中文摘要：

在构建随机效应面板数据Biprobit模型和部分可观测Biprobit模型的基础上，采用基于Halton序列的模拟极大似然法估计这类模型的参数．相比于采用传统的数值积分公式处理似然函数中二重积分的方法，依赖于MonteCarlo积分的模拟极大似然法具有不依赖于积分节点选取的数值稳定性，且无需过多抽样就可以保证求解的精度．模拟实验结果说明了算法的有效性，对农户消费信贷约束的实证结果表明，不同抽样次数下参数估计结果并无明显差别，算法具有稳定性．

英文摘要：

On the construction of random effects panel data Biprobit models and partial observability Biprobit models, we use simulated maximum likelihood method based on Halton sequences for parameter estimation. Compared with Gauss-Hermite quadrature, simulated maximum likelihood estimation relying on Monte Carlo integral is more stable and it doesn＇t need too much draws to achieve high precision. Simulation results confirm the~ effectiveness of the algorithm. Finally we give an application of this method.