Speeches by McEvoy.

“Would you have to only infer if you had the HMRC data annually? HMRC will have the tax returns.”

“How much would real-time PAYE, HMRC and DWP data support the work that you are doing to get a more accurate forecast? Do you think you are overweighted on the LFS?”

“You mentioned a few times that your model is only as accurate as the data you put into it. The Committee has spent quite a lot of time on the labour force survey and the ONS data. How much is the reliability of that data keeping you up at night?”

“But is the AI meeting the human decision? Are they aligned most of the time? How is the breakdown working?”

“Are they?”

“I will ask the same question to everyone.”

“That is really interesting. How far along your AI journey are you? Have you had any red flags so far in regard to unconscious bias or wrong decisions?”

“We are going to move on to AI. Some 95% of the insurance sector is using AI. We know that there have been questions around these models, their bias, unconscious bias, and conscious bias, which we all hold as a society. What are your firms doing to ensure that your use of AI does not discriminate against groups who need…”

“It is a lower overall number than others, but it is a higher overall rate.”

“Is postcode taken into account?”

“The final question is to Mr Storah and Mr Hargreaves on travel insurance. Does postcode come into your travel insurance premium? If you are going abroad, does it matter which part of the country you live in as to how expensive your travel insurance is?”

“Is it regularly updated data?”

“That is good. Mr Walker, you do not do travel insurance, so you cannot answer that. Would you like to answer the broader point on geography?”

“So we are not going to find two houses, with the same price, the same type of house, and the same crime rate in certain parts of the country—the north and the south, for example—that are priced differently.”

“I am trying to see whether there are stereotypes around certain parts of the country or certain parts of a town.”

“I am trying to get to social rather than environmental factors.”

“So it is more to do with safety and the likeliness of the driver having a collision than it is to do with any risk of the car being broken into. You might have crime in a really posh area because they have bigger cars, or whatever, compared to a crime-free estate.”

“I understand that you cannot go into each specific, but, looking broadly at the different variables, let us take two people who have no points on their licence, have been perfect drivers and drive the same car. How variable is the insurance premium if somebody lives in a Darlington postcode versus one in Westminster? I…”

“I am just interested because, if you live in a certain area, there are health conditions and outcomes associated with that. That goes to the poverty premium point of whether you are paying more for your life insurance, house insurance or car insurance because of your postcode. Does anybody have any information on any o…”

“The final question from me is on geography. On this Committee, our constituencies are widely distributed around the UK. Do you take geography into account for life insurance and, if so, to what level? How granular is the postcode variable for life insurance?”