Some model misspecification problems for time series: A Monte Carlo investigation

  • Dong-Bin Jeong (Department of Statistics, Kangnung National University)
  • Published : 1998.04.01


Recent work by Shin and Sarkar (1996) examines model misspecification problems for nonstationary time series. Shin and Sarkar introduce a general regression model with integrated errors and one system of integrated regressors and discuss the limiting distributions of the OLS estimators and the usual OLS statistics such as $\hat{\sigma^2}$t, DW and $R^2$. We analyze three different model misspecification problems through a Monte Carlo study and investigate each model misspecification problem. Our Monte Carlo experiments show that DW and $R^2$ can be in general used as diagnostic tools to detect spurious regression, misspecification of nonstationary autoregressive and polynomial regression models.



  1. Journal of Econometrics v.2 Spurious regression in econometrics Granger, C. W. J.;Newbold, P.
  2. The Annals of Probability v.12 A functional central limit theorem for weakly dependent sequences of random variables Herrndorf, N.
  3. Ph. D. dissertation. Department of Statistics Testing for a unit root in an AR(p) signal observed with MA(q) noise and model misspecification Jeong, D. B.
  4. The Annals of Statistics v.20 Asymptotics for linear processes Phillips, P. C. B.;Solo, V.
  5. Technical report #130, Dept. of Statistics Model misspecification problem for nonstationary time series Shin, D. W.;Sarkar, S.