Hanover Special Report #15

Compensation Study - UK Non Life Qualified Actuaries

Michael Stefan

Last update: Sep 22, 2020

Introduction

This is Hanover’s third annual salary survey for non life actuaries. I am grateful, as always, to all respondents as well as the dozen or so reviewers who have provided excellent feedback over the years.

The project grew out of a desire to have a more granular handle on pay for actuaries in the UK non-life market. It has since expanded to other areas such as exposure management. Pay (and in particular pay equality) has grown in importance over the last decade and whilst this survey does not claim to have all the answers, I hope it at least sheds some light on a very opaque area.

As in previous years, we have only surveyed UK GI Fellows. By “Fellows” we mean anyone who is a Fellow of the UK Institute and Faculty of Actuaries, as well as a UK-based Fellow of a major, exam based actuarial system such as the US (FCAS) or Australia (FIAA). We have not included Fellows of European-based actuarial systems unless they are also an FIA.

The set of questions has been updated to take into account feedback received last year. We have actually removed one of the questions (the year of fellowship) as we found the number of years experience correlated very strongly with the qualification year and was thus redundant.

Whilst total pay is important, it is by no means the only factors that affects recruitment and retention; work/life balance, company culture, team and departmental dynamics, relationship with supervisors etc all impact on whether someone joins or leaves. The covid-19 crisis has also completely changed work-life dynamics; working from home has rendered “distance” meaningless and it will be extremely interesting to see what effect this has on salaries and pay in 2021.

Uses

I foresee three types of uses for this salary survey:

This guide should NOT be used to benchmark pay for non-Fellows, ie actuarial students, nearly qualified actuaries, Associates or those “qualified by experience”. I was recently sent another firm’s salary survey and noticed they freely mixed figures for both students and Fellows, a questionable practice at best.

I will usually provide both average pay (usually referred to as mean) although I prefer to use median pay. Every compensation study I have completed has included a small number of outliers that unduly influence average pay. Median pay, on the other hand, is much more robust against outliers.

Content

The compensation study is meant as as a reference guide, meaning you do not need to read the whole document from start to finish. The report is structured as follows:

Last but not least, this guide is best read on a desktop, or perhaps a larger tablet.

What’s new for 2020

For 2020, the focus is on simplicity and readability. The aim is to have a salary survey that can be understood by any insurance executive - a HR manager, a CFO and of course a Chief Actuary.

For this year, I have changed the way I analyse the data. In previous years I used a combination of Excel, Tableau, PowerPoint, Google Sheets and Paint. This year I have transitioned to the “R” programming language, using an integrated workflow called RMarkdown. The details of how Rmarkdown works are not important, however, the advantages are:

In another break with previous editions, I have chosen to host this presentation online rather than as a pdf. The advantage is that one perma-link will always lead to the presentation, and any changes can be incorporated anytime. As such, if you are reading this online, you are reading the most up to date version.

I also need your help! If you have filled out this year’s survey already, thank you. If you haven’t, I would urge you to complete it; here’s the link. The average completion time this year was 4 mins 20 seconds, and by completing it you are helping to bring more clarity to a very opaque area. I am particularly keen to get more responses from women, as it will help to better quantify the gender pay gap.

Terminology

Earnings

This section will focus on total earnings (ie P60 + LTIPs), adjusted for 100% full time equivalent pay. I will cover the following:

Earnings are affected by many factors, and considering sector, gender, years experience etc individually in turn is the simplest way to identify any glaring gaps.

However, for more nuanced analysis, a higher level of statistical analysis is required. In Appendix 2 I will cover two additional types of analysis: a linear regression and a regression tree. Both are fairly straightforward to explain to non-specialists, and I am particularly keen on regression trees as they capture non-linearities in an intuitive, graphical way.

The overall average for all respondents is £200,605 with a standard deviation of £130,885.

Total earnings by sector

Sector Mean Minimum Median Top Quartile Maximum Avg Yrs Exp Respondents
Consultancy £ 215,872 £ 87,500 £ 169,126 £ 250,000 £ 830,000 15 17
Intermediary £ 160,543 £ 83,000 £ 130,000 £ 153,200 £ 359,500 11 11
LLM £ 213,554 £ 80,000 £ 172,550 £ 270,000 £ 815,000 14 85
Other £ 194,184 £ 130,000 £ 142,000 £ 247,091 £ 340,000 16 6
PLSME £ 176,132 £ 70,000 £ 145,500 £ 211,915 £ 518,667 16 36

Comments:

Total earnings by gender

All respondents:

Gender Mean Minimum Median Top Quartile Maximum Avg Yrs Exp Respondents
Female £ 144,876 £ 73,000 £ 130,000 £ 150,000 £ 330,000 12 37
Male £ 218,079 £ 70,000 £ 170,000 £ 279,864 £ 830,000 15 118

Respondents with 10 years experience or less:

Gender Mean Minimum Median Top Quartile Maximum Avg Yrs Exp Median Yrs Exp Respondents
Female £ 97,576 £ 73,000 £ 99,000 £ 110,000 £ 136,400 7 7 13
Male £ 129,458 £ 77,227 £ 110,650 £ 135,000 £ 382,000 8 8 36

Comments: Sadly the gender pay gap is very much present; male and female earnings (recall, they are adjusted to full time equivalent pay) diverge even at the earlier stages of their careers. Granted, the average number of years experience for female respondents is less than men, but even so, there is a difference. We will examine the gender pay gap in subsequent slides.

Total Earnings by years experience

Please note that for very experienced actuaries (20 years plus), the low number of responses in some cases suggest some caution.

Years Mean Minimum Median Top Quartile Maximum Respondents
5 £ 85,742 £ 80,000 £ 83,000 £ 90,000 £ 94,711 5
6 £ 115,430 £ 73,000 £ 103,000 £ 128,200 £ 220,000 7
7 £ 100,883 £ 80,000 £ 97,000 £ 110,000 £ 131,475 9
8 £ 112,009 £ 73,000 £ 107,630 £ 125,000 £ 155,000 9
9 £ 121,833 £ 92,000 £ 115,000 £ 124,000 £ 192,000 9
10 £ 167,970 £ 77,227 £ 139,750 £ 171,694 £ 382,000 10
11 £ 153,895 £ 98,000 £ 132,000 £ 210,000 £ 235,850 9
12 £ 172,735 £ 100,000 £ 131,648 £ 180,000 £ 440,000 13
13 £ 190,280 £ 100,000 £ 152,926 £ 195,448 £ 433,825 6
14 £ 259,950 £ 142,087 £ 250,000 £ 340,000 £ 455,817 9
15 £ 190,348 £ 120,000 £ 177,500 £ 227,364 £ 280,000 8
16 £ 196,337 £ 70,000 £ 203,950 £ 217,000 £ 369,000 9
17 £ 306,014 £ 141,000 £ 290,000 £ 402,250 £ 530,000 7
18 £ 229,353 £ 126,200 £ 255,000 £ 300,000 £ 330,000 5
19 £ 259,595 £ 85,000 £ 266,179 £ 345,000 £ 425,210 6
21 £ 315,700 £ 160,000 £ 270,000 £ 308,000 £ 660,498 5
22 £ 240,000 £ 130,000 £ 240,000 £ 295,000 £ 350,000 2
23 £ 623,001 £ 224,003 £ 815,000 £ 822,500 £ 830,000 3
24 £ 181,735 £ 118,470 £ 181,735 £ 213,368 £ 245,000 2
25 £ 255,715 £ 100,000 £ 216,408 £ 325,500 £ 518,667 22

Total Earnings by sector for new(ish) Fellows

The table below provides a comparison of total earnings for FIAs with 10 years experience or less. Assuming a mean qualification time of 5 years (+/- 2 years), this would suggest around 5 years post qualification experience (+/- 2 years).

Sector Mean Minimum Median Top Quartile Maximum Avg Yrs Exp Median Yrs Exp Respondents
Consultancy £ 124,488 £ 87,500 £ 104,150 £ 155,170 £ 192,000 8 8 6
Intermediary £ 118,146 £ 83,000 £ 125,000 £ 131,106 £ 136,400 7 6 6
LLM £ 131,581 £ 80,000 £ 107,630 £ 124,500 £ 382,000 8 8 27
Other £ 150,000 £ 150,000 £ 150,000 £ 150,000 £ 150,000 9 9 1
PLSME £ 85,608 £ 73,000 £ 80,000 £ 95,136 £ 116,500 8 8 9

Please note again that median and mean years of experience for the respondents differ slightly.

Cost Ratio by gender

Recall that we define the “Cost Ratio” as the ratio of Total Pay (adjusted for 100% FTE) divided by the number of years experience that individual has. For the cost ratio calculations, we have used the original number of years experience provided by respondents.

We have a high number of male respondents who are either Chief Actuary PC holders and/or hold an SMF position. This could “bias” our results as effectively we would be mixing “leaders” with “doers”. Excluding all respondents with a Chief Actuary PC or who hold an SMF position, we have an “apples for apples” comparison for male vs female “doers”. In this case we have a male cost ratio of £ 14,613 and a female cost ratio of £ 12,767.

Cost Ratio by gender (2)

One argument for the disparity is that our results contain a number of very senior (all male) respondents in consulting, likely director or partner level. We can exclude them and only focus on the “in-house” actuaries at underwriting organisations. Our results are as follows:

Conclusion 1: The numbers show that the earnings are highest for London Market males and not consultants, since removing the consultants increased the average cost ratio for males but reduces the average cost ratio for females.

Conclusion 2: The real disparity between male and female earnings are in the Lloyds and London Market. There, the average female “doer” earns 70% of what a male “doer” earns. Within personal personal lines and the SME market, the average female “doer” earns 109% of what a male “doer” earns. In other words, women actually earn more in the PL & SME market.

SMF Earnings

The following table contains a summary of earnings for those who declared they hold a controlled function (more recently referred as a “Senior Management Function” by the PRA). Our survey did not ask respondents to identify the exact Senior Management Function role they held, so we don’t know whether the individual was the SMF20 Chief Actuary, SMF4 CRO, SMF22 CUO etc. It is also possible that some held multiple SMF roles.

SMF Role Mean Minimum Median Top Quartile Maximum Respondents
SMF Role £ 331,511 £ 100,000 £ 290,000 £ 407,605 £ 815,000 19
The rest £ 182,316 £ 70,000 £ 142,544 £ 213,165 £ 830,000 136

Another way to look at the data is to segment those who hold either type of Chief Actuary PC and also held an SMF position (these are the respondents that the PRA would recognize as the “Chief Actuary”). In other words, “The Rest” in the table below are the FIAs who hold a Chief Actuary PC but are NOT undertaking an SMF role.

SMF Role Mean Minimum Median Top Quartile Maximum Respondents
SMF Role £ 309,004 £ 100,000 £ 285,000 £ 427,364 £ 530,000 12
The rest £ 252,773 £ 134,000 £ 211,276 £ 309,750 £ 660,498 14

Chief Actuary Earnings

The first table covers all holders of a Chief Actuary Practicing Certificate (“PC”), the second and third segment by whether the individual holds a “with Lloyds” or “without Lloyds” PC.

Chief Actuary Mean Minimum Median Top Quartile Maximum Respondents
All Chief Actuaries with PC £ 278,726 £ 100,000 £ 234,298 £ 376,000 £ 660,498 26
The rest £ 184,860 £ 70,000 £ 141,000 £ 217,660 £ 830,000 129
Chief Actuary, Lloyds Mean Minimum Median Top Quartile Maximum Respondents
Chief Actuary, with Lloyds £ 339,770 £ 180,000 £ 310,000 £ 427,364 £ 660,498 16
The rest £ 184,586 £ 70,000 £ 143,000 £ 216,000 £ 830,000 139
Chief Actuary, non-Lloyds Mean Minimum Median Top Quartile Maximum Respondents
Chief Actuary, non-Lloyds £ 181,055 £ 100,000 £ 154,500 £ 200,000 £ 390,000 10
The rest £ 201,953 £ 70,000 £ 152,000 £ 250,000 £ 830,000 145

As an aside, I am, as ever, exceptionally grateful that 26 Chief Actuary PC holders have elected to complete the survey. It is a testament to the level of engagement we have with the actuarial community.

Total Earnings by Gender and Staff Managed

There is a clear upward trend between the size of team managed and total earnings. It is worth noting that we have truncated the 5 largest teams in the market, as showing these on there risks identifying the individuals involved. If you are running a team of > 45 staff, please feel free to contact me directly for a confidential discussion.

Newly qualified fellow earnings

Sector Mean Minimum Median Top Quartile Maximum Respondents
Consultancy £ 88,750 £ 87,500 £ 88,750 £ 89,375 £ 90,000 2
Intermediary £ 117,719 £ 83,000 £ 125,738 £ 132,706 £ 136,400 4
LLM £ 107,971 £ 80,000 £ 97,000 £ 106,500 £ 220,000 11
PLSME £ 77,153 £ 73,000 £ 77,806 £ 80,000 £ 80,000 4

Clients often ask about salaries (or earnings) for newly qualified actuaries. Given the average qualification time is 5 years (+/- 2 years), we have selected all respondents with 7 years of experience or less. Given the low number of responses, I feel somewhat wary of over generalising but it does seem like £80,000 is the MINIMUM. Recall, this is a P60 so it will be inclusive of any (presumably small at this level) cash bonus.

I personally feel median salaries are a better indicator of “where the market is at”. In this case, total earnings are in the region of £90,000 - £100,000.

Also please recall that the overall average for all respondents is £200,605.

Conclusion

First of all, thank you for taking the time to read this report.

I concluded last year’s survey with a discussion of flexibility and work life balance as a differentiator. Little did I think that I would be writing the next year’s follow up against the backdrop of virtually 100% remote working. To some extent it is too early to really assess the pandemic’s long term impact on earnings, although I would like to briefly cover two themes.

The first theme is the general increase in supply of actuaries. Since 2007, approximately 2,200 people have passed SA3, the general insurance fellowship exam. Assuming (conservatively) that 60% of those are in the UK, and that 90% of those are now Fellows, we are still looking at an increase of around 100 new GI fellows per year in the UK. Solvency II (and more recently data science) has occupied the bulk of these new Fellows, but I do wonder whether the constant increase is sustainable for the long term. That being said, very few of my clients are complaining about having “too much slack” - if anything, teams seem to be as busy as they’ve always been. Perhaps (to paraphrase Parkinson’s law), “work expands so as to fill the team available for its completion.” In some ways, the massive increases in data availability over the last decade has helped maintain actuarial employment at basically full levels, especially in areas like personal lines where the amount of data ingested by insurers has taken off almost exponentially. Whether or not it will continue remains to be seen, but one trend that I think will definitely continue is the inflow of newly qualified GI actuaries.

I wonder whether many of these will end up in what could be termed “Data Science Actuary” roles. The UK in particular has had a demarcation of actuarial roles – you are either a “life” actuary or a “GI” actuary or maybe even a “healthcare” actuary. If one were to put these areas together (perhaps under the banner of “consumer insurance”) it is entirely feasible that a new breed of actuary could emerge, one versed in data science methodologies, programming and statistics, able to work on consumer-related analytics across multiple fields of “insurance”. On the commercial side, the Internet of Things could even lead to another deluge of sensor data and anecdotally, a number of Lloyds managing agents have started to recruit “Portfolio Manager” roles. These roles seek to combine traditional actuarial expertise (eg providing judgment into the profitability of deals) with non traditional work (advising on new data sources, being a key member of reinsurance purchasing decisions, evaluating commercial opportunities including M&As). The launch of three follow only (“algorithmic”) syndicates in the summer of 2020 can only be considered good news. I’m certainly very positive about the profession’s long term future, though perhaps without the rapid earnings progression of the Solvency II boom.

About me

I have spent the last 15 years recruiting for senior actuarial, catastrophe modelling and analytical positions in the UK, and abroad (principally the Americas and Bermuda). My client base ranges from consumer insurance, commercial lines and Lloyds of London to more unusual operations involved in insurance linked securities, private equity and broking. I also research and write virtually all of our research reports, including compensation surveys.

Prior to joining Hanover Search in 2010, I spent 6 years working as lead consultant for the insurance and financial services division of Hays, a global FTSE-250 listed recruitment group.

Prior to university I spent 2 years working in sales for Churchill Insurance, one of the legacy companies of Direct Line Group.

I have a degree in Economics and Mathematics from York University and have completed a number of development courses, including a Strategy and Finance module with INSEAD and a psychometrics certificate with Cambridge University. In the summer of 2020 I completed two R-based courses with Essex University’s Summer School in Quantitative Social Science.

I have 2 young children under 5 so I don’t really have any spare time, but when I get a moment, I usually spend it reading.

Get in touch

Other than my research work (eg salary surveys), I am an active member of Hanover’s search and selection team. I have spent the last 15 years recruiting actuaries for the UK and Anglo-America (re) insurance markets. My client base ranges from consumer insurance, commercial lines and Lloyds of London to more unusual operations involved in insurance linked securities, private equity, captives and broking. I have significant experience of completing difficult searches, especially where the “candidate pool” is very small.

My recruiting philosophy can be summarised as very simply:

I am always happy to:

For all queries, please email me:

Appendix

I have always been very interested in being able to predict (to a reasonable degree of accuracy) what someone’s total earnings are. I have fielded numerous questions from both HR and actuarial leaders over the years, usually in relation to benchmarking a role they want to recruit for (and sometimes benchmarking themselves or their own team members). When conducting a search, it’s also useful for me to estimate what someone might be on as it allows me to do a “package sense check” in my own mind before approaching them. As such, I attempt a linear regression model every year, with the key caveat that I want a model that’s reasonably predictive whilst being as parsimonious as possible.

The following slide shows the model summary using two predictors - the number of years experience and the number of staff managed.

The slide after that contains the output from a regression tree. There are many methodologies for constructing regression trees but one of the oldest is known as the classification and regression tree (CART) approach developed by Breiman et al. (1984). Basic regression trees partition a data set into smaller subgroups and then fit a simple constant for each observation in the subgroup. The partitioning is achieved by successive binary partitions (aka recursive partitioning, implemented in R using the function rpart) based on the different predictors. The constant to predict is based on the average response values for all observations that fall in that subgroup.

As with other nonparametric techniques, CART does not require any assumptions for the underlying distribution. Another advantage is that the graphical output is very interpretable.

Linear Regression

For me the simplest possible model is one that includes both years experience and number of staff managed. The model summary is shown to below. I have been able to increase the adjusted R-squared to 0.49 by experimenting with various combinations of interaction terms and higher level polynomials. However, I am concerned by both over-fitting and explainability and would rather stick with something simple and intuitive that still explains a great deal of variation. The model below also has the added benefit that both the Years and Staff terms are statistically significant.

Observations 155
Dependent variable Total
Type OLS linear regression
F(2,152) 56.64
R² 0.43
Adj. R² 0.42
Est. S.E. t val. p
(Intercept) 70080.90 20596.78 3.40 0.00
Years 6646.09 1399.48 4.75 0.00
Staff 4017.20 571.52 7.03 0.00
Standard errors: OLS

Decision Trees

In the chart below, “Cert.CAL”, is a binary variable that is equal to 1 if the individual holds a “Chief Actuary, with Lloyds” PC and 0 otherwise. “Staff” is number of staff managed and “Years” is number of years experience. I have chosen the minimum number of splits per node to be 10 (the default in rpart is 20 but I lowered it as I don’t think ~150 responses is that much data to work with). A higher number of splits can result in a smaller tree with fewer rules, and consequently different numerical predictions.

One disadvantage of CART is that rules can be unstable if new information is added (eg if we have say an extra 10 responses). A few changes in the training dataset can create a completely different tree. This is because each split depends on the parent split. And if a different feature is selected as the first split feature, the entire tree structure changes.

Other reports

Since early 2018, I have authored a number of reports/briefs and a selection is listed below. All are available to download directly from my Box folder (right click, save as, no registration needed).

-Demographic Analysis of Actuarial Teams in the London Market

-Demographic Analysis of Cat Modelling Teams in the London Market

-Compensation Report 2018, UK General Insurance Actuaries

-Compensation Report 2018, Catastrophe & Exposure Modellers

-Compensation Report 2018, Risk Compliance and Governance

-2019 Market Conditions for Actuarial, Catastrophe and Data Science

-Salary Survey 2018, SIMF4 CROs

-AI & ML Expertise in the major insurance Companies

-Compensation Report 2019, UK General Insurance Actuaries

-Compensation Report 2019, Catastrophe Modelling & Exposure Management

I also maintain an up to date online Tableau viz of all “The Actuary” print magazine job ads in the GI sector (whether qualified or not). Whilst I won’t be developing the viz any further, I update the underlying data on a monthly basis when “The Actuary” arrives and you can view it here.

Last but not least, if you recruit or manage cat modellers, you will want to see my 2020 Cat Modelling Salary Survey