Investigation 3 (provisional plan)
Recap: What is Investigation 3?
Investigation 3 relates to the following question:
To what extent did any modelling (including the communication of results) used to support the UK government’s decision-making during the Covid crisis adhere to what the UK actuarial profession would consider best practice?
What is my provisional plan for approaching this investigation?
I plan to approach this investigation with the following steps:
Step 1: Identify the key modelling/communications referred to by the UK government during the Covid crisis.
Step 2: Identify a subset from this material which covers the most important items and can feasibly be reviewed by me.
Step 3: Secure a copy of the modelling/communications, reaching out for access if any of this material is not in the public domain.
Step 4: Review this modelling/communications against the Technical Actuarial Standards which an IFoA actuary would expect their work to meet.
Step 5: Produce a summary of the findings for discussion and invite input from appropriate members of the CMI and the legacy Covid-19 Actuarial Response Group. Take appropriate account of any feedback.
Step 6: Produce a written report summarising the exercise. Produce a slide deck presentation reflecting the report and deliver a webinar covering this content.
Please note that this is a provisional plan and it is possible that I may make some minor amendments as things progress. However I do not intend to deviate significantly from this provisional plan and I judge that this commitment is important because this is what I will have a democratic mandate to pursue.
Note: I have made the decision that I will consider pushing for a working party to accomplish Investigation 3. I am not committing to this approach, but Investigation 3 is likely to be the most challenging and the one where assistance from others could be especially valuable. I will assess the situation during my work on Investigation 1 and Investigation 2.
What difficulties do I anticipate and how will I tackle these?
I do not expect any insurmountable obstacles with this investigation. However I do anticipate that I may run into some challenges. I have detailed these below, together with how I intend to handle them:
The sheer magnitude of material
I need to be realistic about how much of the modelling and associated communications I can meaningfully review. I plan to tackle this by identifying the breadth of material and then rationalising my investigation to focus on a subset consisting of the most important items.
One approach which has immediate appeal is restricting myself to the modelling/communications around each significant variant (i.e. Wuhan, Alpha, Delta and Omicron); however I will need to wait until I have had the chance to review the consolidated position of potential avenues to explore before committing to this area. Of course one immediate challenge will be to produce this consolidated position in the first place: in order to accomplish this efficiently I propose to:
request that the critics of the Covid response provide me with their assessment (bearing in mind they are likely to be well-motivated to produce this).
consult appropriate members of the CMI and the legacy Covid-19 Actuarial Response Group for their feedback on this assessment provided by the critics.
The complexity of the models or software
The models may be challenging to understand, either due to the fact they depend on advanced mathematics or they are written in specialist software code. Fortunately I do have experience of contending with such challenges in my actuarial career and I also know individuals who may be able to assist me if I run into difficulties. I am confident that I have sufficient technical expertise to overcome this challenge but there is a good chance that it will slow me down. Equally it should be noted that if the modelling is not easily intelligible to someone with my background/experience then arguably this is a potential problem in itself: it might suggest the model is overly complex or there are shortcomings with its audit trail, both of which would warrant comment in the results of the investigation.
Lack of access to the models and/or communications
Where the models or the communications are not in the public domain I may struggle to obtain access to the information I need to conduct this investigation. In cases where this applies I will endeavour to negotiate access, pointing to how the actuarial community has signalled the importance of this investigation. If this proves fruitless then I will not be able to review these aspects of the modelling or communication and my takeaway message for inclusion in the report will be that there is opposition to the review.
As noted above, I may opt to push for a working party on this specific investigation as a way of surmounting the challenges posed by the work.
What are the deliverables?
As with the other investigations, I will produce a report that consolidates the answer I have reached with respect to that specific question. Each report will be accompanied by a slide deck presentation that I will create and deliver in a webinar format.
I will try to be as transparent as possible. I anticipate that Investigation 3 may be the one where this is most challenging: I think the models and communications I will be reviewing will be in the public domain, but clearly I will not be able to share anything which is not. Obviously I would adhere to the Actuaries’ Code and respect confidentiality. Nonetheless I hope to be transparent with the background correspondence related to this investigation.
For the avoidance of doubt, I will make clear to all parties that my intention is to publish the correspondence and I will obtain their permission to do so. I will also redact any personal details (e.g. email addresses) when making the correspondence available.
What deadline will I work towards?
I am aiming to produce all deliverables for Investigation 3 during the first 24 months of my election. My expectation is that this is a deadline I should be able to meet. However I will be dependent on others providing their correspondence in good time and any delays from these third parties could lead to the delivery being pushed back. In addition I am planning to prioritise Investigation 1 and Investigation 2 – if these run behind schedule then it is likely Investigation 3 will suffer delays as well.