Say I have the following model: $y_t = 0.5y_{t−1} +x_t +v_{1t}$, and $x_t = 0.5x_{t−1} +v_{2t}$, where both $v_{1t}$ and $v_{2t}$ follow IID normal distribution ∼ (0,1).

How would I go about showing whether the following statement is true or false?

'If a sample of 1000 observations were simulated from this model for $y_t$, then at least five values will be expected to be larger than 6.'

I have been struggling with this question for many days, is it an old exam question, and I have never seen a question similar to this before... does this type of model have a specific name? I am struggling to find anything online to help me understand.


The first part of the question:

I think you should have calculated the unconditional mean and variances. These are time-constant, because the processes are stationary: $$E(y_t ) = E(y_{t-j} ),$$ and $$var(y_t ) = var(y_{t-j} ),$$ and the same for $x_t$.

Theses are my results: $$E(y_t ) =E(x_t )=0, \quad var(x_t ) = \frac{4}{3}, \quad var(y_t ) = \frac{28}{9} $$

The unconditional standard error of $y_t$ is $1.7$ and is normally distributed. This means that $0.02$ % of observations will be larger than 6. In other words, in a sample of 10000, I expect 2 observations to be larger than 6.

  • $\begingroup$ $$\mathrm{var}(y_t)=\frac{116}{27}\qquad P(y_t\geqslant6)\approx0.13\%$$ $\endgroup$ – Did Jan 17 '17 at 9:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.