In social lending, in which an individual lends or borrows money using an SNS network, a person who lends money must take a risk that the money won’t be returned. Since social lending is a comparatively new field, very few studies have been made. Therefore, we present an experimental assessment of the influence of the updating of an interest rate using Bayesian estimation, which takes into consideration the influence of groups with agents. Our method decreases dispersions of the delay of the borrower in payment with the increasing loan history of the borrower. As a result, when the lenders are risk-averse (risk means the dispersions of the delay of the borrower at each interest rate), the number of transactions increases. Therefore, our method is effective because it can cause the transactions of lenders who are risk-averse to increase.