Analysis of a continuous-time proportional hazard model using discrete duration data
研究了经济持续时间数据仅以区间形式获得时，将离散数据当作精确观测的近似方法渐近偏差，以及正确考虑数据离散性的最大似然估计效率。
Many economic duration variables are often available only up to intervals, and not up to exact points. However, continuous time duration models are conceptually superior to discrete ones. Hence, in duration analyses, one faces a situation with discrete data and a continuous model. This paper discusses (i) the asymptotic bias of a conventional approximation procedure in which a discrete duration is treated as an exact observation; and (ii) the efficiency of a correct maximum likelihood estimator which appropriately accounts for the discrete nature of the data.