Zopa’s default rates up to 40 per cent below expectations

Zopa’s chief executive Jaidev Janardana has said the peer-to-peer consumer lending platform’s default rates were up to 40 per cent lower than expected during the pandemic.

Speaking at UK Fintech Week, Janardana (pictured) said that he was “positively surprised” to see default rates were 30 to 40 per cent below expectations since Covid-19 struck.

He said during the crisis the platform also offered a quick service to customers, for example, setting up systems to answer phones within minutes.

Read more: Funding Circle and Zopa urge chancellor to back fintech sector

“We think we have succeeded last year in terms of our ability to lend and compete and take market share from incumbents, shown from growing customers and lenders getting positive returns from us during uncertain economic times,” Janardana said.

He said as a huge number of people saved more during the pandemic, lending will become more prominent than saving in the UK in the foreseeable future.

Read more: Zopa secures £20m in private fundraise to grow business

“A rise in investing provided they understand risks is always a good thing,” Janardana said.

“If one has money, they should seek a diversified portfolio.

“Covid has probably axled away the trend of wealth inequality even more, so there’s a bigger saving spool out there and even if some of that gets put into investments, who can lend better will be the winner rather than who can attract the deposits.”

Janardana said machine learning can be useful in aiding lending but is needed alongside good judgement.

Read more: Zopa exec “very excited” about embedded finance for lenders

“To make lending work is hard, it takes time but you also need to bring together multiple pieces of ecosystem, how to find the right customers and who to lend and not to, that comes from experience and having the right data and the customer value proposition and how to avoid adverse selection,” he said.

“A lot of people want to jump in wanting to think machine leading can solve problems but not appreciating that machine learning only works if it’s exposed to the right data and in the right environment and without applying that judgement be that machine learning or otherwise, will not work.

“You’ve seen some examples of that not working but those who stay disciplined and prudent and put different sets together can succeed but it’ll take time.”