5. Benchmark Model Setting
With reference to
Qiu et al.
(
2018
) and
Li et al.
(
2020
), the benchmark model of this
paper is set as follows:
Risk
it
=
α
0
+
α
1
Fintech
+
α
2
Bank
it
+
α
3
City
mt
+
α
4
Policy
t
+
δ
i
+
θ
t
+
ε
it
(23)
Among them, the subscript
i
indicates the
i
-th bank, and t indicates the t-year. The
explanatory variable risk indicates the bank’s risk-taking, including Z-Score and Volatility
of Return on Assets (SDROA). The core explanatory variable Fintech represents the degree
of Fintech development in the city
m
where the bank
i
is registered, including the breadth
of the digital financial index (Fintech 1) and credit level (Fintech 2). Bank is a control
variable at the bank level, including bank size (Size), asset-liability ratio (DAR), capital
adequacy ratio (CAR), net operating margin (Netprf), cost-income ratio (CIR), non-interest
income ratio (NIRR), saving-asset ratio (SAR). City represents city-level control variables,
including regional per capita GDP (PGDP), and regional deposits and loans total wage
ratio (FinDev). Policy is a macro-level control variable, including broad money growth rate
(M2) and economic policy uncertainty index (EPU). Where
δ
i
is the individual fixed effect
of the bank, because the regional fixed effect will be absorbed by the individual fixed effect
of the bank, so it also essentially controls the regional fixed effect. Where
θ
t
represents the
fixed effect of time, and denotes factors not being observed that change over time (except
Policy). The random error term is represented by
ε
. According to this hypothesis, the
expected coefficient
α
1
significantly negative.
Dostları ilə paylaş: |