Abstract
We propose an alternative bivariate zero-inflated negative binomial (BZINB) regression model based on a copula. The empirical result shows that the proposed model performs better than the existing BZINB models in terms of the maximum log-likelihood and the AIC.
Original language | English |
---|---|
Pages (from-to) | 183-185 |
Number of pages | 3 |
Journal | Economics Letters |
Volume | 113 |
Issue number | 2 |
DOIs | |
State | Published - Nov 2011 |
Keywords
- Bivariate negative binomial
- Copula
- Heterogeneous dispersion
- Negative correlation
- Zero inflation