Estimating Demand Function

• Where do demand functions come from?

• Sources of information for demand estimation
• Cross-sectional versus time series data
• Estimating a demand specification using the
ordinary least squares (OLS) method.
• Goodness of fit statistics.
• Forecasting demand by using regression
equation
 The Purpose of Estimating Demand Function

1. To determine factors that influence the demand of the

product (empirically significant factors)
2. The analyze the impact of the factors on demand of
the product (elasticity)
3. To forecast demand of the product that are crucial for
planning and also in managing our business
 Demand function: Recall

Lets say that the demand function was as follows:

Q = 20 + 2Y + P” – 3P
Where Q is the quantity demanded/units of goods sold;
Y is a level of income,
P” is the price of our rival product,
and, P is the price of our own product.

The issue is how we estimated

this demand equation?
 Questions managers should
ask in estimating demand equations

1. What is the “best” equation that can be

obtained (estimated) from the available
data?
2. What does the equation not explain?
3. What can be said about the likelihood
and magnitude of relationship between
variables?
4. What are the consequences of forecast
errors? (if want to use for forecast)
 How do get the data to estimate
demand equations?

• Customer surveys and interviews.

• Controlled market studies.
• Uncontrolled market data.
• sales record/Time Series Data
Survey pitfalls
 Sample bias
 Response bias
 Response accuracy
 Cost
Types of data
Time -series data: historical data--i.e., the data sample consists of a series of
daily, monthly, quarterly, or annual data for variables such as prices, income ,
employment , output , car sales, stock market indices, exchange rates, and so
on.
Cross-sectional data: All observations in the sample are taken from the same
point in time and represent different individual entities (such as households,
houses, etc.)

Which one is better?

Panel Data/longitudinal Data!
 Example of Time series data: Daily
observations,

Year Month Day Won per Dollar

1997 3 10 877
1997 3 11 880.5
1997 3 12 879.5
1997 3 13 880.5
1997 3 14 881.5
1997 3 17 882
1997 3 18 885
1997 3 19 887
1997 3 20 886.5
1997 3 21 887
1997 3 24 890
1997 3 25 891
Example of cross sectional data

Student ID Sex Age Height Weight

777672431 M 21 6’1” 178 lbs.

231098765 M 28 5’11” 205 lbs.

111000111 F 19 5’8” 121 lbs.

898069845 F 22 5’4” 98 lbs.

000341234 M 20 6’2” 183 lbs

 Estimating demand equations
using regression analysis

Regression analysis is a statistical technique that allows

us to quantify the relationship between a dependent
variable and one or more independent or explanatory
variables.

If only one independent variable – simple regression

If more than one independence variables – multiple
regression

Which one is better?

 Specifying a simple regression model

Our model is specified as follows:

Q = f (P)

where Q is ticket sales and P is the fare

Q is the dependent variable—that is, we think

that variations in Q can be explained by
variations in P, the “explanatory” variable
Estimating the single variable model of demand function
(simple regression)

Since the data

Q i   0   1P i points are unlikely to fall
exactly on a line, (1)
must be modified
to include a disturbance
Q i   0   1P i   i term (εi)

0 and 1 are called parameters or

population parameters.
We estimate these parameters using the
data we have available (data from
sample)
 Simple Linear Regression Equation
(sample)

 The estimated simple linear regression equation

ŷ  b0  b1 x

• The graph is called the estimated regression line.

• b0 is the y intercept of the line.
• b1 is the slope of the line.
• ŷis the estimated value of y for a given x value.
Estimation Process
Sample Data:
x y
Regression Model
y = b0 + b1x +e x1 y1
. .
Regression Equation
. .
E(y) = b0 + b1x
xn y n
Unknown Parameters
b0, b1

Estimated
Regression Equation
b0 and b1 ŷ  b0  b1 x
provide estimates of Sample Statistics
b0 and b1 b0, b1
Estimation Method: Ordinary Least
Square
Y
(OLS)

X and Y are not

perfectly correlated.
However, there is
on average a positive
relationship
between Y and X

0 X1 X2 X
 We assume that
expected conditional values
of Y associated with
alternative values of X
Y fall on a line.
E(Y |Xi) = 0 + 1Xi

Y1
1
E(Y|X1) 1 = Y1 - E(Y|X1)

0 X1 X
OLS: Line of best fit

The line of best fit is the one that minimizes

the squared sum of the vertical distances of
the sample points from the line
Least Squares Method
 Least Squares Criterion

min  (y i  y i ) 2

where:
yi = observed value of the dependent variable
for the ith observation
y^i = estimated value of the dependent variable
for the ith observation
Least Squares Method
 Slope for the Estimated Regression
Equation

b1   ( x  x )( y  y )
i i

 (x  x )
i
2
 Least Squares Method

y-Intercept for the Estimated Regression Equation

b0  y  b1 x

where:
xi = value of independent variable for ith
observation
yi = value of dependent variable for ith
_ observation
x = mean value for independent variable
_
y = mean value for dependent variable
n = total number of observations
The 4 steps of demand
estimation using regression

1. Specification
2. Estimation
3. Evaluation
4. Analyzing/Forecasting
 Specification

How to determine that we are correctly specify

the demand function? (linear or non-linear)

Simple linear regression begins by plotting Q-P

values on a scatter diagram to determine if
there exists an approximate linear relationship
Year and Average Number Average
Ticket Prices and Ticket
Quarter Coach Seats Fare
97-1 64.8 250 Sales along an Air Route
97-2 33.6 265
97-3 37.8 265
97-4 83.3 240
98-1 111.7 230
98-2 137.5 225
98-3 109.6 225
98-4 96.8 220
99-1 59.5 230
99-2 83.2 235
99-3 90.5 245
99-4 105.5 240
00-1 75.7 250
00-2 91.6 240
00-3 112.7 240
00-4 102.2 235
Mean 87.3 239.7
Std. Dev. 27.9 13.1
 Scatter plot diagram
290

280

270

260
Fare

250

240

230

220

210
20 40 60 80 100 120 140 160

Passengers
Scatter plot diagram with possible line of best fit

Average One-way Fare

Demand curve: Q = 330-P

$ 27 0

26 0

25 0

24 0

23 0

22 0

0 50 100 150
Number of Seats Sold per Flight
 Computing the OLS
estimators

We can estimated the equation using the statistical

software package SPSS (Excel also can). It generated
the following output: SPSS!!!!!
Coefficientsa

Standardi
zed
Unstandardized Coefficien
Coefficients ts
Model B Std. Error Beta t Sig.
1 (Constant) 478.690 88.036 5.437 .000
FARE -1.633 .367 -.766 -4.453 .001
a. Dependent Variable: PASS
 Reading the SPSS Output

From the table we see that our estimate of 0 is 478.7

and our estimate of 1 is –1.63.

Thus our estimated demand equation

is given by:

Qˆ i  478.7  1.63Pi
 Evaluation

Now we will evaluate the estimated equation

using standard goodness of fit statistics,
including:
1. The standard errors of the estimates.
2. The t-statistics of the estimates of the
coefficients.
3. The standard error of the regression (s)
4. The coefficient of determination (R2)
 Standard errors of
the estimates

• We assume that the regression coefficients are normally distributed

variables.
• The standard error (or standard deviation) of the estimates is a
measure of the dispersion of the estimates around their mean value.
• As a general principle, the smaller the standard error, the better the
estimates (in terms of yielding accurate forecasts of the dependent
variable).

Rule-of-thumb is useful: The standard error of the

regression coefficient should be less than half of the
size of the corresponding regression coefficient.
 Coefficientsa

Standardi
zed
Unstandardized Coefficien
Coefficients ts
Model B Std. Error Beta t Sig.
1 (Constant) 478.690 88.036 5.437 .000
FARE -1.633 .367 -.766 -4.453 .001
a. Dependent Variable: PASS

By reference to the SPSS output, we see

that the standard error of our estimate
of 1 is 0.367, whereas the (absolute value)our
estimate of 1 is 1.63 Hence our estimate is about 4 ½
times the size of its standard error.
 The t test

To test for the significance of our estimate of 1, we set the

following null hypothesis, H0, and the alternative hypothesis, H1
H0: 1 0
H1: 1 < 0 ˆ 1   1 1.63  0
t   4.45
The t distribution is used to s 1
ˆ 0.049
test for statistical significance of
the estimate:
Coefficient of determination (R2)

The coefficient of determination, R2, is

defined as the proportion of the total
variation in the dependent variable (Y)
"explained" by the regression of Y on
the independent variable (X).
 ANOVAb

Sum of Mean
Model Squares df Square F Sig.
1 Regression 6863.624 1 6863.624 19.826 .001a
Residual 4846.816 14 346.201
Total 11710.440 15
a. Predictors: (Constant), FARE
b. Dependent Variable: PASS

Model Summary

Std. Error
Adjusted R of the
Model R R Square Square Estimate
1 .766a .586 .557 18.6065
a. Predictors: (Constant), FARE

We see from the SPSS model summary

table that R2 for this model is .586
Notes on R2
Þ Note that: 0 R2 1
Þ If R2 = 0, all the sample points lie on a horizontal
line or in a circle
Þ If R2 = 1, the sample points all lie on the regression
line
Þ In our case, R2  0.586,
58.6 percent of the variation in the dependent
variable is explained by the regression. Is it good?
 Standard error of the regression
Model Summary

Std. Error
Adjusted R of the
Model R R Square Square Estimate
1 .766a .586 .557 18.6065
a. Predictors: (Constant), FARE

Þ The model summary tells us that s = 18.6

Þ Regression is based on the assumption that the error term
is normally distributed, so that 68.7% of the actual values
of the dependent variable (seats sold) should be within one
standard error ($18.6 in our example) of their fitted value.
Þ Also, 95.45% of the observed values of seats sold should
be within 2 standard errors of their fitted values (37.2).
 Forecasting
Recall the equation obtained from the
regression results is :

Qˆ i  478.7  1.63Pi
At the most basic level, forecasting consists of inserting forecasted
values of the explanatory variable P (fare) into the estimated equation
to obtain forecasted values of the dependent variable Q (passenger
seats sold).
 In-Sample Forecast of Airline Sales
Year and Predicted Actual
Quarter Sales (Q*) Sales (Q) Q* - Q (Q* - Q)sq
97-1 64.8 70.44 5.64 31.81
97-2 33.6 45.94 12.34 152.28
97-3 37.8 45.94 8.14 66.26
97-4 83.3 86.77 3.47 12.04
98-1 111.7 103.1 -8.6 73.96
98-2 137.5 111.26 -26.24 688.54
98-3 109.6 111.26 1.66 2.76
98-4 96.8 119.43 22.63 512.12
99-1 59.5 103.1 43.6 1900.96
99-2 83.2 94.94 11.74 137.83
99-3 90.5 78.61 -11.89 141.37
99-4 105.5 86.77 -18.73 350.81
00-1 75.7 70.44 -5.26 27.67
00-2 91.6 86.77 -4.83 23.33
00-3 112.7 86.77 -25.93 672.36
00-4 102.2 94.94 -7.26 52.71

Sum of Squared Errors 4846.80

 In-Sample Forecast of Airline Sales
160

140

120
Passengers

100

80

60

40 Actual

20 Fitted
97.1 97.3 98.1 98.3 99.1 99.3 00.1 00.3

Year/Quarter
 Can we make a
good demand forecast?
Our ability to generate accurate forecasts of the dependent variable
depends on two factors:

• Do we have good forecasts of the explanatory variable?

• Does our model exhibit structural stability, i.e., will the causal
relationship between Q and P expressed in our forecasting equation
hold up over time?
• While the past may be a serviceable guide to the future in the case of
purely physical phenomena, the same principle does not necessarily
hold in the realm of social phenomena (to which economy belongs).
Single Variable Regression Using Excel

Class Discussion: by using excel

estimate an equation and use it
to predict home prices in two
cities. The data set is on the
next slide.
 City Income Home Price
Akron, OH 74.1 114.9
Atlanta, GA 82.4 126.9
• Income (Y) is
Birmingham, AL 71.2 130.9
average family
Bismark, ND 62.8 92.8
income in
Cleveland, OH 79.2 135.8
2011
Columbia, SC 66.8 116.7
• Home Price Denver, CO 82.6 161.9
(HP) is the Detroit, MI 85.3 145
average price Fort Lauderdale, FL 75.8 145.3
of a new or Hartford, CT 89.1 162.1
existing home Lancaster, PA 75.2 125.9
in 2011. Madison, WI 78.8 145.2
Naples, FL 100 173.6
Nashville, TN 77.3 125.9
Philadelphia, PA 87 151.5
Savannah, GA 67.8 108.1
Toledo, OH 71.2 101.1
Washington, DC 97.4 191.9
Model Specification


HP  b0  b1Y
 Scatter Diagram: Income and Home Prices

200
180
Home Prices

160
140
120
100
80
50 60 70 80 90 100 110
Income
 Excel Output Regression Statistics
ANOVA Multiple R 0.906983447
  df SS R Square 0.822618973
9355.71550 Adjusted R
Regression 1 2 Square 0.811532659
2017.36949 Standard Error 11.22878416
Residual 16 8 Observations 18
Total 17 11373.085

Standard
  Coefficients Error t Stat

Intercept -48.11037724 21.58459326 -2.228922114

Income 2.332504769 0.270780116 8.614017895

Equation and prediction

HP  48.11  2.33Y

City Income Predicted HP

Meridian, MS 59,600 $ 138,819.89
Palo Alto, CA 121,000 $ 281,881.89