中文摘要：

Chang（2002）基于非线性工具变量估计，提出了对截面相关综列数据进行单位根检验的SN检验，在综列数据生成过程不舍飘移和时间趋势时具有良好的表现。但我们的研究表明，在非平稳综列数据含有确定性趋势时，SN检验是有偏的，不再服从标准正态的渐近分布，而需要进行修正。本文基于蒙特卡罗仿真试验提出了一种简便的修正方法，修正后的SN检验统计量渐近分布为标准正态分布。仿真结果表明，修正后的SN检验具有更好的有限样本性质和较高的检验势。

英文摘要：

The SN unit roots test proposed by Chang （2002） for panels with cross-sectional dependency based on the nonlinear Ⅳ estimation performs better than the existing tests. However, we found, for nonstationary panels with deterministic trends, the SN test statistics would be biased and its asymptotic distribution would not be standard normal. We introduce a simple but effective way to correct the SN statistics, and the asymptotic distribution of the modified SN statistics is standard normal. According to the result of Monte-Carlo simulation, the modified SN test performs very well in terms of both finite sample sizes and powers, but the unmodified does not.