FRM Part I · FRM Exam Part I · Stationary Time Series
An analyst fits a regression to monthly data using an intercept and 12 monthly dummy variables, one for each month. What is the most likely problem?
The problem is perfect multicollinearity, the dummy variable trap: twelve monthly dummies always sum to one, duplicating the intercept column. The regression cannot be estimated uniquely unless one month is omitted as a base or the intercept is removed.
- APerfect multicollinearity, because the dummies sum to the intercept columnCorrect
- BHeteroskedasticity, because dummies are binary
- CNonstationarity, because dummies are deterministic
- DOmitted variable bias, because seasonal effects cannot be captured by dummies
Explanation
The 12 dummies sum to 1 in every period, identical to the intercept column, creating the dummy variable trap and perfect multicollinearity. The fix is to drop one dummy (base month) or drop the intercept.
Did you get it right without looking?
One question tells you little. A timed set on Stationary Time Series shows your real accuracy, how long you take and where you lose marks.
More Stationary Time Series questions
- A monthly series with annual seasonality is modeled as y_t = 0.5 y_{t-12} + e_t, where e_t is white noise with variance 1.5 and the process …
- A stationary AR(1) process has a long-run mean of 4 and is given by y_t = 1.2 + phi y_{t-1} + e_t. What is the autocorrelation at lag 3?
- An ARMA(1,1) model is Y_t = 2 + 0.7 Y_{t-1} + e_t + 0.2 e_{t-1}. The last observation is Y_T = 8 and the last residual is e_T = 1.5. What is…
- A white noise process ε_t has variance σ² = 4. A series is built as Y_t = 3 + ε_t + 0.5 ε_{t-1}, where the ε_t are white noise. What is the …
- A risk analyst models monthly changes in a credit spread index as a zero-mean AR(1) process: Y_t = 0.6 Y_{t-1} + e_t, where e_t is white noi…
- A covariance stationary AR(1) process is Y_t = 2 + 0.6·Y_(t-1) + ε_t, where ε_t is white noise with variance 1.28. What are the unconditiona…