Project

ECO 309

Economic Forecasting

Project Part 2

Due Saturday, February 22

 

 

This paper requires content knowledge in Chapters 3 – 5. Please listen to the lectures and read your chapters before starting your work.

 

You need to choose at least 2 (preferable 3) macroeconomic variables that you think will work well with your revenue variable. You can choose your macroeconomic variables from the Excel file entitled “Macroeconomic Variables” under DocSharing . Each macroeconomic variable should start and end on the same date as your revenue variable and should have the same sample size.

 

Copy and Paste the macroeconomic variables to your Minitab. To successfully complete your PP2, you will only need your X variables; we will not use the company revenue variable for this project part.

 

Please write a narrative for PP2 and do not use numbers.

 

1. Plot your X variables (Stats>Time Series>TS plot) and describe what you see. For each X variable, explain if it is stationary vs nonstationary, multiplicative vs. additive, seasonal or not, trending vs. not, if trending, is the trend linear vs non-linear?

 

2. Plot ACF of your X variables and identify trend/seasonality if any. Make sure to do your hypothesis testing for significance of autocorrelation. Do you have trend and/or seasonality?

 

3. For each X variable, run the Naïve method. Start making your table and making note of the error summaries.

 

4. For each X variable, run all three smoothing methods and enter error summary to your table.

 

5. Go to Time Series>Trend Analysis and try all 4 trend models. Pick the best one and tell me why it is the best one.

 

 

6. Go to Time Series>Decomposition, run both “SEASONALY ONLY” and “SEASONAL + TREND”. Also run both the additive and multiplicative models to confirm what you find in question 1. Which one is the best model, why?

 

7. Make a table that summarizes the errors (MSD, MAD, or MAPE) of each model for each of the X variables. Below is a partial table to give you an idea. Once you complete your table, pick the best overall model for each X variable. You will have as many rows in your table as methods you are using. See the partial incomplete table** example below.

 

8. Using the best method for each X variable, generate 10 out-of-sample forecasts.

**

MSD, MAD or MAPE	X1 (name of the 1st X variable)	X2(name of the 2nd X variable)	X3(name of the 3rd X variable)
Naive
Single smoothing
Double smoothing
Winters smoothing
Trend
All other methods