Guide-linked quiz · Medium Level

Vector Autoregression Quiz

Review VAR lagged equations, parameter growth, Granger predictability and assumptions for multivariate time-series analysis.

Related guide pathsVector Autoregression
Question 1 of 8

In a VAR(p), what predictors appear in each equation?

Preview every question8 questions

Answers and explanations appear as you take the quiz.

  1. In a VAR(p), what predictors appear in each equation?
    • 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. In a VAR(2) with three variables, how many lag coefficients appear in each equation before intercepts?
    • 3
    • 6
    • 2
    • 9
  3. What does a Granger-causality finding support?
    • 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. Why assess cointegration before differencing several nonstationary series?
    • 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. Why can a high-dimensional VAR overfit a short series?
    • 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

Keep testing yourself

More quizzes picked for your level.

View all quizzes