Will fintech development increase commercial banks risk-taking? Evidence from China

Abstract

The development of financial technology (Fintech henceforth) has brought both opportunities and challenges to commercial banks’ risk management. Here, we discuss the interactive relationship between Fintech and traditional finance. We first construct a mathematical model and deduce that the synergy between Fintech and traditional finance can more effectively help banks measure the creditworthiness of customers, thereby reducing risk-taking. In order to verify the deduced result, we then conduct an empirical analysis using panel data of Chinese commercial banks from 2011 to 2020. The empirical results support the theoretical outcome that the synergy between Fintech and traditional finance has indeed effectively reduced commercial banks’ risk-taking. Our research also shows that Fintech development significantly reduces the risk-taking of commercial banks, and this mitigation effect is more prominent for larger-scale banks and those with a higher-developed traditional finance basis. Moreover, this mitigation effect is more significant after 2016, a period marked by an increased level of Fintech development.

Appendix

Proof of Formula (3):

First, suppose there are ${y}_{1}$ and q that obey the following distribution:

$$\begin{array}{*{20}c} {y_{1} = q + \varepsilon_{1} } \\ \end{array}$$

(A–1)

$$\begin{array}{*{20}c} {q\sim N\left( {\overline{q},\overline{\sigma }^{2} } \right)} \\ \end{array}$$

(A–2)

$$\begin{array}{*{20}c} { \varepsilon \varepsilon_{1} \sim N\left( {0,\overline{\sigma }_{1}^{2} } \right)} \\ \end{array}$$

(A–3)

$$\begin{array}{*{20}c} {y_{1} \sim N\left( {\overline{q},\overline{\sigma }^{2} + \overline{\sigma }_{1}^{2} } \right)} \\ \end{array}$$

(A–4)

Secondly, the covariances of $y_{1}$ and q is:

$$\begin{aligned} \sigma_{{y_{1} ,q}} &= {\mathbb{E}}\left( {y_{1} - \overline{q}} \right)\left( {q - \overline{q}} \right) \\ &= {\mathbb{E}}\left( {q - \overline{q} + \in_{1} } \right)\left( {q - \overline{q}} \right) \\ &= {\mathbb{E}}\left( {q - \overline{q}} \right)^{2} + {\mathbb{E}}\left( { \varepsilon_{1} } \right)\left( {q - \overline{q}} \right) \\ &= \overline{\sigma }^{2} \\ \end{aligned}$$

(A–5)

Finally, by introducing the expression of covariance (Eq. (5)), it can be deduced that the conditional expectation of q is:

$$\begin{aligned} {\mathbb{E}}\left[ {q\;|\;y_{1} } \right] &= {\mathbb{E}}\left[ q \right] + \frac{{\sigma_{{y_{1} ,q}} }}{{\sigma_{{y_{1} }}^{2} }}\left( {y_{1} - \overline{q}} \right) \\ &= \overline{q} + \frac{{\overline{\sigma }^{2} }}{{\overline{\sigma }^{2} + \overline{\sigma }_{1}^{2} }}\left( {y_{1} - \overline{q}} \right) \\ &= \overline{q} + \frac{{\tau_{1} }}{{\overline{\tau } + \tau_{1} }}\left( {y_{1} - \overline{q}} \right) \\ &= \frac{{\tau_{1} y_{1} + \overline{\tau }\overline{q}}}{{\overline{\tau } + \tau_{1} }} \\ \end{aligned}$$

(A–6)

Thus, Formula (3) has been proved.