70
significant when the values of τ vary. For example, “Size” in both scenarios.
Furthermore, parameter estimates vary when the correlation between the depen-
dent censoring and events changes. In some cases, the estimates vary slightly. For
example, “Age” from the second scenario. The estimates may vary quite a bit. For
instance, “Profit” in the first scenario. Detailed results are shown below.
• Tests 3, 4, 5 and 6
Tables 4.3 - 4.12 present parameter estimates and related standard errors and 95
percent confidence intervals for both bankruptcy and acquisition scenarios. Figures
4.1 - 4.10 illustrate such data graphically. In those figures, “Beta” represents the first
scenario and “Betac” represents the second. Р-values are calculated based on the fact
that the following statistic (-g∙‰)2 follows the χ2 distribution.
Figure 4.11 shows the cumulative hazard curve of bankruptcy obtained using the
proposed method and its 95 percent confidence interval. It will be further discussed
in Chapter 5.
Table 4.3: Results for Covariate 1 - Growth for Bankruptcies
|
τ = —0.5 |
τ = 0 |
τ = 0.2 |
τ = 0.5 |
τ = 0.8 | |
|
Parameters |
-0.870 |
0.116 |
0.204 |
0.169 |
0.166 |
|
SE of Est |
1.127 |
0.440 |
0.232 |
0.247 |
0.208 |
|
Upper Bound |
1.338 |
0.978 |
0.658 |
0.654 |
0.575 |
|
Lower Bound |
-3.079 |
-0.746 |
-0.251 |
-0.315 |
-0.242 |