Table 3.
Multi-group interrupted time series analysis of monthly relative-value units per employee
| Periods/Parameters | Central town and street PHIs vs. Remote countryside PHIs | Other PHIs vs. Remote countryside PHIs | Central town and street PHIs vs. Other PHIs |
|---|---|---|---|
| Period I (Before the pilot reform, n = 24 months) | |||
| Trend of control (β1) | 4.24 (2.69–5.78)*** | 4.24 (2.69–5.78)*** | 4.64 (2.50–6.77)*** |
| Level difference (treatment - control, β6) | 8.64 (−37.96–20.66) | 28.52 (−1.45–58.50) | −37.57 (−70.52 to −4.61) * |
| Trend difference (treatment - control, β7) | 1.12 (−2.21 to 4.45) | 0.40 (−2.24–3.03) | 0.72 (−2.92–4.36) |
| Period II (After the pilot reform, n = 36 months) | |||
| Trend of control (β1 + β3) | 11.04 (8.82–13.25)*** | 11.04 (8.82–13.25)*** | 4.93 (3.64–6.21)*** |
| Trend of treatment (β1 + β3 + β7 + β9) | 2.53 (1.55–3.52)*** | 4.93 (3.64–6.21)*** | 2.53 (1.55–3.52)*** |
| Trend difference (treatment - control, β7 + β9) | −8.50 (−10.93 to −6.08)*** | −6.11 (−8.67–−3.55)*** | −2.40 (−4.02 to −0.78)** |
| Period III (After COVID-19 pandemic, n = 36 months) | |||
| Trend of control (β1 + β5) | −5.11 (−9.27 to −0.95)* | −5.11 (−9.27 to −0.95)* | 1.16 (−2.13–4.45) |
| Trend of treatment (β1 + β5 + β7 + β11) | 4.16 (0.56–7.77)* | 1.16 (−2.13–4.45) | 4.16 (0.56–7.77)* |
| Trend difference (treatment - control, β7 + β11) | 9.27 (3.77–14.78)** | 6.27 (0.97–11.58)* | 3.00 (−1.88–7.88) |
| Total impact | |||
| Trend of control (β1 + β3 + β5) | 1.69 (−1.41–4.79) | 1.69 (−1.41–4.79) | 1.46 (−0.82–3.73) |
| Trend of treatment (β1 + β3 + β5 + β7 + β9 + β11) | 1.34 (−0.88–3.56) | 1.46 (−0.82–3.73) | 1.34 (−0.88–3.56) |
| Trend difference (treatment - control, β7 + β9 + β11) | −0.35 (−4.16–3.46) | −0.23 (−4.10–3.61) | −0.12 (−3.30–3.06) |
Note: * p < 0.05, ** p < 0.01, *** p < 0.001; regression with Newey–West standard errors-lag (1)