Mod 2 - Topic 3 - Demand estimation and forecasting

Question 1

Because data for analysis may be difficult to obtain, what options do businesses have? (4)

Answer 1

• purchase data from providers
• performance of customer surveys
• acquire from focus groups
• acquire from technology (point of sale / barcodes)

Question 2

What is regression analysis?

Answer 2

It is a statistical technique to find the best relationship between a dependent variable and selected independent variables.

Question 3

Regression analysis can be simple or multiple, what do these mean?

Answer 3

• Simple regression - Where one independent variable is used

* Multiple regression - Where more than one independent variable is used.

Question 4

What are the two types of regression analysis?

Answer 4

• Cross-Sectional - analyse a variable for a single period of time, eg. pizza demanded on each campus on a specific date.
• Time-series data - analyse a variable over multiple periods of time, eg. pizza demanded each week, annual income per capita

Question 5

What is the regression equation?

Answer 5

Y= a + b1X1 + b2X2 + b3X3…
Where:
Y is the dependent variable;
a is the constant value (y-intercept);
Xn is the independent variables used to explain Y
bn are the regression coefficients (measure the impact of the independent variables)

Question 6

What does a negative coefficient mean in the following circumstances:

• Price of the product (P)
• Price of a related product (PR)
• The Incomes of those who may purchase the product (I)

Answer 6

All move in the opposite direction:

• P = increase in price = decrease in demand
• PR = product is complementary: increase in demand in one = decrease in related product
• I = product is inferior: increase in income = decrease in quantity demanded.

Question 7

What does a positive coefficient mean in the following circumstances:

• Price of the product (P)
• Price of a related product (PR)
• The Incomes of those who may purchase the product (I)

Answer 7

All move in the same direction:

• P = increase in price = increase in demand
• PR = product is a substitute: increase in demand in one = increases demand for the other
• I = product is normal or superior: increase in income leads to an increase in demand

Question 8

What is the importance of the magnitude of the regression coefficients?

Answer 8

They are a measure of elasticity for each variable.

Question 9

From the regression equation, how do we determine the elasticity?

Answer 9

We need to determine Y (quantity demanded), so each variable must be given values, then we can apply the formula:
E = -b x (P/Q)

Question 10

What is the t-test and why is it important?

Answer 10

Because results are based on a sample the t-test provides the test of statistical significance of each estimated coefficient, which indicates if the result is reflective of the population.

Question 11

What is the ‘Rule of 2’?

Answer 11

It is where t is greater than 2 and means that the estimated coefficient is statistically significant at the 0.05 level.

Question 12

If the t-test is passed, what can we say about the variable?

Answer 12

It has a true impact on demand.

Question 13

What is R squared?

Answer 13

It is the coefficient of determination and is the percentage of variation in the variable (Y) which is accounted for by variation in all the explanatory variables (Xn)

Question 14

What is the range of R squared and what does the placement in this range mean?

Answer 14

R squared ranges in value from 0-1 and the closer to 1 the greater the explanatory power of the regression.

Question 15

What is the impact on R squared when more independent variables (Xn) are added?

Answer 15

It will increase.

Question 16

What is an F-test?

Answer 16

The f-test measures the statistical significance of the entire regression as a whole (not each coefficient) ie. it measures the statistical significance of R squared

Question 17

What are the steps for analysing regression results?

Answer 17

1) check the coefficient signs and magnitudes
2) compute implied elasticities
3) determine the statistical significance

Question 18

What are some common regression problems? (3)

Answer 18

• Identification problem
• Multicollinearity problem
• Autocorrelation problem

Question 19

What is the identification problem?

Answer 19

The estimation of demand may produce biased results due to simultaneous shifts of supply and demand curves.

Question 20

What is the solution of the identification problem?

Answer 20

Use advanced correction techniques such as two-stage least squares and indirect least squares.

Question 21

What is the multicollinearity problem?

Answer 21

Where two or more independent variables are highly correlated making it difficult to separate the effect each has on the dependent variable.

Question 22

What is the solution of the multicollinearity problem?

Answer 22

A standard remedy is to drop one of the closely related variables from the regression.

Question 23

What is the autocorrelation problem?

Answer 23

(Serial correlation) - it occurs when the dependent variable relates to the Y variable according to a certain pattern. Causes include omitted variables or non-linearity.

Question 24

What is the solution for the autocorrelation problem?

Answer 24

Consider transforming data into a different order of magnitude or introduce leading or lagging data.

Question 25

What is demand estimating used for?

Answer 25

To understand the effect on quantity demanded given a change in one or more independent variables.

Question 26

How does demand forecasting differ from demand estimating?

Answer 26

It places less emphasis on causes of demand change and more on future levels of sales given assumptions on the independent variables.

Question 27

What does a good forecast require? (4)

Answer 27

• consistency with other parts of the business
• be based on knowledge of the relevant past
• considers the economic and political environment
• be timely

Question 28

What are the two primary different forecasting classifications?

Answer 28

• Qualitative forecasting - forecasting based on the judgement of individuals or groups
• Quantitative forecasting - forecasting that examines historical data as a basis for future trends.

Question 29

What are the two different types of quantitative forecasting methods?

Answer 29

• Naive forecasting - projects past data without explaining the reasons for future trends.
• Casual (explanatory) forecasting - attempts to uncover functional relationships between independent variables and the dependent variable.

Question 30

What is ‘expert opinion’ forecasting?

Answer 30

A qualitative technique based on the judgement of knowledgeable people.

Question 31

What is ‘opinion poll and market research’ forecasting?

Answer 31

A qualitative technique which are conducted among survey populations (not experts) to establish future trends or consumer reponses

Question 32

What is ‘survey of spending plan’ forecasting?

Answer 32

A qualitative technique which is concerned with consumer sentiment and is based on replies to questionnaires or interviews.

Question 33

What is ‘economic indicator’ forecasting?

Answer 33

A technique which uses a number of economic indexes to forecast the short-run movements of the economy, including changes in direction.

Question 34

What is ‘trend projection’ forecasting?

Answer 34

A quantitative technique which employs historical data to project future tends (it is naive because no causes for trends are usually identified).

Question 35

What is ‘econometric model’ forecasting?

Answer 35

A quantitative technique (using a casual/explanatory model) that identifies which independent variables will influence the statistic to be forecast.

Question 36

What techniques are used to create a trend projection? (3)

Answer 36

1) Compound growth rate
2) Visual time series projection
3) Time-series projection using the least squares method

Question 37

What does time series forecasting examine? (4)

Answer 37

• Trends
• Cyclical fluctuations
• Seasonal fluctuations
• Irregularities (departures from the normal)

Question 38

What are three smoothing techniques that may be used?

Answer 38

• Moving average - uses actual average of past results
• Exponential smoothing - assigns greater importance to more recent data
• Box-Jenkins forecasting - uses an iterative procedure to arrive at the best result.

Question 39

What are the 5 steps to applying regression analysis on the estimation of demand?

Answer 39

1) Specification of the regression model of demand
2) Collection of the relevant data
3) Estimation of the regression equation
4) Analysis of the regression results
5) Assessment of regression findings for use in making policy decisions