**Appendix B. Robustness Check**

Since the dataset is from 2010 to 2014, the change in macroeconomic environment in these years may influence the decisions of the investors and the behavior of the borrowers. As China has 36 different regions, regional differences may be found in financial behavior. Thus, we added region and year dummy variables into the model to control the fixed effect of time and region.

The loan application distribution by region and year are listed in Tables A2 and A3 accordingly.


**Table A2.** Loan Application Distribution by Region.

**Table A3.** Loan Application Distribution by Year.


The regression result with region and year dummy is presented in Table A4. The results are in line with original regression, in that most of the hard information variables have opposite results in the two models while most of the soft variables have consistent results. This proves the existence of TYPE II errors in the investors' decision-making process.


**Table A4.** Robustness Test with Region and Year Dummy.


Heteroscedasticity-Robust, standard errors in parentheses. \*\*\* *p* < 0.01, \*\* *p* < 0.05, \* *p* < 0.1.

To control for multicollinearity, we analyzed the variance inflation factors (VIF) of our chosen variables. The results<sup>1</sup> show that all the independent variables' VIFs are within 2 and with an average of 1.27. In other words, the variance of the estimated coefficients is inflated with very low factors and within a reasonable rule-of-thumb of 10. For verification, we also calculated the square root of VIF, the R square for the correlation between the given independent variable and the rest of the independent variables, and the tolerance indicators, which are computed as 1- R square. The results prove the non-existence of multicollinearity.

#### **Note**

<sup>1</sup> We checked the variance inflation factor, the R square for the correlation between the given independent variable and the rest of the independent variables, and the tolerance indicators for each independent variable. The results show that all variables have VIFs lower than 2, R square less than 0.2, and tolerance less than 1.

#### **References**

Akerlof, George A. 1970. The market for "lemons": Quality uncertainty and the market mechanism. *The Quarterly Journal of Economics* 84: 488–500. [CrossRef]

Arrow, Kenneth J. 1964. The role of securities in the optimal allocation of risk-bearing. *The Review of Economic Studies* 31: 91–96. [CrossRef]

Barasinska, Nataliya, and Dorothea Schäfer. 2014. Is crowdfunding different? Evidence on the relation between gender and funding success from a German peer-to-peer lending platform. *German Economic Review* 15: 436–52. [CrossRef]

