Problem 2: (about 20 minutes). This question has 3 sub-questions, 8 points each. Consider the AR(1) model below: It = Bu

[answerhappygod]

Problem 2 About 20 Minutes This Question Has 3 Sub Questions 8 Points Each Consider The Ar 1 Model Below It Bu 1 (121.41 KiB) Viewed 29 times

Problem 2: (about 20 minutes). This question has 3 sub-questions, 8 points each. Consider the AR(1) model below: It = Bu + Butt-1+ ut (1) where B1 <1 and u is a white noise normal error term. 1. Show that the iterated two-period-ahead forecast for 2, labeled as #t|t-2, can be written as 01/2-2 = 80 + 0171-2 (2) and derive the value for 80 and 81 in terms of the AR(1) coefficients in equation (1), Bo and B1. Given the model in equation (1) and the forecast in equation (2), what would the structure of the forecast error be? 2. If you know the population parameter values and the model in equation (1) is the correct model, do iterated multiperiod forecasts differ from direct multiperiod forecasts? Provide a supporting argument. 3. Would your answer to question 2 change if you did not know the population parameters and were supposed to estimate them? (Hint: if you have a correctly specified linear model, which approach would you pick for forecasting if you were to estimate the parameters?) Explain your answer