Gauging the Tort Temperature and the Tort Temperature Scorecard

Transcription

1 Gauging the Tort Temperature and the Tort Temperature Scorecard Prepared by Stephen Lee, Geoff Atkins, and Francis Beens Presented to the Actuaries Institute General Insurance Seminar November 2012 Sydney This paper has been prepared for Actuaries Institute 2012 General Insurance Seminar. The Institute Council wishes it to be understood that opinions put forward herein are not necessarily those of the Institute and the Council is not responsible for those opinions. Finity Consulting Pty Ltd 2012 The Institute will ensure that all reproductions of the paper acknowledge the Author/s as the author/s, and include the above copyright statement. Institute of Actuaries of Australia ABN Level 7, 4 Martin Place, Sydney NSW Australia 2000 t +61 (0) f +61 (0) e actuaries@actuaries.asn.au w

2 Gauging the Tort Temperature and the Tort Temperature Scorecard Stephen Lee, Geoff Atkins, Francis Beens Abstract It is ten years since Tort Law Reform swept across Australia and the post reform honeymoon has concluded. There is growing concern that liability claim costs continue to increase and the profitability of liability products is being tested. An unresolved question is what an insurer should do to manage this risk. The impact of a heated tort environment has many consequences for an insurer writing liability insurance, ranging from reserve adequacy to the Board s responsibility to manage its business within its risk appetite. The long tail nature and diversity of risks for this class leads to difficulty in understanding what the current experience is telling us. On the flip side if we wait for more experience so that we can be more confident of our analysis, then we are at risk of acting too late. In this paper, we will explore the current evidence in the Australian Liability market. There is often a lot of information on this topic and it is difficult to make sense of the range of views. We will propose that a Tort Temperature Scorecard and a regular framework for updating and implementing the scorecard be put into place. Having such a framework would mean that an insurer is better able to alter its strategy based on its view of the cost of future Tort settlements. Keywords: Gauging the Tort Temperature in Australia; Tort Temperature; Tort Temperature Scorecard; Superimposed inflation; Personal Injury; Tort Law Reform; Civil Liability; NCPD; Liability 2

3 1. Summary The term tort temperature describes the current social and judicial environment, and the influences this has on the cost of claims. We would say that a hot tort temperature would be one that favoured the plaintiff, either by awarding larger settlements or stretching the law i by expanding the duty and standard of care owed by one party to another. In cooler tort temperature environments, there would be no extensions either through larger settlements or the duty and standard of care required. When measuring social and judicial inflation, actuaries have traditionally thought in terms of superimposed inflation. However, the usefulness of this analysis is limited by its measurability and relevance. Measuring past superimposed inflation is complicated by a high degree of random variability and difficulties identifying like-for-like claims. Even if it could be measured reliability, it remains a backward looking measure and does not answer what we really want to know, that is, what level of superimposed inflation will occur in the future (Taylor, 2002). Instead, analysis should turn to understanding the Root Cause drivers of cost, such as whether more cases are going to court, changes in the behaviour of claimants and the courts, changes to social attitudes, or any precedents being set. In doing so, we can attempt to better describe whether these trends may continue and form a view on what the impacts might be for an insurer. There is no single source of information that will allow us to understand the tort temperature. To bring the analysis of root causes together, organisations should implement a Tort Temperature Scorecard (TTS) and a framework for regular review of the Tort Temperature. The TTS is a simple but effective tool to achieve a consistent view across an organisation and therefore enable a consistent response across a range of activities. The earlier sections of this Paper set out our Root Cause analysis of the Australian Industry by going through recent High Court judgements, the number of cases being filed in court, experience from the NCPD, experience from a typical portfolio, and discussions with experts in the field. In summary, our key findings from looking at these sources were: No indications from the High Court of heating or cooling tort temperature. Litigation is more prevalent now than it was post Tort Reforms in There is evidence that average claim costs have been on the rise, though the number of claims appears to be stable. We have applied the TTS based on these observations of the Australian Industry and the likelihood of the trends continuing into the future. On a scale of nil to ten and where a score of five suggests a stable tort temperature in the future, we estimate that the Australian environment is around seven. We note that this is the 3

4 view of the authors of this Paper, and other assessments of the Tort Temperature may lead to different conclusions. 2. Outline of this Paper The remainder of this Paper is set out as follows: 3. History of the liability class 4. Discuss what we mean by Tort Temperature by introducing the idea of a Tort Marketplace. 5. Analyse the evidence in the Australian environment regarding Tort Temperature from various sources. 6. Discuss why it is important for insurers to be aware of the Tort Temperature. 7. Consider traditional actuarial thinking on superimposed inflation and its limitations. 8. Propose a move from traditional analysis to one focused on understanding Root Causes. 9. Propose a Tort Temperature Scorecard (TTS) to bring the sources of information together in a manner that is useable to an insurer s senior management. 10. Show an example of how the TTS might look, based on an assessment for the Australian Industry at this point in time. 11. Discuss how the TTS could be implemented into a control cycle framework and into various functions of an insurer. 12. Discuss practical applications of the TTS for an insurer. 13. Concluding remarks 3. History of the Liability Class The following figure shows public and product liability loss ratios (financial year basis, net of reinsurance) over the past 30 years. 4

5 Net Loss Ratio Gauging the Tort Temperature and the Tort Temperature Scorecard 160% Figure 1 Public and Product Liability Net Loss Ratios 140% 120% 100% 80% 60% 40% 20% 0% Sources: APRA; Office of the Insurance Commissioner; Insurance and Superannuation Commission The figure shows financial year loss ratios, which can be distorted by prior period reserve releases or strengthening, and changes in discount rates. However, it is still a useful comparison over the medium term, and matches the claims cost result that insurers actually report each year. Until the mid to late 1990s the loss ratios trended in a band from 60% to 90%. However, loss ratios spiked during the late 1990s to reach around 140% in 1999 and Following tort reforms the loss ratio fell to historically low levels in the 30% to 50% range. In more recent years the loss ratio has crept up to around 70%, though this is partly due to reductions in the discount rate during the Global Financial Crisis (GFC) which began in 2007 with continuing effects on the global economy to this day. What drives performance in this class? Most claims by number are for property damage, which tend to be reasonably simple to assess and reserve for. There are examples of very large and complicated property damage claims, such as those arising from liability policies following the Melbourne Bushfires in 2009, but these types of claims are infrequent. In general, property damage claims make up only about a third or less of the total cost, with the majority of costs therefore arising from personal injuries. Personal injury claims, particularly when serious, often take in excess of five years to finalise. Ultimately, it is the court system that sets the level of compensation for personal injuries based on the common law of negligence as modified by the Civil Liability laws in each jurisdiction. The Liability Crisis of 2001 There is a marked reduction in the loss ratio shown in Figure 1 following 2002 and Much of this reduction can be attributed to tort reform. Below we 5

6 summarise the history of liability insurance in Australia over the 10 years leading up to the liability crisis and the decade afterwards, with reference to the loss ratios. 1990s Public liability insurance costs rose during 1990s, due to superimposed inflation and some increase in claim frequency. Market prices fell, however, and some companies (HIH and FAI) continued to provide cover in certain segments at prices that other insurers could not sustain Collapse of HIH in March 2001 led to an insurance crisis. Total unavailability of cover for some segments, particularly the community and non-profit segments in what was termed the "death of fun". Loss ratios remained in excess of 100% to to onwards The Australian Government commission a report to review the Law of Negligence (commonly known as the Ipp Report ). Following the Ipp Report reforms to civil liability laws were implemented throughout Australia. While there were differences in legislation between States, some common changes were: - caps and thresholds introduced for pain and suffering damages - caps on economic loss - limits placed on defendant liability for obvious and inherent risks - return of the Bolam Test (though in modified form), intended to limit medical liability and damages - restrictions on claims for mental harm APRA established the National Claims and Policies Database (NCPD) to satisfy data requirements for insurers and government to better understand liability and professional indemnity classes. After tort reform, the claims experience in public liability portfolios has tended to be characterised by: - Much lower claim frequency - Moderate claim size, in contradiction to the expectation that claim sizes would increase due to the impact of thresholds Actuaries were slow to recognise the good experience. Once recognised, however, this revealed the extent of the tort "honeymoon" where loss ratios were extremely favourable as insurers gradually adjusted to the new environment. Reserve releases from liability classes were common, and this supported many insurers profits. Being a long tail class, it is difficult to be certain about what has happened in these years. However, there is a general consensus that claims costs are on the rise. Continuing reserve releases from prior years suppress the financial year reported loss ratios. We believe that without these releases, liability loss ratios would be showing a greater trend upwards. 6

7 4. What is Tort Temperature? The term tort temperature is used to describe the current social and judicial environment, and the influences this has on the cost of claims. We would say that a hot tort temperature would be one that favoured the plaintiff, either by awarding larger settlements or stretching the law ii by expanding the duty and standard of care owed by one party to another. In cooler tort temperature environments, there would be no extensions either through larger settlements or the duty and standard of care required. We first became aware of the Tort Temperature term from the US market. Product and pricing actuaries for liability policies there are well aware of the cost impacts due to jurisdictional differences, and in some instances talk of specific counties or courts that are trouble-spots, known as Judicial Hellholes iii. Attempts have been made to measure the relative Tort Temperature between US states by looking at both the rules governing Torts, and the settlement outcomes by jurisdiction. However, discussions around Tort Temperature and Judicial Hellholes are focused towards pushing for tort law reforms and do not appear to contain anything further than a subjective assessment. In the context of this Paper, we look at the Tort Temperature affecting liability, and more specifically, personal injury claims. We aim to build on the concept of Tort Temperature as used in the US market and turn it into something more tangible and of use to insurers, with a focus on the Australian market. The Tort Marketplace It is helpful to consider the environment in which personal injury cases operate. We can think of claims as operating in a marketplace, as represented by Figure 2 below. 7

8 Figure 2 The Tort Marketplace Only a handful of personal injury claims will come to the High Court each year. Fewer than 100 claims each year will reach appeal. A few hundred claims will be finalised by court judgment. The rest will settle before this. Around 3,000 personal injury claims are filed in court each year. Around 30,000 liability claims per year enter this market. We estimate around 10,000 of these are personal injury claims. Source: Finity Consulting Design: Vetro Design When a claim is made by the plaintiff to the defendant, each participant has some expectation of the market value of the claim. For the majority of claims, lawyers have a good idea of what level of compensation to expect based on similar recent cases. These claims will generally settle through negotiation out of court as minor differences in expectations are not worth the expense and risk of litigating the claim. Decisions at this level are rarely public, but the same lawyers in this area work across many claims and will know what is generally happening with settlement costs. Over time the expectations of these lawyers (or the market) may shift, and if they do, insurers will see the impact of any shift in the market in their claims costs over time. There are situations where the expectations of each party differ significantly. In these cases the difference is large enough to be irreconcilable and also justify the cost of legal fees needed to initiate court action. Most of these claims will settle before or during the case, while the remainder will be decided by court judgement. The outcomes of any individual case may not raise eyebrows, but over time will change the expectations in the civil liability claims market. A small proportion of civil liability claims will ultimately end up at the Courts of Appeal, and a handful at the High Court. At this stage, arguments are usually around specific points of law which may only affect other cases in a limited manner. However, the outcomes do have an influence on the litigation climate. If decisions tend to be consistently pro-plaintiff, then this will mean plaintiffs are more willing to wear the cost of appealing court decisions, and vice versa. 8

9 The Tort Temperature can be influenced at each step of the litigation process. When we describe Tort Temperature, what we are trying to do is to measure the temperature of this system, i.e. of the whole tort marketplace. 5. Tort Temperature indications for the Australian Insurance Industry To gauge the Tort Temperature in Australia, we have looked for indications from the experience of the claims that are appealed in the High Court down to those that get settled without any court intervention. In this section, we will: (a) (b) (c) (d) Summarise judgements from the High Court relating to Personal Injury cases and see what we can learn from these Look at indications from the number of cases going to court Analyse the NCPD and a typical portfolio Summarise our discussions with players in the Tort Marketplace We note that a limitation in undertaking any sort of analysis of the liability class will be the availability of data. The long tail nature of these claims makes timely analysis difficult, while large variability in claim sizes means that we must always use our judgement in interpreting the outputs. However, this process should be more about gathering evidence from a range of different sources rather than relying on any one source alone. a) High Court Indications Over the last three years, we identified seven non-asbestos civil liability cases relating to personal injury that have reached the High Court. In Figure 3 below, we simplify the outcomes of these cases by successful party, before and after the appeal. 9

10 Successful Party BEFORE Appeal Figure 3 Summary of recent High Court decisions Successful Party AFTER Appeal Defendant (4 cases) Appeal Dismissed (1 cases) Appeal Successful (3 cases) Defendant (3 cases) Plaintiff (3 cases) Source: Finity Consulting Appeal Dismissed (1 cases) Appeal Successful (2 cases) Plaintiff (4 cases) Of the seven liability cases, four favoured the defendant and three favoured the plaintiffs prior to the appeal being lodged. After the appeals, the opposite is true with three favouring the defendant and four favouring the plaintiffs. These outcomes do not imply any particular trend in High Court decisions, i.e. the decisions do not appear to favour plaintiff over defendant, and vice versa. The most significant observation may be that five of the seven appeals were successful (possibly in itself a result of selective granting of leave to appeal). This analysis loses much of the specific details for each case, which are important as the circumstances surrounding each case influence the outcomes. The judgements handed down by the High Court highlight that these cases are determined on very specific legal arguments which have limited relevance to the majority of personal injury claims. We have reviewed the specific circumstances of each of these seven cases, and have not found suggestions that the High Court is observably pro-plaintiff or pro-defendant. More importantly, we do not expect these judgements to have ramifications in respect of larger quantum of awards or extending the duty of care for other cases brought to the courts. This is in contrast to the 1990s when both the circumstances in which negligence would be found to have occurred and the scope of damages recoverable in the event of such a finding were more consistently pro-plaintiff (a long term trend noted by then Chief Justice Spigelman of the NSW Supreme Court in his 2002 speech Negligence: The last Outpost of the Welfare State in which he also referenced the then recent reversal of that trend in judicial decision-making). 10

11 Personal Injury Matters Filed Gauging the Tort Temperature and the Tort Temperature Scorecard b) Cases going to Court The number of cases filed in court is one indication of whether or not plaintiff and defendant expectations are consistent or not. As discussed earlier, most claims settle before litigation would commence, because the parties agree on what the market value for a certain level of injury is. Where the parties have an irreconcilable gap between what each considers fair and reasonable, the plaintiff will commence litigation. Figure 4 below shows the number of personal injury matters filed each year in the Victorian County Court. Unfortunately, data such as this is neither timely nor comprehensive; for example, we are not aware of commencement data of this type being publicly available for NSW or Queensland. However, we would expect Victoria, as the second largest Australian jurisdiction, to be reasonably representative of Australia in total. Figure 4 Victorian County Court personal injuries matters filed by year Source: Victorian County Court Statistics as at 12 November 2010 Note: We expect that the number of matters filed for 2004 and 2005 was lower as the lodgement of some matters would have been brought forward to 2003, while other matters were delayed until the implications of Tort Reform were better understood. Before tort reform there were in excess of 700 personal injury claims commencing through the Victorian County Court each year. More than 1,100 were filed in 2003 in anticipation of the tort reforms that took effect in October of that year. Immediately after tort reform, the number of cases being filed dropped dramatically to around 200 but has been steadily increasing each year reaching over 550 in If filings have continued to increase at the same rate then we estimate that in 2012 they would near pre tort reform levels. c) Indications of the General Marketplace There are a number of ways to look at the experience in the Liability Claims Marketplace. In this section, we look at the following two sources: 11

12 Ultimate Claims Per Thousand Risks Gauging the Tort Temperature and the Tort Temperature Scorecard (i) (ii) Industry data from APRA s National Claims and Policies Database (NCPD) An example analysis of an insurer s own data - Indications from the NCPD In Figure 5 below, also using NCPD data, we show our estimate of the ultimate (i.e. once all claims are reported) claim frequency iv Figure 5 Claim Numbers Accident Year Sources: APRA NCPD; Finity Consulting analysis Claim frequency has been steady at around 14 claims per thousand policies written since While each of the States has different frequency levels, no particular State has observed a persistently increasing or decreasing frequency. Figure 6 shows the ultimate average claim size v of liability claims. We have included high and low estimates to apply to the more recent accident years, as these periods are undeveloped and more dependent on selected development factors. 12

13 Average Claim Size ($'000) Gauging the Tort Temperature and the Tort Temperature Scorecard 40 Figure 6 Average Claim Size Sources: APRA NCPD; Finity Consulting analysis Accident Year Average Claim Size High Low The average claim size has been increasing by accident year since 2004 and From 2005 to 2008 the average increase was about 9% per annum appears to be a bit of an outlier at this stage with an average claim size around 30% higher than other accident years. The average claim sizes in 2010 and 2011 have flattened out, though if the claims develop towards the higher range of our estimates then this will mean a continuation of the upward trend in claims costs. It is too early to tell for these accident years. If we ignore 2009, our analysis of the NCPD data suggests that claims costs are increasing by 3% per annum. At the lower and higher end of our estimates, the increase in average claim size ranges from around 2% to 5% per annum respectively. As frequency is flat, this means that total claims costs are also increasing by this amount. If 2009 is included in our analysis, then the rate of increase in average claim size is approximately doubled. To illustrate what this means for insurers, in Figure 7 and Figure 8 we show the exposure, average premiums, and loss ratios for the public liability class. 13

14 Gross Loss Ratio Number of Risks (millions) Average Premium per Risk ($) Gauging the Tort Temperature and the Tort Temperature Scorecard 3.0 Figure 7 Risk Numbers and Premiums 1, , Number of Risks Written Average GWP per Risk Source: APRA NCPD The average premium per risk written has been falling, and is now 25% below the levels in 2003 and In a stable market, where claims costs increased with inflation each year, we would expect the premium to also increase in line with inflation. Allowing for inflation over this period the premium reduction is approximately 7% p.a. in real terms, giving a total 40% real rate reduction over eight years. 80% Figure 8 Accident Year Loss Ratios 70% 60% 50% 40% 30% 20% 10% 0% Accident Year The recent loss ratios over the last 10 years in Figure 1 appear benign. However, they are accounting year loss ratios, which means they can mask the underlying claims costs for some time due to prior period reserve releases or strengthening. The loss ratios shown in Figure 8 are accident year loss ratios, which gives a better view on recent profitability. 14

15 With a reducing average premium, loss ratios are therefore rising even faster than the increase in average claim sizes. Our analysis suggests that the accident year loss ratio across the whole liability class has increased from around 35% in 2004 to around 55% in 2010, or about 7% per annum. The loss ratio was more than 20% higher in 2009 than in 2008, or around $300 million more in claims. This poorer performance appears across a number of construction and property related classes and in NSW and Victoria. The high loss ratio is due to an increase in claim size, while frequency remained similar to 2008 levels. The limitations in the data provided in the NCPD make it difficult to understand this spike more, though we suspect that economic conditions may be a factor vi. There is some NCPD data available by State. All States are showing an increase in average claim sizes, but NSW and Queensland appear, at this stage, to have slightly bigger increases than the other States. As an aside, we note that the NCPD contains significantly more information than is made publicly available to the market and which would be useful for insurers such as breakdown into bodily injury or property damage. - Example Analysis from an Insurer s Own Data We can get an idea of whether plaintiff and defendant expectations are in the same ballpark by looking at how insurers set case estimates in the period leading up to settling of claims. In an environment with unexpected escalating claims costs we would expect case estimates to be less than sufficient to cover the ultimate cost of the claim. In an environment where claims costs are stable we would expect case estimates to be more than sufficient (savings of up to 20% against estimates would be typical at this stage). We have been able to examine the data for a homogenous portfolio of bodily injury claims for a single entity. The causes of claims are also quite similar, and consist of high frequency but lower severity injuries. This isolates for claims variability (by different causes and also the skewing impact of large claims) which can confound trends in the portfolio. In Figure 9 below, we show the ratio of the finalised cost to the case estimates 12 months prior to the claim finalising. Liability claims are inherently volatile so we have shown the 12 month rolling average. In this figure, a ratio above 100% means that the finalised cost was greater than the case estimate, while a ratio below 100% means the finalised cost was less than the case estimate. 15

16 Ratio of Finalisation / Prior Estimated Cost Gauging the Tort Temperature and the Tort Temperature Scorecard Figure 9 Finalised cost relative to case estimate 12 months prior 250% 200% 150% 1. Liability crisis, escalating costs Case Estimate Sufficiency - 12 Months Prior 3. Claims activity gradually heating up 100% Hotter Cooler 50% Finalisation Period Source: Finity Consulting analysis of confidential data 2. Tort Reform honeymoon period The difference is stark between the period before tort reforms (up to 2002) and the period since tort reforms (2004 and later). Prior to the tort reforms, case estimates set 12 months before a claim finalised were consistently too low the values around 200% suggest that claims were finalising for double what the case estimates had been set at a year earlier. This is a sign that claims managers are playing catch up to escalating costs. Following tort reform, in the honeymoon period from 2003 to 2005, we observe that estimates 12 months prior to settlement are close to the eventual settlement amount. Tort reform had provided pleasant surprises not anticipated by claims managers. Since 2006 the ratio of finalised cost to estimate has been higher than in the honeymoon period, and has arguably been on a gentle upward trend. This trend suggests that the period of good news for claims managers has ended, and claims are now coming in higher than the expectations of insurers. While this analysis doesn t tell us about the level of inflation, it is nevertheless a consistent message with the upward average claim size trends observed from the NCPD data. One of the main lessons from this analysis is that Tort Temperature studies need to focus on finalised claims and the relativity to estimates. Traditional actuarial approaches based on triangles and development are not effective. We note that the usual limitations in relying on case estimates apply for this analysis, such as change in case estimation philosophies or staff. We have taken some care to consider these impacts judgementally. 16

17 d) Opinions of Marketplace Players The market for personal injury liability claims is a system that includes insurers, lawyers, and the courts. The numbers can only go so far in telling the story and the views and expectations of the stakeholders are important to get a complete picture. We have spoken to a range of stakeholders and have summarised their key views below. Who said it? Claims Managers What did they say? Injuries attracting higher benefits Claims managers have reported that even common injuries are attracting much higher settlements. However, each of these claims remains quite small individually and it is not worthwhile for the insurer to appeal. Increased litigation Previously, letters of demand will be received around 18 months following first being notified of the claim, with the aim to settle the claim. This is no longer true and claim managers are now receiving letters of demand earlier as plaintiffs become increasingly litigious, increasing the costs to the insurer. Differences between states NSW is seen as the problem State, while Victoria has been relatively benign. However, there are inklings that Victoria is starting to be more litigious with personal injury claims. Lawyers (Defendant) Psychiatric Element Damages for psychiatric injury are usually overlaid on top of other settlement costs. The level of damages paid for these injuries has increased notably. Impact of judgements There were claims being settled at higher than expected levels in first instance decisions up to around These higher settlements have led to plaintiffs being more willing to roll the dice by taking matters to court. More recent judgements have come down from these levels as the courts self-correct in line with community expectations of appropriate amounts. Differences between states Generally echo the opinions of claims managers. In addition, the lawyers expressed to us that there is an element of forum shopping due to differences in state legislation on torts, with the differences diverging further over time. Different legislation by State also increases insurer costs, as specialised claims teams are required for 17

18 each jurisdiction. High Court decisions High Court decisions have been viewed as fair and have brought back decisions by lower courts which expand the scope of torts. Actuaries Superimposed inflation nervousness Actuaries have been concerned with superimposed inflation since around Before this, the feeling was that it was under control. The view is that superimposed inflation will continue at around 2% to 4% per annum (adding expected wages inflation gives total claims cost inflation of between 6% and 8% per annum), but it is difficult for actuaries to be comfortable with forecasting at this level. Profitability of the liability class Profits continued to be made on liability classes, and of the actuaries we spoke to, none consider this class loss making yet. However, the general view is that the days of super profits are gone and this class is either making target returns or less. Insurers SME vs Corporates SMEs have been performing well, and the poorer performance is coming from the bigger corporate end of the market. In addition, SME claims are coming in quicker and there is less uncertainty with costs. Pockets of good performance There remain pockets of good performance. Some insurers we spoke to have loss ratios well below those we see from the NCPD (of course the implication is that some have loss ratios well above NCPD average). The qualitative information we received from our meetings is consistent with the quantitative assessments from above. More specifically: We have observed increases to claim sizes. Our discussions with experts suggest this is due to larger benefits being paid for common injuries, more damages awarded for psychiatric injury, and from increased legal fees from earlier engagement of lawyers. There is no suggestion that the frequency of claims has increased. Confirmed our view that the High Court judgements do not appear to be either pro-defendant or pro-plaintiff. While our focus is on understanding the tort environment, there was fairly consistent feedback from insurers that the liability class remains profitable and that the tort environment has not eroded profitability as it did just before tort reform. 18

19 This is supported by our projected loss ratio of between 50% and 60% for 2011 from NCPD (our low and high estimates range), which is profitable but at a lower return on capital than in recent history. There is more concern on liability covers for larger corporate risks. It was suggested that the claims costs for these policies have been increasing more rapidly to the point that these policies may no longer be profitable, or at least to the point where target return on capital levels may not be met. The views offered by these discussions are by no means conclusive; however they do help to paint a fuller picture of what has been happening. Indeed, we have learned things about the experience from talking to the experts that would not be possible through any form of data interrogation. 6. Why is it important for Insurers? Understanding the current environment on recent and future claim costs is the focus of every insurer s management and Board. In Figure 10 below, we show some areas where the Tort Temperature may be of interest to an insurer. The Tort Temperature has implications from balance sheet adequacy to directors in understanding the risk profile of their company. Figure 10 Why Tort Temperature is important for an insurer Source: Finity Consulting Design: Vetro Design The susceptibility of property classes to large catastrophic events is generally well understood by insurance executives, with portfolios and reinsurance arrangements structured to manage the exposures to natural disasters. Concentration risks also exist for liability policies, but are somewhat less transparent than understanding the location of insured properties. For example, while the Tort Temperature remains hot all liability claims may settle above expectations during this period leading to an elevated claim cost for the insurer, analogous to a catastrophe event in property classes. However unlike property classes, after such a catastrophe event a higher watermark is set which will lead to elevated claim cost in future even if the Tort Temperature returns to neutral levels. 19

20 A property insurer may look at weather patterns such as El Nino to indicate future risk levels and adjust their strategy accordingly. Liability insurers would ideally do something similar, but rather than understanding weather patterns, the focus would shift to gauging the Tort Temperature and estimating its impact on the cost of liability claims. While less of a science than measuring weather patterns, we can still try to measure the temperature across the liability claims market and use this to inform the strategy for liability products. 7. Limitations of Traditional Actuarial Analysis When measuring social and judicial inflation, actuaries have traditionally thought in terms of superimposed inflation. However, the usefulness of this analysis is limited by its measurability and relevance. Measuring past superimposed inflation is complicated by a high degree of random variability and difficulties identifying like-for-like claims. Even if it could be measured reliability, it remains a backward looking measure and does not answer what we really want to know, that is, what level of superimposed inflation will occur in the future (Taylor, 2002). Measurability: How good is current Superimposed Inflation analysis? One thing we observed from talking to actuaries is that there is a range of estimates for prospective superimposed inflation from 2% to 6%, with considerable uncertainty around what the correct level is. The range of opinions could be justified based on differences in the portfolios, though in our experience it stems more from the difficulty in analysing historical superimposed inflation using traditional actuarial techniques. In looking for superimposed inflation, actuaries may measure the average payments by payment or settlement period and the trend over time. From these trends, the actuary would generally select an assumption for reserving or pricing applications. There are various refinements to this approach, such as operational time adjustments to adjust for differences in the speed of settlement. An example of superimposed inflation analysis of a high frequency and low severity class looking at changes to PPCFs by finalisation year is shown in Figure 11 below. 20

21 Year on Year Change in Average Claim Sizes (%) Gauging the Tort Temperature and the Tort Temperature Scorecard 50% Figure 11 Example of current superimposed inflation analysis 40% 30% 20% 10% 0% -10% -20% -30% -40% Finalisation Year Change in Average Claim Size 3 Year Moving Average Source: Finity Consulting analysis of confidential data In the example liability portfolio, the year on year variability in settlement outcomes is evident. In some finalisation years, the PPCF changed by 30% from the previous year, with both upward and downward changes of this magnitude observed. The problem with this analysis is that the settlement sizes are quite volatile year on year. The analysis is sensitive to the presence (or absence) or large claims. Capping these large claims is not particularly helpful either, because you risk suppressing the very thing you want to measure. Some more sophisticated approaches attempt to split claims into more homogenous groups. However, this process is rarely able to reliably identify like for like claims as each claim will be unique with its own set of circumstances. Even for more homogenous classes where we know a lot about the claims and have large volumes of data, such as CTP, traditional actuarial analysis will show large swings in superimposed inflation year on year. The actuary is still required to exercise a large amount of judgement in selecting a Superimposed Inflation assumption (which will typically be somewhere between 2% to 6%). As a consequence, the traditional approach will not be able to discern between a deterioration of Superimposed Inflation experience or random year on year variability, and as such it is not reactive to changes in the tort environment. Relevance: Backward looking analysis Traditionally when we considered Tort Temperature we thought of superimposed inflation. The problem with this is that superimposed inflation analysis attempts to measure what has occurred. The measurement problems were discussed above, but even if we could reliability measure past superimposed inflation, it is not an indicator of future experience. 21

22 2013 Premiums Observed Loss Ratio Observed History Gauging the Tort Temperature and the Tort Temperature Scorecard Figure 12 illustrates that when we look at the experience for a long tail class like liability, we don t actually know very much about the experience. For the latest accident year, we only ever have the one development year of experience to look at, and it is what might happen in the next few years (starting from 2012 onwards) that is important to an insurer s insurance liabilities. When setting premium rates for next year, it is the Tort Temperature for 2013 and onwards that impacts on premium adequacy. Figure 12 The data we see versus the periods we are concerned with How we observe liability experience What does it really show us DY1 DY2 DY3 DY4 DY Accident Year It is the tort temperature for 2013 onwards that we need to consider for premium rates The tort temperature for 2012 onwards needs to be considered for reserving. The point of Figure 12 is to illustrate a very distinct break between: 1. the observed historical superimposed inflation, which we can attempt to measure but the cost will have already been absorbed, and 2. the Tort Temperature impacting on insurance liabilities and future policies, which can t be measured but where we still have an opportunity to change our premium rates or how we manage the claims (settlement years around two to three years from the time of measurement). The use of superimposed inflation analysis is discussed by Taylor (2002) in a Guest Editorial of the British Actuarial Journal: how should you allow for [the historical] trend in forecasts of future claims experience? The distinction between observed experience and future experience means that it would be incorrect to simply extrapolate recent observed superimposed inflation. If superimposed inflation has been high over a number of years, assuming that this continues may lead to uncompetitive premiums and over-reserving of insurance liabilities. The reverse may also be true. Taylor poses two questions in his Guest Editorial to which he states he has no answers: Is it in any way reasonable to damage an insurer by carrying out financial forecasts on the basis of an extrapolated trend when one can fairly confidently assign it a probability of virtually zero? 22

23 In the absence of any clear guidance as to when and how that trend will be interrupted, how is one justified in any lesser forecast? (Taylor 2002) We too do not have a response to these questions, though in this paper we hope to offer an approach that moves measurement of social and judicial inflation in personal injury claims one step closer to a solution. 8. Moving from Traditional Actuarial Analysis to Root Cause Analysis As the deficiencies of traditional numerical analysis are evident, we suggest that efforts should turn to understanding the Root Cause drivers of claims cost. This could include understanding whether more cases are going to court or whether there have been changes in the behaviour of claimants. In doing so, we can postulate on whether these trends will continue and the likely impact for an insurer. Figure 13 below highlights some of the benefits of understanding Root Causes. Figure 13 Benefits of understanding Root Causes Traditional actuarial analysis Looking backwards Hard to measure Impossible to predict and we generally extrapolate historical trends Doesn t offer any insight directly useful for the business Lacks a control cycle Move from outcome to root cause Links the cause to the outcome Meaningful assessments can be made Could offer insights that can be used by the business to make decisions An immediate benefit of this approach is transparency. We move to measures which are more easily understandable and observable, and away from only looking at outcomes. The ease of communication of observed experience will help to foster constructive discussion within an organisation. 23

24 One criticism we have of traditional superimposed inflation analysis is that it generally lacks a useful Control Cycle. The observed superimposed inflation outcomes are notoriously volatile, and an actuary would focus on long term trends rather than any movement over the shorter term. This might be fine for reserving purposes, but not particularly helpful for strategic or pricing decisions. 9. Bringing it all together with a Tort Temperature Scorecard There is no single source of information that will allow us to understand the tort temperature. What we can do is to consider information from a range of sources and form a view of the temperature. Different views will exist across an organisation and so will each individual s view of its importance for the future. What is needed is an approach to consolidate the range of views, and once an agreed view is settled upon the insurer may modify its strategy consistently across the organisation. We propose that a Tort Temperature Scorecard (TTS) process can be used to achieve this. Aim of the Tort Temperature Scorecard It is not possible to predict future superimposed inflation or Tort Temperature with any kind of certainty. There are simply too many factors which influence the outcomes, some of which have not yet occurred or cannot be foreseen. Some aims of the scorecard approach include: Summarise into one place what we know and consider important. The scorecard will be a combination of qualitative and quantitative assessments. Relatedly, it should be a discussion point where all stakeholders can contribute in a meaningful way. Help form a view of a possible outcome based on piecing together what we know now and judgementally assessing what might occur in the future and the impact this might have. It is designed to make clear the delineation between what we see in experience and the future. Give an indication of future Tort Temperature which management, or other decision makers, can act upon. It will be a single consolidated view across an organisation so that there is some consistency in management action or responses. Be something that can be tracked over time, with previous judgements back tested and improved upon. What the Tort Temperature Scorecard might look like We have come up with a sample TTS for the Australian Industry below, but the design of the TTS should differ to suit a particular organisation. However, we suggest that the TTS should at least have the following key features: 24

25 At least consider claim frequency and claim severity separately. Some more detailed scorecards might separately look at defence costs or key heads of damage. Clear delineation between observed experience and the forward looking impacts. This might include some measure of likelihood and impact. Not try to be too accurate. Our sample TTS gives a score between nil and ten. Arguments over whether a certain score is 6.5 or 7.0 are unhelpful. The level of precision is not needed (and not possible) and loses sight of the main aims of the TTS. Some other features that organisations might include: Considering the score separately for reserving and pricing purposes. Having separate scorecards for different product lines. E.g. SME and corporate liability products might be considered separately. Below we give a view of the Tort Temperature score for the Australian Industry as an example of how the TTS might work. To do this, we have focused on industry level information and fairly high level comments. However, the TTS should take into account the specific nuances facing a particular organisation, and as such the factors that are included in the TTS should reflect this. Our suggestions here are a starting point. We expect that the look of the TTS will develop over time as we get better at identifying what factors are important for an organisation to look at. 10. How the Tort Temperature Scorecard may look for the Australian Industry In the earlier sections of this article we have shown the type of analysis and discussions that are inputs into the TTS. Our attempt to summarise the Australian industry tort temperature using a TTS is shown in Figure 14 below. 25

26 Figure 14 A view of the Tort Temperature for the Australian Industry Factor Observed trend Impact future underwriting periods? Claim Frequency Impacts Rationale Reported claim numbers Flat Yes No evidence to suggest reported claim numbers will increase 75% 5 Social attitudes Policy terms / wordings External claim drivers (e.g. Economic Factors) Limited impact observed so far No changes impacting on claim frequencies Likelihood of trend continuing 100% = Certain 0% = Won't occur 10 = Detrimental impact 5 = No impact Potentially General view that we are again becoming a more litigious society 25% 6 Yes Unlikely to change in the near future 90% 5 Recent economic slump Maybe Reduced economic activity might reduce claim frequency 25% 4 Magnitude of Impacts 0 = Beneficial impact Claim Severity Impacts Claim Frequency Score 5 Claims going to litigation Steadily increasing, leading to higher expenses Yes (and maybe settlements) Steady trend observed. Lawyers are advertising more. 80% 8 Court decisions Legislative changes Policy terms / wordings Recent court decisions have backed off slightly No recent talk of additional reform No changes impacting on claim severity Possible Lawyers attribute this to new judges being appointed. 50% 4 Yes Not a key political topic at this moment. Lack of 'political will' likely to continue, adding to the tort temp. 75% 7 Yes Unlikely to change in the near future 90% 5 Claim Severity Score 7 Expected Tort Temperature (10 = Hot; 5 = Neutral; 0 = Cold) 7 26

27 On a scale of nil to ten and where a score of five suggests a stable tort temperature in the future, we estimate that the Australian environment is around seven. This is the view of the authors of this Paper only, and other assessments of the Tort Temperature may lead to different outcomes. 11. Framework for Implementing the Tort Temperature Scorecard for an Insurer The process of filling out the TTS must be one that is shared and discussed across an organisation. The benefits of using a TTS approach arise from the journey as much as the destination. The TTS is a simple tool and designed to involve all stakeholders in its construction, yet is also effective from a communication perspective. Once the organisation comes to a view of the tort temperature on its own liability portfolio, it can then be used to inform or influence planning across the insurer, including risk appetite, pricing, and underwriting (along with all the other areas we mentioned earlier). Figure 15 Framework for implementing the Tort Temperature Scorecard The process of populating the TTS should be completed to feed into an insurer s business planning process. The TTS may require substantial elapsed time in assessing the numerical experience and talking to experts. 27

28 How often should you review the TTS? Repeating the full framework annually may not be necessary as factors impacting liability claim costs are slow to develop. The full process might only be performed once every two years. During the years in between, you might just look at experience for anything particularly abnormal or consider if there have been changes in legislation or notable judgements. Analogy to Economic Forecasts Banks and Investment Managers will generally have an economics team which issues forecasts of future economic conditions. They do this by looking at a range of past economic information and current conditions. The result is that there will be a view on factors such as future interest movements or equity returns. Investment decisions in turn will be influenced by these forecasts. Every economics team will have its own view of important leading indicators, and the economics teams with models which better assess future outcomes will have a competitive advantage against their peers. We see the TTS as a similar tool for competitive advantage. In trying to guess the future Tort Temperature, an organisation will develop a way of estimating potential future impacts on its liability portfolio. Companies will each come up with ways to improve their own models, and use the results of their forecasts to influence whether they need to increase/decrease liability premiums, to decide the segments in the market to target, or whether corrective actions needs to be taken. Those organisations that invest in it and do it better than their competitors will in turn perform better. 12. Practical applications of the Tort Temperature Scorecard framework This section gives examples of a few practical applications of using the Tort Temperature scorecard in the areas of: (a) (b) (c) Reserve adequacy Underwriting decision making Risk appetite and risk management (a) Reserve adequacy The actuarial reserving process will look firstly at claim frequency the TTS should add a little confidence to interpretation of recent trends but is not of high importance. Trends in PPCF, PPCI and ICD are critical diagnostics. The TTS seriously informs judgements about whether the movements over the last two to three years are systemic or random, and allows establishment of a history that has calendar year growth varying by year, rather than just needing a linear trend. 28

29 The process of adjusting historical values to current levels can be done with more confidence. The basic current value assumptions for the reserving will adapt more quickly to recent conditions, and will move from year to year based more on established reasons rather than guesswork. Future superimposed inflation assumptions will be clearly delineated from historical observation. Lower or higher assumptions may be used for the first couple of years of projection, reverting to long term assumptions after that time. As a result reserving should be more responsive. Prior year reserve releases or increases will still occur, but should occur more quickly. Risk factors regarding reserve adequacy will be more specific to recent trends and conditions, possibly leading to variations in the monitoring priorities for the coming year. If a certain amount of company resource is invested in monitoring, it is logical that the use of that resource be prioritised based on high risk areas from time to time. Lastly, communication of reserve changes and adequacy issues to senior management and Board will become more effective because it is related to issues in the external environment (albeit still with a degree of uncertainty) rather than solely being a function of observations on claim outcomes. (b) Underwriting Decision Making The annual TTS process will inform the tilts in underwriting appetite and price movements across States and industry segments. Occasionally a particular segment will be called out for policy wording changes or defensive underwriting (related for example to psychological injury trends). The underlying data collection (qualitative and quantitative) will be undertaken by State for the major markets of the insurer, and underwriting stance will receive more focussed State differentials than before. Confidence is added to State variations in profitability analyses (or alternatively unusual results are more confidently discarded). (c) Risk Appetite and Risk Management To the extent that the Risk Appetite articulated for the company has specific reference to long tail business, the TTS update will be referenced back to the Risk Appetite at each update to identify areas where the company might be getting close to the wind. It would be expected that superimposed inflation and/or tort environment appear as significant items in the risk register in one form or another. The TTS update will inform the review of ratings (the heat map) for these risks, and serves as the occasional deep dive on those risks that feeds back into the risk management cycle. The risk management function will then act as a discipline and quality control on the TTS updates, and may have useful suggestions to make. 29

30 It is likely that this process (integrating risk management with the TTS) will lead from time to time to specific further investigations or actions in response to increases in perceived risk. Benefits of the Tort Temperature Scorecard and Implementation Framework Figure 16 below summarises some of the key benefits we see of implementing a TTS framework. We have already discussed the benefits of the TTS framework as we ve gone through this Paper, so we will not discuss them further here. Figure 16 Benefits of the Tort Temperature Scorecard framework Transparent rationale Distinction between observed trends and the future Better communication Benefits of the TTS Incorporates views from various experts Can review changes year on year Fosters consideration of factors driving costs 13. Concluding remarks Understanding the Tort Temperature has always been challenging, and it is tempting to put it into the too hard basket. However, the tort environment is an important risk factor for insurers writing liability or indemnity policies where claims are often decided by legal action; and in particular those involving bodily injury. The indications are that the Tort Temperature is higher now than it has been over the last decade. How much higher and the implications for the future are what should be important to all insurers. In this Paper we have suggested that a Tort Temperature Scorecard process could be implemented as a way to bring together all the knowledge which resides 30