Table A2.
The matching quality of propensity score matching (PSM).
| Panel A. Propensity Score Prediction-Dependent Variables: (=1 if the Hospital Taking Digital Business as the Largest Business Except for Common Hospital Services, Otherwise = 0) |
|||||
|---|---|---|---|---|---|
| Parameter | Standard Error | ||||
| Matching variables | |||||
| Capital input | −0.4492 ** | (0.0681) | |||
| Human input | 0.0974 | (0.0830) | |||
| Cash flow | −0.8048 ** | (0.2186) | |||
| Leverage | 0.0001 | (0.0002) | |||
| Intercept | −2.8063 ** | (0.0495) | |||
| LR χ2 (p-Value) | 229.66 (0.0000) | ||||
| Matching period (before legislation) | 2013–2015 | ||||
| Observation size | 15,098 | ||||
| Panel B. Balance check [nearest-neighbor, N = 3] | |||||
| Mean | t-test | ||||
| Treated | Control | Bias % | T-value | p > t | |
| Capital input | 0.5117 | 0.4839 | 0.3 | 0.22 | 0.8260 |
| Human input | 0.3126 | 0.2710 | 1.2 | 0.52 | 0.6020 |
| Cash flow | 0.0222 | 0.0328 | −0.9 | −1.31 | 0.1900 |
| Leverage | 4.8218 | −3.4042 | 5.0 | 0.80 | 0.4230 |
| Panel C. Balance check [nearest-neighbor, N = 7] | |||||
| Mean | t-test | ||||
| Treated | Control | Bias % | T-value | p > t | |
| Capital input | 0.5117 | 0.4839 | 0.3 | 0.22 | 0.8260 |
| Human input | 0.3126 | 0.2540 | 1.6 | 0.77 | 0.4420 |
| Cash flow | 0.0222 | 0.0344 | 2.5 | −1.07 | 0.2860 |
| Leverage | 4.8218 | 0.7040 | −0.7 | 0.60 | 0.5460 |
Notes: Logistic model is applied for the propensity score prediction. Standard errors are reported in brackets. Capital input is measured by total assets, and human input is measured by total costs of employees of hospitals (both in €10 million), * p < 0.05, ** p < 0.01.