Skip to main content

Table 3 Hausman-Taylor estimation at the cohort level

From: The effect of activity-based financing on hospital length of stay for elderly patients suffering from heart diseases in Norway

  HT
Age2/1000 0.150*** (0.019)
Female -0.046 (0.069)
Activity-based component -0.128*** (0.026)
DRG weight 0.308*** (0.019)
No. of comorbidities 0.068*** (0.014)
Non-elective 0.345*** (0.065)
Angina pectoris1 -0.083 (0.051)
Congestive heart failure -0.266*** (0.067)
Group123 0.061 (0.068)
Group24 -0.022 (0.057)
Time trend -0.069*** (0.003)
_cons 137.5*** (5.511)
N 864
Spec. Test56 χ 9 2 = 18.91
  1. Standard errors in parentheses.
  2. *p < 0.05, **p < 0.01, ***p < 0.001.
  3. 1Myocardial infarction is the reference dummy for both angina pectoris and congestive heart failure.
  4. 2Group 1 consists of catchment areas where the intervention center is the only hospital.
  5. 3Group 3 which consists of catchment areas with only local hospitals is the reference dummy for both Group 1 and Group 2.
  6. 4Group 2 consists of catchment areas with both intervention centers and local hospitals.
  7. 5We choose the exogenous variables X1 = (ABF, Group1, Group2, Year), Z1 = (Female) to implement the HT estimator.
  8. 6Hausman’s test based on the difference between the within and HT estimator gives an observed x 9 2 = 18.91 . Compared to the Hausman’s test for fixed and random effects which gives x 10 2 = 89.98 , we can see that HT approach does improve the efficiency in the fixed effects model.