Never Mind Your Penalty Points – Where You Live is Key Factor in Calculating Car Insurance

Where you live can negatively affect your car insurance premiums. Living in a more deprived, more ethnically diverse location increases your chance of receiving a higher quote, a new survey on the Irish insurance industry has found.

A new report jointly produced by CeADAR, Ireland’s Centre for AI, and Idiro Analytics used AI methods to analyse almost 40,000 quotes from 10 car insurance companies across 20 locations in Ireland.

Researchers found that driving experience and the length of time a driver has had a policy are key factors that are modified by home address when a quote is being calculated. This can lead to extreme cases in which 25-year-olds in areas of south Dublin are receiving lower quotes than 60-year-olds with fewer penalty points living in poorer locations.

The report also found that the extent to which your premium increases if you receive penalty points varies depending on your location. For example, drivers in Dublin 17 who receive penalty points can expect an average increase in their premium of 107%, from €792 to €1,642, compared to 68% for drivers in Mallow Co. Cork, from €560 to €941.

Meanwhile, a driver’s occupation increases their quote even if they don’t use their car for business purposes. This is then compounded by where they live. Retail workers are most affected by this, with drivers who work in retail and live in Balbriggan, Co. Dublin, seeing their premiums increase by 35%, from €555 to €747, compared to 5%, from €512 to €540, for a retail worker living in Glenageary, Co. Dublin.

The year-long research analysed almost 40,000 quotes from 20 locations around Ireland, including counties Dublin, Cork, Longford, Roscommon, Wicklow and Donegal. Multiple samples of quotes were collected to isolate single changes to accurately measure the effect of each change – for example, a driver’s name, home address or gender – on the quoted figure.

The research did find significant differences across the 10 insurers whose quotes were analysed. The full findings of the report can be analysed here.

Other key findings in the report include:

— Residents of Longford Town are quoted the highest premiums in the study while residents of Crookstown, Co. Cork, receive the lowest quotes
— There was no difference in quote between those with traditionally Irish names and those with non-Irish-sounding names. — There was also no significant difference between males and females
— Drivers who make claims even when they are not at fault receive higher quotes

The aim of the study is to understand how insurance companies’ algorithms work and the extent to which unconscious bias is embedded in them. The companies could have legitimate reasons for quoting higher prices for some areas – higher rates of car-related crime, for example – but bias or unconscious bias could arise if those areas are also more ethnically diverse and economically deprived.

European legislation due to be introduced later this year will see essential service providers like car insurance companies penalised for failing to guard against bias in their AI systems. Companies found to have broken the rules of the EU AI Act will find themselves liable for administrative fines of up to €30m or a sum equal to 6% of their annual turnover.

Dr Adrian Byrne, a Marie Sklodowska-Curie Career-FIT PLUS Fellow at CeADAR and Lead Researcher AI Ethics Centre, Idiro Analytics, says:

“It appears to be a virtuous circle versus a vicious circle in terms of where you live, i.e. living in a more deprived, more ethnically diverse location not only increases the chances of receiving a higher quote but if you add penalty points, claims and zero NCB into the mix then we have seen unequal treatment between different locations. In extreme comparisons, we’ve seen 25-year-olds in the most well-heeled locations be predicted to receive a lower quote than 60-year-olds living in the poorest locations despite having much more driving experience and policies in their own name.

“If car insurance companies are employing AI more and more, they need to guard against harmful bias that may lead them to be non-compliant with the new EU AI Act because data features that are important to their business are also correlated with other aspects of society so can encroach on social determinants of success in life.

There could be unconscious bias through their AI channel because they might not understand everything their AI is doing – aside from the specific outcomes they want – but the Act means they must guard against negative bias. To do that they have to detect it and to do that they need to have processes in place.”