Following the introduction of ‘pension freedoms’ in April 2015, many people have chosen to shift their pension pots into income drawdown products before or at retirement, rather than buying annuities as was typical before the reforms.
 
Income drawdown products can be relatively complex products with multiple features and often with an array of fees. Charges and fees vary with pot size and how the product is used, so the same drawdown product could be relatively cheap for one consumer, but relatively expensive for another. In order to get the best deal, consumers need to compare products across multiple dimensions, usually with no single price or fee to focus on.
 
There are different ways of presenting the cost of an income drawdown pension product. In this context, the Financial Conduct Authority (FCA) wished to understand which way of summarising the cost of the product was best able to help consumers to identify, from a range of product options, which was the most cost-effective for them. The FCA therefore commissioned Oxera and the Centre for Experimental Social Sciences (CESS) to conduct a behavioural experiment to assess the effectiveness of different summary cost metrics. This report presents the findings of the behavioural experiment.
 
In this online experiment, participants were presented with a set of drawdown products (designed on the basis of products available on the market) and a consistent decision-making environment to see how the provision of different summary cost metrics affected their product choices, with all other factors held constant.
 
The results of the experiment show that two of the five personalised summary cost metrics we tested had a statistically significant positive impact on the product choices of participants, resulting in them selecting lower-cost products on average, and being better able to select the cheapest product. These two cost metrics were ‘pension savings available after costs’ and ‘average cost per year’. The other three summary cost metrics, ‘total cost’, ‘reduction in yield’ and ‘cost rating’, did not have a statistically significant effect on improving product selection performance.