7/4/2022

6.1 Hedonic Pricing Model

CHAPTER SIX
HEDONIC AND CONTINGENT MODELS  The hedonic regression method is a regression technique
used to determine the value of a good, service, or asset by
fractionating the product into constituent parts or
characteristics.
 It is done to determine the contributory value of each
6.1 Hedonic Pricing Model characteristic separately through regression analysis.
 It is based on the theory that the price of an asset is a
6.2 Contingent Valuation Model function of its quantifiable characteristics, which can be
plotted in a regression model to determine how the price
changes with regard to changes in each characteristic.

1 2

6.1 Hedonic Pricing Model 6.1 Hedonic Pricing Model

 The data should include the following:
 Asset characteristics differ according to the asset but can  Property selling prices and the respective locations of
include various elements, such as weight, color, speed, the properties
 Property characteristics that affect property selling
power, size, location, form, etc.
prices, such as type of property, size, number of
 Hence, the hedonic model is used to estimate the effects of rooms, size of rooms, etc.
changes in the quality of a product on its price.  Neighborhood characteristics that affect prices, such as
 Hedonic methods are largely used in real estate pricing to property tax, crime rate, scenic views, quality of
estimate the value of properties. schools, etc.
 Real estate pricing is determined by a variety of factors that  Accessibility characteristics that affect prices, such as
make hedonic regression the perfect estimation tool. proximity to shopping centers, malls, and workplaces,
 To apply hedonic regression methods to property pricing, availability of public transport, etc.
 Environmental characteristics that affect prices, such as
participants should first carry out a data-gathering exercise
quality of water and air, etc.
on property sales for a predetermined period.
3 4

1
 6.1 Hedonic Pricing Model Hedonic Regression Function
7/4/2022
 The hedonic regression function illustrates the relationship
between the price of the asset (being the dependent
variable) and the components/characteristics of the asset
(being the independent or explanatory variables).

pi = j (ci)

Where: p is the price of a variety i of a good

ci is a vector of characteristics associated
with the variety of the good

5 6

Hedonic Regression Function Hedonic Regression Function

 We will use a real estate example to demonstrate the hedonic  Example: Determinants of Residential House Price in Bahir Dar
regression function in an applicable format. City.
p = (loc, str, acc, env, nei)  The dependent variable is the price of residential houses in bahir
Where: Dar city ( continuous variable measured in Birr).
p is the price of a property  The explanatory variables that are expected to determine the price
 The explanatory variables are the characteristics that determine the of a residential house are;
price of a property being:  FLA: Floor area in m2
 loc is the location characteristics, i.e., urban, rural, distance from the city  GPA: Gross plot area in m2
center, etc.  AGE: Age of the house from its construction in years
 str is the structure of the property, i.e., number and size of rooms, size of  TRANAV: Transport availability: 1 if it is with in 10
the stand, property age, etc.
 acc is the accessibility of the property, i.e., proximity to social amenities,
minutes walking distance to road; 0 otherwise
public transport accessibility, etc.  MARAV: Market Availability: 1 if it is within 10
 env is the environmental quality, i.e., quality of air, quality of water, etc. minutes walking distance to market; 0 otherwise
 nei is the neighborhood characteristics, i.e., crime rate, scenic views, quality  SCHAV: School availability: 1 if it is within 10 minutes
of schools, etc. walking distance to market; 0 otherwise
7 8

2
 Hedonic Regression Function Contingent Valuation Method
7/4/2022
 CVM has become a useful tool in determining willingness to pay
(WTP) when markets are missing and preferences cannot be
revealed through market responses.
 Contingent valuation is probably the most widely used method for
placing monetary values on public goods.
 The contingent valuation method (CVM) involve directly asking
people, in a survey, how much they would be willing to pay (WTP)
for a given property.
 Contingent valuation (CV) method is based on consumer surveys
whose questions elicit consumer preferences for public goods by
 The results indicate that price of a residential house is determined by constructing a hypothetical market for the public goods.
the size of floor and the size of the total plot area.  The aim of a CV study is to estimate consumers WTP for public
 A 1m2 increment in the size of floor resulted in 14069 Birr increment goods by asking them how much they would pay for certain
in the price of a house. government actions.
 A 1m2 increment in the size of the total plot area resulted in 6568 Birr
increment in the price of a house. 9 10

1.1 Introduction 1.1 Introduction

 Suppose you wants to study determinants of  The logit model can be specified as
individuals WTP for a certain public good.
 In this case a respondent is asked a question if s/he is willing
to pay a certain amount of money for a given public good  It could be written in terms of the odds ratio and log of odds ratio,
(CVM). which enable one to understand the interpretation of the
 Hence, the dependent variable is dummy which takes coefficients.
1=If an individual is willing to pay  The odds ratio can be computed as;
0=Otherwise
 The probability of an individual to be willing to pay may be
determined by various variables (X1, X2,…., Xn).  It can be written in terms of logs of odds ratio;
 Logit regresion can be used to estimate the influence of key
parameters on the explanatory variable.

11 12

3
 1.1 Introduction 1.1 Introduction
7/4/2022
 Assume that WTP for constructing a new road is expected to be  Logistic Regression model (odds ratio)
influenced by Income, Education and age of households.
 Primary data is collected from 40 households and the estimated logit 
model is given as;

 As the monthly income of a household increase by one birr, the

 As the monthly income of a household increase by one birr, the odds ratio to be willing to pay increase by 1.001 times.
loggs of odds ratio towards willingness to pay increase by 0.001.  The odds ratio to be willing to pay for literate households is greater
 The logs of odds ratio to be willing to pay for literate households is than illiterates by 17.85 times.
greater than illiterates by 2.88. 13 14

1.1 Introduction Evaluating the overall performance of the Logit

 Logit Model (Marginal Effects) model
1. Hosmer-Lemeshow goodness of fit test
 It is a goodness of fit test for logistic regression
 It tells you how well your data fits the model
 The Hosmer–Lemeshow test specifically identifies subgroups
as the deciles of fitted risk values. Models for which expected and
observed event rates in subgroups are similar are called well
calibrated.

 As the monthly income of a household increase by one birr, their

probability of being willing to pay increase by 0.0095%.
 The probability of being willing to pay for literate households is
greater than illiterates by 30.62%.
 large p-values (usually > 5%) mean that your model is a good fit
15 16

4
2. Percent of Correct Predictions
7/4/2022
3. likelihood ratio (LR) statistic

 To test the null hypothesis that all the slope coefficients are
simultaneously equal to zero, the equivalent of the F test in the
linear regression model is the likelihood ratio (LR) statistic.
 The small p-value from the test, <0.05, would lead us to
conclude that at least one of the regression coefficients in the
model is not equal to zero.

Number of obs = 40
LR chi2(9) = 33.14
Prob > chi2 = 0.0000
The result indicate that overall 90% percent of the  Hence, the null hypothesis that all coefficients of interest are
observations are correctly predicted. Hence, the data well fit
the model.
simultaneously equal to zero is rejected.