Machine learning (ML) can help banks “screen” risky borrowers, but its overall benefit for credit conditions depends on the structure of the market, research published by the Bank of England finds.

In the staff working paper, Peter Eccles, Paul Grout, Paolo Siciliani and Anna Zalewska develop a model of how ML might impact credit markets. They note that while several studies have shown ML can improve screening, the overall impact on the market remains ambiguous.

Their models sheds light on the