1. Objectives and background
The Employers' Liability (EL) market has come under recent scrutiny, with premium increases being cited by businesses as generating significant increases in their own costs. The Department of Work and Pensions has published an interim report as part of policy research aimed at evaluating this claim and, more broadly, the market for EL insurance (DWP, 2003) [Endnote 1] : this report has been described as an "agenda for action" by Nick Brown, Minister for Work [Endnote 2] , and it seems clear that policy is envisaged in this area.
A potentially important component of the above premium increases may be the costs to insurers of defending EL claims, and these costs are the focus of our report. Information on the costs of pursuing legal cases is, in general, in notoriously short supply. This is despite the fact that Lord Woolf's procedural reforms (implemented in April 1999), CFAs (1999), recoverability of additional liabilities (April 2000) and a recent set of cases [Endnote 3] have all had implications for legal costs. To the extent that some of the above have aimed at reducing the level of costs, while others have aimed at changing the nature (or the burden) of these costs, it is possible that the combination of such changes may have altered the nature of EL claim costs and, therefore, fed through to premiums [Endnote 4] .
Against this background, we were approached by the Department for Constitutional Affairs (DCA) to undertake data collection and analysis with the following objectives:
To estimate the extent to which total costs paid to claimants' solicitors for the settlement of low value EL claims has changed over the last five years.
To indicate which components of total costs may have contributed most to any changes observed in total costs paid over this period.
To indicate whether a reliable estimate can be made of the current base cost of settling standard low value EL claims.
As in our recent report to the Civil Justice Council (CJC; see Fenn and Rickman, 2003), it is perhaps important to make clear what we were not seeking to do. In particular, we were asked only to describe the trends in costs over our sample period, not to explain them, although we feel that inferences may be drawn from the results reported below on grounds of timing and logic.
The structure of the report is as follows. We first set out our methods and data collection strategy. Then, we describe our data. Section 3 presents our findings from these data and introduces findings from data we have collected elsewhere (which enables us to corroborate our findings). The final section provides conclusions and some caveats.
2. Methods and data
2.1. Methods: Statistical inference
We collected data on EL claims as part of our previous project for the CJC and, therefore, had information similar to that used in this earlier work. As before, our population of interest was completed claims with a final settlement value less than £15k in England and Wales from 1997-2002. Our sampling strategy also followed from the CJC project: it collected claim level data from defendants' payment records and claimant solicitors' records. We asked the Association of British Insurers (ABI) and Lloyds to facilitate collection of defendant data and the Association of Personal Injury Lawyers (APIL) and Thompsons to do likewise for our claimant data. We felt that these stakeholders would provide us with a reasonable spread of data.
Finally, we sought to test hypotheses about how costs (and components of costs) paid on the above claims have varied over the period in proportion to damages, claim duration or type of claim. It is important to assess costs relative to various types of "output", such as those mentioned here.
2.2. Data
We now describe our data in more detail. Having sent data templates for comment to a number of the parties who had agreed to supply data, we sent them to each source to indicate the variables we were keen to have information on (see the Appendix to our previous CJC report). The sources for our data were defendant payment records, claims negotiators and claimant records. We were often directed towards claims negotiators by insurers who did not keep data of the type we were requesting. In practice, this had the twin benefits of allowing relatively consistent definitions of variables and providing large data sets.
The variables we were interested in were
Not all sources provided data on all the above. Typically, individual defendant insurers were only able to provide a figure for total costs agreed, whereas claims negotiators were able to breakdown this figure into its component parts. Where a detailed breakdown was available, we asked for base costs to be defined as costs agreed in relation to the claimant solicitor's input net of any success fee or additional liabilities. Defendant solicitors tended to supply cost data inclusive of VAT whereas claimants supplied data net of VAT. For consistency, we have chosen below to report data inclusive of VAT
Our final database consisted of very full information from two sources: a major EL defendant insurer who supplied data on cases with damages below £15,000 that were closed between 1997 and 2002; and a pooled dataset from claims negotiators for cases closed between 2000-2002. Although we cannot determine the extent of the market share covered by these sources, we are aware that our defendant insurer is a 'major player' in the EL market, while our pooled database covers a variety of significant sources: thus, we believe that our information is a satisfactory representation of the market for the purpose at hand [Endnote 5].
3. Results for EL claims
3.1. Company-specific data set
We begin by exploring the insurer dataset that permits us to identify trends from 1997 onwards.
3.1.1. Trends in costs
The insurer provided data on 95,775 EL claims with damages below £15,000 over the whole period; we have analysed the data by month - mean damages paid, and mean total costs agreed (combined profit costs, disbursements and additional liabilities). Figure 1 indicates how the firm's monthly claim count has varied over the period. Although there is a reduction in the number, this is hard to interpret without detail on the rest of the EL market over this period: Figure 1 may indicate a declining market share for our insurer, or a general decline in claims over the period.
Figure 1: Monthly claim count for EL claims 1997-2002;
Defendant insurer
Now consider the trends in agreed costs and damages in the insurer database: see Figure 2. Both mean monthly damages and mean monthly costs (defined by the insurer as claimant costs and disbursements agreed, including VAT) have been on a rising trend for the insurer's overall portfolio of claims (recall that we only have data from this insurer on claims below £15,000 in damages). There seems to be a rough proportionality between costs and damages, although some sign of increasing costs relative to damages towards end of the period.
Figure 2: Agreed costs and damages for EL claims 1997-2002;
Defendant insurer
3.1.2. Claim duration
We now consider the duration of claims from the defendant insurer (Figure 3). Over the period, duration appears reasonably constant, averaging between 700 and 800 days. Unlike the RTA data in our CJC report, there is less clear evidence of 'spikes' at around the time of Lord Woolf's reforms, or some of the more recent landmark rulings on success fees [Endnote 6].
3.1.3. Jurisdictional results
The defendant insurer database allows us to compare the above variables across Scotland and England and Wales (EW). Thus, Figures 4 - 7 do this. As expected, Figure 4 shows a larger monthly claim volume in EW than Scotland, with declining trends in both cases. Mean costs are rising in both jurisdictions (Figure 5): those in Scotland appear a little lower than in EW while their higher variability stems from a smaller number of claims. Mean damages are rising in both cases; again there is more variability in Scotland though apparently a similar mean to that in EW. Finally, although case duration in Scotland exhibits more variation (as above), there is some evidence of an increase - in comparison with EW.
Figure 4: Monthly count by jurisdiction 1997-2002
Defendant insurer
Figure 5: Mean costs by jurisdiction 1997-2002
Defendant insurer
Figure 6: Mean damages by jurisdiction 1997-2002
Defendant insurer
Figure 7: Mean case duration by jurisdiction 1997-2002
Defendant insurer
3.1.4. Results by claim-type
Figures 8 - 11 break down claim volumes, costs, damages and durations by the nature of the claim: in particular, we focus on 'Accident', 'Deafness' and 'Other' claims (which includes asbestosis and vibration white finger claims). The trends in the number of claims in Figure 8 help explain the falling volumes observed in Figure 1 above: there has been a dramatic fall in the number of deafness claims, which were the majority in 1997, and have more or less vanished in 2002 [Endnote 7].
In Figure 9, we can see a reason for the steeply rising trend in the insurer's claim costs; as we have seen, deafness claims were predominant in the early years, and declined quickly thereafter. Figure 9 shows that deafness claims were also relatively low cost. Hence, a portfolio effect is taking place, whereby relatively low-cost claims are being replaced by an increasingly high proportion of relatively higher cost ones, thereby pushing up the average cost of claims in the insurer's portfolio.
Taking Figures 10 and 11 together, there is some evidence of a selection effect in the remaining deafness claims: they appear more complex (higher duration) and larger (higher damages) than before.
Figure 8: Monthly count by claim type 1997-2002
Defendant insurer
Figure 9: Mean costs by claim type 1997-2002
Defendant insurer
Figure 10: Mean damages by claim type 1997-2002
Defendant insurer
Figure 11: Mean case duration by claim type 1997-2002
Defendant insurer
Figure 12 presents cost/damage data on accident claims alone in order to control for the portfolio and selection effects noted above. For these claims, combined claimant costs and disbursements for sub-£15,000 claims have increased from approximately £2000 to approximately £2800 between 2000 and 2002 (an increase of 40%), during a period when mean damages were relatively constant at just over £3,000 on average.
Figure 12: Mean costs and damages 1997-2002
Closed EL accident claims under £15,000
Defendant insurer
Mean costs by jurisdiction 1997-2002
We move now to consider the pooled dataset obtained from several claims negotiators, some directly and some via their insurers. The companies represented in this pooled dataset of almost 150,000 personal injury claims over three years are: Norwich Union, CGU, London and Edinburgh, RSA, Axa, Avon, Fortis, NFU, Highway, Ecclesiastical, CIS, Cornhill, Methodist Insurance Plc, MIB, Ensign Motor Policies, Westminster, Chaucer, Town and Country, Zenith, Endsleigh Insurance Services Ltd and Orion. Clearly, these companies provide a significant degree of market coverage (though not all of their EL cases will be present in the data set) and, because of the numbers of claims involved, allows results to be produced with considerable statistical confidence. As mentioned earlier, because of the recent growth of claims negotiating (post-2000), these data cover the period 2000-2002; in terms of EL claims only, this yields a data set with 8,236 closed claims in total. This data set is the basis for the analyses presented below.
3.2.1 Overall trends in volume, costs and damages
To begin, Figure 13 shows the build-up of settled cases on the claims negotiators' books - this reflects their growing market presence. This growth can be observed in claims with damages above and below £15,000.
Figure 13: Monthly count by claim value 2000-2002 Closed EL claims
Closed EL claims
Pooled defendant database
Figures 14 and 15 look at agreed costs and damages for all claims (including those over £15,000 damages) and for claims with damages under £15,000 respectively. The costs here are "profit costs" [Endnote 8] - that is the hourly fees not including either disbursements or any success fee or ATE premium where those are relevant. It is apparent that while damages have increased, then decreased, slightly over the period, this trend has been reversed for the smaller (sub-£15,000) cases. Overall, the gap between damages and costs in Figure 15 appears to have narrowed over the period 2000-2002, while their ratio has remained more constant (on average) in the total data set.
Figure 14: Mean costs and damages 2000-2002
Closed EL claims
Pooled defendant database
Figure 15: Mean costs and damages 2000-2002
Closed EL claims under £15k
Pooled defendant database
3.2.2 CFA vs non-CFA claims
A convenient feature of the pooled data set is that it distinguishes between CFA and non-CFA claims. As Figure 16 shows, the relative use of these fee schemes has switched since 2000, with CFAs now constituting the bulk of EL claims. (This was not the case in our previous analysis of RTA claims - presumably because of the role of BTE insurance in these claims.)
Figure 16: Monthly count by claim type 2000-2002
Closed EL claims
Pooled defendant database
In Figure 17, we see that a similar trend is apparent in claims with damages below £15,000, although CFAs are now used in an even greater proportion of these claims.
Figure 17: Monthly count by claim type 2000-2002
Closed EL claims under £15k
Pooled defendant database
Next, we consider the elements of costs in small value (i.e. below £15,000) CFA and non-CFA cases. Figure 18 reports these for non-CFA cases, and Figure 19 does the same for CFA cases. Looking first at the trend in base costs for low value non-CFA claims, we see that these have increased from just over £2000 to just over £3000; disbursements seem relatively more stable at just under £1000.
Figure 18: Mean costs by category 2000-2002
Closed non-CFA EL claims under £15k
Pooled defendant database
Turning to CFAs in Figure 19, the picture is complicated by the existence of several categories by which costs are recovered, including success fees and ATE premiums. Disbursements seem very similar to the non-CFA cases. If we add the base costs to the success fees and ATE premiums, we would get a similar picture here to the previous chart: the indication is that there may be little difference between overall costs recovered from CFA and non-CFA cases.
Figure 19: Mean costs by category 2000-2002
Closed CFA EL claims under £15k
Pooled defendant database
We can confirm this by looking at total costs (base costs plus disbursements, success fees and ATE premiums) for 'small' CFA and non-CFA cases in Figure 20. It is not clear that there is any statistically significant difference between CFAs and non-CFAs in terms of the overall costs of running a claim (if anything CFA costs seem slightly lower).
Figure 20: Mean total costs by claim type 2000-2002
Closed EL claims under £15k
Pooled defendant database
Interestingly, despite their (roughly) equivalent costs, Figure 21 indicates that CFA cases have, since 2000, been associated with lower damages than their non-CFA counterparts: awards appear to be falling with CFAs, whilst remaining stable with non-CFAs (although this conclusion is tentative only as the numbers of CFA claims in the early part of this period is relatively low).
Figure 21: Mean damages by claim type 2000-2002
Closed EL claims under £15k
Pooled defendant database
3.2.3 Litigated vs non-litigated claims
The pooled database has the added advantage of allowing a break down of claims by whether or not they were litigated (i.e. pre- and post-issue). Figures 22 and 23 provide the evidence for all claims, and those with damages below £15,000, respectively. It is apparent in both figures that a larger proportion of EL cases are litigated (regardless of damages) than in our RTA study. In the case of small-value claims, approximately 40% of closed claims were litigated in 2002 (the trend having been increasing since 2000); the equivalent figure in our RTA study was 18-20%, with little evidence of a rising trend since 2000.
Figure 22: Monthly count by claim stage 2000-2002
Closed EL claims
Pooled defendant database
Figure 23: Monthly count by claim stage 2000-2002
Closed EL claims under £15k
Pooled defendant database
Interestingly, as Figure 24 indicates, whilst the litigation rate (i.e. issued versus non-issued claims) arrives at the 40% mentioned above by 2002, it was actually falling quite fast until the end of 2001.
Figure 24: Monthly litigation rate 2000-2002
Closed EL claims under £15k
Pooled defendant database
Comparing mean total costs (combined base costs, disbursements, success fees and ATE premiums) across litigated and non-litigated claims, Figure 25 shows that the former were fairly constant over the period; in contrast, mean total costs for non-litigated cases showed an increase from around £2,000 to around £3,000 over the period. Figure 26 shows that the increase in damages for both litigated and non-litigated cases were, if anything, falling over the period concerned; consequently the increase in non-litigated costs cannot be explained by a trend towards more serious claims. These relative trends in costs are consistent with those in our RTA work, where we found a 50% increase in total costs of non-litigated cases - possibly due to Woolf front-loading.
Figure 25: Mean total costs by claim stage 2000-2002 Closed EL claims under £15k
Closed EL claims under £15k
Pooled defendant database
Figure 26: Mean damages by claim stage 2000-2002
Closed EL claims under £15k
Pooled defendant database
3.2.4 Costs in relation to damages
Finally, we ask how base costs (net of VAT) and damages are related. In particular, Figures 27 and 28 mirror those in our RTA work, for all claims, then for claims closed pre-issue, respectively [Endnote 9]. Both figures exhibit a significant degree of proportionality up to a case value of approximately £10,000, with a levelling off for claims with a higher value than this. Figure 29 shows the pre-issue costs by claim type (CFA or non-CFA), and as before there do not seem to be very marked differences in costs between types of claim. It is important to note here that these diagrams relate to mean costs from the whole period 2000-2002, and therefore will underestimate the most recent level of costs.
Figure 27: Mean base costs by damages
Closed EL claims under £15k
Pooled defendant database
Figure 28: Mean base costs by damages
EL claims closed pre-issue under £15k
Pooled defendant database
Figure 29: Mean base costs by damages and claim type
EL claims closed pre-issue under £15k
Pooled defendant database
4. Conclusions
We can summarise our findings as follows:
A decline in the overall volume of EL claims against one EL insurer has coincided with an increase in average claim costs, partly because of a portfolio effect as the, relatively cheaper, deafness claims have declined in number. This portfolio effect can be removed by focussing on accident claims alone, which appear to have been fairly constant in volume terms over the period 2000-2002.
Combined claimant costs and disbursements for sub-£15,000 accident claims against the EL insurer increased from approximately £2000 to approximately £2800 between 2000 and 2002 (an increase of 40%), during a period when mean damages for such claims were relatively constant at just over £3,000 on average.
Drawing on evidence from a pooled (defendant) claims database, the increase in total costs including disbursements was much larger in non-litigated EL cases than litigated ones between 2000 and 2002, perhaps indicating a 'Woolf-effect' due to front-loading.
There is little difference between CFA and non-CFA EL claims with respect to the level and growth rates of mean agreed total costs including disbursements and success fees and ATE premiums. Thus, the cost increases we have observed cannot readily be ascribed to recoverability rules introduced in April 2000.
While costs have risen in similar fashion for CFA and non-CFA EL cases, there is some tentative evidence that damages in CFA cases have fallen.
Base costs of EL cases rise proportionately with damages in a fairly predictable manner up to a case value of approximately £10,000.
While we believe that the findings listed above are robust, the following caveat might be raised regarding the data analysis and the need for further work. We have made no direct statistical tests of the impact that procedural and funding reforms have had on costs, even though we have suggested that the Woolf reforms (rather than changes to the recoverability rules) seem to be a candidate explanation on grounds of logic and timing. More analysis and more detailed data collection are necessary to identify the causes of these trends with statistical confidence. This is on our agenda for future work.
References
DWP (2003) http://www.dwp.gov.uk/publications/dwp/2003/elci/dw2583_employers_review.pdf
OFT (2003) http://www.oft.gov.uk/Market+studies/default.htm
Fenn, P. & Rickman, N. (2003): Costs of Low Value RTA Claims 1997-2002: A report prepared for the Civil Justice Council, January 31.
Endnotes
In parallel, the Office of Fair Trading has recently produced a summary of key findings from a fact-finding study of UK liability insurance markets (OFT, 2003)
E.g. Callery vs Gray, Sarwar vs Allam, Halloran vs Delaney, and the Claims Direct and Accident Group cases.
It should be borne in mind, however, that the EL insurance market is prone to cyclical fluctuationsin premiums and underwriting profits which may make it difficult to separate out long term from cyclical influences on premiums.
The claims negotiator database covers a shorter period than the insurer one for the straightforward reason that claims negotiation only began to appear as a significant aspect in personal injury litigation after the changes in recovery rules in April 2000.
Of course, these rulings mainly concerned RTA cases and would not therefore be expected to have such a significant impact on other types of claim.
In the time available, we have not sought to investigate the reason for this decline. It is plausible that deafness claims were initiated by some legislation permitting them in the 1990's, and the high numbers in 1997 were a reflection of the backlog of pent-up claims being worked through.
"Profit costs" are defined in this dataset as solicitors' costs excluding disbursements, success fees and ATE premiums - these are referred to as "base costs" elsewhere in this report.
Note that our pre-issue figure in the RTA study also show base costs net of VAT.