🔧 What Was Developed
A method to estimate spatial-lag duration models when the outcome is subject to right censoring, the most common form of censoring in duration data. This fills a gap because existing spatial duration models do not account for censoring in the likelihood function.
🔍 How the Method Works
- Adapts Wei and Tanner's (1991) imputation algorithm for censored (non-spatial) regression to spatially interdependent duration models.
- Treats unobserved duration outcomes as censored values and alternates between:
- multiple imputation of the incomplete (right-censored) duration values, and
- estimation of the spatial-lag duration model using those imputed values.
- Focuses on an estimator for log-normal duration outcomes within a spatial-lag framework.
🧪 How the Approach Was Tested
- Performance evaluated via Monte Carlo simulations.
- Simulations vary the degree of right censoring to assess estimator behavior under different censoring levels.
🌍 Empirical Application
- Provides empirical examples by estimating spatial dependence in states' entry dates into World War I.
📌 Why It Matters
- Enables researchers to account for right censoring when modeling interdependent durations, improving the fidelity of spatial duration inference.
- Makes it possible to combine spatial dependence modeling with standard techniques for handling censored duration data.
