By guest contributor Simon Russell, Director, Behavioural Finance Australia
The new product design and distribution obligations are due to come into effect in October 2021. If your organisation issues or distributes investment or credit products, now is your chance to prepare. The obligations will apply to a range of financial products, including investments, super, insurance and various credit products. The obligations are particularly relevant for super funds, investment product providers, banks, credit unions and insurers, among others. They appear to indirectly impact financial advisers and mortgage brokers too.
This article is the third in a 4-part series that discusses what you need to know about behavioural finance and associated decision-making research to implement these new requirements successfully.
How can you use data and technology to understand consumers’ needs?
Rather than asking questions, an alternative approach is to use data and technology. This is a possibility that ASIC specifically allows for. Measuring consumers’ actual decisions and behaviours can mitigate the risk of their fallible memories and self-assessments. And the data doesn’t even have to be perfect – ASIC will be satisfied so long as the issuer has made a reasonable assessment of consumers as a group, not an assessment that is necessarily correct for every one of them.
But while the data presents opportunities, it also creates its own range of decision-making challenges. One of the challenges is how to deal with uncertainty. When making an inference about a consumer based on some demographic information (such as their postcode, say), one needs to be very careful about straying too far from what psychologists refer to as relevant ‘base rates’ (ie what applies for most people most of the time).
In psychological experiments and in my own workshop demonstrations people often fail to properly account for this uncertainty. When I give people a string of semi-random coloured squares and ask them to predict what follows, people find beguiling patterns in the noise; they naturally and often unknowingly ‘overfit’ a narrative to fit the patterns they find.
In the context of customers, this can mean creating an apparently compelling story about a stereotypical customer that we actually know very little about. These stories can be infectious. One of the most often-repeated stories relates to the retail store Target, which reportedly used data analytics to predict a young customer’s pregnancy before her angry father was aware of his pending grand-parenthood.
As I discussed in my recent book, while the story of the Target customer appears compelling, having read mixed accounts of it, I’m still not sure if it is true and, if it is true, the reliability of the prediction. For example, did Target ‘predict’ that 10,000 of their female teenage customers were pregnant, each with a 10% probability of being correct? Was the success story plucked from a much larger pool of failures, all shrouded in a murky cloud of uncertainty? When an inference is beset with uncertainty, it is easy to feel you know more about a customer than you do.