1: \begin{abstract}

2: We consider a multivariate time series model which represents a high

3: dimensional vector process as a sum of three terms: a linear

4: regression of some observed regressors, a linear combination of some

5: latent and serially correlated factors, and a vector white noise. We

6: investigate the inference without imposing stationary conditions on

7: the target multivariate time series, the regressors and the

8: underlying factors. Furthermore we deal with the the endogeneity

9: that there exist correlations between the observed regressors and

10: the unobserved factors. We also consider the model with nonlinear

11: regression term which can be approximated by a linear regression

12: function with a large number of regressors. The convergence rates

13: for the estimators of regression coefficients, the number of

14: factors, factor loading space and factors are established under the

15: settings when the dimension of time series and the number of

16: regressors may both tend to infinity together with the sample size.

17: The proposed method is illustrated with both simulated and real data

18: examples.

19: 

20: 

21: 

22: \end{abstract}