1: \begin{abstract}

2: The eigenvector Empirical Spectral Distribution (VESD) is adopted to

3: investigate the limiting behavior of eigenvectors and eigenvalues of

4: covariance matrices. In this paper, we shall show that the Kolmogorov

5: distance between the expected VESD of sample covariance matrix and the

6: Mar\v{c}enko--Pastur distribution function is of order $O(N^{-1/2})$.

7: Given that data dimension $n$ to sample size $N$ ratio is bounded

8: between 0 and 1, this convergence rate is established under finite 10th

9: moment condition of the underlying distribution. It is also shown that,

10: for any fixed $\eta>0$, the convergence rates of VESD are $O(N^{-1/4})$

11: in probability and $O(N^{-1/4+\eta})$ almost surely, requiring finite

12: 8th moment of the underlying distribution.

13: \end{abstract}