GI TORP Automation Survey 2017

Download the results of the survey

Summary of Results

• The total number of survey participants was 39, among which:
  • 50% are personal lines insurers
  • 40% are Lloyd’s/London Market insurers
  • 10% are reinsurers or health insurers
     
• About 2/3 of the participants are doing business mainly in the UK (some of which also in other European countries), and about 1/4 have US operations
   
• There was a good balance of small, medium and large firms in terms of Gross SII TP volumes
  • 25% less than £100m
  • 20% between £100m and £500m
  • 20% between £500m and £1bn
  • 35% more than £1bn
     
• Regarding their reserving process, the participants currently use:
  • An-off-the-shelf product from a software supplier -> about 50%
  • Excel spreadsheets (exclusively) -> about 25%
  • An internally-built/coded platform (potentially combined with spreadsheets and other software) -> about 25%

It is worth noting that, within the group of firms that only use Excel spreadsheets for their reserving, half of them are actually very large firms from a gross solvency 2 technical provision perspective (SII TPs > £1bn) while the others are composed of small and medium firms.

• Automation: Regarding the level of automation of their current reserving process, the responses have been:
  • 50% as being '50 manual and 50% automated'
  • 30% as being mostly manual or not automated at all
  • 20% as being mostly automated with little manual work needed

It is worth noting that approximately 1/3 of the participants using an off-the-shelf reserving platform have described their process as being 'mostly manual' while the remaining 2/3 have responded '50% manual and 50% automated.' However, within those exclusively using spreadsheets, only 1/4 have considered their process to be 'mostly manual', which might show that using a dedicated software package doesn't seem to solve most of the challenges related to automation even compared to spreadsheets. This appears to be consistent with the point that 90% of all the participants have responded 'Yes' or 'Maybe' to the question, ' Given time, would you like to invest in further automation of your reserving process?', disregarding what tools or platforms they are using.

• Based on the other results (incl. open-ended responses) also received, the main outcomes to highlight relate to the following items:
   
  • Data Quality – This appears to be the main issue for most participants. Data quality is often not very good, it often needs to be processed at a significant level (for 3/4 of the participants) when coming from Management Information/Information Technology or from other teams, and it is not always consistent between quarters (e.g. quarter on quarter data formatting changes are common place). Data Quality arguably represents the main area of improvement as evidenced by the responses, with  of the participants considering 'data processing as an essential area of improvement regarding automation'
     
  • Level of Automation – There remains a significant volume of manual work needed by reserving actuaries (e.g. regarding data processing, ultimate claims selections, model blending, movement analyses from a quarter to another). Most participants would like to see an increased level of automation to allow for more time to do analyses and in-depth reserve review (but also reduce manual mistakes due to “tedious” tasks that must be achieved). Nevertheless, a few participants have said that there shouldn’t be too much automation – some is needed to provide first cuts/validation, but it always will be overlaid by manual adjustments and elements of expert judgement
     
  • Reporting – Reporting processes are mostly manual at the moment for almost 50% of the participants.  This seems to be the other major challenge regarding automation alongside data quality. In particular: QRTs, translation from GAAP/IFRS reserves to SII TPs and PowerPoint presentations (to various committees) are the most manual tasks to do for reserving actuaries in terms of reporting. One of the participants mentioned the potential of data visualisation software like Tableau for reporting purposes.
     
  • IFRS 17 – This will be a major event which will increase the impetus for more automation within reserving processes for 75% of the participants. In particular, participants appear to want to try to use as much work as possible from SII for upcoming IFRS 17 implementation, and have a consistency between SII and IFRS 17 (e.g. regarding cash-flows)
     
  • Machine Learning – About 50% of the participants have responded 'Yes' or 'Maybe' when asked if they had already given any considerations to machine learning techniques applied to reserving (that includes both personal lines and Lloyd's/London Market insurers). It is seen as a great opportunity for them. However, a large majority have said that they are still in the very early/conceptual stages onl, and that they they are still a bit scpetical about it. This is mainly because f the level of complexity involved and the the technical skills that would be required to implement and maintain such models
     
  • Consequences of Automation – Most participants have said that automation would help do the 'tedious' work faster, reduce time/manual mistakes and allow for more time for deep analyses. A few of them, however, mentioned that currently most of the work performed by reserving actuaries actually relates to data/processing/manual tasks within their organisation, and on that perspective, automation could potentially reduce the need for reserving actuaries in the long term (hence the question of a possible reduction of Reserving sizes over time