<ARTICLE>Exact Distributions of Dickey-Fuller Statistics For Unit Root Testing
スポンサーリンク
概要
- 論文の詳細を見る
Let N(ρ-1), N(ρ_μ-1) and N(ρ_τ-1) be Dickey-Fuller Statistics for unit root testing, where N is size of sample and ρ,ρ_μ and ρ_τ are the ordinary least square estimates for coefficients in the time series models AR (1). In this paper, the exact distributions of those statistics, for N = 10 (1) 30 and 30 (5) 50, are numerically calculated by Imhof (1961) procedure. Our distributions for N - 25 and 50 are almost equivalent to the results of the Monte Carlo method (Fuller 1976, Table 8. 5. 1 and Dickey 1976, Table 5. 2). Using those distributions, the power of the testing for unit root hypothe- sis is numerically evaluated for N = 25 and 50. The unit root testing is bi- ased when true ρ is close to unity.
- 大阪府立大学の論文