, you have these large organizations. You can't just assume that it's just
going to disappear and then there's nodownstream impact. So there's something I find

(20:11):
interesting. So let's get into themeat of it now. No, we
can talk about we set their toneagain. Talk about climate scenario analysis.
So how would you describe climate scenarioanalysis? Yeah, So, because I
know a lot of your student listenerslisteners are students and potentially in my students,
I'm going to start a little bitbasic with how I see the whole

(20:33):
suite of stress and scenario testing tools, and build up from stress testing to
the details of scenario analysis. So, in general, lots of vectorial problems
are solved by building a model.Models take data, make some assumptions about
the future, do some calculations,and give a result. But that's about
it. You have inputs, yourinputs of the data, which you know,
the assumptions, which the bits youdon't know when you're you're filling it

(20:56):
with a guest or an educated guess. Sometimes that's just because you don't have
the data, or sometimes it's becauseit's longer term projections where you won't know
the answer until the future. Butyou make a sensible assumption, you then
do a bunch of maths to thatand you get your answer. It's kind
of easy, right, But modelsare a representation of reality, they aren't
reality itself. So whatever answer youget from your model, no matter how

(21:18):
much time you spend with it,that answer is wrong. And that's a
thing that actually are quite used toexplaining and understanding, is that all of
our models are wrong. They willnever get the right answer. The important
question is how wrong might our modelbe? Well, the thing in any
model that's most likely to be wrongis the assumptions. Everything else can be
validated, tested, checked, Thecalculations can be checked, that data can

(21:40):
be reviewed against other sources. Butthe assumptions, they're the educated guess bit
and the bit that's most likely tobe wrong. So we can see how
wrong my model might be and howwrong my results might be by seeing how
much my answer changes when I makea variation in the assumptions. So the
first, the most basic type ofvariation, you can run as a univariate
stressor that means changing just one ofmy assumptions. So I might have a

(22:03):
pension scheme projection where I think futureinflation is going to be three percent,
and I say, okay, ifit's not three percent, it's actually four
percent. What happens to my results? And I can run the model again
with four percent in ste of threeand see what the impact is. The
next is a multivary stress test,so changing two or more assumptions at once
but otherwise works in exactly the sameway. You change the assumptions Rundom numbers

(22:25):
get the results. Scenario analysis isabout creating more of a narrative. You
can be as simpler as complex withanalysis as you like, but the whole
point is creating an internally consistent assumptionset that follows it the storyline that can
be explained to people. So withclimate scenarios, nobody really has a clue
exactly how things will play out.We're reasonably confident over what many of the

(22:48):
climate impacts will look like over certainemissions pathways. That's been done to death
over years. There's some really cleverpeople working on that, and it's pretty
well understood that there's very there.But we can take that as quite well
understood. But trying to then predictwhat emissions pathways will look like, trying
to predict national and company level behavioris really really hard. There's a US

(23:11):
election in about six months, andemissions pathways for the US and even globally
could wind up being entirely different dependingon who wins that elections. So the
way we solve that in our modelsis by creating lots of different projections for
lots of different possible futures. Sowe create a broad narrative that describes that
possible future, and within each narrative, we describe a logically consistent set of

(23:33):
assumptions. So if I take,for an example, a scenario that's like
my best case scenario in which allcountries in the world take immediate and concerted
action to keep global warming to withinthe one point five degree Paris targets,
then you need to create an assumptionset that aligns to that. So it
has to include a drastic reduction emissions, because there's no other way of getting

(23:55):
there. Even if you believe thattechnology is really really valuable here, you
still need to drastically reducer emissions fromtoday to keep us within one point five
degrees. So then you have tothink about how that translates in the real
world. And the first and mostobvious thing is, Okay, if we're
cutting emissions, where are we goingto get pup from. As we've said,
some countries are really ahead on renewableenergy, some countries are behind,
and the laggards, So you needto come up with an energy mix for

(24:18):
each country that's grounded in the currentreality of what energy mixers currently look like,
but then projective over time to allowfor each country's commitments, whilst also
taking into account growing economies things likeelections that might change things over time and
potentially be tipping points for an individualcountry and to what their energy mix looks
like, and that might get youan idea for what the energy need looks

(24:38):
like and what the energy supply lookslike for each country, or at least
each major economy. So that's oneassumption that we'll go into something like that,
and then you might think, okay, what about electrification of transport.
That's going to be a big partof any transitions with low carbon economy in
an our scenario where emissions are beingdrastically reduced, we're going to need to
have an assumption about, Okay,what proportion of vehicles will be electric,

(24:59):
how efficiently all those vehicles run,how will that improve over time, and
how what will be doing with allthe old carts that we're getting rid of.
And then more broadly, we needto start making assumptions about where the
resources are coming from, so miningintensiveness, deforestation and replanting assumptions, and
you need to try and take allof these to make a view of the
world and be able to project thatforwards with with your macroeconomic assumptions as well,

(25:22):
and try and work out what thatmeans for your organization and the organizations
you invest on a line by linebasis. So it starts to get very,
very complicated, but we can alwaysdraw it back to that core broad
narrative of Okay, this is aone point five degree world and everything that
needs to happen is built up inthose assumptions. That's extremely helpful of the

(25:42):
very thoughtful articulation. I like theway you build it up from stress testing,
and you kind of can increase thecomplexity, so it makes the art
of sense. I think one naturalfollow up I have for that is,
I think of it fundamentally, whatare some of the key inputs in terms
of like what you use in termsof the mothers and the scenarios, and
what are some of the outputs,so whether that be metrics, KPIs,

(26:03):
ranges, reports, and how dothose support the disclosures. So if you
think the inputs outputs or they supportdisclosures, yeah, absolutely, Again,
it depends from company to company.I mean a lot of the first challenges
really just working out what your omissionsare for the company that is doing the
reporting, and it's not that hardto do, but it requires data and

(26:25):
that's not data that company has necessarilybeen set up to capture in the past,
because why would it. It's notsomething that's been required historically, and
generally you don't do things you're notrequired to do unless you're really far ahead
of the curve, so that that'sa challenge. They're just in working out
what your emissions are. Then,in terms of the other inputs to some
of the more broader scenariouts and stuff, I mean, as I've said,
you need to try and work outto future pathways for the world. So

(26:49):
the most common scenarios use of theNNGFS scenarios, the Network for Greening and
Financial Services. They have limitations andI might come on to a few of
those later on in the episode anddiscuss those, but essentially they're the ones
the most commonly used in industry,and they describe three broad pathways to the
transition. One that transitions very effectivelyand quickly, one that puts this off

(27:12):
for a decade and then transitions,and one where nobody in the world does
anything, and we all let globalwarming happen. And each of those comes
with a set of underlying assumptions aboutwhat carbon pricing might look like, what
energy mixes like look like. Sothose are really the key inputs. When
it gets to company level inputs,you need to start looking at the individual

(27:33):
supply chains. What does this companyactually rely on to produce what it produces.
If you're looking at Apple, forinstance, you need to be looking
at lithium as a key sort ofinput into that organization. I tend to
work with financial services where actually you'reone step removed because you're investors in the
companies and it's then have for supplychains rather than the asset manager itself or

(27:53):
the insurance company itself. Great,So that's the inputs in terms of the
outputs, and our work comes upto that, Like what's shared in terms
of you know, the disclosures.Yeah, absolutely, so again still at
an early stage for a lot oforganizations and disclosure and a lot of what
they're talking about is Okay, thisis what our emissions look like, this
is how we're going to reduce ourown emissions. Trying to get their arms

(28:15):
around what emissions in the supply chainlook like, which is a lot harder
to do. And actually, whenmost organizations the majority of their emissions come
from, especially in financial services,a lot of organizations are using this modeling
stuff and going quite a lot furtherin what they can put out. So
you might see some of the oneswho are very exposed to physical risk,
you might see some global heat mapsof sort of where those risks are going

(28:37):
to crop up, where they're mostexposed in terms of both where their biggest
property investments are and parts of theirsupply chain that rely on physical presence,
and also how that relates to increasedrisk in those areas, so potential sort
of loss as a result of physicalclimate risk. And then on the transition
side, typically much much simpler interms of the outputs visualization, because you're

(29:00):
looking at charts of by industry,how much do you think this might lose,
how much do you think this mightgain? What could the potential impacts
be under each scenario? Okay,great, so I think that was like
the broad spectrum of the scenarioanalysis.But now we're in on that a little
bit more in terms of what andthat's going to involve several parties, But
how specifically, do actuaries get involvedwith climate scenario analysis. Yeah, I

(29:23):
think actories are in a really valuableposition for a number of reasons, and
not just to be involved in scenarioanalysis, but to be involved in climate
and sustainability work as a whole.So, firstly, I think actories are
typically very well positioned as highly respecteddecision makers at many of the key points
and pivots around which finance flows,so asset managers and asset owners, pension

(29:45):
schemes, insurance companies, people knowwho actuaries are respect actories decision making and
modeling and ability to understand and projectmodels into the future, and that means
that the decisions that actories make orat least contribute to, can affect quite
a lot of capital. So that'sone reason why actually are quite important to
this discussion. Secondly, I thinkit's also that we're very used to modeling

(30:07):
uncertainty. I've learned throughout my careerthat many professionals have a really hard time
seeing value in something that they knowis wrong. And one thing you can
say about every single actuarial projection evermade is that it's wrong, and wrong
doesn't mean useless, and that's quitea filch, a big philosophical chasm to
leap, but it's just natural tothe way I actually think we do this
all the time. One of myfavorite examples of this is actually a comic.

(30:32):
It was to do with with Brexit, actually, but someone is Essentially
the comic is a guy who's askingthree people whether they should jump off that
cliff over there, and the firstperson says, oh, that cliff looks
to be about ten feet high.You'll probably get hurt if you jump off
it. The second person says,I think it's around fifteen feet. You
could easily break a leg, andthe third person says, I reckon,
it's twenty feet. You're risking seriousinjury if you jump, which the guy

(30:52):
who's asked the question says, well, you guys can't agree on anything.
Is it ten to fifteen or twentyfeet? I'm just going to jump,
And it doesn't matter that they wereall wrong and it could have been twelve
feet or eighteen point sixty five ninetyor whatever it was. The answer was
the same, don't jump. It'sa bad idea. And so the actual
conclusion of the models that we docan be a lot more useful than the
final number and that the exact resultcomes out. So I think that understanding

(31:18):
of uncertainty and projecting things that areuncertain is a really valuable skill that actuallys
have. What I would then dois take a step back and potentially discourage
actories from getting too involved in thedetail because I think, and I say
this with as much modesty as Ican, muster actories are very clever people
and it can be very easy toget into a mindset of being able to

(31:38):
solve every problem you've ever faced.And a real turning point for me where
I realize that actually I'm not theright person to be doing this detailed climate
modeling. I don't have the expertiseis when I was talking at dinner with
someone who has a PhD in modelingflooding and I was asking him some questions
about river flooding and he just lookedat me like I was crazy and said,
I don't do rivers, I docoast flooding. You want a guy

(32:00):
with a different PhD for river flooding. And that made me realize the level
of expertise that people go into inthis industry is there's so much there,
and we need to be able totake advantage of that and help them translate
it to people who make the decisions. But there's no way I'm going to
go and get a PhD in riverflooding myself to fill that gap. I

(32:21):
need to be able to work withthese people, and I think I need
to understand where to take a stepback from that and say, Okay,
I'm bright, but I'm not asbright as you at river flooding, and
I'm going to trust your results here. I was actually going to ask,
but just to purely or to curiosity, is like what kind of modeling?
So it's not like you have veryspecific domain experts doing particular different types of

(32:43):
modeling. So yeah, So thisone in particular was looking at the UK
real estate and this was a bigreal estate book all over the UK and
trying to get an understanding of whatpotential flood risk would look like in the
future. And we're very lucky inthe UK that we have really, really
good flood maps. A government agencywas set up decades ago to sort of

(33:06):
work out what all the floodplanes aremonitorate over the years, see when things
flood, when they don't flood,relate that to rainfall and things. So
we've got a really good data setto build it off, and then these
guys have started projecting that in thefuture for different rainfall amounts, difference of
wind patterns, different levels of soilerosion, and loads of other stuff to
see what it might look like inthe future. And that that's a level

(33:27):
of detail that no actuary is goingto get into. Yeah, floodplanes are
in the US are interesting, andwell leave with that that. Sure.
No, you know, as yousaid, this is an emerging discipline when
we talk about scenario analysis. Sowhat are some of the key gaps to
date in climate scenario analysis? Yeah, so I think it's it's really important

(33:51):
to recognize that no scenario is perfect, no matter how much time you invest
in it, and no scenario everwill be perfect. So there's there's gaps
that are fatal flaws that we shouldworry about, and there's gaps that we
can learn to live with. Weknow they're there, we can describe the
limitations to decision makers and carry onwith our flawed scenario. So it's becoming
i think, generally accepted, orat least increasingly accepted, and this is

(34:12):
largely thanks to a paper that theIFA institu in factory actualies did with the
University of Exeter that was called theEmperor's New Climate Scenarios that took a really
deep look at the NNGFS scenarios andput a lot of work into understanding and
more importantly, articulating the limitations ofthose scenarios. There was also a recent
bulletin from the Bank of England thatwas about measuring climate risk that actually references

(34:36):
that IFoA paper as well, butgoes a bit further into talking about some
of the key limitations we should thinkabout. And I think it's generally the
way I would summarize them are theydon't capture the full speed of risks.
No possible scenario projection can. Butsome models, and particularly the NNGFS Holothouse
World scenarios, are quite implausible inwhat they show. Going back to the

(35:00):
Seabees exercise of three years ago,some of the first scenario modeling I did
showed that the best economic scenario wasfour degrees of global warming. That's just
completely implausible. Economically, that woulddestroy so much of any market. It's
just completely unbelievable, but that's whatthe modeling was showing. So we had

(35:20):
to recognize that is a serious flawand error in the model. I think
calibrating the risks correctly is a keylimitation and issue with the model. We
tend to only really have historically data. Historical data, and historical data can
fatally underestimate nonlinear relationships, so risksmight develop more quickly than we realize and
more quickly than scenarios suggest. Andthen we've talked a little bit about how

(35:44):
difficult it is to factor in newtechnologies, whether you should go all in
and say, well, technology isgoing to be a silver bullet and we
should allow for every kind of newtechnology that's out there, versus allow for
nothing and say, look, there'sno technological answer that we can take into
a car of our scenarios. LikeI said, the answers probably somewhere in
the middle, but your scenario willneed to take a call on to okay,
what do you allow? How muchdo you allow carbon capture and storage?

(36:06):
To eat up that last bit,what do you allow in terms of
offsets? And offsets are getting alittle bad press because I think they are
a fundamental, important part of thesolution, but they're not being used in
the right way right now, andthey're not very robust and not very validated.
So again the truth is always somewherein the middle there. And the
final key limitation of scenarios is climatetipping points. So those are really hard

(36:30):
to factor into scenarios and the sciencereally, at least from my perspective.
I'm sure it's well understood by lotsof people, but in terms of the
actuarial world and the clients that I'veworked with, only just really beginning to
be taken into account. And essentially, these climate tipping points are key of
global issues where if you hit acertain point of global warming, things will

(36:52):
get much much worse. So theexamples they're often given are melting of specific
eyed sheets are the Greenland sheet orthe Western Tarctic sheet. If those melt
after we hit a certain level ofwarming, that will push warming up further
than no matter what we do,and it becomes much much harder to stuff,
and you almost get this snowball effect. I think I prefer snowballing points

(37:12):
myself rather than tipping points. Buttipping points is that generally use term,
and lots of things like that thatare some that are well understood, like
melting sea ice, some that aremuch much harder to predict like prevailing sea
currents being put off course or couldhave huge impacts that are really hard to
factor into scenarios. So, inlight of those challenges, like, what

(37:34):
are your thoughts on addressing some ofthose key challenges and addressing some of those
gaps? Sorry? Yeah, soI think, like I said i started
this section, there's gaps you shouldaddress, and there's gaps where it's okay
to say, look, we don'tknow the answer. It's a recognized limitation
of this model. We need tomake make our choices with the best data
we have. And I always comeback to my favorite quote from Sherlock Holmes

(37:58):
here, which is where Sherlock isis talking to wat He says, data,
data, data, I cannot makebricks without clay. Data is always
the answer here. We need moreof it. We need better data,
We need more validated data, broaderdata sets to really get to grips with
what the answer can be. AndI think the Bank of England bulletin gets
into quite a lot of that,because what they're doing is encouraging firms to

(38:21):
extend the scenarios that they use toimprove the granularity and improve resolution and physical
risks derives related variables I've talked aboutin our scenario analysis, how you have
to need to have that logically consistentset of variables. Some of those can
be derived from others, and youneed to understand how those go together,
and data can be part of thatsolution, and then determine the extent to

(38:42):
which long term risks can be pricedin is the other key part of the
model. So how much do assetprice is already taken into account of all
this stuff? If everyone knows thatthis is going to be an issue,
then to some extent assets are beingpriced on that basis. And so the
answer to all of those is data. Looking at just one example that the
Bank of England looked at, whichwas how to do this for residential mortgages

(39:05):
and how to improve the modeling there. What the Bank of Ingland suggesting is,
Okay, you need data at theindividual aset level for each of the
houses in your portfolio. You needthings like energy performance certificates to understand how
much price shock might impact individual householdsand what it might mean for what they
do going forwards. You also thenneed data on macroeconomic variants, so you
need to understand things like household disposableincome, what defaults rate, default rates

(39:29):
look like in different macroeconomic climates,what household debt to solvency ratios look like,
and how that can impact things.And then finally the big data on
the macro climate variables, so theflood risk data that we talked about,
if we're talking about sort of mortgagesand UK property outside the UK, it
gets a lot broader and you're notjust having to deal with flood but having
to deal with much bigger risks,especially if you look at coastal areas.

(39:53):
I think scenario analysis is powerful evenwith its limitations. Any refinement that we
make scenarios and reductions in those limitationsonly makes it more powerful. It's fair
to conclude that improving data capture,improving data sourcing working with a variety of
different third party data providers, andthere are some great organizations that are getting

(40:13):
set up purely to try and fixthis data gap. I think that's a
key step for any firms to take. And then also just remember being decisioned
useful can be about helping the usersto understand where the flaws are and where
the limitations are. There's always thatthat classic quote or models are wrong,
but some are useful. Always comeback to that and remember that as long

(40:34):
as we know where our model iswrong, we can still use it great.
Now we talk a lot about it. You mentioned data a few times,
and well, and it's several timesthroughout the course of this conversation.
And it's just something that based onwhat I've seen you up here to be
passionate about. Is data visualization.Now, data visualization that's an important tool
kit factuaries in the modern world.So if you think of this topic broadly,

(40:58):
what are the most some of themost important considerations regarding data visualization.
Yeah, so for me, datavisualization is a tool for persuading. I
think if you spend enough time withdata, it starts to tell a story.
And actualis who spend all day everyday in that data become very familiar
with that story. They're able toglean insights quite quickly just from looking at
the data sets or the calculations basedoff them. The problem is, and

(41:22):
the thing that a lot of peopleforget, is that decision makers in the
business aren't spending all day every dayin the data. They don't have that
familiarity, and so they can missthat story quite easily, and data visualization
is almost a shortcut. It tellsthe story in a way where you don't
need to have seen the data orunderstood the data to get the insights and
to make the right decisions. Potentially, it can even reveal new insights and

(41:45):
allow people to ask really interesting questions. And I think that shortcut to storytelling
is vital. It's almost like,if you were asked to explain to me
about Roman architecture, you have achoice between giving me two really dry textbooks
on architecture. I could spend hoursreading or we could spend five minutes walking
around. I know the second oneis the one that allowed me to really
understand the point and understand what Romanarchitecture is about. Data visualization is kind

(42:07):
of like that. It's give peoplethe information they need in a way that
they can engage with it. Great. Yeah, you know, this picture
is worth a thousand words, exactly, very simple. So tying that back
to the broader theme of scenario analysis, now you know, data visualization,
I think just given some of thethings we talked about availability of data or

(42:29):
lack of availability in some cases,so visualization is one of the key challenges
our own scenario analysis. So otherthan I know we mentioned availability, what
are there any other reasons why thatchallenge exists? Is it like form of
data too like? And I knowthere's different forms, some of these structured
unstructured, you know, are thereany other reasons for that? Yeah?
Absolutely, I mean data availability isthe first one. As I said earlier,

(42:53):
it's stuff that it's there that justhasn't been captured the past, because
why would you, you know,if I think about I often sort of
talk about Crow as an organization herebecause it's a relatively easy organization in terms
of calculating its own emissions. Consultanciesare really easy because they don't make anything.
We only sell people's time. Ouremissions are largely the offices that we

(43:15):
work in, people traveling to andfrom those offices and to client offices,
you know, any other supplies thatwe need for the office, the coffee,
beans and so on. So we'rea really easy organization to do this
for And it is so so hard, nonetheless, and takes so many people
so much time to work it out, and a large part of that is
just that no company has ever structuredthemselves thinking about this stuff. We have

(43:37):
eight offices at Crow and each ofthose offices has a different coffee machine for
staff to use, and under atraditional company structure, the way that you
would set that up is you'd askeach office manager to source coffee machines coffee
beans for that office and they wouldfind probably the cheapest deal, and you
might end up with eight different suppliersjust the coffee machion coffee beans to the
different offices. That means that anyonelooking at the supply chain and trying to

(44:00):
understand that sustainablewth ofupply chain has totalk to those eight different suppliers and has
to do the same investigation eight timesand follow eight different supply chains. What
companies need to start thinking about doingand what we're already on our way to
do with CROW, it's quite along journey is draw that together and actually
get the suppliers together and get yourarms around that. And it might be

(44:23):
more expensive if you go for onesupplier, but that means you have so
much less to look at and somuch less to understand. You can actually
start to work out what the emissionsare from that supply chain and whatever issues
then might be with that supply chain. And I think we're seeing lots of
different opportunities to use data visualization inthat not just in the data capture part,

(44:45):
but then in terms of how weinterrogate and how we start to move
things forwards. And one thing I'vebeen working on recently that I've found really
interesting is reducing emissions for employee commutingbecause that sits within a company's own emissions,
that the emissions that employees generate gettingto and from work are part of
the company emissions, and companies havebeen trying to cut those pages. Even
before you know, reducing emissions wasparticularly popular, and there's lots of ways

(45:07):
that companies have tried that, anda very common way of doing that is
investing money in cycle to work schemesand encouraging people to cycle in and use
lower carbon form of transport. Andthat sounds like a great idea. We
did some data visualization at Crow wherewe took everyone's commuting patterns in we wrote
some code and r to visualize itlooks really cool. What we found when

(45:28):
we drove it as a picture wasthat even though most our staff work in
London, those people aren't generating emissionson their commutes. Everyone in London takes
public transport there might be a coupleof people who drive once a week to
a train station nearby and then getthe train into London, but generally it's
a train, tube, bus,cycle walk. Those are the main method
of transport into London. And thenfor the staff that are outside of London,

(45:50):
well, okay, Manchester has apretty good public transport network, who
don't worry about those. But wehave regional offices where people are driving thirty
forty miles to get to the officeevery day, and we've been saying for
years cycle to work, that's thescheme, that's the solution. You're not
going to do that thirty forty milesthere and back every day. So that's
not a solution that works for allthe offices. And we would never have

(46:12):
realized that until we started visualizing this, from plotting it out and seeing those
actual pathways that people take and realizingLondon is probably going to be okay because
people are taking the tube. Ifwe want to reduce our commuting emissions,
we need to figure out what thepeople in the Cheltenham office need and the
other regional offices need and see howwe can reduce their emissions. And that's
a really simple bit of visualization,but really really powerful and immediately decision makers

(46:37):
can see what the issue is,can see how to solve the problem,
and can start taking actual action basedoff it. You mentioned a just one
quick follow up by it on itis like, what are some of the
most common tools and packages you're seeingbeing used for data visualization? Yeah?
I mean so, I'm a bigperonent of open source software. I'm quite
a big R user myself. Myteam prefers Python and will always tell me

(46:59):
it's better, but I'm just terribleat it, which is probably why I
don't prefer it. I think onething that many people outside of financial services,
any of the students listening, don'trealize, is just how reliant everything
in financial services on Excel. Excelis most of what you see used for
every kind of data management, datacapture, data visualization, with data,

(47:22):
even just people writing lists and meetingnotes, it will go and Excel and
I have never understood why on thatmatter thing, it's helpful that Excel is
getting better. I think Microsoft haverealized the key pivotal role that they play
in financial services globally with Excel andtrying to actually invest in it and make
it more robust you might remember.You may not not being in the UK,

(47:45):
but listeners in the UK might rememberhow during COVID there was data being
copied from one Excel spreadsheet to another, and Excel wasn't big enough and we
lost a bunch of COVID deaths asa result of that, and had to
readjust it a few weeks later.It just got dropped from the day because
it wasn't big enough. That kindof thing isn't going to happen going forward,
And you know, spreadsheets are gettingbigger and more about every day,

(48:06):
but they're still not the best toolfor most purposes. They're very Excel as
a very good general purpose tool,very poor at specific problems that needs to
be solved. But I think oneof the reasons it's so prevalent is that
everyone knows how to use it Andeven though I've talked to all about sort
of giving people the picture that tellsa thousand words and the story they need
to make the decision, there's stillpeople at the most senior level who like

(48:29):
to just get in and play withthe data a little bit, and if
I give them a bunch of ourcode, they're just gonna not know what
to do. With it. IfI give them a spreadsheet, they can
at least have a play around andplay with some of the assumptions themselves.
So yeah, unfortunately, a lotof what I do still tends to be
in Excel. And I've occasionally builtstuff in our in the first instance,
and then built it in Excel whensomeone asked for it. It's just just

(48:50):
the way the world is. Yeah, welcome to my world, you know,
trying to get people to move beyondExcel. But yeah, we'll we'll
end on that. Not it isactual worries. Yes, we know you're
comfortable, but they're other, youknow, more powerful appropriate tools out there,
so we'll just leave with that that. Oh yeah, thanks Louda.
This was very insightful and you know, looking forward to sharing this with the

(49:10):
community, and you know again wantto thank you for your time and just
wishing you wonderful rest of the day, the even probably the evening for you.
You know, you're a little bitahead of me, so afternoon now
still still sunny. Fortunately you finallygot the sun here. Spring took until
this weekend to finally hit the UKSex. Thank you so much for having
me on. You're welcome. Havea wonderful one. You too,