Impact of COVID-19 Defence Measures


Iris Deng, Michelle Dong, Helena Hayes, Abbey Huang, Richard Lyon and Colin Yellowlees

Different countries and sub-national regions have adopted different responses to the threat posed by Covid-19. There are strong differences of opinion about these choices and their impacts, including those on health and the economy.

Even if this question can ever be settled, it is likely to take some time, partly because the virus is still rampant and partly because of the inevitable time lags involved. However, it is worth attempting to sketch an objective picture based on available data from different countries.

This paper will be the result of research on two fronts. First, data will be gathered for about 20 countries, in relation to the economy, Covid-19 and a consolidated measure of government responses. This data will be used to plot broad relativities between countries. If it is possible to do so, there will be deeper dives on a subset of these countries.

Secondly, the team will search for existing published academic research addressing this topic. The aim will be to synthesise this research and make it accessible. If there are clear common themes and messages, these will be drawn out and compared/contrasted with what the data suggests...

View Paper

Grasping at the invisible – Bayesian models for estimating latent economic variables


Hugh Miller and Isabella Lyons

Actuaries have long used official statistics and market data such as the unemployment rate or the traded price of bonds to set assumptions. However, correct interpretation of these quantities often depends on elements that cannot be directly observed. For example:

▪ Short-term interest rates are expansionary or contractionary depending on the ‘neutral’ interest rate, which is generally not observed

▪ The severity of unemployment is usually measured relative to the non-accelerating rate of unemployment (NAIRU).

Such quantities are increasingly of interest to actuaries – for instance IFRS 17 allows more judgement about the composition of the yield curve. The challenge in estimating these is heightened by the fact that quantities such as the neutral interest rate evolve over time.

Despite actuaries being pioneers in early work on credibility theory, Bayesian approaches have largely given way to frequentist approaches. This is to our loss, since such methods have advanced substantially over the past couple of decades and they are well-suited to problems with latent variables or noise around underlying ‘true’ values.

The paper gives an introduction to modern Bayesian model-fitting using the stan package and then builds and applies models to estimate the latent economic variables listed above. Findings in the paper are be supported by corresponding published code so results can be reproduced. By having public models that can be routinely updated, these models will fill a gap in the Australian market, where regular estimates are not routinely available.

View Paper

Introduction to Formal Evaluation of Probabilistic Forecasts in Actuarial Practice


Dimitri Semenovich

Actuaries are often called upon to make probabilistic forecasts, however it is relatively uncommon that these are tracked and evaluated formally. In this presentation we survey techniques developed in meteorology and statistics for this purpose and highlight those of particular practical relevance. The key feature of formal forecast verification is that it is agnostic to the forecast generation process and as such can be applied even in those situations where no formal stochastic models are available. We will also demonstrate the connection between forecast verification and traditional regression residual analysis, which will hopefully help to further elucidate both concepts..

View Paper

Leadership and Professionalism Plenary  


Was the misconduct identified in the Banking Royal Commission enabled by directors who didn't understand the complexity of the businesses they directed? 

Ian Laughlin, Barry Rafe

On 8 May 2018 the Financial Review reported that a number of directors of AMP were stepping down because they no longer had the support of the institutional shareholders. The resignations were directly attributable to the fallout from the banking Royal Commission which revealed that the behaviour of financial services organisation often did not meet community standards and expectations. In some cases, such as the AMP there appeared to be conduct that broke the law.

The board is responsible for the actions and success of the corporation and hence the composition of the board would appear to be relevant to the corporation’s success. Ideally, directors are selected for board positions based on their technical and behavioural competencies. Boards often develop a skills matrix that sets out the broad set of competencies needed for the corporation to meet its aims. The expectation is that directors are appointed to fill or bolster various competencies required.  In addition, the Corporations Act requires that all directors be financial literate i.e. they need to understand the financial drivers of the organisations they direct.

The questions to be asked in this session are:

  • Was the misconduct revealed by the banking Royal Commission caused by directors who did not understand the businesses they directed?
  • Who should be on the boards of complex financial services businesses
  • How do complex financial services businesses manage the need for diversity of skills on the board with the need for directors to understand the complexity of the businesses they direct.

View Paper

Super Plenary: Gender Inequality in the Retirement Savings Gap Invest&WM_IconSuperannuation_Icon

Michael Rice, Richard Dunn, David Knox

Over the last decade, there has been a body of research into the gap between the average male and female retirement balances. The Report of the Retirement Income Review looked at gender differences and confirmed the gap, which is a function of differences in working life (since superannuation contributions are linked to wages). The relative income in retirement by gender is further impacted by women retiring earlier and living longer. However, the Age Pension significantly reduces the gap in average retirement incomes. Nonetheless, Australia has been criticised for its high poverty levels amongst elderly single renters (the majority being female) and for not addressing the emerging gap for working Australians.

This Paper looks at the retirement savings gender gap in the light of the findings of the Retirement income Review and it also assesses where Australia stands internationally in terms of supporting female retirement incomes. Some options to address the issue will also be discussed. The authors have undertaken past work into female retirement inadequacy and will examine whether progress is being made.

View Paper

Australian Investment Performance 1959 to 2021 (and Investment Assumptions for Stochastic Models) Invest&WM_Icon

Colin Grenfell

Every two to four years since 2005, the Institute has published papers by Colin Grenfell on Australian investment performance and these papers have been discussed at Institute conferences and Insights meetings and at the 2010, 2014 and 2018 ICA's.  Each of these papers tabulates and summarises annualised average investment results for (a) risk margins (b) coefficients of variation (c) skewness (d) kurtosis (e) cross-correlations and (f) auto-correlations.  The analysis spans 15 "sectors" covering eleven investment classes and four key financial indicators.  Colin's latest paper (86 pages) entitled " Australian Investment Performance 1959 to 2021 (and Investment Assumptions for Stochastic Models)" has just been uploaded to the Virtual Summit website.  The paper will be discussed as part of the 12.00pm Monday 10 May Investment and Wealth Management plenary session "Markets Panel: Where Next?".  Pre-reading of the paper is suggested.".

View Paper

Excess Mortality in 2020 Health_Insurance_IconLifeInsurance_IconGlobeIconPINKLead&Professionalism_Icon 

Richard Lyon, Jennifer Lang, Karen Cutter

In June 2020, the Australian Bureau of Statistics (ABS) published its first analysis of provisional monthly mortality statistics, covering weekly doctor-certified deaths in the first three months of 2020, compared with the five previous years.  Subsequent reports were produced broadly once a month, with the latest one (December) updating the analysis to 27 October.

The authors have produced a series of small papers adjusting the ABS reports for better comparability and discussing the implications of the revealed mortality experience.  For example, we identified the relevance of spikes in deaths from pneumonia, diabetes and stroke at the same time as the first wave of Covid-19 in Australia.

This paper will be a deeper dive into Australian mortality in 2020 and will also draw comparisons with experience overseas, such as the UK and the US, to the extent that reliable data is available.

The paper will explore the implications of the observed differences..

View Paper

Deep Fried Bananas: From wholesome to gruesome (and back again?) DataAnalytics_IconGInsurance_Icon

Tim Spicer

This paper looks at the thorny issue of Add-On Insurance sold at car yards, examines and provides insights into the key historical behaviours and outcomes that gave Add-On Insurance its poor reputation, the various regulatory actions and reforms, and asks whether this market can be transformed into one that offers quality products, at reasonable prices, that meet identifiable consumer needs.

View Paper

Framework for Determining Marketable Price of Retail Insurance GInsurance_Icon
Mudit Gupta

In pricing of retail insurance products there are many steps involved from initial analysis of data to arriving at final premiums – with different stakeholders involved along the process. Typically, the initial part of the process is quite scientific when data is used to identify opportunities for change, however, as the process moves towards the final ‘marketable’ price, the decision making becomes less scientific and more reliant on judgment of distribution and marketing experts. This paper attempts to bring the end-to-end process of pricing retail insurance products within a structured framework to guide sound decision making during a pricing exercise combining both technical processes as well as expert judgment to arrive at the marketable price. The underlying hypothesis in this paper is that three parts of product development process – product, pricing and distribution strategy – are intrinsically linked to each other. Any change in one will influence outcomes of the other parts. It is therefore, not possible to consider pricing in isolation of product design and distribution when trying to determine marketable prices. A framework is proposed to deal with these in a structured manner.

View Paper

Predictive Total Loss: Using AI and Automation to Improve Motor Vehicle Claims in a Responsible Way DataAnalytics_Icon
Chris Dolman, Hannah Sakai and William You

In this paper, we outline a recent project conducted by IAG to use AI and automation techniques to improve the customer experience for motor total loss claims. We outline three aspects which should be of interest to actuaries undertaking similar AI and/or automation projects: 1. Technical aspects of the data and modelling process 2. Holistic aspects of the project delivery, including effective interdepartmental collaboration and customer delivery 3. Ethical aspects of the project, including customer testing and monitoring We hope that by outlining this project in some detail, we can assist others involved in the delivery of AI solutions in their businesses.

View Paper


A Mutual Renaissance GInsurance_Icon
Charles Pollack

In an environment of drastically increased pricing uncertainty across many lines of business, and a demand for customer focus, the origins of general insurance carry useful lessons and opportunities. Mutual insurers have a structural advantage in the delivery of outcomes for customers.

The Australian GI market is somewhat unique globally in its concentration and composition and noteworthy for its lack of resident general insurance mutuals. With recent regulatory change, an opportunity has emerged to go back to the origins of insurance with a next generation overlay and create a viable alternative to consumer insurance offerings currently in the market.

This paper looks at the global and Australian context, the recent changes and what that means for local mutuals, as well as the different types of mutuals that can be established (including takaful structures) and their respective strengths and weaknesses. It then explores the deeper aspects of the customer benefits of mutuals and how actuaries’ approaches to product design, pricing, underwriting, capital management and organisational design may look different in a mutual environment.

View Paper