1' Estimating demand relationships

1
1. Estimating demand relationships

• From theory to estimation
• Overview of regression analysis
• Estimating market demand

2
2. Quick review

• Quantity demanded is a function of
• own price (substitution, income)
• price of other goods
• income
• expectations
• population
• advertising and tastes

3
3. Demographics

• Big focus in marketing on age, gender, type of
  household, etc.
• Buying patterns
• New products
• Product development

4
4. Network Externalities

• Up to this point we have assumed that peoples
  demands for a good are independent of one
  another.
• If fact, a persons demand may be affected by the
  number of other people who have purchased the
  good.
• Examples fads, snob appeal, new technology

5
5. Network Externalities

• A positive network externality exists if the
  quantity of a good demanded by a consumer
  increases in response to an increase in purchases
  by other consumers. (Fads, new technology)
• Negative network externalities are just the
  opposite. (Snob goods)

6
6. Positive Network Externality
Price ( per unit)
D200
At one point, there were only 200,000 people
who belonged to AOL.
Quantity (thousands per month)
200
400
600
800
1000
7
7. Positive Network Externality
Price ( per unit)
D200
D400
However, if another 200,000 people join, the
value of chat rooms and instant messaging
increases, so the demand curve shifts right.
Quantity (thousands per month)
200
400
600
800
1000
8
8. Positive Network Externality
Price ( per unit)
D200
D400
D600
D800
D1000
The more people subscribing to AOL, the further
to the right the demand curve
Quantity (thousands per month)
200
400
600
800
1000
9
9. Positive Network Externality
Price ( per unit)
D200
D400
D600
D800
D1000
The market demand curve is found by joining the
points on the individual demand curves. It is
relatively more elastic.
30
Demand
Quantity (thousands per month)
200
400
600
800
1000
10
10. Positive Network Externality
Price ( per unit)
D200
D400
D600
D800
D1000
Suppose the price falls from 30 to 20. If there
were no bandwagon effect, quantity demanded
would only increase to 480,000
30
20
Demand
Quantity (thousands per month)
200
400
600
800
1000
480
Pure Price Effect
11
11. Positive Network Externality
Price ( per unit)
D200
D400
D600
D800
D1000
But as more people buy the good, it becomes
stylish to own it and the quantity
demanded increases further.
30
20
Demand
Quantity (thousands per month)
200
400
600
800
1000
480
Pure Price Effect
Bandwagon Effect
12
12. Estimating Demand Parameters

• Back of the envelope estimates of elasticities
• Consumer interviews and surveys
• Producer goods cost savings
• Market experiments
• Uncontrolled market data

13
13. Out of control data?
14
14. Models of behavior

• Ydependent variable
• MSM GPA
• Salary
• Odds of car purchase
• Intel stock returns
• Sales

• Xindependent variable
• GMAT
• Schooling, region
• Mileage of current car
• Market returns
• Advertising price

15
15. Types of data

• Time series
• Cross section
• Hybrids
• Continuous variables
• Dummy variables

16
16. Data on sales advertising
17
17. Simple regression

• Goal estimate a and b in YabX
• a Y when X0 (vertical intercept)
• b change in Y for 1 change in X (slope)
• Add error term
• Yi a bXi ei
• Estimate by least squares

18
18. Data on sales advertising
19
19. Multiple regression

• Allows Y to be a function of gt 1 variables
• Y a bX1 cX2 . . . kXk e
• In 2D interpretation of Y, X1 , role of
• X2 . . . Xk is to shift vertical intercept
• New issue multicollinearity

20
20. Precision of estimates

• Standard errors
• Confidence intervals
• T-statistics

21
21. Heteroskedasticity
22
22. Serial correlation
23
23. Other issues

• Goodness of fit
• -- Standard error
• -- R2
• Common sense
• Forecasting

24
24. Conceptual model

• Ideal Q f(P, I, P of other goods, A)
• Compromises often necessary because
• - data not available
• - data not independent
• - data analyst should keep sanity
• - past value of Q might be best predictor

25
25. Econometric model

• Choose functional form and X variables based on
• Theory or common practice in economics
• Scatterplots
• Intuition
• Avoid all combos of all variables

26
26. What is this?