Abstract
In multivariate hazard rate models, the duration variables are dependent if their unobserved determinants are dependent on each other. This paper derives properties of association measures for these duration variables. Sharp bounds for correlations are derived in the general case as well as in cases in which the bivariate distribution of unobserved heterogeneity belongs to specific families. Similar results are derived for Kendall's tau, which does not depend on the shape of the baseline hazard. The results are useful for empirical analyses, as they can be used to compare the flexibility of different heterogeneity distributions.
Original language | English |
---|---|
Pages (from-to) | 221-245 |
Number of pages | 25 |
Journal | Journal of Econometrics |
Volume | 79 |
Issue number | 2 |
Publication status | Published - Aug 1997 |
Keywords
- Bivariate duration models
- Competing risks
- Correlation
- Duration models
- Kendall's tau