Default or profit scoring credit systems? Evidence from European and US peer-to-peer lending markets

Financial Innovation

Table 4 Out-of-sample performance of credit and profit scoring models—lending club

Model	% invested loans	Performance across invested loans		Performance across all loans			Total profit in mil. USD
		Average return	SD	Average return	SD	MCS
Panel A: Credit scoring models
LR	60.8	8.66	18.20	5.27	14.81		1.42
\(LR^{{\lambda_{\min } ,\alpha = 1}}\)	61.0	8.53	18.28	5.20	14.87		1.39
\(LR^{{\lambda_{\min } ,\alpha = 0}}\)	61.4	8.57	18.15	5.26	14.81		1.40
\(LR^{{\lambda_{\min } ,\alpha_{\min } }}\)	61.2	8.49	18.38	5.19	14.96		1.39
RFC	78.5	7.46	21.46	5.86	19.26		1.62
NNC	70.1	8.21	20.31	5.75	17.42		1.58
Panel B: Profit scoring models
LM	68.6	8.52	20.63	5.84	17.54		1.61
\(LM^{{\lambda_{\min } ,\alpha = 1}}\)	72.5	8.94	20.43	6.48	17.85^†		1.84
\(LM^{{\lambda_{\min } ,\alpha = 0}}\)	72.2	8.69	21.12	6.27	18.35^†		1.79
\(LM^{{\lambda_{\min } ,\alpha_{\min } }}\)	72.8	8.93	20.42	6.50	17.87^†		1.84
RFR	73.3	8.47	21.69	6.21	18.95^†		1.82
NNR	80.1	7.83	22.76	6.27	20.61^†		1.79

\(LR^{{\lambda_{\min } ,\alpha = 1}}\), \(LR^{{\lambda_{\min } ,\alpha = 0}}\), \(LR^{{\lambda_{\min } ,\alpha_{\min } }}\) are lasso, ridge and elastic net version of logistic regression, \(LM^{{\lambda_{\min } ,\alpha = 1}}\), \(LM^{{\lambda_{\min } ,\alpha = 0}}\), \(LM^{{\lambda_{\min } ,\alpha_{\min } }}\) are lasso, ridge and elastic net version of linear regression
SD standard deviation, LR logistic regression-based models, LM linear regression-based models, RFC random forest classification, NNC neural network classification, RRR random forest regression, NNR neural network regression
^†Denotes a model that belongs to the set of superior models