When the error term forms a martingale process, the asymptotic covariance matrix can be consistently estimated by the HAC covariance matrix estimator. 14. Autoregressive conditional heteroskedasticity (ARCH) process is an MDS. 15. Generalized autoregressive conditional heteroskedasticity (GARCH) process is not an MDS. 16. The parameters in the GARCH process can be estimated by the OLS estimation. 17. ARCH process is used to model time-varying volatility. 18. If the explanatory variable is an MDS, the asymptotic covariance matrix of the OLS estimator can be consistently estimated by the heteroskedasticity consistent (HC) covariance matrix estimator. 19. The limit distribution of the OLS estimator is the same irrespective of error serial correlation. 20. The HC covariance matrix estimator constitutes to the HAC estimator. 21. If the model is dynamically misspecified, the residuals obtained by the OLS estimation has not zero population mean. 22. Independent data sets are necessarily ergodic. 23. The trend-stationary process is stationary. 24. If the coefficient of Yt-1 is estimated by the ordinary least square (OLS) method, then it's super-consistent, when Y₁ = Yt-1+ut and u~i.i.d. N(0, 1). 25. Let B(-) stand for the Brownian motion, and a data set, {Xt}, is collected from B(-) when t is integer-valued from 1 to n. That is, {Xt = B(t): t = 1,2,...,n}. {Xt}=1 must be a stationary process. 26. The spurious regression occurs when the data set of interest is stationary. 27. Unit-root process is a stationary process. 28. Generalized autoregressive conditional heteroskedasticity (GARCH) process is a non-stationary process. 29 CARCH process forms a martingale difference sequence