The Actuary - April 2024

INTERVIEW

The UK’s first female Government Actuary

INSURANCE

Once society’s do-gooders – are we now baddies?

AN EYE TO THE FUTURE

Dystopia or utopia – what’s the outlook for DB pensions actuaries?

AUp Front

4 Editorial

We pose some of the big questions facing actuaries, says Yiannis Parizas

5 President and CEO comment Kalpana Shah and Ben Kemp praise actuarial innovation, past and present

6 IFoA news

The latest IFoA updates and events – plus membership survey results

12 News Focus

The ‘planetary solvency’ idea behind a new IFoA climate change paper

Features

14 Interview: Fiona Dunsire

The first female Government Actuary gives us the inside track on doing her job on behalf of the whole country

18 Data science: Mixed blessing Brundha Krishnamoorthi assesses how mixture models fare at predicting claim counts

20 Life insurance: Great British Take Off

What will it take to truly unlock UK capital? Kyle Audley and Brandon Choong on the necessary reforms

22 Data science:

The Bayesian revolution Penny Drastik and Bradley Shearer discuss how to make your modelling more accessible to stakeholders

25 IFoA: A long life A tribute to the CMI’s important history as it hits a century

28 Pensions: An eye to the future

What will the DB jobs space look like in 2040? Mark Williams predicts

31 General insurance: Are we the baddies? Has the insurer’s role morphed in the modern world, asks Steven Fisher?

34 Data science: Efficiency model

It’s not just for modelling: data science can help us in all sorts of other ways, says Ralph Clayton

36 Technology: The power of two

How digital twins are revolutionising the insurance industry. By Neha Agarwal and David Basson

38 Data science:

Blurring the boundaries

GLM, GBM, ANN, and a blend –different lost cost models tested

At the Back

41 People and society news

News from the actuarial community, including celebrating David Wilkie

42 Extra-curricular Priya Shah, actuary and online ‘travel content creator’

44 Soft skills:

From hybrid to ‘mybrid’ Jenny Segal on how to embrace the push to return to the office…

46 Student

Insurers should work with climate innovators, argues Adeetya Tantia

47 Puzzles

This month, continued on page 48 for a full page and a half of teasers

Get the app

You

The Actuary: theactuary.com

IFoA: actuaries.org.uk

PUBLISHER

Redactive Publishing Ltd

9 Dallington Street, London EC1V 0LN

PUBLISHING DIRECTOR

Anthony Moran

MANAGING EDITOR

Melissa Walton +44 (0)20 7880 6246 melissa.walton@redactive.co.uk

SUB-EDITOR

Kate Bennett

DISPLAY SALES

theactuary-sales@redactive.co.uk

+44 (0)20 7324 2753

RECRUITMENT SALES theactuaryjobs@redactive.co.uk

+44 (0)20 7324 2777

ART EDITOR

Sarah Auld

PICTURE EDITOR

Akin Falope

SENIOR PRODUCTION

EXECUTIVE

Rachel Young +44 (0)20 7880 6209 rachel.young@redactive.co.uk

PRINT

PCP

SUBSCRIPTIONS

EDITOR

Yiannis Parizas editor@theactuary.com

FEATURES EDITORS

● Travis Elsum: Environment and sustainability

● Alex Martin: Environment and sustainability

● Sachal Gandotra: General insurance

● James Reeder: General insurance

● Blessing Mbukude: Life

● Ruolin Wang: Pensions and investments

● Lily Hallett: Pensions

● Lisa Morgan: Health and care

STUDENT EDITOR Adeetya Tantia student@theactuary.com

IFOA COMMS LEAD

Sonia Sequeira sonia.sequeira@actuaries.org.uk

IFOA EDITORIAL

ADVISORY PANEL

Peter Tompkins (chairman), Chika Aghadiuno, Nico Aspinall, Naomi Burger, Matthew Edwards, Jessica Elkin, Richard Purcell, Sonal Shah, Nick Silver

Subscriptions from outside the actuarial profession, per annum: UK: £125 Europe: £160 Rest of world: £195

To subscribe, email: editor@theactuary.com

Changes of address, email: membership@actuaries.org.uk

Delivery queries, email: rachel.young@redactive.co.uk

Published by the Institute and Faculty of Actuaries (IFoA), Staple Inn Hall, High Holborn, London WC1V 7QJ.

The editor and the IFoA are not responsible for the opinions put forward in The Actuary. No part of this publication may be reproduced, stored or transmitted in any form, or by any means, without prior written permission. While every effort is made to ensure accuracy of the content, the publisher and its contributors accept no responsibility for any material herein.

© Institute and Faculty of Actuaries, April 2024

All rights reserved ISSN 0960-457X

Throwing it open

Welcome to the April issue. As usual, we have enjoyed traversing the evolving actuarial landscape with features on a wide range of topics.

In so doing, we pose some big-hitting questions. Has the DB pension actuary got a future? Find out Mark Williams’ prediction on page 28. What does the worst-case climate change scenario look like? Sandy Trust outlines a key concept from his latest IFoA collaborative research paper on page 12. Are we the baddies? Insurers used to be a force for social good, says Steven Fisher on page 31 – but in today’s complex world, is that still the case? And how can we embrace the fact that our bosses want us back in the office? Jenny Segal throws light on this hybrid-working dilemma on page 44.

So much easier to answer is: how can technology help us in our work? On page 34, Ralph Clayton challenges the conventional view of data science in insurance pricing, urging teams to embrace it beyond modelling, also using it to enhance efficiency and collaboration. On page 36, Neha Agarwal and David Basson explore the potential of digital twin technology in the life industry, highlighting its ability to drive real-time insights and predictive capabilities.

Our technical articles this time test different motor loss cost modelling techniques (p38) and advocate Bayesian methods (p22), to enhance model interpretability and stakeholder involvement.

In a face-to-face interview with the UK’s first female Government Actuary, Fiona Dunsire, former colleague Stephen Hyams finds out her impressions of the place and the job (p14).

We may have just closed a reader survey but we always welcome your thoughts and contributions –use the email address just on the right here.

KALPANA SHAH

BEN KEMP

Professional, passionate

The gift of feedback

IF

n the IFoA’s latest member survey, 82% of respondents said the actuarial profession was a good career choice and 68% said they loved being an actuary. This passion for the profession is something I want to focus on in this column.

Recently I had the opportunity to visit Liverpool, Kent, City and LSE universities in the UK and the University of Mumbai in India, sharing insights on the actuarial profession and the diverse career opportunities for students entering it. These visits gave me the chance to talk to students and academic staff about the skills modern actuaries need. The IFoA, and I as president, have an important role in attracting and supporting the next generation of actuaries and collaborating with universities in developing research. I have been discussing key takeaways with Council, resulting in exciting new ideas being proposed for testing. I will visit more universities, regional societies and employers in the coming months, and look forward to seeing many of you there and hearing firsthand the opportunities and challenges we face.

eedback is a gift, as many have said. Many thanks to those of you who took the time to share your views in our latest membership survey; you can read the results on page 10 of this magazine. Please be assured that we read, hear and value everything you say – the good, the bad and the indifferent. We have just agreed our plan for 2024/25, and our priority will be to improve the membership experience for all members when you interact with the IFoA as your professional body, and in the service and support we provide for you. We also hope to make the survey itself easier to take.

One way in which we aim to support you is by providing the opportunity to contribute to, and benefit from, cutting edge thought leadership. Our collective brainpower and voice are definitely greater than the sum of our parts. Led by our members, we are making profound contributions to thinking on some of the most important issues affecting our profession, and society at large.

Together with the University of Exeter in the UK, the IFoA has recently published the latest in a series of high profile reports challenging conventional assumptions around the impact and pace of climate change. Climate Scorpion: The sting is in the tail (bit.ly/ IFoA_Climate_scorpion) advances the case for using financial services risk management techniques to evaluate and communicate climate risk. It introduces the idea of ‘planetary solvency’ as an assessment of different ecological threats, including those beyond climate change, to determine the risk of planetary ruin.

KALPANA SHAH is the president of the Institute and Faculty of Actuaries

This month I am delighted to be speaking at a conference at the University of York to celebrate David Wilkie’s 90th birthday. David’s contributions to our profession are vast. He is known internationally for the Wilkie model, a stochastic investment model for long-term actuarial applications; he also played a pivotal role in constructing mortality tables and developing models for valuing maturity guarantees, asset/liability modelling, income protection insurance and projecting the impact of the spread of AIDS. His achievements have been recognised with honorary degrees from Heriot-Watt, Waterloo and City universities, gold medals from the Institute of Actuaries and the Faculty of Actuaries, and honorary memberships of the Swiss, Italian, Swedish and South African actuarial societies. He spent almost 60 years volunteering for the profession on various committees and has written an extensive number of papers – the first in 1960. I’m sure you agree that his passion and commitment is an inspiration to us all.

We have also launched an artificial intelligence (AI) thought leadership series. AI’s rapid advance has captured global interest; this series seeks to address what it means to the actuarial and financial sector, drawing on expertise from policy, science, academia and financial services to demystify this field, its growing role in society, and its risks and opportunities. We seek to tackle the big debates around AI ethics and regulation. Please join us and get involved – more minds are definitely better.

BEN KEMP is interim chief executive of the Institute and Faculty of Actuaries

CHARITY

An active part in giving back

With its sporting fundraising successes of last year, the IFoA Foundation is now working with the organisation Run4Charity to provide more of these opportunities for members and supporters in 2024.

Last year, actuaries Lloyd Richards, Leonard Mapfumo and Masimba Zata (chair of the IFoA Foundation Board of Trustees) took part in individual long-distance running challenges to raise money for the IFoA Foundation, each wearing ‘Actuaries Supporting Others’ branded vests.

“I completed the Royal Parks halfmarathon to raise money for the IFoA Foundation, as it provides educational support to disadvantaged students, which is a cause very close to my heart,” Lloyd shared.

Leonard said: “I’m thrilled that I completed a half-marathon – thank you to everyone

RESEARCH

who supported me and the IFoA Foundation on this incredible fundraising mission to make a significant impact on someone’s life and future trajectory.”

Masimba, meanwhile, enthused that he completed the race wearing “a vest proudly showcasing our wonderful profession, hoping to inspire at least one person in the running crowd to discover the world of actuaries.”

Running is by no means the only athletic fundraising challenge. Perhaps you could face your fears with a bungee jump or persuade your colleagues to throw themselves into a muddy obstacle course? We’ll back you all the way. Please contact us to arrange your challenge, email foundation@ actuaries.org.uk

For Run4Charity activities inspiration, scan this code

Working Party scoops 2023 Brian Hey prize

The IFoA’s annual Brian Hey award, for the best general insurance research paper of the year, has been won for 2023 by the IFoA General Insurance Climate Change Reserving Working Party for its report Reserving for Climate Change

Members of the working party, co-chaired by Alex Marcuson (top right) and James Orr (right), will present a summary of the paper –which centres on climate litigation, communicating uncertainty and quantitative tools – at a sessional meeting at Staple Inn on 8 May (book your place at bit.ly/CCRWP_sessional). They said, on behalf of all the contributors: “With our paper, we sought to engage reserving actuaries, whatever their views on climate change, in responding to the challenges it presents to them in their professional capacity. Not only did the award provide a focus for us to complete our work, it has also provided a high-profile platform from which to promote this critical issue and the central role of the profession in responding to it.”

The judging committee said: “The paper offers valuable insights and presents a practical framework that could be immediately applied to assess individual contracts and aggregate portfolio analysis. Though primarily targeting reserving actuaries, the content also resonates with a broader audience.”

Submissions for the 2024 Brian Hey award are now being accepted, until 30 August. For more information, visit bit.ly/Brian_Hey_Prize

Scan the QR code to read the winning paper

IFoA gives government input on duties

On 21 February, IFoA Pensions Board chair Debbie Webb gave evidence to the House of Commons Work and Pensions Committee on fiduciary duties (bit.ly/Parliament_live_fid_ duties), including managing climate risks relating to pension schemes. In the session, she highlighted the findings of our The Emperor’s New Climate Scenarios report (bit.ly/ Emperors_climate_scenarios) and stated that the recent Financial Markets Law Committee report on fiduciary duty and sustainability investment (bit.ly/FMLC_ trustees_fid_duties) was a helpful clarification for trustees. She said that guidance from The Pensions Regulator to support this, to make it easily understandable and accessible, would be welcome.

New home found for gilts database

The Economic Statistics Centre of Excellence (ESCoE) in London is now to look after the historical British Government Securities Database.

Originally supported by the IFoA, the database was developed and maintained from the 1990s by actuary David Wilkie (see page 41) and actuarial professor Andrew Cairns, and hosted on the Heriot-Watt University website. Both will remain as consultants in the new set-up, which will go by the name of the Heriot-Watt-IFoAESCoE Database.

Scan the code for the database

Governance reforms and blue-sky thinking

Council’s meeting on 7-8 March was a significant milestone in progressing the reforms to our governance. We reviewed the work that has been done following the changes agreed at our December meeting (shared with members in January) and reconfirmed our commitment to implement these changes – barring any unanticipated circumstances – on 1 April.

Building on the discussions we had at our strategy day in November about developing a vision for the IFoA and the profession – which will provide an essential steer for the IFoA Board as it develops strategy – we discussed a ‘strawman’ vision statement, and explored what we consider

should be of greatest importance to the IFoA in future. This work will require considerable refinement, ahead of agreeing a high-level vision at our meeting in June.

Work is ongoing to develop the brand of the IFoA and the ‘brand of actuary’, to promote the profession to a wider audience and to articulate better who we are and what the IFoA does. Council provided feedback on the current draft ‘sales pitches’ for this work, as developed by a steering committee comprised mostly of IFoA members working in conjunction with external consultants. We expect to see final versions of this work at our meeting in June.

We received an update from the Council Working Group on

the further progress of its ‘blue-sky thinking’ on the future of Council. By the time you are reading this, Council will have met again to consider a draft consultation, planned to go out to members shortly.

At what was a very busy meeting, Council also received updates on numerous other topics, including: the IFoA’s risk appetites; the forthcoming introduction into the Actuaries’ Code of requirements regarding diversity, equity and inclusion (DEI); the upcoming mid-point review of the IFoA’s five-year DEI strategy; and the Council-led working group looking at generative AI. Council’s next meeting will be on 13 June. Email us at presidents@ actuaries.org.uk. For minutes of meetings, scan this code

IFoA president discovers more about the profession in India

Last month, IFoA president Kalpana Shah attended the Global Conference of Actuaries, hosted by the Institute of Actuaries of India, in Mumbai.

She participated in the presidential roundtable on the topic “navigating actuarial frontiers globally”, sharing her views on the latest trends, challenges, and opportunities facing the actuarial profession in both a local and global context. She also contributed in a panel discussion aimed at students, about remote working skills.

Shah has always made it clear that one of her

main presidential priorities is to strengthen professional relationships and explore potential areas of collaboration. To this end, she also attended a networking event with senior Fellows, and visited accredited educational partner the Institute of Actuarial and Quantitative Studies to interact with more than 200 students.

The visit reaffirmed the IFoA’s commitment to promoting excellence, professionalism and innovation within the profession in India, and throughout the world.

DEI

The Actuaries’ Code –guidance consultation

We will be launching a consultation later this year on draft guidance to support the forthcoming diversity, equity and inclusion (DEI) changes to the Actuaries’ Code .

The IFoA provides guidance on the principles and amplifications in the code, to help members understand the nature of their obligations under it, and how to apply these in practice.

The IFoA’s Regulatory Board has already begun engaging with stakeholders to help shape the draft DEI guidance, ensuring that key themes highlighted in last year’s consultation on changes to the code are addressed.

When the draft guidance is ready for publication, we will launch an eight-week consultation, in which members and other stakeholders will be able to share their views. We will host a series of webinars, where members can learn more and ask any questions.

We would like to encourage you to take part in the consultation once it launches, to ensure your views shape the guidance to support the DEI provisions in our regulatory framework.

POLICY SPRING BUDGET: SUMMARY

Last month, chancellor Jeremy Hunt delivered the Spring Budget. He stated that the measures announced would ensure long-term growth in the economy while enhancing the UK’s resilience against future economic shocks, such as the ones experienced due to Covid and war in Ukraine. In the run-up, IFoA president Kalpana Shah called on the chancellor to deliver a Budget “for the long-term” and to address the issues that often sit in the “too difficult box”.

The flagship measure of the statement was the widely trailed cut in National Insurance (NI) by 2%, with the chancellor signalling that a future Conservative-led government might abolish NI entirely. Other significant announcements included the replacement of the ‘non-dom’ tax rules with a residencebased regime; a vaping products levy from October 2026; and the introduction of a new ‘British ISA’.

Measures of specific interest to actuaries included a pledge to bring forward requirements for defined contribution pension funds to disclose their asset allocations, including UK equities; a £100m investment to the UK’s new AI Safety Institute; and continued work on a new Value-for-Money framework for pensions.

Scan the code for the IFoA’s Budget briefing Scan for the relevant IFoA webpage

Smaller fortunes

The gender pensions gap is the focus of a new paper in the IFoA’s ‘think’ thought leadership series. Henry Thompson asks paper author NOW:Pensions some key questions

Henry Thompson Head of public affairs at the IFoA

The IFoA’s ‘think’ thought leadership series promotes debate on topics from the actuarial world and beyond, allowing members and stakeholders to share opinions that may differ from the IFoA’s ‘house’ view.

Its third paper, Tackling the gender pensions gap – the road to financial equality in retirement, builds on a report published by NOW:Pensions in February, which highlighted that, for women to retire with the same amount in their pension pot as a man, they would need to work an average of 19 years longer. Here, NOW:Pensions answer questions on the topic.

What is the gender pensions gap?

Unlike the gender pay gap, there is no clear consensus in terms of definition, magnitude or

potential solution. While the gap is widely acknowledged to exist, organisations record and report on it in a variety of ways. Our February report viewed it as the amount saved into a pension by the average man (£205,000) and the average woman (£69,000), then extrapolated that into the additional number of years women would have to work to bridge that gap – 19 years.

People are also living longer, and women typically live longer than men, so they must make their pension provision go even further. Two-thirds of pensioners in poverty are women, and half are single women. And with an aging society, the phasing out of defined benefit pension schemes, and fewer people owning homes or other assets, the gap may widen further.

What are its causes?

These are complex, touching on areas such as government policy, employer practices, educational outcomes, individual choices and gender roles. Broadly speaking, they can be placed in two areas. The first is the difference between earnings. Women’s average pay was equivalent to 75% of men’s in 2023, giving a gender pay gap of 25%. Women’s average annual incomes stood at around £24,800, compared with £33,000 for men.

The second area is employment patterns, including the impact of caring responsibilities – women are more likely to take time out to raise a family or look after

relatives or friends. Even if the gender pay gap is eradicated, a woman will still only acquire 66% of a man’s pension wealth on average if these patterns continue.

Does it vary between industries?

Female-dominated industries tend to have smaller gaps than male-dominated industries. There are also differences between the private and public sectors. Women might be ‘over-represented’ in healthcare and education, men in finance and technology. Male-dominated fields tend to have higher incomes, perhaps due to the perception that they are ‘breadwinner’ roles. The imbalance is greatest at the most senior levels when earnings are higher, in roles that typically pay more than the national average.

How can we tackle the issue?

It’s often said that men and women gravitate towards different professions, and while organisations can’t vouch for personal preference, industries such as financial services still have a reputation for being male dominated. Sectors and firms can always do more to encourage and support women. It will take a collective effort from industry leaders and individual firms to address the structural issues.

Flexible working policies and childcare support, along

Two-thirds of pensioners in poverty are women, and half are single women

with equitable parental leave policies, can make it easier for mothers to return to work if they wish. There should be more shared learning within the industry from organisations that are doing well. Indeed, learning from those outside the industry with regard to flexible working, mentorship and networking opportunities can allow senior executives and business leaders to examine their own working practices.

What is the ‘takeaway’?

Solving the gender pensions gap is a collective effort. In an aging society, equitable pension provision affects us all. The actuarial field has its own role to play in making sure its employees can save for the retirement they hope for, by giving them equal opportunities to pursue fulfilling and rewarding careers.

Find out more about NOW:Pensions and its research at bit.ly/NowPensions_gender_gap

membership survey

Listening and learning

The results from last year’s membership survey give us clear pointers as to how we might make IFoA membership work better for you

We asked

At the end of 2023, we asked you to complete a survey to rate the membership experience, with some extra questions seeking deeper insights into how you feel and what we might do to make improvements.

Your replies

The response rate was 14.4% – down on our 2022 survey (21%).

Our actions Interim CEO Ben Kemp, said: “It’s clear the IFoA’s recent difficulties have been felt by you, as well as the increase in fees. But your responses also point to plenty of positive actions we can take to make IFoA membership better for you in future. Thanks to all of you who responded to this survey.”

SATISFACTION

63% You gave us this satisfaction rating – down 2% from 2022. We froze our fees for the previous three years but were unable to do so last year –63% of those dissatisfied said this was an issue

36% say membership is improving (rising to 48% of students)

90% of you have used our new online member portal

4,857 MEMBERS RESPONDED TO THE SURVEY

2,207 answered our in-depth questions to help us improve the member experience

82% agreed that the actuarial profession was a good career choice for them –with 68% saying they love being an actuary

83% think it’s important that the IFoA connects with government decisionmakers to influence policy

70% of those who use the ‘IFoA communities’ forum like to see what their peers are thinking

54% of those not already on it are considering joining

You gave us some great guidance on what new benefits would be valuable to you:

78% BENEFITS think e-learning courses would be helpful VALUE

70%

66% said that an IFoA mentoring programme would be of worth

61% want to see IFoA outputs around artificial intelligence and 57% around data science are interested in career support tools

62% value being part of the wider professional world that IFoA membership provides – we need to do more to make the rest of you feel the same way

61% of respondents are proud to belong to the IFoA

57%  agree that the IFoA helps to ensure that their expertise is widely recognised

Worst case scenario

Sandy Trust explains the concept of ‘planetary solvency’– a new way to think about what we’ve done, and are doing, to the Earth

The IFoA recently published its latest paper on climate change, Climate Scorpion –the sting is in the tail. The last in a series of three, it was produced by me and a team, in conjunction with the University of Exeter, and follows on from Climate Emergency and The Emperor’s New Climate Scenarios. In it, we coined the phrase ‘planetary solvency’, exploring how we could use actuarial techniques to help society manage climate change and other risks. What do we mean by this, and how can the concept help us successfully navigate the challenges of the Anthropocene?

A new geological era

Humans are now so numerous and so dominant that, for the first time in history, the activities of a single species are driving planetary outcomes. Scientists call this latest geological era the Anthropocene.

Since 1950, the global population has soared from 2.5 billion to nearly eight billion, with life expectancy rising from age 46 to 74. Around $500trn of wealth has been created, but it is unevenly spread, with the richest 1% owning more than the poorest 55%. This has been supported by huge increases in human material and primary energy consumption – in 2022 we consumed more than six times the amount of energy we did in 1950.

Despite progress, we face many problems, including climate change, nature loss, antibiotic resistance, plastics, space junk, obesity and rising geopolitical tension.

The mass of human-produced items is estimated to have exceeded that of all planetary biomass in 2020. Habitats have been destroyed and approximately 25% of species face extinction. Human activity is polluting the biosphere that provides food, water, air and raw materials, and we are using up resources faster than they can be replenished – we would require 1.7 Earths to satisfy current consumption rates on an ongoing sustainable basis.

What is planetary solvency?

We’re used to the concept of solvency for an insurance company or pension scheme. A financial entity has assets, which must be able to meet its liabilities (claims or pensions payments) – sometimes many years in the future. Regulatory regimes protect customers and society from financial entity insolvency, providing a framework to assess risks and cut the chance of insolvency to a small probability – equivalent to a one in 200-year event, under EU regulation.

Actuaries seek to understand risks and avoid insolvency or the risk of ruin. We consider scenarios that might cause insolvency in a process we call reverse stress testing, which helps to inform management on risk mitigation.

Before we get to planetary solvency, a quick reminder on ecosystem services –the goods and services provided by nature. These can be split into three categories:

Regulating services – those that regulate environmental conditions such as climate, air, water and oceans

Material or provisioning services – those that provide material goods such as energy, food, medicines and raw materials

Non-material services – such as opportunities for learning, inspiration, recreation and spirituality.

In planetary solvency, nature is the asset – the balance sheet that provides ecosystem

services. These services are the flows from nature that provide our essentials: food, water, raw materials, a stable climate and so on. Just as financial solvency assessments look at an entity’s ability to pay claims now and in the future, planetary solvency would assess nature’s ability to continue providing ecosystem services.

To apply this approach, we would need to derive:

A set of most likely and worst-case outcomes for nature, climate and other risks

A ruin scenario exploring how nature, climate and other risks could lead to societal ruin

A set of temperature thresholds and other limits, and how to determine societal fragility under stressed conditions.

It would also be important to derive an acceptable probability of ruin. One insurance company failing will not impact all of society,

Planetary solvency

FIGURE 1: The Anthropocene reality and planetary solvency

WWIIPost-war consensusNeo-liberalism Renaissance or decline?

Now – Anthropocene reality – risk of insolvency Planetary Boundaries breached – a need actively to steward Earth system

Good Anthropocene – solvent Low carbon; nature positive, just

Turbulent Anthropocene – insolvent

Holocene Earth system – solvent Humanity operating within planetary boundaries

Abundance, prosperity

Disruption, shocks, uncertainty

IMPACTFINANCIAL IMPACT

CATASTROPHIC ≥25% ≥25% > 2 billion deaths

DECIMATION ≥10% >$10trn annual losses ≥10% > 800 million deaths

SEVERE ≥5% >$5trn annual losses ≥5% > 400 million deaths

LIMITED ≥1% >$1trn annual losses 100% > 8 million deaths

but there is a level of global warming or nature loss to which it may be challenging to adapt. This suggests it would be prudent to aim for a lower chance of failure than one in 200.

If we can carry out this analysis rigorously, dispassionately and meticulously, we can gain data on climate change, nature loss and other risks that could be used to inform policy decisions. Planetary solvency would combine nature, climate and societal risk assessments, leveraging the Planetary Boundaries framework to assess risks to ecosystem services – and thus to society and the economy.

A planetary solvency risk matrix

A standard financial services risk matrix could be adapted for planetary solvency to communicate the risks, likelihoods and potential impacts. It could use a standard likelihood scale but with an adapted impact scale that is suitable for assessing societal impact, adapted from the thinking in Luke Kemp et al.’s 2022 paper ‘Climate Endgame: Exploring catastrophic climate change scenarios’ (bit.ly/Climate_endgame). This defined a set of novel impact terms for society, including:

Systemic risk – the potential for individual disruptions or failures to cascade into system-wide failure

NON-FINANCIAL IMPACT

Worst case warming Significant breakdown of most ecosystem services and Earth systems. Mass extinction of majority of higher-order life on Earth

3°C or more by 2100, multiple climate tripping points triggered, tipping cascade

2°C or more by 2100, high number of climate tipping points triggered, partial tipping cascade

Global warming limited to 2°C by 2100, several climate tipping points triggered

Global warming limited to 1.5°C by 2100 following overshoot, some proximate climate tipping points triggered

Global warming below 1.5°C by 2100 with limited overshoot, climate tipping points largely avoided

Breakdown of several critical ecosystem services and Earth systems. High level of extinction of higher-order life on Earth

Breakdown of some critical ecosystem services and Earth systems. Major extinction events in multiple geographies. Ocean circulation severely impacted

Severe reduction in several critical ecosystem services. Major extinction events in some geographies. Frequent global food and water crises

Some impacts to critical ecosystem services. Ongoing species extinction. Regular global food and water crises

Mass extinction avoided and ecosystem services largely functional. Occasional global food crisis and widespread water crises

Extinction threat – a plausible and significant contributor to total extinction risk

Societal collapse – significant sociopolitical fragmentation and/ or state failure, along with the relatively rapid, enduring and significant loss of capital and systems identity; this could lead to large-scale mortality and morbidity increases

SANDY TRUST

works in financial services on climate change and sustainability

Global catastrophic threat –a plausible and significant contributor to global catastrophic risk; the potential for climate change to be a global catastrophic threat can be referred to as ‘catastrophic climate change’ Global decimation risk – the probability of a loss of 10% (or more) of global population and the severe disruption of global critical systems (such as food) within a given timeframe (years or decades).

The ‘Climate Endgame’ paper focused on climate change; planetary solvency would expand this analysis, assessing risk likelihoods and impacts across GDP, mortality, climate change, nature and society. It would work backwards from ruin, rather than forwards from a benign past that does not represent future risks. We could then define a planetary solvency risk impact and likelihood matrix, and carry out a combined risk assessment.

N/A

Significant socio-political fragmentation worldwide and/or state failure with rapid, enduring and significant loss of capital and systems identity. Frequent large-scale mortality events

Severe socio-political fragmentation in many regions, low-lying regions lost. Heat and water stress drive involuntary mass migration of billions. Catastrophic mortality events from disease, nutrition, thirst and conflict

Severe socio-political fragmentation in regions exposed to climate and/or nature impacts. Failure of vulnerable states and mass mortality events in impacted areas

Some socio-political fragmentation in most vulnerable states where adaptation has been limited. Fragile states exposed to climate risks see mass migration and mortality events from heat, water stress and weather events

Ongoing significant climate impacts with many hundreds of $1bn+ loss events annually and associated mortality and socio-political stress

mitigated, leading to further global temperature increases and increasingly severe climate impacts, which overwhelm societies’ ability to adapt.”

In 2023 we saw an increase in $1bn-plus loss events and 10,000+ mortality events. We might assess the impact as Limited but Trending to Severe in the short term, with a likelihood rating of Highly Likely (more than 90%).

The risk trajectory for climate (based on analysis in Climate Scorpion) would be Catastrophic (Highly Likely), with emissions and greenhouse gas levels implying more than 2°C of warming by 2050. Extreme warming might be assessed as Possible to Likely.

As with financial solvency risk assessment, planetary solvency’s purpose is to inform management of actions (policy decisions) that can be taken to avoid risks that exceed risk appetite; there has been limited consideration of the impacts we face under worst-case scenarios. Planetary solvency would show policymakers the full range of potential impacts, allowing them to understand how high the stakes might be and informing long-term policy actions.

Extreme climate change – a mean global surface temperature rise of 3°C or more above pre-industrial levels by 2100 Scan the

Taking climate change as an example, we define the risk as: “Climate change is not

Climate Scorpion paper

FIGURE 2: Planetary solvency risk impact and likelihood definitions (illustrative)

CAUSE AND EFFECT

What’s it like moving from a long career in global private practice to the biggest actuarial job in the UK public sector? As the first woman in the role, to boot? The new Government Actuary, Fiona Dunsire, tells Stephen Hyams how she’s embracing it

iona Dunsire was appointed Government Actuary in late 2023 for a five-year term – the first woman to take on the role. She leads the Government Actuary’s Department (GAD), which provides advice and analysis to other government departments and some public sector organisations.

“GAD resembles a private sector consultancy in terms of culture, the use of modern processes and tools and even time recording,” Dunsire explains. However, it faces a high level of external scrutiny because most of its work is published, meaning she also has a significant hands-on technical role – much more than a private sector leader would. This was one of the things that attracted her to the job, on top of the public service opportunity, she says.

Dunsire previously enjoyed a long and successful career at Mercer. Having qualified as a pensions actuary, she then moved into investment consulting and went on to lead the firm’s investment business. She was appointed CEO in 2012 – a position she held until 2019.

In her final years at Mercer, Dunsire was responsible for growing its investment and retirement businesses across Asia, the Middle East, Africa and Latin America. “It was a very diverse experience, having to adapt my style to different cultures, and there were many

government-related clients – similar to GAD,” she observes.

Access all areas

GAD is quite a small team, with 220 people across locations in London and Edinburgh, and some rotation between teams to provide staff with wider experience. As part of the civil service, it benefits from the Catapult mentoring scheme, designed to progress people from less socially advantaged backgrounds. Employees can also move to other departments, sometimes temporarily. Even at very senior levels, staff can find opportunities to perform leadership roles at other types of organisations.

Secondments from the department, of which there are many, last for one to two years, although Dunsire notes that it has “an almost permanent secondee at NHS Resolution, the body that pays clinical negligence claims”. In general, she explains, “secondment is a really good development tool and means of sharing analytic capabilities”.

Bringing pensions up to date

“GAD’s work is far more diverse than might be imagined from the outside,” Dunsire says – it even analyses premium bond numbers to check they are truly random. However, about a third of its work relates to public sector pensions, mostly unfunded pension schemes such as those of civil service staff, teachers, and NHS, fire and police employees. These cover around 15 million lives in total.

For the 2024 actuarial valuations, the department will be using more effective techniques – including artificial intelligence (AI) – to process the data more quickly. It is currently exploring how it can use a large language model to identify the correct regulations for various complex pension calculations, thus providing quality assurance (QA) across public sector pension administrators. The team would then write Python code to perform the calculations and check the administrator’s result.

‘How do you get the most out of people? Be optimistic, resilient, motivational and humble’

GAD is an important part of the Government Analysis Function (AF), a cross-departmental network for all civil servants working in government analysis whose aim is to foster more collective problem-solving and consistency across careers. In one recent development, it has made some of its large datasets accessible to different teams for various purposes, streamlining the data collection process.

The AF can throw resources at any problem – during Covid, for example, GAD was able to second a senior actuary to jobshare with the deputy director of healthcare analytics at the Department of Health and Social Care. The secondee was also part of the team that developed the QCovid model, which predicted who was most vulnerable to the virus.

However, GAD is “unique within the AF”, according to Dunsire, because all government actuaries are employed by it, “whereas economists, statisticians, data scientists, operational and social researchers and so on are spread across departments”.

Data analysis can also throw light on strategic issues. For example, when the Local Government Pension Scheme (LGPS) Advisory Board was investigating its gender pensions gap, GAD was able to amalgamate data from across all 87 funds.

“It was an interesting example of Simpson’s paradox, where the perceived gender pension gap at an individual local authority level was much less than when you combined the data,” Dunsire reveals.

“While GAD provides some input, key assumptions for its actuarial valuations are set centrally by the Treasury,” she continues. The 2020 valuations were held up due to the McCloud judgment, which ruled that the public schemes had discriminated against younger members and needed to devise remedies.

The department is currently working on the Section 13 report into the LGPS funds to comment on regulatory compliance, consistency of approach, solvency and long-term cost effectiveness. In previous reports it commented on the inclusion of climate analysis, which is now embedded within the local valuations.

IMAGES: FLEA PURICE AND HARRY WASSELL AT DESIGN 102

Strong safety net

GAD’s insurance reserving client base has grown significantly. In 2024, it will have been involved for 10 years with the risk protection arrangement for schools, in which the government provides an alternative to commercial insurance. GAD is involved in the scheme’s set-up, design, regular provisioning, pricing and ongoing best practice. “It’s expanded over time to include, for example, cyber risk,” shares Dunsire.

The department’s international work has also grown, especially in disaster risk finance, risk pooling and activating private sector capital. Its work mainly involves carrying out due diligence on financial structures: for example, via the Foreign, Commonwealth & Development Office, it supports the African Risk Capacity through a sequence of risk pools to ensure funding is available ahead of natural disasters. Such parametric insurance releases money when pre-agreed parameters are hit, targeting those places that need it.

“I have been very impressed with our modelling capabilities,” says Dunsire. “Our Analytical Solutions Team does all the analytical work, which amounts to about 40% of what we do. Formed only a few years ago, it has grown to about 80 people with diverse backgrounds. They are continuously seeking improvements and some interesting modern approaches, such as the use of dashboards to help communicate the output of models, are being taken at a pace of change one would expect in the private sector.”

A mark of quality

GAD’s QA work is also growing.

“We don’t have the resources to build models for all government departments, but we can bring incredible value because of the way we think about the problem – in particular testing assumptions and their consistency,” says Dunsire. The department’s position within government means it’s easy to share data, and as a non-profit it is especially good value.

QA involves three stages:

verification (is the model performing as expected?), validation (are the data, assumptions and methodology appropriate?), and governance (how is the model documented, updated and used, and are its limitations and uncertainties clear?).

“We carry out all three stages in amounts depending on the project,” Dunsire explains.

GAD often carries out QA for compensation scheme models. During Covid it provided QA for the model built by the British Business Bank, responsible for the Covid lending scheme supporting small and medium-sized entities. It also has four

‘Secondment is a really good development tool and means of sharing analytic capabilities’

permanent secondees at the Contingent Liability Central Capability, an analytical and advisory unit within UK Government Investments that aims to strengthen government contingent liability expertise.

Eyes on the horizon

GAD’s current strategy runs to 2025 and it is in the early stages of its next plan. What are some of the main themes? “Our horizon scanning will need to look at the fiscal challenges and systemic risks we face in the UK, such as changing demographics, the ageing population, long-term care and the impact of climate change,” says Dunsire. “We must be able to support key policy decisions in the next five years, such as the review of the state pension age due within two years of the next parliament.

“I am especially keen to see us continue to embed innovation across the whole organisation. We need to carry out our regular work as efficiently as possible, to make space for new areas of work.” Use of AI will be key here, she notes.

“We also need to continue to develop the leadership skills of our people. An effective organisation has strong leaders all the way down.” How do you get the most out of people?

“Be optimistic, resilient, motivational and humble.”

It’s not just leadership skills on her mind. “The mix of skills within GAD needs continual monitoring as well, with an increasing need for people with more diverse backgrounds. I’ve mentioned analysts, but potentially people with more credit risk capabilities.”

She is keen to see GAD diversify its work and client base, “in order to debunk what people think we do and for us to be the go-to partner across government for any work involving financial risk and data analytics, and to be included in early discussions”.

Dunsire’s term looks set to coincide with a particularly interesting period of evolution and innovation – we look forward to seeing her in action.

MIXED BLESSING

When it comes to predicting claim counts, mixture models seem to provide a better fit than traditional GLMs, says Brundha Krishnamoorthi

ctuaries have traditionally used Poisson regression to model claim counts for personal lines business and some commercial lines business. While the Poisson response distribution is the default go-to option under the generalised linear model (GLM), it does not give a good fit to the data in most cases – despite choosing the most sensible predictors for the model based on initial exploratory one-way analyses and judgment, removing outliers, segregating attritional and large claims, and making sure the categorical predictors have an optimal number of categories. This is because two fundamental assumptions in the Poisson regression model are often violated in real life:

Equi-dispersion – the mean of the Poisson random variable equals its variance Independence – the observations are independent of one another.

In most real-life settings the claim count response variable is overdispersed, with the variance of the response being greater than its mean. Dealing with outliers can tackle overdispersion to some extent, but overdispersion may persist due to excessive nil claims in the data, unobserved heterogeneity in the data that is not

captured by predictors used in the model, and correlation between observations (claim events).

Exploring alternative models

To deal with these issues, the following options may be explored:

1 Negative binomial regression model (NegBin) – more robust to correlation between observations, as it primarily addresses the overdispersion problem

2 Zero-inflated Poisson regression model (ZIP)– if the overdispersion is driven by excessive nil claims

3 Zero-inflated negative binomial regression model (ZINB) – if the overdispersion is driven by excessive nil claims and correlation between the observations

4 Poisson mixture model (PMM) –a more flexible modelling approach in which a mixture of Poisson distributions is fitted to the claim count response to deal with the overdispersion and unobserved heterogeneity better.

While the NegBin regression model is a conventional approach like the Poisson regression model, the rest are all mixture models. Zero-inflated models are two-part models with a binomial model for the zeros and a Poisson or negative binomial model for the counts. The binomial model uses a

logit link, and the Poisson or negative binomial model uses a log link. Here, the first part models whether there is a claim or not; if there is a claim, the second part models the counts.

The PMM is a k-compartment Poisson model, used to describe data that is assumed to come from a mixture of k different Poisson distributions. The observations come from k sub-populations. This model helps identify to which group each observation is likely to belong.

At the start, k prior probabilities are attached to all the observations in the dataset for belonging to each of the k Poisson sub-populations. Using an expectationmaximisation algorithm, the model (by looking at the observations) calculates the posterior probabilities of each observation belonging to each of the k-Poisson subpopulations. k-Poisson models are then fitted to the data and each datapoint has k-posterior probabilities attached to it. Taking a weighted average of the k fitted responses with the k posterior probabilities as weights gives the final fitted response from the PMM model.

Apart from the chosen predictors, this stratification of observations into k groups helps to quantify the unobservable or latent factors to some extent, thereby reducing the residual variance.

Testing it out

The data used for this analysis comes from a health insurer in the Turkish market. The response variable is claim count and the predictors used in all the models are age group, BMI group, occupational status, gender, new business or renewal flag, and city group (all cities are grouped into six classes). Log-link function is used for all the models as it ensures the predicted counts are non-negative and facilitates easier interpretation of the model relativities. The exposure variable that is the duration of cover is treated as an offset variable rather than a predictor variable to capture it as a known effect in the model, so log(exposure) is used as an offset term in all the models. The number of compartments (k) in the PMM model here is two, as this is the optimal number of compartments for the given response dataset.

The models are evaluated based on the goodness of fit metrics such as Akaike information criterion (AIC), Bayesian

Density

information criterion (BIC), log-likelihood and root-mean-square error (RMSE). We also look at the numerical summaries of the fitted response, namely the mean, variance, minimum, maximum, median, 25th and 75th percentiles, and compare them against the corresponding numerical summaries of the actual response.

Table 1 shows that the zero-inflated negative binomial model has the lowest AIC and BIC, and the highest log-likelihood. However, the PMM has the lowest RMSE and provides a closer fit in terms of all the numerical summaries. This is also confirmed visually by Figure 1, which shows the actual and fitted response distributions under each model. The fitted response under the PMM has the highest variance, showing that it captures the heterogeneity better. The PMM thus provides a better fit to the data due to its flexible structure.

The fitted response under the Poisson mixture model has the highest variance, showing that it captures the heterogeneity better

When the PMM model is fitted to the data, it gives k fitted responses and corresponding k posterior probabilities. This is the dataset on which the model is trained. When the model is applied to a new dataset (test dataset), the model can give only the k predicted responses. As it doesn’t have visibility of the actual response, it cannot calculate the likelihood to derive the posterior probabilities. In that case, only the k prior probabilities will exist, which can be refined by a combination of underwriting and actuarial judgment to mimic the would-be posterior probabilities. As a result, the predictive accuracy of the model will depend on expert judgment and mixture of loss count distributions chosen.

Features Data science

While the negative binomial model and the mixture models (zero-inflated model and PMM) have their limitations, they are still better alternatives to the standard Poisson model in most cases, as the assumptions underlying the Poisson model are usually violated. However, care should be taken when selecting a model to ensure it is relevant for the underlying response data being modelled.

BRUNDHA KRISHNAMOORTHI is a general insurance consultant at 4most

Claim count response

ACTUAL POISSON NEGBIN

ZIP ZINB PMM

FIGURE 1: Actual vs fitted claim count distributions

TABLE 1: Goodness-of-fit metrics and numerical summaries

Life insurers are a key ingredient in the UK’s aims to become an investment showstopper, say Kyle Audley and Brandon Choong, as the Solvency UK reforms move closer

Great British Take Off

he UK remains one of the world’s leading financial centres; having survived the thematic headwinds of global inflation and limited fiscal headroom, as well as systemic shocks brought on by geopolitical turmoil and the pandemic, it is looking to new opportunities.

The future presents its own challenges, such as the imperatives to reach net zero and mitigate climate change impacts, provide affordable homes for a growing population, and deliver progress so everyone can benefit from the opportunities created in a thriving economy. With this in mind, the UK government has outlined several objectives: reaching net zero by 2050; investing in infrastructure to ‘level up’ economic growth; and driving regeneration and housing delivery to create affordable homes.

Ambitious targets require commensurate capital, and meeting these objectives will depend, in part, on the government’s ability

to access private long-term capital funding for large-scale underlying projects and initiatives that drive progress.

The UK life insurance industry could be the single biggest provider of long-term fixed capital. Key players, particularly the major bulk purchase annuity companies, manage liabilities and invest assets within their matching adjustment (MA) portfolios – a regulatory mechanism that recognises the positive risk-management effects of cashflowmatching illiquid, fixed and highly predictable liabilities with similar-natured assets. The annuity providers have long-term liabilities (in some cases beyond 50 years), so they need assets with similar time horizons to back these liabilities, and lend to a diverse set of high-credit, quality public and private borrowers that seek debt for their own long-term projects and obligations. As a result, annuity providers are suited to act as long-term senior debt finance providers.

The bulk purchase annuity market is thriving, buoyed by the higher interest rate environment, and is expected to increase further. Facing potentially significant flows of

new liabilities, firms will need to find large amounts of long-term investments. However, the life industry’s role in supporting broader government objectives could be enhanced beyond long-term fixed debt through reforms on both the supply side and the demand side – after all, it takes two hands to clap.

Supply-side reform

Recognising this opportunity and seeking to secure the UK’s position as a global financial leader, the government unveiled the Edinburgh Reforms in December 2022. These included plans to reform Solvency II to unlock more than £100bn of long-term productive asset investments from UK insurers. Driven by collaboration between the government, the Prudential Regulation Authority (PRA) and the life industry, individuals and firms scrutinised how such advances could be realised while preserving policyholder security and the UK stability.

This culminated in the release of a PRA consultation last September on reforms to the MA’s asset eligibility requirements. It built on the existing requirements of fixed and certain

There is a role for the life insurance sector in guiding harmonious reform across institutions, borrowers and advisers

productive investment in the UK economy”. This enhances annuity writers’ ability to provide senior fixed capital, and we hope to see more investment in real assets in the UK.

The government has also made reforms in pensions to stimulate growth and innovation in UK business and technology. The Mansion House Compact, launched in July, commits a major portion of the UK defined contribution workplace market to allocate at least 5% of its default funds to unlisted equities by 2030, increasing further investment flows into growth and productive assets. The Long-term Investment for Technology and Science initiative also seeks proposals for new vehicles to enable pension schemes to invest in UK science and technology companies.

Demand-side reform

Supply-side reforms may yield more flexible institutional investors with vast amounts of capital – but to unlock a new era of productivity, further reforms are required to provide a pipeline of suitable investment opportunities for this capital. And here we encounter the demand side of the UK investment solution: the perennial challenge for annuity writers to originate suitable private market investments capable of backing the annual volumes experienced in bulk annuity markets. These reforms cannot be implemented in isolation; there is a role for the life insurance sector in guiding harmonious reform across institutions, borrowers and advisers, spanning a diversity of UK opportunities, to realise potential.

Central government’s role

bank as lender of last resort, and with other public institutions that provide specific risk facilities (such as Flood Re). A Treasury able to underwrite new risks until capital markets can accept and efficiently price those risks could catalyse many large-scale projects.

Local government’s role

Generation of productive assets is not limited to Westminster; local governments work to drive growth and opportunity in their regions. However, the scope and investment expertise of such borrowers often varies. Reforms to improve consistency and structure could lead to a more compelling landscape of empowered regional borrowers.

National institutions’ role

National institutions play a key role in mediating capital supply and demand –they can enhance borrowers’ capabilities, provide debt and equity capital, and coordinate institutional investors. Each institution has a specific focus, often aligned with a broader government objective, such as UK company growth, infrastructure development or affordable home provision. This fragmentation of purpose and capability is difficult to navigate. A cohesive single body could promote government objectives better, simplifying engagement for borrowers and investors and allowing expertise to be combined.

Two hands to clap

The impacts of the Solvency UK reforms on UK insurers’ overall investment profile are yet to be realised. The reforms could lead to a broader spectrum of productive assets eligible for the MA, unlocking more investment capital. However, the measurable success of these reforms will depend on the availability of projects in which to invest.

Current MA eligibility proposals broaden the assets eligible for inclusion within the MA portfolio to those with highly predictable cashflows. Per the PRA, “the proposals will allow the life sector to play a bigger role in

Attracting private investors is a key element of fiscal budgets and government spending plans, particularly as higher interest rates have increased the cost of government borrowing. However, there are risks that capital markets cannot efficiently digest, as they are beyond the expertise of investment professionals and/or exceed institutions’ risk limits and appetites. A fiscal budget limited in capital can support this investment by instead acting as an underwriter to those risks, insulating institutional lenders from tail risks and black swan events. Markets are familiar with the central cashflow, and developed the wider regulatory and risk management framework around MA portfolios to make it consistent with this new flexibility. Many productive investments are not fixed and are not currently eligible for the MA – often because of their means of revenue generation, uncertainty around the timing of key events, or structural debt requirements that challenge the fixity requirements needed for annuity matching.

As we have said, it takes two hands to clap, and the case for insurance companies to support productive investments at an even larger scale can be made more compelling by demand-side reforms. Insurers and regulators can become a guiding influence in the future of UK investment by continuing their efforts here.

THE BAYESIAN

As our models get more and more sophisticated, how can we ensure we bring our stakeholders along? Try probabilistic programming, suggest Penny Drastik and Bradley Shearer

s actuaries adopt machine learning and artificial intelligence techniques to improve pricing sophistication, the gap is widening between our models and non-technical stakeholders. Probabilistic programming builds interpretability into model code and can help us bring these stakeholders on the modelling journey. Here, we introduce probabilistic programming by applying Bayesian methods to price testing.

The goal: modelling demand without data

Suppose that a team wants to construct a demand model for a product to inform future price changes. This would generally be done using market data to measure competitiveness. However, a different approach is required if market data is unavailable or of low quality. Such situations could include the launch of a new product, a sector where little or no market data is available, or a period of instability that has made past data a poor predictor of the future.

The solution: price testing

In price testing, distinct quotes are randomly assigned a percentage loading to collect conversion data, which can

then be used to model elasticity. To maximise efficiency, we need to determine the minimum number of quotes over the minimum period required for a credible result.

The frequentist method

In the frequentist method, quotes are split into two groups: one priced as usual, the other given a fixed percentage loading or discount. It is assumed each group has a certain degree of homogeneity such that the number of sold policies from group i follows a Binomial(ni = number of quotes, pi = conversion rate) distribution. The analyst can vary the number of quotes split between the groups and the point estimate of conversion in the loaded group.

Using a hypothesis test for difference of proportions, we can determine whether the loading has made any significant impact – or, just using the loaded group, we can form a confidence interval for its conversion rate.

The Bayesian method

The Bayesian method is identical to the frequentist approach, except that instead of using a point estimate for the conversion rate, we assign it a prior distribution. This incorporates both the elasticity estimate and the extent of the uncertainty surrounding this estimate.

Traditional Bayesian methods rely on an analytically convenient choice of statistical distribution: the ‘conjugate prior’. But this does not always reflect the context in which these methods are applied. In our price testing example, we assumed that the number of policies sold followed a binomial distribution – so, under the standard approach, the conversion rate should follow a beta distribution. This gives us a relatively high degree of flexibility in our choice of prior, as we can vary two parameters to set a desired mean and variance. In the most extreme case, when we have no insight into conversion, we can use the noninformative prior, a Uniform(0,1)=Beta(1,1) distribution, to assume conversion is equally likely to take any value between 0 and 1.

Objections would be that for reasonable data volumes, conversion is unlikely to fall to 0, and when we increase premiums, we will probably see conversion decrease. A more suitable choice of prior may therefore be Uniform(a,b) – which cannot be reframed as a beta distribution. This is where probabilistic programming comes in, allowing the use of a wide range of distributions in statistical modelling via simulation and numerical solutions.

Available Python packages include PyMC, Pyro and TensorFlow. The programming language Stan interfaces with all standard data science languages to provide options for Bayesian analysis. These platforms make Bayesian inference accessible by reducing mathematical demands without sacrificing rigour. An example of the Uniform(a,b) prior / Binomial likelihood model described earlier, set up with PyMC, can be found at bit.ly/Heapsort_PPER

Features Data science

Unburdened by the limitations of mathematically convenient solutions, we can use priors that more accurately capture the real-world behaviour that we expect, or that may be observed in extreme cases. By changing one line of code we can test many priors, so it is easy to run multiple scenarios for more robust analysis.

Data science as storytelling

A step-by-step price testing approach:

1 Decide on prior(s) to be used and the exercise’s business constraints (for example, maximum permissible loading, proportion of book to be loaded).

2 Determine how many quotes must be loaded to produce a statistically significant result for the chosen conversion assumption.

3 After the test is complete, redo the analysis using observed quote volumes and conversion. Produce Bayesian credible intervals and other metrics using one or both groups’ posterior distribution.

Using a probabilistic package, steps 2 and 3 are easy to implement thanks to simple, intuitive syntax. The true value comes from step 1 – choice of prior and the domain expertise driving it. This is where we can leverage the skillset of market-facing stakeholders such as underwriters and product analysts, as well as conversion data.

Collaboration is crucial; stakeholders may be able to contribute insights not captured in internal data. For example, an underwriter may know that a new competitor is likely to join the market. Here, the actuary would revise the choice of prior distribution, reducing mean conversion (to account for lower market share) and increasing variance (as market conditions are less certain).

Another use of probabilistic programming is the hierarchical model (Figure 1). Suppose the underwriter thinks there is a 30% chance that a competitor will successfully enter the market. We can start by simulating this event, which will influence the mean and variance of

FIGURE 1: Documentation under the probabilistic paradigm

our prior distribution (and consequently its parameters), at which point we carry on with our Bayesian analysis as before. We can add any number of steps to the hierarchy (something not feasible by hand), although at some point the additional complexity adds little value.

The ecosystem of packages surrounding PyMC aims to boost accessibility and interpretability –model visualisations with a structure similar to Figure 1 can be produced with Graphviz using a single line of code. Easy-to-use tools and readable code reduce the effort required in the documentation and peer review processes. Technical teams often neglect documentation; the packages showcased here present a path of least resistance for communication with stakeholders, making models and their output interpretable by design.

Figure 1 exemplifies the principle that modelling processes should tell a story. Often, the story is obscured by code or specialised software – but just as the best books resonate with a wide audience, the best models are accessible to many stakeholders to encourage diversity of thought and perspective. While an actuary’s statistical knowledge is needed to guide the analysis, other team members are actively involved.

Another example is the visualisation of the posterior distribution. The first plot in Figure 2 shows the process by which we adjust our prior beliefs using observed data, while the second zooms in on a corresponding Bayesian credible interval. These plots are easily made with the ArviZ package and can be interpreted intuitively. Given the observation, there is a 94% probability that the true conversion rate falls inside the highest density interval (HDI), which picks out the most likely values rather than insisting on equal tails.

Just as the best books resonate with a wide audience, the best models are accessible to many stakeholders

The value-add

PENNY

BRADLEY SHEARER (FIA & CFA) is executive director of Protagion Active Career Management

The power of a statistical test depends on the magnitude of the effect and the sample size. For products with high quote volumes and elastic demand, a frequentist hypothesis test is sufficient to obtain a statistically significant result. However, for low-volume products or those with relatively inelastic demand, we need another approach to make price testing viable. A well-chosen prior in a Bayesian analysis allows us to obtain a significant result while collecting lower volumes of data, as we compensate for the reduced information via distributional assumptions.

This approach can even provide a risk reduction technique for larger products with greater elasticity –testing large proportions of the book or experimenting with large loadings poses a risk due to the impact on a product’s performance (for example, affecting loss ratios). We can also extend this method to test segments (for example, a given region), perform A/B testing (comparing two ratesets more generally, not just comparing percentage loadings) or compare multiple loadings.

Join the Bayesian revolution and add value to your business through accessible and interpretable modelling practices!

DRASTIK is an actuarial student and pricing analyst at ERS

FIGURE 2: Visualisation of the posterior distribution

A long life

The Continuous Mortality Investigation has been helping actuarial work for a century now – as this timeline of its key milestones reveals

For 100 years, the Continuous Mortality Investigation (CMI) –a wholly-owned subsidiary of the IFoA – has carried out regular research into mortality and morbidity experience, producing tools and tables for use as industry-standard benchmarks. Today, it has two key areas of focus:

1 Analysing historic experience and periodically producing mortality and morbidity tables, using data supplied by UK life assurance companies and actuarial consultancies.

2 Considering future changes in mortality experience, largely relying on population data.

The CMI’s impartial analysis of mortality and morbidity is widely appreciated among insurers, reinsurers, pension schemes and others, both in the UK and internationally. The ongoing nature of its analyses provides valuable insight into how experience has developed over time, and acts as a ‘common currency’ within the industry.

The CMI’s success is down to the following:

The support of the insurers and actuarial consultancies that share with their data with the CMI for its purposes. It would not be possible to undertake analyses without this data, the provision of which supports the wider actuarial community

The commitment of around 50 volunteers, who give their time and share their expertise to oversee the running of the CMI and deliver outputs from its five investigation committees

An experienced CMI Secretariat, which carries out the CMI’s day-to-day work. The Secretariat provides continuity and support to the volunteers in delivering outputs and acts as a neutral party in handling and analysing the data submitted

All the firms who subscribe to the CMI for its outputs and information. These subscriptions finance the CMI’s work.

For regular updates from the CMI and to hear about its recent publications, you can follow it on LinkedIn or sign up for its newsletter by logging into your IFoA account online and opting in. The CMI also publishes blogs on the IFoA website and periodically invites subscribers to attend webinars and user groups, providing the opportunity for them to comment on the outputs being discussed.

To access the CMI page on the IFoA website, scan the QR code

1924 THE START OF COLLECTING DATA ON A CONTINUOUS BASIS

This year is taken as the start of the CMI (officially known then as the CMI Bureau, or CMIB), as that is when it began collecting data on a continuous basis. The committee behind it, though, would have met before then, its first chair being William Palin Elderton, who went on to become president of the Institute of Actuaries from 1932 to 1934.

1936 PUBLICATION OF THE CMI’S FIRST MORTALITY TABLES

These first tables, the A1924/29 Permanent Assurances tables, were followed by ‘Light’ and ‘Heavy’ tables, based on the seven lightest and five heaviest insurers’ experiences respectively.

FIRST SECRETARY

Roland Clarke, an actuary working for the Prudential, is appointed the CMI’s first secretary.

1951 FUTURE PROJECTIONS

The first CMI tables to incorporate projections of future mortality rates, the a(55) tables, are published.

1956 A1949/52

The CMI publishes the successor mortality tables to the a(55) annuitant tables, the A1949/52 Permanent Assurances tables.

EXECUTIVE COMMITTEE MEMBER

David Wilkie becomes a member of the CMI’s Executive Committee. He is the longest serving volunteer in the CMI’s history, having stepped down (from the CMI Income Protection Committee), in his eighties, in June 2021.

1968 REPUBLIC OF IRELAND ASSURED LIVES

The CMI begins investigating the mortality of lives assured under policies written in the Republic of Ireland. This preceded the formation of the Society of Actuaries in Ireland (in 1972). At this time, many of the offices contributing data were branches or subsidiaries of UK offices, so were existing contributors to CMI investigations.

1970 INCOME PROTECTION INVESTIGATION STARTS

Then called Permanent Health Insurance, this required the CMI’s constitution to be amended from “mortality” to “mortality and morbidity”.

1991 THE CMI’S FIRST SOFTWARE PROGRAM

The CMI releases its first software, the Standard Tables Program, alongside the ‘80’ Series tables. This was intended to replace the previous practice of publishing monetary values for a range of interest rates – the A1967/70 tables were published in six volumes totalling more than 1,500 pages.

1995 CRITICAL ILLNESS INVESTIGATION BEGINS

…However, the initial volume of data submitted did not make it feasible to produce any meaningful results. A key reason for this was that neither of the two offices that had the highest early sales success had previously submitted data to other CMI investigations.

1999 75TH ANNIVERSARY

This was celebrated at Staple Inn Hall with an event that also marked the launch of the new ‘92’ series tables.

1973 FEMALE ASSURED LIVES

The CMI begins investigating the mortality of female assured lives; previously, few policies were taken out by females. And the first CMI Report is published – before, CMI work was submitted to the Institute of Actuaries in London and/or the Faculty of Actuaries in Edinburgh as sessional papers.

1982 IMPAIRED ASSURED LIVES

The CMI begins collecting data on impaired assured lives for a number of specified groups of diseases, including hypertension, ischaemic heart disease, nervous disorders, diabetes mellitus, respiratory disorders and overweight.

1988 SMOKING STATUS

The CMI begins collecting data differentiated by the smoking status of the policyholder at the outset of a policy. This followed UK life insurers offering premium discounts to non-smokers. The results showed that the differences in mortality were much higher than had previously been recognised by actuaries.

2002 THE CMI’S FIRST ‘WORKING PAPER’

The CMI releases Working Paper 1, describing the “interim cohort projections”. Previously, it provided a single projection basis that was embedded into mortality tables; now, actuaries had to decide between three projections – Short, Medium and Long. This paper was awarded the Memorial Prize by the Institute of Actuaries in 2003.

To date, there have been more than 180 working papers. They were initially intended to complement CMI Reports, which took a long time to produce, but over time became the main vehicle for publishing new CMI work.

Peter Nowell, Colin Kirkwood and David Wilkie at the anniversary event

THE MISSION AND VISION

MISSION: To produce high-quality impartial analysis, standard tables and models of mortality and morbidity for long-term insurance products and pension scheme liabilities on behalf of subscribers and, in doing so, to further actuarial understanding

VISION: To be regarded across the world as setting the benchmark for the quality, depth and breadth of analysis of industry-wide insurance company and pension scheme experience studies

2004 THE CMI SHORTENS ITS NAME

The CMI drops the ‘Bureau’ part of its official name and becomes known as just the CMI.

2006 SAPS COMMITTEE AND THE CMI MERGE

The Self-Administered Pension Scheme (SAPS) Mortality Committee formally becomes part of the CMI. Prior to this, life insurers had submitted all of the data received by the CMI; hereafter, consultancies submitted data for large pension schemes.

2009 FIRST MORTALITY PROJECTIONS MODEL RELEASED

The first version of the CMI Mortality Projections Model, CMI_2009, is released. This quickly becomes established as the industry-standard way to communicate assumptions for future mortality improvements in both financial reporting and underpinning transactions, and is updated annually.

The first CMI tables based on pension scheme data – the ‘S1’ series – are also published. Before, all CMI tables were based on life insurers’ data, even though the tables were often also applied to pension schemes.

2013 BECOMES A LIMITED COMPANY

The CMI becomes a limited company following a major review of its structure and operations. This removes its reliance on voluntary donations, introducing a subscription model for those wishing to access its publications.

2013 COMMITTEES RECONFIGURED

The Life Office Mortality Committee and the Critical Illness Committee are reconfigured into the Annuities Committee and the Assurances Committee, to align better with the internal structure of most life insurers and with volunteers’ expertise.

2018 WINS PETER CLARK PRIZE

The CMI High Age Mortality Working Party wins the Peter Clark Prize for Working Paper 100, which illustrates alternative methods of estimating population exposures at high ages, presents analysis of mortality at the oldest ages and describes a framework for setting mortality rates at high ages for portfolio graduations.

THE COVID-19 PANDEMIC

When publishing CMI_2019, the CMI warned that “Covid-19 could lead to an increased number of deaths in 2020 outside the range of typical annual volatility”. The CMI began publishing weekly mortality monitors, noting more than 6,000 excess deaths registered by 3 April 2020. Worse was to come later in the month, with two weeks each having more than 10,000 excess deaths. In the summer of 2020, the CMI set up the COVID-19 Working Party to consider how to set best-estimate mortality and morbidity assumptions in light of the pandemic.

2021 MORTALITY PROJECTIONS MODEL MODIFIED

This is so that no weight was put on the exceptional data for 2020 in CMI_2020. It similarly put no weight on data for 2020 or 2021 in later versions, but has put some reduced weight on later data. Investigation committees begin monitoring pandemic experience, comparing it with prior years and the general population.

2024 CMI REACHES ITS CENTENARY YEAR

Compiled by Dave Grimshaw (CMI secretary 2006–2021), Vivienne Maclure (current CMI secretary) and Jonathan Hughes (current CMI chair)

Does the demise of DB mean a dystopia for pensions actuaries? Is the clock ticking on careers? Or is it less Orwell, more all well? Mark Williams presents his vision

et’s imagine life in 2040. Has artificial intelligence rendered accountants and taxi drivers obsolete?

Is there a permanent base on the moon? Are suits and ties only found in dressing up boxes? Has global warming reached 1.5°C? Is the Olympic Games taking place via PlayStation 8? Are we all vegan? Is Gregg Wallace still, inexplicably, hosting MasterChef?

And most importantly: what is working life like for a pensions actuary in 2040?

Diminishing returns?

The demise of actuarial work in relation to defined benefit (DB) pension schemes has been predicted since before the turn of the millennium.

The closure of most private sector DB schemes put a finite life on that branch of work for many of us. However, the pace of change was wildly overestimated; in the actuarial discussion forums of that era, it was commonly asserted that all DB schemes would wind up within a decade or two.

Far from declining, though, the number of actuaries working in pensions has been fairly stable – indeed, more than half of scheme actuaries are under 50 years old. Moreover, pensions actuaries’ expertise is firmly in demand, as evidenced by the scramble for resources in the recruitment market, overflowing meeting

agendas, and strong revenue growth reported by employers across the industry. There is a reasonable argument to be made that pensions actuaries have never been busier.

The number of DB schemes has fallen by around 25% (Figure 1) during the past 10 years as schemes have bought out in the insurance market, entered the Pension Protection Fund or merged with other schemes. However, the amount of actuarial work required for each scheme has multiplied. An analysis of trends in effort expended and an informal survey of pensions actuaries suggests that this increase is around 50%-75% on average, more than offsetting the reduction in the number of schemes and explaining the current resource squeeze.

What’s driving this high workload? It’s not traditional actuarial work, such as funding valuations – if anything, the amount of that work has shrunk because schemes are further advanced in their funding strategies and core actuarial calculations are more efficient. Instead, the factors contributing include:

Complexities due to volatile economic conditions (such as high inflation)

Guaranteed Minimum Pension (GMP) equalisation

Increased involvement in other ‘pseudo-actuarial’ projects, such as data cleansing

Risk transfer projects, such as buy-in transactions Navigating the technical complexities of run-offs.

KEY: Actual Trend line (overall best fit) Trend line (based on 2018-2022) Trend line (example acceleration)

FIGURE 1: Number of UK DB pensions schemes

Not only has all of this created a larger workload, but the pivot towards project work has also led to more diverse and challenging workloads, providing a fantastic foundation for career development.

The future of pensions

What of the future? To understand, we need to consider how DB schemes could evolve.

Pensions actuaries’ workloads remain high, but we may have reached an inflection point: DB liabilities are likely to have peaked, and increasing buy-out funding positions led to a record year of insurance transactions in 2023. Figure 1 shows three progression scenarios for DB schemes – based on data from The Pensions Regulator (TPR) – indicating there could be around 3,500 schemes remaining in 2040 in a moderate scenario; two-thirds of the current number.

However, the current high workload is set to continue until and beyond 2040, with GMP and data projects only just getting started, the introduction of the government’s new Funding and Investment Regulations, TPR’s new funding code and pensions dashboards, and the inevitable spread of risk transfer work over a long period.

Additionally, the schemes remaining in 2040 will include many of the largest and most complex, including open schemes (which still represent around 30% of all DB schemes, according to TPR statistics). All this points to the effective number of DB schemes remaining, in ‘2012 workload terms’, being more than 5,000.

Clearly it is possible that the decline in scheme numbers may further accelerate, for example if the government’s consolidation drive gains momentum. However, even if this occurs, the liabilities still need to end up somewhere – and wherever that is, actuaries’ services will be required. This means significant opportunities and an extensive career lifespan for pensions actuaries. A survey conducted as part of the IFoA’s ‘Pensions and a changing economic outlook’

The pivot towards project work has led to more diverse and challenging workloads, providing a fantastic foundation for career development

webinar showed that around 62% of attendees believe most current pensions actuaries will be working in the pensions sector in 2040.

As shown by the IFoA’s Savings Goals for Retirement research (bit.ly/Savings_goals_ retirement), the inadequacy of defined contribution (DC) pensions savings is one of the largest issues facing the UK population, with the savings level required to achieve a moderate retirement income being around one-quarter of pay – unaffordable for many. Actuaries’ role in helping employers navigate this challenge will only grow as whole workforce cohorts reach DC decumulation. Moreover, the Pension Schemes Act 2021 paved the way for collective DC (CDC) schemes, and it is expected that appetite for

such schemes will grow over the period to 2040. Responses to the webinar survey mentioned earlier suggest that DC-related pensions work may account for around 9% of work for current pensions actuaries in 2040.

Generation Z dynamism

By 2040, our industry will be dominated by Generation Z (those born between the late 1990s and the early 2010s) – typically described as entrepreneurial and quick to embrace innovation, change and flexibility. Far from craving certainty and routine, a typical Generation Z pensions actuary will embrace the dynamic career provided by an evolving project-based work profile, fostering an even broader range of transferable skills to apply within or outside the pensions landscape.

In addition to the pensions sector’s increasingly diverse work profile, the IFoA survey points to a wide range of related areas in which pensions actuaries will increasingly apply their skills, including banking, financial advice, consulting, insurance and risk management (including climate change). This aligns with the IFoA’s Vision, Skillsets, Mindsets and Domains strategy, the benefits of which the Generation Z pensions actuary is ideally placed to realise.

Far from being a pessimistic outlook, current and prospective pensions actuaries look set to have strong and varied careers well beyond 2040.

MARK

WILLIAMS

is a member of IFoA Council

TABLE 1: What could a pensions actuary’s working life look like in 2040? One possible vision

ARE WE THE BADDIES?

t’s an uncomfortable question. But with the insurance industry’s motives sometimes treated with scepticism, it’s a question that actuaries need to be prepared to confront.

It’s also a question that the IFoA’s Insurance as a Force for Social Good Working Party has been thinking about, and we have a clear and simple answer: no, we are not the baddies. Indeed, we can go much further than this: throughout human history, insurance has been a critical force for the reduction of human suffering and the promotion of societal progress.

At heart, insurance mitigates negative impacts for those who suffer misfortune. It’s a fact of life that bad things happen – and when misfortune strikes, insurance plays a key role in reimbursing the financial cost or providing an income in the event of a major accident or illness. It can also prevent some of the wider ripple effects: think of the job losses that would follow if a business suffered losses without recourse to insurance.

But the social benefits of insurance are so much greater than this. To explore this further, let’s begin at the beginning.

A brief history

years ago, i

From the earliest days of human civilisation, societies have developed structures that are recognisable today as insurance. Nearly 4,000 years ago, in ancient Assyria, the Code of Hammurabi included provisions under which a merchant seeking a loan to fund a shipment would pay the lender an additional sum in return for a guarantee that the lender would cancel the loan if the shipment sank or were stolen.

In later ancient sources, we see the emergence of the concept of general average: rules were developed under which losses would be shared among traders if a portion of a ship’s cargo needed to be jettisoned to prevent a shipwreck, or if a caravan of merchants were robbed.

In both instances, the aim was to oil the wheels of economic development. Trade was already recognised in the ancient world as a driver of wealth generation but the risk inherent in long-distance land or sea journeys was a potential barrier. And so these early societies developed risk mitigation mechanisms – which we now call insurance – to remove these barriers.

To this day, the modern insurance industry continues to play its role in promoting economic progress. Whether it’s through the direct facilitation of trade or other major capital investments, or simply by freeing up capital that would otherwise be required to protect against risk, it plays a key role in encouraging the entrepreneurship upon which so much of our modern society depends.

Other benefits gradually emerged as the modern insurance industry began to bloom during the 17th century. Insurers did not just compensate losses that had already occurred but also began promoting measures to prevent or reduce potential losses.

For example, the first fire brigades that sprung up in London following the Great Fire of 1666 were private operations run by local insurers. Research indicates that these insurer-owned fire brigades did not limit themselves to protecting their own insured properties; rather they recognised – perhaps through self-interest – the contagion risk of fire spreading to their own policyholders.

Around the same time, the new marine insurance market, centred around Edward Lloyd’s Coffee House, was to play its own role in the development of marine safety. Lloyd’s underwriters were instrumental in popularising the use of the Plimsoll line on ships – a simple innovation that ensured ships did not leave the dock dangerously overloaded.

In a similar vein, we see modern insurers playing an active role in promoting risk management, from fire safety measures as a condition of insurance to behaviour-changing technological innovations such as telematics (‘black’) boxes that incentivise safer driving.

As the world entered the modern era, a new wave of innovation saw the principles of insurance applied for mass social benefit. The 19th century saw the creation of workers’ compensation, sickness insurance and state pension schemes, which evolved during the 20th century into the modern welfare state.

In the 21st century, insurance is again at the forefront of social development, with micro-insurance schemes making insurance accessible to some of the world’s poorest citizens. Disaster relief is also being revolutionised by new forms of parametric insurance, which triggers payments automatically to ensure cash can arrive immediately where it’s most needed.

When insurance gets it wrong

Notwithstanding the benefits of insurance (see The big four, facing page), we must acknowledge that insurers don’t always get it right. There are some examples where insurers could do better by customers:

Poor service – With complaints of mis-selling or poor customer service, justification can sometimes be hard to find. As an industry, failings like this reflect poorly on us all, so we have a collective interest in treating customers fairly Communication breakdown – Many people declare a preference for the lowest possible premium, others lament that they’ve had a claim denied because of a policy exclusion. But how often do people spot the connection? There’s a balancing act to strike between the lowest price and maximum coverage, but there is perhaps more work to do in explaining these trade-offs to the public

Unfair pricing – Who should pay more and who should pay less? Should loyal customers be charged more than those who frequently switch insurers? And is it fair if higher risk policyholders are consistently charged more than individuals who are deemed lower risk?

Buying an expensive car is an active choice, so may merit a higher premium. But living in a high crime area when nowhere else is affordable is not an active choice

This last issue generates perhaps the toughest dilemma. As we stand on the threshold of an artificial intelligence revolution in pricing, with insurance premiums becoming ever more personalised, these questions will become more pertinent.

Let’s dig a little deeper.

The perils of personalisation

Our actuarial DNA instinctively guides us to seek the most risk-reflective pricing possible. After all, an insurer that doesn’t distinguish sufficiently between risks will quickly find itself at a competitive disadvantage, falling victim to the downsides of anti-selection.

So, risk-reflective pricing is always best, right? Not necessarily. We must consider the customer perspective. How does an insurer respond to someone who discovers they have a genetic predisposition for developing cancer? Equally, how does the industry respond to an individual on a low income who lives in an area with a high crime rate? Insurers need to consider the outcomes for social solidarity in situations such as these, where individuals may be charged a higher premium that is viewed as commensurate with the underlying risk.

All of which leads us to a follow-up question: can we develop a framework that allows us to address the pros and cons of personalisation? These three principles can help us do just that:

1 Risk rating versus risk pooling

We need to acknowledge the tension between two conflicting objectives: on the one hand, a desire for ever-more-accurate risk rating; on the other, the traditional insurance principle of risk-pooling, which can allow for cross-subsidies between risks. Cross-subsidy ought not to be anathema to the insurance world: it’s a concept that’s already implicitly acknowledged in the prohibition of discrimination in pricing, and in schemes such as Flood Re in the UK, designed to ensure the availability of affordable flood insurance.

2 Behaviour versus luck

We need to know when, and when not, to personalise. Perhaps the guiding feature should be one of agency. Are we distinguishing between behaviours, which is a matter of policyholder choice, or between good and bad luck, over which policyholders

THE BIG FOUR: MAIN WAYS INSURANCE BENEFITS SOCIETY

1 Mitigating impacts for those who suffer losses

2 Oiling the wheels of the economy by encouraging entrepreneurship

3 Promoting risk management measures

4 Facilitating scientific and social development.

have no control? Buying an expensive car is an active choice, so may merit a higher premium. Conversely, living in a high crime zone when nowhere else is affordable is not an active choice, and ought not to be penalised with higher insurance costs.

3 A role for regulation

We must recognise that these principles may run contrary to competitive pressures. A single insurer that acts alone to apply cross-subsidies will quickly find itself the victim of anti-selection. All insurers must act together if such outcomes are to be averted –and realistically, such collective action can only happen if and when mandated by the regulators. Price walking in the UK could be effectively tackled only with regulatory intervention, and the same applies for the less healthy implications of personalisation.

A light on the horizon

The emergence of new technology brings ethical challenges and threats – but also an opportunity to innovate. The examples below are but a small selection of current initiatives that are already making a difference, each of which could merit an article of its own.

Together they give a flavour of what’s possible.

The industry is narrowing the insurance gap through micro-insurance programmes, extending coverage to vulnerable communities most exposed to natural disasters, which have historically had no access to such cover. Parametric insurance is being applied to disaster relief efforts, enabling cash to reach the places where it’s most needed as quickly as possible

Insurers are embracing a role as a positive force for the environment, prioritising underwriting capacity to focus on sustainable risks, and steering away from more harmful activities. Meanwhile, the claims process is being re-engineered to prioritise repair over replacement, to implement eco-friendly recycling or disposal practices, and to build back better in post-claim reconstruction – for example by replacing carbon-based technologies with renewable alternatives

Insurers are increasingly adopting a customer-centric approach, leveraging artificial intelligence to revolutionise their services while image recognition software expedites claims assessment and settlement. Telematics technology is incentivising safer driving habits

Insurers are fostering public and private sector partnerships to share industry expertise, boosting awareness of risk mitigation opportunities. By aligning their strategies with societal and environmental considerations, insurers aim to serve society, people and the planet better.

If insurers can successfully capitalise on the opportunities for positive innovation, and avoid the pitfalls that can come with new technology, then its future contribution to the social good can be as powerful as it has been during the past 4,000 years.

Thanks to Insurance as a Force for Social Good Working Party members Jamie Brennan, Mathilde Haran, Rachel O’Connell and Richard Winter.

STEVEN FISHER

is a director at Deloitte and chair of the IFoA’s Insurance as a Force for Social Good Working Party

Features

Data science

Efficiency model

Data science isn’t just about machine learning models –pricing teams can use it in so many other ways to streamline their work processes, says Ralph Clayton

In insurance pricing, the term ‘data science’ is generally associated with the use of machine learning models. In other industries, though, data science is a very broad term, its core aspects being the use of data to aid or facilitate business decisions through analysis or predictive models. This is essentially what a pricing team does; pricing teams were data scientists before the term existed.

Data science is a fast-moving field that has changed rapidly in the past 10 years as more and more businesses sought to use data. This has resulted in significant investment into new tools and techniques that aim to improve the data science workflow.

Despite this, pricing teams’ approaches have remained relatively unchanged. Given the overlap in underlying concepts, there are plenty of opportunities to look at data science teams in other industries to find better practices beyond modelling techniques.

Data science in other industries

The past decade has seen an explosion in the use of data science workflows in

other industries, with tech giants such as Netflix, Amazon, Google, Uber and Airbnb becoming known for their complex machine learning systems.

Compared with the rest of the workflow, modelling is relatively straightforward. The most powerful modelling tools are open source and can be used fairly easily. The main challenges lie with the rest of the infrastructure, managing, storing and processing large volumes of real-time data – and then, once a model is built, validating, deploying and monitoring it.

Many techniques, tools and best practices have been developed to meet this demand. A lot of them originate in software engineering and focus heavily on parallelism, automation, collaboration, continuous integration and delivery, testing and monitoring, all while ensuring processes are as simple as possible to ensure understandability and maintainability. These practices are common in other industries, but are rarely seen in insurance pricing.

Automated data pipelines

Data processes upstream of pricing teams will involve automation, batch jobs and automated testing, along with efficient database and pipeline design. When data is further processed within pricing teams, though, these practices are rarely kept in mind: many processes involve manual steps, analytical projects built from scratch in an ad hoc manner, duplicated work and pipelines, all increasing the workload complexity.

If a data pipeline can be automated upstream, it is also feasible to automate processes within the pricing remit –data for reporting, analysis, modelling, quote batches and any other data produced can be scheduled as jobs, with many pipelines able to share the same core transformations.

RALPH CLAYTON is a pricing/data science consultant at Pricing Frontier

It is worth investing time in automating processes, as it saves time in the long run and reduces opportunities for human error. The ability to share elements across multiple pipelines also reduces duplication and aids in data reconciliation.

Pricing teams commonly default to Excel for data processing and transformation. However, while Excel is easy to set up, it often requires manual steps (again, increasing the chance of human error), is hard for others to follow, and is difficult to implement for other projects. Using a programming language such as Python means processes can be automated and reusable modular tools and functions can be built.

Documentation

High quality project documentation can save a lot of time when analysts are working on an existing project, replacing meetings or time spent figuring out how a process or project was developed.

Well-documented processes and codes also clarify what the process is doing, allowing other analysts to maintain and improve on it easily.

Documenting data is also useful: a data catalogue enables analysts to see exactly what the data contains, making it easier to find data and understand the data lineage and context.

Repeatable analytics vs ad hoc analysis

Data analytics projects are often built ad hoc, with aspects built from scratch, data assembly performed manually, and many components unusable elsewhere as they have been built specifically.

Inefficiencies become apparent quickly, especially when manual tasks need to be repeated during iterations or reviews. Designing projects to be repeatable from the outset allows for easier updates and reassessments, ensuring findings remain valid over time but also meaning the project can be adapted for additional uses. This is best achieved through automation and code rather than manual processes.

Data analytics teams should aim to develop a system that provides analytics or makes it easy to carry out analytics. If an ad hoc piece of work is required that the system doesn’t currently enable, look to develop the functionality into the system rather than doing it as a standalone project. This further improves the system so that similar projects in the future can reuse the functionality.

Version control

With repeatable analytics in mind, using version control allows for better

Features Data science

collaboration across codebases and provides an audit trail of data analysis –an important concept in a highly regulated industry.

When multiple analysts are working on a project, multiple file versions are often created – on which only one analyst can work at a time, requiring them to save and close the file if another analyst needs to work on it. This makes it easy to lose changes, means changes have to be rebuilt into new versions, and generally makes it hard to keep track of development.

By using Git when developing code, analysts can develop on the same codebase at the same time and integrate changes. It also provides a full version history, tracking what changes occurred, when, why and by whom, and includes the ability to roll back changes.

Data can also be version controlled – data changes over time and it may be useful to reproduce analysis using the same data from which the original analysis was produced; Delta Lake is an open-source framework that supports this functionality.

Models can also be version controlled using tools such as MLFlow, which not only have version histories of registered models but also track each experiment during the model building process, again providing a complete audit trail of analysis.

The cloud

Many data-heavy companies are migrating to the cloud. There are pros and cons to this, and different costs to evaluate, but on the cloud it is generally easier to implement better data practices. Platforms such as Databricks are all-in-one data science platforms that include tools for all the aspects discussed earlier.

Many pricing teams think data science is simply using machine learning techniques for modelling. While it is true that data science teams in other industries will often using machine learning algorithms, there are many other tools, techniques and practices that could vastly improve pricing teams’ efficiency, beyond modelling capabilities.

The power of two

Digital twin technology is set to soar in use.

Neha

Agarwal and David Basson explain the energy it could bring to the insurance industry

Digital twins have been hiding in plain sight.

The Covid app, for example, uploaded its data to a digital twin that allowed decision-makers to identify virus patterns, replace assumptions with evidence and understand the implications of proposed policy changes. When building a smart city, you would first build a digital version – by digitally simulating traffic flows, for example, we can work out how to reduce wrong-turn collisions. And if you were carrying out complex surgery, you would want to do it digitally before doing so in real life.

Insurers are rapidly catching on to digital twins – because whether you’re plotting a route out of Covid, planning a smart city or creating a new product, it’s crucial to identify and reduce risk.

What are digital twins?

A digital twin is a virtual representation or simulation of a physical object, process or system; it contains all the data and characteristics required to mimic its real-world counterpart, including behaviour, performance and interactions. It provides an advanced visualisation through which you can run what-if scenarios and simulate processes or risks –generating synthetic data, training and testing artificial intelligence algorithms, and making predictions.

Even the largest vaults of historic data have gaps. The impact of climate change, for example, means that the past is an unreliable predictor of the future: parts of Greece and Italy that have never burned or flooded before are now burning and flooding; Australia’s warming climate means that snakes are no longer entering brumation (a dormant period for reptiles that is similar to hibernation in mammals).

We live in a world that is abundant in data – whether from SCADA systems, the Internet of Things (IoT), GPS, social media, drones, satellites or other sources. Digital twins are

able to ingest this abundance of data and combine it with historic data to provide more accurate predictions.

Levels and maturities

Digital twins operate at different levels and maturities (Figure 1):

Level 1: Descriptive – This focuses on the visual representation and engineering analysis of existing systems. Think visual mastery – 2D/3D wonders and virtual reality models helping us to grasp designs, assess sites and train personnel. For example, an L1 DT can conduct system analysis of a wind turbine and digital twin airflow mechanism via virtual reality.

Level 2: Informative – This seamlessly integrates IoT sensor and maintenance data into a dynamic 2D/3D dashboard, providing real-time insights and analytics. It triggers alarms, integrates with enterprise asset management or enterprise

NEHA AGARWAL is co-founder and partner at AInsurCo

DAVID BASSON is business developer and adviser at NTT UK

2D/3D visualisation: System to detailed view Design analysis: Engineering analysis, computational fluid dynamics, finite element analysis etc

DESCRIPTIVE

Visual representation of the physical system (2D/3D/augmented reality/ virtual reality) and design analysis

Use cases: Digital walkthroughs, personnel training, engineering design analysis

IoT connectivity: Operational technology data + anomaly detection + alarms

INFORMATIVE

Asset management: IoT + asset configuration + service history

Integration of IoT sensor data and maintenance information to understand the current state of the physical system

Use cases: Asset monitoring and alarms, asset management, root cause analysis, personnel training

Asset performance management: Operational planning and predictive maintenance

PREDICTIVE

Fleet management: Optimal fleet usage

Predictions of current state of unmeasured quantities or future state under continued operations

Use case: Operational planning and predictive maintenance

Stage of digital representation

resource planning systems, and excels in asset management, rootcause analysis and personnel training. For example, an L2 digital twin can showcase a wind turbine’s gearbox temperature anomalies on an IoT dashboard or offer an augmented reality overlay for technicians during repairs, seamlessly merging data and service history.

Level 3: Predictive – This anticipates a system’s present and future states using predictive models. These models may be entity-specific or fleet-wide, grounded in scientific principles, purely data-driven (machine learning) or a hybrid approach. Key applications include predictive maintenance (asset performance management) and fleet management. For example, an L3 digital twin can use physics-data hybrid models to predict a wind turbine’s main bearing wear (using a virtual sensor) and the remaining useful life of its gearbox.

Level 4: Living – This is the pinnacle of the digital transformation journey, delivering dynamic data-driven insights at individual entity level. Unlike predictive models, an L4 digital twin continuously updates with real-time IoT data, ensuring accurate simulations and enabling scenario analysis. This adaptability helps in, for example, predicting the remaining useful life of a wind turbine’s gearbox, dynamically adjusting parameters such as shaft misalignment and bearing wear for precise scenario analysis and optimal preventive maintenance guidance.

Digital twin insurance applications

How does all this apply to what we do as actuaries? Given all of the above, this is how…

Underwriting – Digital twins can streamline and enhance the underwriting process. For instance, if an applicant applies for a life insurance policy, their digital twin is activated. This can rapidly analyse medical data, family history and lifestyle factors to assess the applicant’s risk profile, providing insights into potential health risks that might affect policy eligibility or premium pricing.

Claims processing – Digital twins are reshaping claims by replicating policyholder data for scenario analysis, expediting assessments and improving decision-making. Real-time monitoring using digital twins

Scenario analysis: DT updatable models for what-if analysis

LIVING (DT)

Features Technology

Actionable insights: Guidance on best course of action

Updatable models for forward predictions to drive actionable insights (prescriptive analytics) at the individual asset level

Use case: Decision support for actionable insights, stepping stone to autonomous systems

enhances efficiency, cuts processing time and boosts customer satisfaction in claims processing.

Customer engagement – Using digital twins, we can analyse and aggregate large datasets to enhance customer engagement and provide better customer service.

Regulatory compliance – Digital twins can streamline and enhance regulatory compliance and reporting processes, leading to more accurate, efficient and transparent dealings with regulators.

Insurance distribution – Integrating digital twins into distribution channels can enhance efficiency, personalisation and customer engagement, leading to more effective distribution strategies and better agent-customer interactions. Imagine a scenario where an insurance agent uses digital twins in their distribution efforts. As the agent prepares for a virtual customer meeting, the digital twin provides insights about the customer’s life stage, preferences and recent interactions. The digital twin can simulate scenarios relevant to the customer’s situation, showing the benefits of various insurance products. It generates real-time quotes and personalised coverage recommendations, ensuring a seamless and engaging customer experience. It also analyses customer feedback and engagement, helping to refine strategies for future interactions. This holistic approach, driven by digital twins, allows agents to connect with customers more effectively, offer tailored solutions and optimise distribution efforts.

Climate change impact assessment – Insurers can use digital twins to assess the impact of climate change on insurance portfolios. By modelling climate-related risks and their potential consequences, insurers can make informed decisions about risk mitigation strategies, portfolio adjustments and product development to address emerging challenges.

Actuaries are in a great position to use digital twins to their full potential. Whether it is to visualise the complex world, present information to stakeholders, run scenarios or build a live model that reflect our organisation’s portfolio, we can use digital twins to boost the value we bring to organisations.

FIGURE 1: Digital twin levels

Features

Data science

There are the different model types; and then there’s making a bespoke blend. Is it worth getting creative? Alisha Dhyani, Alinta Wilson and Antonio Nehme test what’s best in a loss cost modelling exercise

ince its early days, insurance has operated on a foundation of risk prediction. Personal computers and the data age have made it crucial to use statistical modelling techniques to understand the complex relationship between risk factors and the potential loss cost associated with an insurance policy. In this context, generalised linear models (GLMs) have become indispensable, capable of predicting a variety of outputs, such as claims frequency and severity.

GLMs are still relevant in insurance but the new risk factors brought about by automation and new technologies add complexity to their construction. This is particularly relevant in motor insurance, where advanced driver-assistance systems and self-driving features are on the rise.

Machine learning (ML) models, on the other hand, have proven useful in capturing intricate relationships between large numbers of features and dependent variables. Gradient boosting models (GBMs) are gaining prominence in pricing, and there have been initiatives to explore ML techniques further.

Blurring the boundaries

Here, we explore four modelling techniques used to predict the loss cost, comparing GLMs’ predictive accuracy with that of GBMs, artificial neural networks (ANNs), and a hybrid model combining a GLM with an ANN.

Pre-modelling

When building a statistical or ML-based model, the predictions’ accuracy and reliability is significantly shaped by data cleaning and pre-processing.

We used the French motor claims dataset due to its size and relevant attributes for risk assessment; it includes 680,000 rows and has been used in several technical reports and academic papers. In the pre-modelling analysis, we examined the following features:

Driver: driver age, bonus malus

Vehicle: power, age, brand, fuel

Geographic: area, region

Pre-processing measures involved removing duplicates and outliers, setting caps to promote more stable and accurate risk assessment, and correlation analysis to avoid fitting highly correlated factors in the models. We then tailored features for different algorithms through log transformations, min-max scaling and one-hot encoding.

Log transformations involve taking the variable’s logarithm to compress the scale of its data. We applied log transformation to the ‘Exposure’, which is used as an offset term in our models

Min-max scaling scales numerical features to a defined range (commonly [0,1]). We performed min-max scaling on numerical features: vehicle power, vehicle age, driver age and bonus malus One-hot encoding transforms categorical variables into binary (0/1) format. We used this technique on categorical features such as area, vehicle fuel, vehicle brand and region for improved model compatibility.

It should be noted that min-max scaling and one-hot encoding were carried out exclusively for the ANN and the hybrid model.

Finally, we split the data into a 70% training subset and 30% testing subset – the first for training the models and the second for validation to avoid overfitting.

Modelling

We built separate frequency and severity models, and later combined them to form loss cost models. We used the product of the two models’ outputs to calculate the predicted loss cost.

To optimise model performance, we carried out hyperparameter tuning for both GBMs and ANNs. Just as tuning a string instrument can enhance the pitch for notes of a particular song, tuning hyperparameters customises the ML model to a particular dataset.

The hybrid model was able to capitalise on the strengths of both algorithms

ALISHA DHYANI

ALINTA WILSON is lead pricing analyst at NFU Mutual

ANTONIO NEHME is a lecturer in computer and data science at Birmingham City University

Recognising the computational intensity of hyperparameter tuning on the entire dataset, we needed to use a representative 10% subset while ensuring it followed the original data’s distribution. We used grid search and random search with 10-fold cross-validation to identify optimal parameters. Grid search systematically tests multiple combinations from a specified range of discrete hyperparameter values; random search is a more effective way to randomly sample combinations of hyperparameters when the search space is vast. Both approaches assess the impact of the hyperparameter values on model performance, to determine optimal values for the model.

The 10-fold cross-validation process involved dividing the subset into 10 groups (folds) containing roughly the same number of instances, then using nine for training and one for validation over each iteration, recording the error rate for every iteration, and calculating the average error rate for the 10 iterations. The optimal combination of hyperparameter values is the one leading to the lowest error rate. Hyperparameter tuning was executed independently for frequency and severity models.

GLMs

We used traditional GLM techniques to build our first set of frequency and severity models. A Poisson model was fitted for the frequency model and a Gamma distribution for severity. These models were built using R’s ‘glm’ package.

GBMs

GBMs aggregate predictions from weak learners (decision trees) through a voting process to create a final prediction model combining the strength of the learners. The decision trees are trained progressively, with each new learner trying to fix preceding learners’ mistakes. The following parameters were found following the GBM’s hyperparameter tuning.

is a pricing consultant at NFU Mutual

Features Data science

For the frequency model:

Number of trees: 515

Learning rate: 0.01

Depth of the trees: 7

Minimum number of observations in terminal nodes of trees: 5

For the severity model:

Number of trees: 278

Learning rate: 0.01

Depth of the trees: 5

Minimum number of observations in terminal nodes of trees: 5

ANNs

ANNs have sparked fascination in insurance modelling, with their ability to learn and adapt like the human brain. Neural networks aren’t confined to crunching numbers – they also grasp patterns and navigate complexities to arrive at decisions.

Inspired by the intricate design of neurons, ANNs are structured into input, hidden and output layers; the network consists of a sequential connection of neurons. Data is shared across the network, and each link between neurons in the input and hidden layers is assigned a weight.

Adam optimiser was used to determine the optimal number of hidden layers and a random search used to identify the number of neurons and activation function for each layer, as well as the model’s learning rate. The frequency model achieved optimal results with three hidden layers, containing 64, 80 and 88 neurons, coupled with tanh, sigmoid and relu activation functions respectively. Optimal hyperparameters for the severity model were four hidden layers with 256, 32, 72 and 88 neurons respectively; sigmoid was selected as the activation function for the first layer and tanh for subsequent ones. The optimal learning rate for both models was 0.01.

Hybrid model: ANN + GLM

We adopted a hybrid model combining an ANN with a GLM as our final algorithm for constructing frequency and severity models, inspired by the combined actuarial neural network framework proposed by Mario V Wüthrich and Michael Merz (bit.ly/Yes_we_CANN).

We used predictions from our frequency and severity GLMs as an additional input to the ANN described earlier. The aim was to test if additional GLM input could improve the ANN model’s prediction. The model was able to capitalise on the strengths of both algorithms – providing interpretability and transparency through GLMs for regulatory compliance, while leveraging neural networks to identify intricate patterns in the data.

Figure 1 illustrates the structural framework of the hybrid model for frequency.

Results

We evaluated the loss cost models’ performance by computing the mean absolute error (MAE) on the test data. MAE is calculated as the average error between actual and modelled loss cost across test data observations. Figure 2 shows out-of-sample MAE values for the different models.

The hybrid model had the lowest MAE, followed by the ANN, GBM and GLM. This implies that synthesis of ANN and GLM captures more underlying data patterns than individual models. While the improvement in the hybrid models was relatively small compared to the ANN model based on the MAE, there is potential to improve results through further hyperparameter tuning with the additional input of the GLM predictions.

The performance of all the algorithms surpassed that of the basic GLM in predicting loss costs. The study’s objective was to transcend the boundaries of conventional modelling techniques as the insurance industry changes. Exploring and comparing advanced modelling approaches is crucial to stay ahead of the curve and ensure continued innovation and adaptability.

FIGURE 1: Structure of the hybrid model

FIGURE 2: MAE values for ANN, GBM, GLM and the hybrid model

OBITUARY

Bob Willis

In October, at the Society of Actuaries in Ireland’s (SoAI) annual convention, a presentation was made to Bob Willis to mark his recent 100th birthday. He responded with a witty and incisive speech outlining the actuarial profession’s growth in Ireland. It was extraordinarily moving, even – perhaps especially – for younger actuaries who would never have known him.

Bob went to The High School in Dublin and took a first class maths degree with gold medal from Trinity College Dublin. He joined the then Irish Assurance Company as an actuarial student in 1945, working in pensions, and qualified in 1950 – the second person in Ireland to do so following independence. He became deputy general manager then managing director in 1966, a post he held until he retired in 1983.

The Irish Assurance Company was formed in 1936 as a merger of some small Irish life companies and the Irish business of several UK companies; the transformation of this hotchpotch into Irish Life is a tribute to Bob and his team. It was early into pensions, unit-linked business and sophisticated investment management, and became a major training ground for young actuaries and other professionals, producing a stream of talent that had a profound effect on Ireland’s financial services industry.

Bob was a founding member and early president of the SoAI, which began with 17 members in 1972; it now has more than 2,000. This remarkable growth says a huge amount about the quality and dedication of those early actuaries.

Bob achieved much outside of his career. He was chair of Cement Roadstone Holdings, the Rotunda Hospital and Rathdown School, and a pro-chancellor of Trinity College, his alma mater. He was married to Elaine for more than 60 years, with three children, Linda, Neil and Erica. A remarkable man, to whom Ireland owes a great deal.

OBITUARY

Joseph Byrne

On 23 November, Joe Byrne passed away peacefully in his 86th year.

After leaving school in Dublin, he began his career at Irish Life in 1955 under Bob Willis and Geoffrey Rowe. In 1962, he became Ireland’s first non-graduate to qualify as a Fellow of the Institute of Actuaries; that same year he married his wife Breda. They had three daughters, Bridgette, Maria and Jacinta, and three sons, Joe, John and Peter. Joe moved to Irish Pensions Trust in 1966, and was a founding member of the Society of Actuaries in Ireland in 1972. In 1985, he set up Byrne Actuaries, which became the largest independent Irish-owned actuarial consulting firm and employs his three sons and his niece Kathy Byrne.

Joe was well known throughout Ireland as an expert witness and in the pensions community. His interests included playing and coaching senior cricket for Pembroke CC and Leinster CC; he also played Gaelic football. He will be missed by all.

At the back People and society

DEATHS

It is with great regret that we announce the death of the following members. We offer our condolences to their families, friends and colleagues.

David Drysdale Fotheringham, a Fellow who joined in 1964

Yash Paul Sabharwal, a Fellow who joined in 1965

David Edwin Arthur Sanders, a Fellow who joined in 1971

Call for your news… If you have any items for this page, email social@theactuary.com

Bonanza birthday

David Wilkie has celebrated his 90th birthday. A giant in the actuarial field, he is best known for his eponymous stochastic investment model, and his mortality tables and models for income protection insurance and the spread of AIDS. Over his long career, he has worked for many big insurance companies, in actuarial consultancy, in research, and as a visiting professor at Heriot-Watt University. He has a CBE, and gold medals from both the Institute of Actuaries and the Faculty of Actuaries. The University of York is holding a one-day conference in his honour on 11 April (booking now closed). And actuary Cathy Lyn is collating messages for a special book to give him – email your contribution to lcathy88@yahoo.co.uk

Foundation party

Last month, the IFoA Foundation held an event – hosted by Moody’s – to celebrate another year of its work and to launch its 2023 Impact Report At it, two Foundation beneficiaries, undergraduates Jenna Kane and Tegan Banks, recipients of the Actuaries of Tomorrow scholarship, spoke of how the financial help has boosted them, enabling them to focus fully on their actuarial studies and university life.

At the back

Extra-curricular

I’M AN ACTUARY AND... TRAVEL CONTENT CREATOR

Priya Shah, head of UK actuarial oversight at Aspen, shares the images of her natureinspired adventures with the world online

How did you get into travel and what do you love about it?

I’ve lived in London since I was 18 but I grew up in Kenya, where local trips got me craving adventure – as well as Duke of Edinburgh expeditions and David Attenborough documentaries on TV. I love discovering the natural world, both wildlife and landscapes. However, people make or break the experience of a place, so countries with warm, friendly and welcoming people easily become the favourites.

How do you fit your travel around your actuarial job?

I use my annual leave and work four days a week, so I have an extra day I can use to travel and keep on top of editing and managing my social media.

Who do you go with?

Mostly with friends, or with expedition groups where you meet like-minded people. I have also been on press trips, including some with the Czech and Georgian tourist boards. The deliverables for these are different each time –sometimes I promote them on my social media, other times I provide them with photos or videos for their marketing.

What kind of trips do you do, and how many?

I do 10 to 12 a year, a mix of long-haul and European destinations. I have been to 71 countries and to all seven continents. My long-haul trips are quite adventurous and involve nature and wildlife – swimming with orcas in the

Clockwise from left: Priya, and her photographs of polar bears in Canada, the Northern Lights in Canada and the stars in Jordan

Follow Priya on Instagram: @journey7 continents or scan the QR code

Arctic Circle, a snow leopard expedition in the Himalayas… I mostly go where I want to go, but I leave some room to work with sponsors if they’re offering a destination I might be interested in.

How do you plan your trips?

When I’m planning my own, as opposed to organised expeditions or press trips, I do it down to the last detail. Tips come from other travellers and guides, and I use TV documentaries; I try to go where they filmed on location. On my snow leopard trip, I used the same guide and lodge as the crew from Planet Earth II. The guide on my recent orca trip is a senior producer and director for the forthcoming Blue Planet III

What have been your highs and lows?

A low was a road accident in Namibia – we spent our first night in a hospital in the middle of nowhere. There have been hundreds of highs: making eye contact with an orca up close underwater; seeing a full lunar eclipse and the Northern Lights together; New Year’s Day in Antarctica surrounded by humpback whales; a snowy Christmas day on South Georgia island with half a million king penguins; so many!

How did you start posting about your travel?

I used to share photos of places I’d visited on Facebook, and friends would ask about them and then book trips themselves. I thought, why not share it with a wider audience,

The guide on my recent orca trip is also a senior producer and director for the forthcoming Blue Planet III

At the back Extra-curricular

maybe more will be inspired to travel off the beaten track? So I started my Instagram account, Journey7Continents. I began taking it more seriously in 2020 (great timing!), and I’m now building a website for a blog and planning to share more videos on YouTube. I have about 15,000 followers.

It’s competitive out there –what got you noticed?

I think it was the quality of the content. I travel with heavy-duty camera gear and a drone, and focus on getting cinematic images and videos. For many brands, the quality of your work and engagement with followers is more important than follower numbers. I love being able to

capture memories to look back on –although I do make a conscious effort to enjoy the moments with my own eyes, as well as through the lens.

How do you find new editorial angles?

Going to lesser-known places, asking locals and doing research via guidebooks or YouTube. With photography and video, you try to find unique angles, get creative with video edits and draw out small details that often go unnoticed.

What would you be if you weren’t an actuary?

A polar expedition guide, or a travel photographer/filmmaker.

At the back Soft skills

FROM HYBRID TO ‘MYBRID’

Your company wants you back in the office more? You may resent that, but it’s got a point – it helps both them and you, and you can work it to your advantage, says Jenny Segal

We all know the score: employers want us back in the office for more days than we want to be in for.

To reach any kind of middle ground, we need to understand each side’s motivation. For employers, the desire for staff to come into the office is driven by positive reasons, such as collegiality, culture, collaboration and mentoring the next generation. But sometimes it’s not: it can stem from a lack of trust or a need to micromanage. Or ego and status – the desire to parade around the office and oversee the troops. Or a concern that expensive office space is under-used.

For employees, there are very good reasons to work from home: better work-life balance, less time and money wasted on commuting, and the ability to focus uninterrupted on a task that requires concentration and flow. But… some people are using it as an excuse to slack off. And some have become fearful: Covid has made them hermitic and they’ve forgotten that the key determinants of happiness are contact with fellow humans and, at work, feeling part of a team. But the most interesting reason is the assertion that they are more effective when they work from home. To which I declare: your job is not your ‘to do’ list! Your job is to enable your company to succeed.

If we have to work on something that requires deep concentration, perhaps writing a report or coding a spreadsheet, the best thing from a personal efficiency stance may well be to stay at home and dive in until the job is done. And our efficiency may drop to, say, 70% when we are in a noisy office with the threat of constant interruptions.

However, when we are in the office we raise other people’s effectiveness – perhaps because we inspire them or set the benchmark for what deep work looks like. Or maybe we just have a laugh and make everyone feel happier so they work that bit harder. Overall, the office’s effectiveness increases. We may lose out because it takes us 30% longer to get our work done but, from the perspective of the office as a whole, that lost 30% is easily made up for in terms of increasing the effectiveness of those around us. Net-net, the company wins.

The office as organism

More than this, the act of coming into the office breaks down silos by allowing us to interact with each other in an unplanned fashion: an overheard fragment here, a glance at someone’s screen there, a random thought provoked by someone’s trip away or their new shoes. Exponential neural pathways form from these serendipitous interactions. The

office ceases to be a collection of silos, with the interconnectedness of the whole being much more powerful than the sum of its parts, working together like the organs of a pulsing, living being.

These neural pathways are where the magic happens, where innovation and blue-sky thinking occur, where we make the paradigm business leap. And great business leaders know it, which is why they’re trying to encourage us back to the office.

One theory as to why such random interactions increase creative output is based on the concept of entropy which, the A-Level physicists among us might recall, is loosely defined as a measure of the disorder in a system. High-entropy systems are chaotic, low-entropy systems are stable.

Dario Krpan, a behavioural scientist at the London School of Economics, has extended this idea to psychology: humans prefer low-entropy environments that are stable, safe and well understood. When we find ourselves in a chaotic situation, our brains don’t like it and seek to remove the entropy from the situation as quickly as possible, sparking rapid creativity in a desperate attempt to seek safety and survive.

The Work Wheel

Given that spending time with each other in the office is a good thing for both our

At the back Soft skills

Reading

Client meetings

Serendipitous interactions

Brainstorming

Market research Spreadsheets Reading

personal happiness and for business innovation, how much time should we be there for? Some companies dictate numbers of days and which ones, some leave it up to individual teams to decide organically. One thing is clear, though: carrots and sticks don’t work long term. Free pizza on a Thursday quickly gets old, and threats of docked bonuses and no pay rises for dissenters is a surefire way to end up with a workforce that lacks diversity, as those who value their work-life balance move elsewhere. Is there a better way?

Figure 4. Staff and managers alike can take this on board because the beauty of it is two-fold. First, it forces a logical thought process based on the drivers of business success, rather than more emotive factors. Second, its clarity enables the rationale behind the policy to be easily communicated – so important in achieving employee buy-in and winning over dissenters’ hearts and minds.

The post-Covid manager

Client meetings

Serendipitous interactions

Brainstorming

Market research Spreadsheets Reading

Client meetings

Serendipitous interactions

Brainstorming

Market research Spreadsheets Reading

Market research

Client meetings

Serendipitous interactions

Spreadsheets

Whether you’re junior, senior or the HR director deciding the back-to-the-office policy, start by considering what will drive business success. Suppose that, to achieve its targets, a business needs to focus on customer growth and product development, and this will involve employees spending time reading reports, writing spreadsheets, brainstorming ideas, meeting clients and carrying out market research. And we want to allow time and space for some serendipitous magic. Some of these functions will be best performed in the office and some best done at home.

Management has become multi-dimensionally more complex: ensuring that, when people come to work, they are clear about why they are there and what they will do

Now imagine a wheel, where each spoke represents one of these successdriving functions (Figure 1).

Consider how effectively you operate in each of these dimensions when you are working entirely from home. For example, you might rate the ability to focus uninterrupted on reading and detailed spreadsheets very highly, but the scope for serendipitous interactions less so. Mark a dot on each spoke at the appropriate point and join them up to get your Home Work Wheel (Figure 2).

Now repeat the process through the lens of 100% office-based working; it might look something like Figure 3

Brainstorming

If we plan our days so that we spend our time working from home on those tasks nearest the Home Work Wheel’s ‘rim’, and ditto for our office life, we will be very effective indeed, as shown in

And we all need to execute the plan effectively by making sure that, when people come into the office, they are focused about how they spend their time – not coming in to sit in back-to-back Zoom meetings, or to interact with parts of the business with which they have no possible synergies. Or to find that no one else from their team is around.

This takes careful management and will require a new set of management skills. Businesses need to recognise the importance of the management function, selecting the right people for the job and giving them sufficient time to do it well.

Management has become multidimensionally more complex: ensuring that when people come in, they are clear about why they are there and what they will do. Creating an open, welcoming environment that encourages serendipity, where people can exchange ideas, forge workplace friendships and have a great time. This will allow us to rediscover the joys of office life so we actually want to come in more. Work should be fun.

JENNY SEGAL (jennysegal.co.uk) is an actuary, investment professional, non-executive director and motivational speaker. She recently published her third book, Board Effectiveness & Culture

FIGURE 1 The Work Wheel

FIGURE 3 The Office Work Wheel

FIGURE 2 The Home Work Wheel

FIGURE 4 The Hybrid Work Wheel

SEGAL

At the back School of thought

Pioneering attitude

Insurers need to be on the side of the innovators and tech revolutionists when it comes to tackling climate change, says Adeetya Tantia

Climate change grew significantly in the public’s consciousness during the 2010s. While some industries took the lead and are truly transforming the way they do business, others are stagnating. Let’s take a look at two industries that are already on their way to becoming greener and see what we can learn from them.

Transport and energy

Over the past decade, the transport industry – specifically personal-use vehicles – has changed radically. Tesla’s electric vehicles (EVs) paved the way, showing established manufacturers that their business could be disrupted. Legacy manufacturers began to follow in the late 2010s, and now provide much-needed competition and consumer choice in the EV space. EV adoption also required huge infrastructure investments, including new factories and material supply chains, new charging stations and electrical grid upgrades.

With the industry now well on its way to electrification, many believe that the climate change aspects here are resolved. It’s never that simple, though – there are concerns around the mineral extraction and supply

required for EV batteries, as well as such batteries’ lack of reparability. It is hoped that new battery technology, still in development, will solve these issues and make the motor industry truly green.

What can we learn from this? Well, when disrupting an entire industry, there is never just one problem to solve – thousands of smaller issues need to be resolved to keep the mission on track. ‘Unknown unknowns’ are to be expected.

The other industry that has garnered a huge amount of attention is energy generation. Fossil fuel use is a major hindrance to sustainability efforts, and it is hoped that solar and wind power will solve this problem; hydropower has various negative externalities, which is why its development has largely remained stagnant.

Renewable sources usually depend on external factors – the sun for solar, wind for wind farms, rivers and other bodies of water for hydropower. It is not usually possible to build an energy grid for such energy sources, as it would need to modulate supply based on demand. This has led governments to re-examine nuclear: many countries around the world, including the UK, have started building new nuclear plants, learning from

past mistakes around reactor vulnerabilities and waste disposal or potential re-use. These new reactors will cost much more than the reactors of yesteryear but be more resilient and less likely to cause incidents.

Nuclear fusion is in its infancy, although there are already companies trying to make it viable. Its potential benefits are immense –a clean energy source with little to no waste products, and by-products such as helium that have important industrial uses and are otherwise in finite supply.

Heavy industries have long been forced to use fossil fuels to attain the constant high temperatures required to generate chemical reactions. Companies have already started to look at this issue and design solutions to electrify such processes; a few of these solutions are set to go into production by the end of 2024. The difficulty is that each industrial process is different and requires a bespoke solution. The process of making industrial processes more climate friendly will therefore be long and arduous, involving a host of corporations and governments.

The role of insurance

How is all this relevant to insurance?

Well, insurers and reinsurers have already

ADEETYA TANTIA, 28, is a risk modelling consultant at Hymans Robertson. He hopes to qualify this year and go into general insurance

started to price climate change impacts into their models, especially those working in catastrophe and property insurance. Environmental, social and governance claims have also been brought under directors’ and officers’ liability policies.

How will new technologies be regarded on a risk and pricing basis? The answer could help or hamper their adoption.

If insurers can cultivate relationships with companies and gain a thorough understanding of new technologies, companies may be able to take more risks on innovative climate solutions without drastically increasing their insurance burden. Insurers have already proved themselves adaptable with the relatively quick release of products for cyber and drone insurance, helping these industries to grow while protecting customers. Insurers can provide innovators with expertise in risk management and potential points of failure, and educating or incentivising customers can go a long way in helping new innovations come to market and electrifying industries more quickly. The insurers that make the climate transition easier and more reasonable stand to gain immensely.

At the back Puzzles

ARE YOU A MATHS MAESTRO OR LOGIC LOVER?

If you have a mind formaths, logical reasoning, cryptic clues and otherconundrums, send your challenges to us and we will publish the most difficult here.

puzzles@theactuary.com

Knights and rooks Mensa puzzle 865

A knight is positioned on the centre square of this ‘chessboard’. Move the knight to each square once only, collecting letters to spell out five types of birds. What are the birds?

At the back Puzzles

England XI Member puzzle 38

Across

1 Join an old salt (6)

4 Charlie with unfinished badge to collect (8)

10 Yob from loud home, almost out of control (7)

11 Customers without advice ran shop haphazardly (7)

12 Full of cargo from US city lair (5)

13 Collection of balls stolen and passed outside (9)

14 Excellent Scottish city announced (8)

15 Penny off cost for food (4)

19 Rubbish charger? (4)

Down

1 He worked with 7 cavities found underneath South Carolina (7)

2 In the past, rum and ale mixed quite a lot (1,4,4)

20 Part of London – Heston, say (8)

24 Pick of the bunch? (3,6)

26 Important game plan I formulated, only partially in retrospect (5)

27 In Hong Kong, dancing topless used to work anywhere (7)

28 Concoctions in cat food discovered by US tax authorities (7)

29 Liberators secure ground before rest, regularly (8)

30 Craftsman with legal claim (6)

7 Liberal boarding sponsor’s jet (5)

8 Change of scene’s including English nature (7)

9 Appropriate? I’m 14! (7)

3 Considering everything 50:50, unopened flavouring rejected (3,2,3)

5 Ship carrying quality gems (6)

6 Emigrants quietly engaged in fiddling taxes (6)

Taking ages Mensa puzzle 866

The combined age of Andre and Matheus is 43. The combined age of Matheus and Oscar is 51. The combined age of Oscar and Andre is 46.

How old are Andre, Matheus and Oscar?

16 Popular, and popular online, but without right meaning (9)

17 Someone who fleeces clients, finally before judge? (7)

18 More calculating corruption of first race,

not second (8)

19 He may kill many, except one-named singer? (7)

21 Sell nut crackers: there’s nothing in it (4,3)

22 Frantically eat a Gu dessert (6)

23 Infantryman, say, like missing one in conflict (6)

25 Biden to end up outside capital of Uruguay (5)

Across: 1 Seaman, 4 Assemble, 10 Hoodlum, 11 Orphans, 12 Laden, 13 Overtaken, 14 Sterling, 15 Rice, 19 Bull, 20 Charlton, 24 Top banana, 26 Final, 27 Hotdesk, 28 Elixirs, 29 Rescuers, 30 Wright

38:

Courtesy of Prime Answers –Member puzzle

Down: 1 Scholes, 2 A good deal, 3 All in all, 5 Stones, 6 Expats, 7 Black, 8 Essence, 9 Impound, 16 Intending, 17 Shearer, 18 Craft ier, 19 Butcher, 21 Null set, 22 Gateau, 23 Walker, 25 Potus Mensa puzzle 865: Eagle, robin, raven, quail and heron 866: Andre –19, Matheus –24 and Oscar –27.

| LEEDS | MANCHESTER | ROTTERDAM | MUNICH | FRANKFURT | HONG KONG | SHENZHEN | SINGAPORE

Location: London

Salary: Up to £140k + Bonus & Benefits

We have partnered with a leading Lloyds business who are looking for a Capital Actuary to

This

c.wright@gravitasgroup.com | 07765 134 727

Non-Life Consultancy Consulting Actuary

Location: London

Salary: Up to £70k + Bonus & Benefits

We are working with a well-known boutique consultancy who are looking to add a part qualified non-life actuarial student to their team. The role offers breadth and diversity of work across the Lloyd’s and London Market. You will have the opportunity to work with high-profile clients across reserving, risk, capital and pricing. You must have experience within the non-life space and be making good exam progress.

l.lanigan@gravitasgroup.com | 07442 577 862

Non-Life Contractors

Location: Hybrid/Remote

Rate: £750 - £1,600/day, Inside/Outside

FTC: £150k - £250k + Benefits

The demand for interim general insurance actuaries has seen a rise over the years and shows no sign of slowing down. We continuously receive high volume of requests across, actuarial transformation, temporary coverage for paternity/ maternity leave, and ongoing support for regular business operations, across Reserving, Pricing, Capital, and Risk functions. If you are nearing the end of your contract or considering transitioning to the interim market, feel free to reach out! rupa@gravitasgroup.com | 07543 176 00

OUR TEAM

Pensions Risk Transfer

Transaction Leader

Location: Hybrid/London

Salary: Up to £100K + Bonus & Benefits

Exciting opportunity in the PRT space! Leading consultancy seeks senior consultant for Transaction Leader role. Spearhead pension transfer initiatives including buy-ins, buy-outs, and swaps. Requires settlement transaction experience and technical expertise. Senior Consultant level, suitable for FIA or CFA certified individuals with 3+ years PQE. Multiple office locations across England & Scotland, hybrid working model.

Pensions Consultancy Scheme Actuary

Location: Scotland

Salary: Up to £150k + Bonus & Benefits

Pensions Consultancy Actuary

Location: London

Salary: Up to £75k + Bonus & Benefits

A specialist Pensions Consultancy is looking to bring on an experienced pensions professional with a scheme actuary practising certificate to provide expert strategic advice for pension schemes, and strong valuation and risk assesment knowledge. This role would suit someone with a vast consulting background and track record of project delivery. You must be a qualified Actuary and have a minimum of 6 years of post qualified experience.

a.gryson@gravitasgroup.com | 07523 342 006

e.nicholson@gravitasgroup.com | 07496 755 470

A prestigious Consultancy are keen to recruit a Senior Risk Transfer Consultant to join their team in London. Ideally, you will be qualified or close to qualification with relevant UK pensions experience and recent experience in Risk Transfer work. You will get the opportunity to work with Senior Advisers, delivering innovative strategic solutions to wide-ranging clients, becoming a specialist in this field.

Leah Lanigan er

k.newton@gravitasgroup.com | 07842 368 630

Karen Newton

LONDON

YOU IN THE LOOP

As a professional, you’ll no doubt want to keep up with the latest industry developments, people and news?

That’s why The Actuary’s weekly email alert brings you a handy round-up of only the most relevant news stories and comment, straight to your inbox every Thursday.

Register for weekly email newsletters at www.theactuary.com

Browse www.theactuaryjobs.com and www.theactuaryjobsasia.com the official jobsites of the actuarial profession. www.theactuaryjobs.com

SENIORRESERVINGACTUARY(12MONTHFTC)

London,upto£120,000

ALloyd’s/LondonMarketinsurerislookingtohireaSeniorReserving Actuaryona12monthfixedtermbasis.Thisisanexcellentopportunity togainexposuretodifferentterritoriesandlinesofbusiness.They requireFIAindividualswhocanstartwithinthenext6weeksandwho haveawealthofreservingexperience,ideallywithResQ.

Contact: hannah.turner@eamesconsulting.com|02070923249

FINANCIALLINESPRICINGACTUARY

London,c.£85,000

WeareworkingwithaleadingLondonMarketcarrierinthesearchfor aPricingActuaryfortheirglobalteam.TherolewillfocusonCyberand CommercialP&Iandwillhaveagoodmixoffront-facingandtechnical elements.OurclientisopenforanyGIbackgroundforthisrole.The preferencewouldofcoursebepricingbutalsohappytoconsider candidatesfromcapitalmodellingandlastly,reserving.Ideallylooking forcandidateswhoarenearly-qualified.

Contact: rafaela.fakhre@eamesconsulting.com|02038465909

RISKACTUARY

London,£95,000+bonus

Ifyou’reanewlyqualifiedactuarywithapricingrolereserving backgroundandinterestedindiversifyingyourexperiencewithinrisk, thisisanexceptionalopportunitytodoso.MyclientisaglobalP&C insurer,withanexceptionalattitudetowardsthedevelopmentoftheir actuarialteam.Theroleitselfwillinvolveriskoversightofreservingand pricing,reinsuranceandassetallocation,modelparameterisation,and validation.

Contact: sam.baker@eamesconsulting.com|02070923230

RESERVINGLEAD

London,£150,000+bonus

EamesConsultingarecurrentlypartneredwithaleadinghome&motor team,whohaverecentlyundergoneamajortransformationproject withintheirreservingteam.Subsequently,they'relookingtohirea qualifiedactuarytosupportarangeofreservingprojects.Thepurpose ofthisroleistoleadasignificantsectionoftheReservingfunction, determiningthestrategicdirectionfortheteam.You'lladviseon actuarialandbusinessplanning,andwillleadtheimplementationand deliveryoflargecomplexprojects.Theidealcandidateisaqualified reservingactuarywithastrongbackgroundinSolvencyIIandIFRS17. Strongstakeholder-facingexposureisalsodesirable.

Contact: sam.baker@eamesconsulting.com|02070923230

CAPITALACTUARY

London,c.£100,000

Ourclientisawell-knownLondonMarket/Lloyd’sinsurerwhoare lookingforaCapitalActuarytojointheirmid-sizedteam.You'llwork closelywiththeHeadofCapitalModellingandthereisanopportunity tomanage1or2analystsifyouwouldliketostepupintoamanagerial position.Wearelookingforcandidateswith5+years’experiencein anyGIbackground(Pricing,Reserving,Capital/ERM)–openforthose withexperiencesoutsidetheLloyd’s/LondonMarketspaceaswell.

Contact: rafaela.fakhre@eamesconsulting.com|02038465909

LONGEVITYPRICING/LONGEVITYRESEARCHACTUARY

London,upto£100,000

AleadingreinsurerislookingtostrengthentheirLongevityteamwith additionalhiresintheirPricingandResearchteams.ThePricingteam offersbroadresponsibilitiessupportingtheabilitytowriteprofitable newbusiness-leadingquoteproduction,improvingpricingprocesses andtools,anddevelopingnewproductsandpropositions.The Researchteamfocusesonsupportingtenderpricingandmanaging longevityriskbyleadinginnovativeresearchprojectsanddeveloping pricingbasesforpensionschemeandannuitybusiness.Keentospeak tonearlyorqualifiedactuarieswithlongevitypricingorresearch experience,orabackgroundinpensionde-risking,andtheabilityto buildinternalandexternalrelationships.

Contact: jo.frankham@eamesconsulting.com|02070923263

ACTUARY(RESERVINGANDPRICING)

London,£competitive

AmarketleadingLloyd’s/LondonMarketInsurerislookingtohirea qualifiedactuaryinarolewhichwillgivepricing,reservingand businessplanningexposure.ThiswillbefocusedonCasualtylinesof businessandyouwillbecomeanexpertinthislineofbusiness.Youwill beinvolvedincasepricingandmanagingthequarterlyreserving.They arekeentospeaktoindividualswhoenjoyunderwriterinteraction.

At the back Appointments

Contact: hannah.turner@eamesconsulting.com|02070923249

SPECIALTYRESERVINGANALYST

London,c.£50,000

WeareworkingwithaglobalLondonMarketinsurerinthesearchfora part-qualifiedactuarytojointheirreservingteaminthespecialty division.Thisrolehasaglobalremit.Weareopentospeakingwith candidateswithanyactuarialbackground(includinglifeandpensions) whoarekeentomoveintoGI.

Contact: rafaela.fakhre@eamesconsulting.com|02038465909

SENIORPRICINGMANAGER

London,£100,000+bonus

Areyoulookingtogainautonomyandtakeownershipoverapricing function?I'mlookingtospeakwithtechnicallystrongpricing professionalswhoareinterestedinjoiningagrowingcommercial motorinsurer.Inleadingtheend-to-endpricingprocessforaselect product,you'lldevelopanddeliveronpricingstrategyandtechnical enhancement,liaisingwiththedatascienceteamtodriveR&D objectives,andpresentingonportfolioperformancetosenior leadership.Theidealcandidatehasstrongpersonal/commerciallines pricingexperience.AbackgroundinEmblemandRadarisbeneficial, althoughnotessential.

Contact: sam.baker@eamesconsulting.com|02070923230

SENIORPRICINGACTUARY

London,£competitive

ALloyd’s/LondonMarketinsurerislookingtomakeaFIAhireintheir teamtofocusontheirpricingtransformationproject.Thisisan excellentopportunityforanindividualwhowantstobeattheforefront ofinnovationandwhohasakeeninterestinbuildingouttheirpython skills.

Contact:hannah.turner@eamesconsulting.com|02070923249

ANALYTICSACTUARY

London,upto£90,000

Wehaveanexcitingopportunityforatechnicaldatapricingor researchactuarytojointheAnalyticsteamofaspecialistemployee benefitsconsultancy.Thebusinesshasexperiencedrapidgrowthover thelast5yearsacross,offeringbespokeandcreativeproductstoa globalclientbase.Lookingforsomeonewithwhocanbalance technicalstatisticalanalysisexpertiseandstrongcommercialskillsabletonotonlyleadthedevelopmentoftechnicalsolutionsbutalso presentthemtoclients.IfyouhaveabackgroundinLife,Health/PMIor EmployeeBenefitsandapassionforinvestigatingexperienceusing dataanddevelopingnewmodelsthenpleasegetintouch.Experience inR/PythonprogrammingandPowerBIwouldbewelcomed.

Contact: jo.frankham@eamesconsulting.com|02070923263

SENIORMANAGER,TECHNOLOGY

London,Edinburgh,Bristol,upto£125,000+carallowance Aglobalconsultancyisseekingaqualifiedactuarywithawealthof experienceinthedesignandimplementationoftechnical/modelling solutions.Thetechnologyteamisagrowing,creativeareaofthe businessspecialisinginthedevelopmentofinnovativetechnology solutionsforclientsintheactuarialspace.Thisisakeyhireforthe team,leadingkeytechnologyprojects,leadingtheteamontheend-toendtechnicaldesignandkeepingabreastofmarkettrends.Youwill workcloselywithseniorstakeholdersandindustrySMEstounderstand theirexistingissuesanddesignbespokesolutionstomeettheirneeds. Theidealcandidatewillbeamodelling/technicalexpertwithan establishednetworkandpeople-managementexperience.

Contact: jo.frankham@eamesconsulting.com|02070923263

At the back Appointments

ACTUARIAL DIRECTOR - CONSULTANCY

Qualified Award-Winning Global Firm

NON-LIFE LOND / MANC / EDIN

LONDON MARKET LEADER

STAR8608

Qualified Award-Winning Consultancy

NON-LIFE LONDON / HYBRID

LONDON MARKET PRICING

CHIEF RISK OFFICER

Qualified Specialist Insurer

NON-LIFE LIFE HEALTH SOUTH WEST

COMMERCIAL LEADER

STAR8527

Qualified / Part-QualifiedLeading-Edge Firm

NON-LIFE LONDON

BPA PRICING ACTUARIES

STAR8653

Qualified Major Insurer

LIFE FLEXIBLE / HYBRID

LONGEVITY PRICING

STAR8711

Qualified Major Firm

NON-LIFE LONDON / HYBRID

HEAD OF ACTUARIAL SYSTEMS

STAR8667

Qualified Major Insurer

LIFE MIDLANDS OR SCOTLAND / HYBRID

GROUP CAPITAL ACTUARY

STAR8569

Qualified Global Reinsurer

LIFE LONDON / HYBRID

STRATEGIC CONSULTANCY

PRICING MANAGER

Qualified UK Insurance Group

NON-LIFE REMOTE

ACTUARIAL TRANSFORMATION

STAR8674

Qualified Leading Global Consultancy

NON-LIFE LONDON / HYBRID

LIFE INSURANCE LEAD

STAR8512

Qualified UK Leader

LIFE LONDON / HYBRID

LEAD A VARIED LIFE

STAR8676

Qualified Top-Tier Consulting Firm

LIFE WIDER FIELDS LONDON / HYBRID

LONGEVITY ACTUARY

STAR8609

Qualified / Part-Qualified Global Leader

LIFE PENSION LONDON / HYBRID

STAR8603

SENIOR ACTUARY - CONSOLIDATION

Qualified Market Leader

PENSIONS NATIONWIDE / HYBRID

STAR8663

TECHNICAL PENSIONS - PART-TIME

Qualified Specialist Consultancy

PENSIONS LONDON / SE / HYBRID

DATA LEAD

STAR8654

Qualified / Part-Qualified Global Pioneer

PENSIONS FLEXIBLE / HYBRID

STAR8698

STAR8556

Qualified / Part-QualifiedLeading-Edge Consultancy

LIFE LONDON OR SCOTLAND / HYBRID

BPA TRANSITION MANAGER

STAR8509

Qualified UK Provider

LIFE PENSIONS SOUTH EAST

FINANCIAL RISK ACTUARY

STAR8640

Qualified / Part-QualifiedFinancial Services Provider

LIFE PENSIONS SOUTH EAST

DE-RISKING DATA SOLUTIONS

STAR8677

Qualified Award-Winning-Advisors

PENSIONS LDN, SE OR EDB / HYBRID

SCHEME ACTUARY

STAR8697

Qualified UK Independent Consultancy

PENSIONS REMOTE

SENIOR DC CONSULTANT

STAR8650

FIA / FFA / CFA Major Global Consultancy

INVESTMENT NATIONWIDE / HYBRID

STAR8708

STAR8549

Qualified Actuarial Solutions Provider

LIFE NATIONWIDE / HYBRID

STAR8572

BUSINESS DEVELOPMENT DIRECTOR

Qualified Data Analytics Firm

LIFE FLEXIBLE / HYBRID

RISK ACTUARY

STAR8638

Qualified / Part-Qualified Major Firm

LIFE EDINBURGH / HYBRID

BPA IMPLEMENTATION

STAR8681

Qualified Leading Insurer

LIFE PENSIONS FLEXIBLE / HYBRID

STAR8714

DIRECTOR - PENSION RISK TRANSFER

Qualified Major Global Consultancy

PENSIONS FLEXIBLE / HYBRID

STAR8636

CORPORATE PENSIONS ACTUARY

Qualified Leading-Edge Firm

PENSIONS FLEXIBLE / HYBRID / REMOTE

SENIOR CONSULTANT - R&D

STAR8694

Qualified Market Leader

PENSIONS FLEXIBLE

HEAD OF INVESTMENT

STAR8646

Qualified / CFA Specialist Firm

INVESTMENT NON-LIFE REMOTE

STAR8683