中文摘要：

Block Bootstrap方法因其适用范围广和操作简单等优点成为面板单位根检验的理想工具之一。然而，由于要求误差项服从独立同分布的假设条件，该方法仍具有一定的局限性。为此，本文发展Wild Bootstrap方法来解决误差项可能具有截面相依性和重尾性等更一般情形下的面板单位根检验问题。Monte Carlo模拟结果表明，当重尾性存在时，Wild Bootstrap检验相对Block Bootstrap检验有更小的水平扭曲和更高的功效。最后对中国股票市场的有效性问题进行了实证检验，并得出其为弱式有效性的结论。

英文摘要：

For the cross-sectional dependent panel data, Block Bootstrap panel unit root test method is attracting the attention of the theory and the applications researchers due to its desired properties such as general model settings and simplicity. Noted that, however, this method needs to assume that the disturbances are independent and identically distributed, which results in some restrictions of this method. So, this paper considers the alternatives bootstrap method for the general settings such as the cases with both cross-sectional dependency and heavy-tailed properties. Monte Carlo simulation results show that Wild Bootstrap test has smaller size distortion and higher power than Block Bootstrap test in the case with heavy-tailed properties. In addition, the Wild Bootstrap panel unit root test producer is then applied to the Chinese securities market weak efficiency hypothesis, and the empirical results imply that the Chinese securities market is general weak efficiency.