Week 3

Slide #

Scene

Narrations

Slide 1

Scene 1

An older cottage style family run business (Katrina’s Candies)

Slide 2

Scene 2

Herb and Renee are in Renee’s office discussing how to select and use forecasting techniques to forecast the behavior of variables Katrina uses to make decisions.

Display:

Regression output for the model.

ECO550_3_2_Herb-1: Good Morning, Renee!

ECO550_3_2_Renee-1: Good Morning, Herb! We should get started because I have another meeting following our meeting. What progress did you and Maria make at your session? I saw your email; however, I’ve been too busy to read it.

ECO550_3_2_Herb-2: No problem, Renee! I will give you an update now. Maria and I made a lot of progress considering we had to revise the model you and I developed.

ECO550_3_2_Renee-2: Revise the model, Why?

ECO550_3_2_Herb-3: Let me explain. The first revision occurred before we estimated the model. Maria was unable to find prices for bottled-water and unable to get data on the number of buyers of sugar-free-chocolate. Maria, however, located proxy data for both of these independent variables; as a result we estimated the demand model you and I formulated using some proxy data.

ECO550_3_2_Renee-3: That’s good news; otherwise, we would have had to reformulate the entire model.

ECO550_3_2_Herb-4: I was quite impressed that Maria made the extra effort to find proxy data to use in order to estimate the demand model. However, after estimating the model and evaluating the significance of regression coefficients, we had to drop caffeinated coffee and bottled-water from the model; as the results were insignificant for both variables. We also added a dummy variable to capture the supply of Katrina’s sugar-free-chocolate on the demand for Katrina’s regular chocolates.

ECO550_3_2_Renee-4: So, did the modifications that were made to the model make a difference in estimation results?

ECO550_3_2_Herb-5: Yes, as a matter of fact, the revised version of the demand model generated significant results for all of the independent variables including the price of Katrina’s sugar-free chocolate, median household income, the export of domestic confectionary merchandise and the dummy variable. Let me show you the regression output for the model.

Slide 3

Scene 3

Herb and Renee are in Renee’s office discussing the first forecasting procedure.

Show linear trend model on projector.

Show line graph example on projector.

ECO550_3_3_Renee-1: Herb, this is a great report--this means we can use the data and the estimated model to forecast the future demand.

ECO550_3_3_Herb-1: How are we going to forecast the demand, Renee?

ECO550_3_3_Renee-2: Since we used historical time-series data to estimate the demand model, we will use forecasting methods for time series data.

ECO550_3_3_Herb-2: That makes sense but where do we begin?

ECO550_3_3_Renee-3: We’ll begin with a simple forecasting method, a linear trend model. In a linear trend model, the dependent variable is regressed against only one independent variable, which is time. For this model, the quantity of Katrina’s sugar-free-candy is the dependent variable and the corresponding year is the independent variable for time.

ECO550_3_3_Herb-3: That’s easy to do, we can use Excel again. Let me open the Excel file Maria created.

ECO550_3_3_Renee-4: Okay, Herb. When you open the dataset, make a smaller dataset consisting of just the quantity of Katrina’s chocolates and year.

ECO550_3_3_Herb-4: Done. That didn’t take long; I just copied and pasted the data onto a new spreadsheet.

ECO550_3_3_Renee-5: Good. Now, let’s look at a time-series graph of Katrina’s chocolates and time.

ECO550_3_3_Herb-5: A graph, what type of graph?

ECO550_3_3_Renee-6: A line graph gives the clearest picture of the relationship between time and a dependent variable. Excel uses the term “chart” instead of graph so look under the Charts option.

ECO550_3_3_Herb-6: Okay, found it! I had to select the “Insert” tab, then the Line option. Please take a look at my graph.

Slide 4

Scene 4

Renee and Herb are in Renee’s office to analyze the graph generated for the trend-line forecast

Show the demand function on projector.

ECO550_3_4_Renee-1: The graph reflects what we would expect; the demand for Katrina’s sugar-free-chocolates is increasing over time.

ECO550_3_4_Herb-1: It also looks as if demand has grown during the last four years. Is that right?

ECO550_3_4_Renee-2: Yes, the line is positively sloped so you correctly interpreted the relationship. Now let’s estimate the regression for the trend line model.

ECO550_3_4_Herb-2: Okay, it will only take a few seconds. [PAUSE] There it is now.

ECO550_3_4_Renee-3: Before we can use the results for forecasting, we have to check for significance.

ECO550_3_4_Herb-3: I’ll do it, Renee. Maria and I performed the same significance tests when we estimated the demand model.

ECO550_3_4_Renee-4: Okay! Can we use this estimation to forecast?

ECO550_3_4_Herb-4: Yes, we can use this trend-line to make predictions. Let me show you.

ECO550_3_4_Renee-5: You can even use this information to create a forecast.

ECO550_3_4_Herb-5: How do we do that?

ECO550_3_4_Renee-6: The data we used was from a span of several years. We can forecast this years demand or even the demand for upcoming years by substituting this year or other years for “t” then solve for the quantity. Here let me show you.

ECO550_3_4_Herb-6: Trend-lining is a useful forecasting methodology, and it’s simple.

ECO550_3_4_Renee-7: Yes it is. Now let’s use the demand function you and Maria estimated to develop another forecast.

ECO550_3_4_Herb-7: The demand function we estimated is here, in Excel, here’s the version I created. I am unsure though of how we will use this function to forecast.

Slide 5

Scene 5

Renee and Herb are in Renee’s office to calculate demand using hypothetical values

ECO550_3_5_Renee-1: You may have already used this approach. All we need to do is hypothesize values for the next year’s independent variables, substitute the hypothetical values into the model and solve to get the forecasted value of demand. This method is called a point forecast.

ECO550_3_5_Herb-1: Yes, I actually do recall this approach. Let’s both solve this problem to make certain I get it correct.

ECO550_3_5_Renee-2: Okay. Use eight-dollars for the price, since price will probably not change within the next year; then use fifty-two thousand-seventeen-dollars for income. We will then have exports measured in pounds per year, so use four hundred eighty-thousand-eight-hundred twenty-six-point-five as the hypothetical value. Finally, use one for the dummy variable.

ECO550_3_5_Herb-2: Are you solving by hand without a calculator? I’m going to use Excel to solve for the predicted level of demand for this year..

ECO550_3_5_Renee-3: Let’s compare solutions. This is the solution that derived.

ECO550_3_5_Herb-3: My solution is the same! I’m glad I did this correctly in Excel.

Slide 6

Scene 6

Renee and Herb are in Renee’s office comparing the results from the demand function and the trend line

ECO550_3_6_Renee-1: Keep in mind that the demand function forecast using hypothetical data is larger than the forecast using the trend-line model. Off-hand, I cannot explain the difference between the two forecasts, however, both forecasts show some decrease but are still attractive numbers. Now we will use smoothing techniques to forecast demand. There are three types of smoothing techniques and they include: moving averages, weighted averages and exponential averages.

ECO550_3_6_Herb-1: Is moving average the method where you combine data points from different time periods then find the average of those data points to forecast a value?

ECO550_3_6_Renee-2: Yes, that’s correct! The forecasted value is based upon an average of two or more values. Since we have only twelve years of data, we will use a two-year moving average to forecast the demand.

ECO550_3_6_Herb-2: Are we using the data we compiled to estimate the demand function?

ECO550_3_6_Renee-3: Yes, that’s the data we’ll use.

ECO550_3_6_Herb-3: Take a look on page four of the report I emailed you, the dataset is there.

ECO550_3_6_Renee-4: I see it now. Make note that since we are developing a two-period moving average, we start by taking the average for the first two years of actual data, and then use each year to forecast the next year.

ECO550_3_6_Herb-4: Ok, I understand! So using two periods, I would compute ninety-thousand plus one-hundred-thousand divided by two to get the demand forecast which is ninety-five-thousand. Then for the next year the demand forecast would be one-hundred-fifteen-thousand and so on using this method. I used the Excel Data Analysis, Moving Average function to finish, let me print the output for you.

ECO550_3_6_Renee-5: Wait; since you’re going to print, we should compute the forecast error, which is the difference between the forecasted amount and the actual amount. Just use Excel to find forecast errors; organize the information into a table so we can see how it looks.

Slide 7

Scene 7

Renee and Herb are in Renee’s office finishing up their discussion on weighted average and exponential smoothing techniques.

ECO550_3_7_Herb-1: According to the moving average forecast, actual demand values are consistently higher than the forecast. What does that mean, Renee?

ECO550_3_7_Renee-1: I know the behavior of forecast error means there is forecast bias. However, other than that I don’t know. We’ll need to ask Ken about this.

ECO550_3_7_Herb-2: I do have one more question. Can Excel also calculate the weighted average and exponential smoothing forecasts?

ECO550_3_7_Renee-2: Let me see if we have anything in the video library. (Pause, clicking) Okay, I found a couple of videos that explain and illustrate how to smooth data using weighted averages and exponential averaging. However, it’s nearly time for my next meeting. I think you can use the videos to learn how to use the other two smoothing techniques.

ECO550_3_7_Herb-3: Thank you for the videos they should great learning tools!

ECO550_3_7_Renee-3: Fantastic! Before I head out, once you finish these videos I would like for you also to participate in a review activity I put together based on key items we discussed.

Slide 8

Interaction Slide

Incorporate iPad to show Videos about Excel and model creation

· What is Demand Forecasting?

· http://www.smetoolkit.org/smetoolkit/en/content/en/416/Demand-Forecasting

· Forecasting Using Regression Analysis

· http://www.youtube.com/watch?v=E73AJ73-S6g

· Using Excel for Basic Forecast Smoothing

· http://www.youtube.com/watch?v=sg7Mv54sISQ

Slide 9

Check Your Understanding

Multiple Choice Questions

Question 1: Time-series methods of forecasting are identified by which of the following characteristics?

a. They are based on the assumption that future events will follow patterns of past economic behavior. Correct Answer*

Correct feedback: Time-series forecasts are based upon an analysis of historic data. Therefore, the logical assumption is that the behavior of the variable will continue, ceteris paribus.

b. They generate data primarily from the opinion(s) of one or more people.

Incorrect Feedback: Forecasts based upon opinions are known as “subjective” forecasts. Subjective forecast are common occurrences yet are considered less reliable than forecasts based upon data. As an example, a subjective forecast is when a person who commutes to work daily states, “today is Wednesday, I know roads will be congested because on Wednesdays there is always a lot of traffic.

c. They incorporate economic theory with quantitative techniques to analyze and forecast the movement of some economic or business variable of interest.

Incorrect Feedback: Economic theory is the basis of some forecasting techniques but not all forecasting techniques.

d. They make use of inter-industry linkages to forecast how changes in demand will affect output by various industries.

Incorrect Feedback: Forecasting techniques are intended to evaluate the behavior of the variable of interest.

Question 2: Which of the following is a qualitative forecasting method?

a. Expert opinion

b. Consumer surveys

c. Delphi method

d. All of the above are forecasting methods.

Correct Feedback: That is correct; all of the listed choices are qualitative techniques.

Incorrect Feedback: Yes, this is a qualitative technique; however, all listed choices are qualitative techniques.

Slide 10

Scene 10

Concluding scene taking place in conference room

ECO550_3_10_Herb-1: Renee, those videos and review activities were very helpful!

ECO550_3_10_Renee-1: I’m glad to hear that. That is all I have for today, Do you want to begin our review of what we accomplished today?

ECO550_3_10_Herb-2: Sure thing! Today we developed forecasts using the trend-line method and moving average smoothing technique. We also used the estimated regression function we formulated for the demand for Katrina’s chocolates. We then used actual data for the trend-line and moving averages forecasts. We later hypothesized data for forecast using the estimated regression model. Based on all of our work, we concluded that the estimated regression model gave us the best estimate of our future demand.

ECO550_3_10_Renee-2: That’s a good summary of what we did today. Remember to document everything we have worked on so we can respond to any questions Ken may ask.

ECO550_3_10_Herb-3: Okay. Do we need to do anything else today?

ECO550_3_10_Renee-3: That’s all for today, Herb, but until we meet again, don’t forget to complete your weekly threaded discussions based on the key concepts we covered this week.

ECO550_3_10_Herb-4: Thanks, Renee and have a great day!