Virtual Summit shorts: Transforming the reserving process

In this edition of the ‘Virtual Summit Shorts’ series, Joey Chen shares her learnings and thoughts on Matthew Webster and Michael Ramsay’s session ‘Transforming the Reserving Process’. The session covered the drawbacks of current state reserving processes, and the journey to defining a better future state and bringing these improvements to life.

Actuaries working in the field of reserving would appreciate the significant amount of time the whole process takes. Additional challenges presented to reserving Actuaries include tight deadlines to prepare data and models, and even tighter deadlines to produce results and opinions. The inefficient processes and challenges constrain the actuaries’ ability to perform actuarial analysis, instead their time is spent on producing the basic results.

Against this backdrop, Matthew and Michael discuss the motivation and set-up for a successful transformation of the reserving process, drawing upon their individual experiences and providing QBE’s journey as a case study.

Why transformation: The motivation

A standard actuarial reserving process requires extensive effort from the actuary for data preparation and model runs. Data issues may come to light during this process. Outsized efforts may be required to configure immaterial assumptions just for the process to run. Furthermore, an unwieldy process limits the actuaries’ ability to quickly re-run the analysis, taking even more time if concerns arise in the results, or senior stakeholders have further questions or demands on the reporting. Finally, results are often presented in a static slide pack.

Actuaries are facing new pressures on the reserving side, creating further difficulties for the current state. These include:

• complexities for example, to factor in and analyse the impacts of COVID-19, multiple financial and regulatory reporting bases, more data and higher granularity;
• speed required to meet ever-shorter reporting deadlines, turn around ad-hoc requests and scenarios, and teams being under pressure to explain movements without performing a full reserving exercise; and 
• transparency, with a broadening user base for the results, auditability, and higher expectations on items to extract from models as well as drill-down style dashboards.

Target state: What an ideal reserving process may look like

To define the target state of the actuarial reserving process, we must first establish what is important. This varies across stakeholders and it depends on an entity’s risk appetite.

The presenters gave an overview of what an ideal reserving process may look like. Some elements to consider include:

• Philosophy – spending time on results rather than process, top-down reserving starting analysis from a full set of results which are based on pre-defined and agreed assumptions, rapid reporting and low cost of model runs, and accessible analytics;
• Tools – modern software, which enables suitable techniques, algorithms, dashboarding, and governance;
• Data – having one version of the truth, automated reconciliations;
• Team roles – defining core responsibilities, wider aims (such as value-adding), and spending time better;
• Modelling – suitable level of complexity, fit-for-purpose, and having the right diagnostics;
• Outputs – aiming for a single source of truth, automated production of reporting, visualisation and self-service.

Ideally, this would mean faster model runs and faster results, with more time for finalisation and performing deep dives, ad hoc analysis, and any tweaks and adjustments.

Realising the vision: The journey to achieve the target state

The project of transforming the actuarial reserving process is a large task, with total time to completion typically taking months or even years. This generally involves many stages, from the initial scope to the final go-live, and the presenters discussed the requirements of each stage.

Michael walked through QBE’s transformation of their short-tail, liability, and professional lines portfolios as a case study, moving from primarily manual excel based models and reporting to a centralised, consistent model. Some of the selected observations he made include:

• Project structure – A balance of internal expertise with an outsider view to challenge existing thinking was helpful. An agile approach was adopted to implementation, to capture flexibility to changing requirements and iteration across build and testing.
• Engagement – ensure that you’re communicating effectively with the key decision-makers and users, as the two major stakeholder groups in a project like this. Overall objectives may be aligned, however specific interests and focus may be different, and balancing these to get the buy-in from both these groups is important to get the right outcome.
• Technical – consistency is key, and the benefits of a consistent approach across different reserving groupings often outweigh the additional benefits of incremental differences and nuances. Not all details need to be finalised in the design phase as these can be progressively developed, but this would require a good feedback loop in order to work.
• Resourcing – In a busy environment, there is a need to be very conscious of the potential additional load on the team through project work and ensure this is supported appropriately. 

A few specific outcomes of the case study were also discussed including:

• Significant reductions in the time taken for what were previously very manual tasks, as an example the annual valuation report was significantly rationalised and significant components of the outputs automated out of the model to reduce the time taken to pull this together during busy periods.
• Improvements in governance and control of models and outputs, for example, the integration of model sign offs directly into the model, rather than as a separate task in the previous model process.
• Better and more timely information to the business, for example, the automation of production of results such as claims development tables meant that these were available earlier and at increased granularity relative to the previous process.

I would like to thank Matthew and Michael for sharing their experiences and considerations. In an area increasingly relevant to an insurance industry, in which we must navigate environmental changes and evolving stakeholder expectations, I have found their observations highly informative, and I hope our readers have too.