An integer-valued random walk process is proposed. Some limit distributions of on sums about the random walk process are obtained. It is shown that the limit distribution of OLS estimators about the autore-ssive coefficient of unit root with no intercept and no time trend is no longer the functional of Weiner process, but convergences to the true value in probability. Monte Carlo simulation is also used to verify the reasonableness of the conclusions.%提出了整数值随机游走过程,推导出该随机游走过程的若干和式的极限分布.证明了无截距、无时间趋势的单位根过程的自回归系数的最小二乘估计量的极限分布不再是Weiner过程的泛函,而是依概率收敛到真值1,最后还用Monte Carlo模拟验证了该结论的合理性.
展开▼