引導式測驗 · 中等 等級

向量自迴歸測驗

回顧多元時間序列分析的 VAR 滯後方程式、參數成長、格蘭傑可預測性和假設。

相關引導路徑向量自回歸
問題 1 的 8

在 VAR(p) 中,每個方程式會出現哪些預測變項?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. 在 VAR(p) 中,每個方程式會出現哪些預測變項?
    • Lagged values of every included endogenous variable
    • Only current values of other variables, with no lags
    • Only each variable’s own lagged values, excluding cross-series lags
    • Only exogenous predictors, with no lagged endogenous terms
  2. 在具有三個變數的 VAR(2) 中,每個方程式在截距之前會有多少個落後係數?
    • 3
    • 6
    • 2
    • 9
  3. Granger 因果發現支持什麼?
    • The variables are contemporaneously independent
    • An intervention has a proven causal effect
    • Lagged values add predictive information conditional on the model
    • The model has no omitted variables
  4. 為什麼在差分多個非平穩序列之前要先評估共積?
    • A standard VAR requires all variables to be independent
    • Cointegration forces every VAR coefficient to zero
    • Differencing creates a seasonal period
    • Differencing may remove a meaningful long-run equilibrium relation
  5. 為什麼高維 VAR 在短序列上容易過度擬合?
    • Lagged predictors use no parameters
    • Each added variable and lag adds many coefficients
    • The residuals become deterministic
    • The model estimates only one mean
  6. What does residual autocorrelation after fitting suggest?
    • The variables are cointegrated
    • All lag terms are causal
    • Some temporal dependence may remain unexplained
    • The series is guaranteed nonstationary
  7. What assumptions affect structural impulse-response interpretations?
    • A zero-valued mean for all observations
    • Only the reported lag order
    • The alphabetical order of series names
    • Identification restrictions on correlated system shocks
  8. Which model may be considered for cointegrated nonstationary variables?
    • A vector error-correction model
    • A standard stationary VAR without checking
    • A k-means clustering model
    • A univariate moving average of residuals only