Articles - Revolution, not evolution, on regulatory reporting horizon

 By Paul Simmons, Senior Consultant, Life Insurance Consultancy, Towers Watson

 The demands are stark. The initial submission deadline for the solo quantitative reporting templates (QRTs) is eight weeks after the quarter end in 2016, which then accelerates to five weeks by 2019. And in reality, this means end-state modelling will need to complete within approximately two weeks of the quarter end to allow time for governance and sign off processes.
  

 Quantum shift
 Such timescales represent a quantum shift for companies whose capital modelling activities have been focused on delivering an annual assessment of capital requirements three, or even six, months in arrears. Only in Sweden, and to some extent the UK, do insurers have experience of comparable pressures.

  

 The problem for most insurers is that their existing processes, where they exist, developed without anything like the time pressures they now face. Practice has evolved to include extensive manual steps, and actuarial and management judgement.

  

 Furthermore, most processes are tolerant of errors and changes, on the grounds that there is sufficient time for re-work where necessary. For example, one insurer noted to us recently that the average run was performed 2.8 times due to set up and processing errors.A

  

 With the arrival of Solvency II’s QRTs deadlines, processes must be transformed. We see the following elements as key to meeting this challenge:

 Renovating process components
 Achieving these goals will require transformation at every step of the regulatory reporting process, to reduce each step from days (or weeks) to hours.).

  

 

 Figure 1: Key building blocks in a Solvency II reporting process

  

 Data cleansing and validation
 Current processes involve manual and ad-hoc handling of multiple sources of asset and liability information, to which the reporting team may not have direct or timely access. These limitations can be addressed by integrating different systems and by automating the manipulation of data.

  

 Data grouping or model pointing
 The process of compressing policy data is typically reliant on expert judgement and manual activity. As with data validation, these processes are algorithmic and can therefore be codified and automated. Furthermore, implementing the process of generating model points into automated software can help to identify techniques to improve the quality and consistency of liability data, such as we have seen with cluster algorithms.
  

 Economic scenario generators (ESGs)
 Most ESGs can now automate calibration and projection tasks and permit batch processing, so that multiple files can be produced in a single process. We see the next stage to be the integration of ESG software into other steps, such as the proxy calibration.

  

 Assumption management
 Our research indicates that for many firms over 50% of run failures are due to manual errors in the set-up of assumptions. Automation using functionality built into modern projection systems, or standalone assumption management tools, can deliver a process that executes quickly and eliminates opportunities for mistakes.

  

 Model runs
 Many companies have already optimised code and updated software to take advantage of high performance computing grids. The new regulatory timetable is likely to drive leveraging of enterprise versions of modelling software to schedule and manage runs, as well as moving from in-house grids to private and public clouds with scalable resources, so that extra modelling capacity can be accessed automatically as required.

  

 Calibration of the capital model (internal model firms)
 The standard approach is to use proxy models to approximate the balance sheet, using mainly polynomial models or, (occasionally) replicating portfolios. While firms implement these proxy models in various ways, the typical implementation remains resource intensive and manual. These blockages force many firms to calibrate off-cycle and roll-forward the proxy model.

  

 However, specialist proxy generation software can eliminate the manual and expert judgement-based processes to render on-cycle calibration feasible.

  

 Capital model runs
 Specialised aggregation tools for either standard formula or internal model, plus appropriate integration with other systems, reduces the production of capital results dramatically.

  

 Don’t forget governance
 No amount of process change will deliver without corresponding governance reform. For many companies this may prove particularly difficult because it involves change to established cultures. For example, many existing reporting processes allow management multiple intervention points and permit assumptions to be revised even after they have been ‘signed off’. In the new world, those are luxuries companies cannot afford.

  

 The status quo is going to change – and revolution, not evolution, will be the name of the game as companies adapt to the Solvency II regulatory reporting timetable.