Table 3 Average and conditional...

Table 3

Average and conditional treatment effects of FS1 and FS2 on ZSCORE

Variable	Group	Effect	95% CI (low)	95% CI (high)
FS1	ATE	54.50	26.46	82.54
FS2	ATE	53.84	18.86	88.82
FS1	SOE	23.47	18.23	28.72
	JSCB	25.18	22.98	27.37
	Large	16.25	14.73	17.76
	Small	33.62	30.26	36.99
	Post-COVID	11.97	11.04	12.90
	Pre-COVID	31.42	28.72	34.11
FS2	SOE	31.46	29.03	33.89
	Private	24.39	23.22	25.55
	Large	29.05	27.58	30.53
	Small	21.74	20.32	23.16
	Post-COVID	30.88	29.31	32.45
	Pre-COVID	22.65	21.35	23.96

Variable	Group	Effect	95% CI (low)	95% CI (high)
FS1	ATE	54.50	26.46	82.54
FS2	ATE	53.84	18.86	88.82
FS1	SOE	23.47	18.23	28.72
	JSCB	25.18	22.98	27.37
	Large	16.25	14.73	17.76
	Small	33.62	30.26	36.99
	Post-COVID	11.97	11.04	12.90
	Pre-COVID	31.42	28.72	34.11
FS2	SOE	31.46	29.03	33.89
	Private	24.39	23.22	25.55
	Large	29.05	27.58	30.53
	Small	21.74	20.32	23.16
	Post-COVID	30.88	29.31	32.45
	Pre-COVID	22.65	21.35	23.96

Note(s): This table reports ATEs and CATEs of FS1 and FS2 on ZSCORE using CDML terminology. Because FS1 and FS2 are system-year exposure variables rather than bank-level treatments, the estimates are interpreted as heterogeneous bank-level responses to common annual FBP changes. The 95% CI (Low/High) reports the lower/upper bounds of the confidence interval. SOE = state-owned commercial banks; JSCB = joint-stock commercial banks. FS1 and FS2 are measured as proportions; thus, a one-percentage-point effect equals the reported estimate × 0.01, while a one-standard-deviation effect equals the estimate × SD(FS1 or FS2). CDML estimates use honest sample splitting, cross-fitting, and out-of-bag diagnostics to limit overfitting

[ViewLarge]

Sharing Unavailable