Consumers'' willingness to pay for plastic recycling in Vietnam the case of Ho Chi Minh city

RESEARCHES & DISCUSSIONS

Plastic recycling can help improving the environment quality by reducing waste discharge as well as keeping the surrounding clean. This study examines the perception of
individual solid-waste generators about plastic recycling and their willingness to pay
(WTP) an extra fee in addition to monthly waste collection charges at their current resident places. A survey-based, contingent valuation, approach with an anchored payment
card technique was used to interview 487 individuals in Hochiminh city. The findings
show that the mean expected WTP is VND43,200 per year from ordered probit model. Marital status, education level, employment and income are significant socioeconomic determinants of consumers’ WTP. Other important behavioral factors are concerns towards
the current ambient environment quality and the threats to human health caused by plastic wastes and the benefit of plastic wastes recycling.
Keyword: willingness to pay, plastic waste recycling, contingent valuation method

1. Introduction
Dangers from solid wastes generation to the
global environment as well as human health have
been remarked and received a great concern of societies recently. Plastic is one of the central concerns because most plastics are non-degradable
and leads to a growing concern about space at sanitary landfill sites. According to Vietnam Environment Situation 2004 – Solid Waste Report, solid
wastes mostly originated from households (60% –
70%) in urban areas. Statistics (2009) of HCMC
Department of Natural Resource and Environment
shows that the solid waste system collects approximately about 5,600 to 6,000 tonnes of wastes
daily, in which plastics account for the second
largest proportion, about 10.8% in households and
19% in schools (The Saigon Times) (1). A large
volume of plastic wastes is not collected for re-use
or recycling but goes directly to landfills and this
causes a great problem for the current ambient en-

52

vironment quality in the city.
Recycling is widely considered as a common
method to deal with such wastes. Recycling plastics can, in many cases, help reducing negative effects on the environment and keeping the

surrounding clean. Although several global environmental programs have been implemented to
improve the country’s environmental status in recent years, solid waste management, especially regarding plastic recycling, seems to have lower
priority than the other issues such as climate
change and water pollution. Despite the fact that
the country has its own Environmental Law and
also follows the global framework of environmental enhancement in some aspects, well-structured
policies and regulations regarding waste recycling
(i.e. plastic waste) have not been enacted. This
shows the need of further research of consumers’
or public perception on plastic recycling to support
policy makers.
Recycling would result in cost for the local gov* University of Economics - HCMC

Economic Development Review - June 2011


RESEARCHES & DISCUSSIONS

ernment and the program could succeed only with
the contribution from the public. This study aims
to examine the individual’s perception on plastic
recycling program at a more practical aspect. And
a hypothesized scenario is used to elicit the financial contribution of individuals or, in other words,
their WTP for plastic recycling. The policy context
is consumers’ preferences for an extra payment in
addition to monthly waste collection charges at
their current living places. Because HCMC is considered as the biggest commercial city and the
most populated in Vietnam, and since such environmental issue is more serious in urban area, the
research scope is limited to HCMC residents only.

2. Theoretical consideration
a. Contingent Valuation Method (CVM):
CVM is a survey-based method of eliciting
WTP for an improvement in environmental quality through direct questions. Owing to its advantage in measuring passive value of public goods
where market price does not exist, the CV method
is considered a capable measure for evaluating
non-marketed goods. The specific eliciting WTP
technique plays an important role in CV research
because each type of question format may bring
different results and associated biasness. The payment card method (PC) does not suffer a starting
point bias associated with iterative bidding and
dichotomous choice. This method presents respondents with a range of ordered threshold values
and requires them to pick a single amount they
are willing to pay. However, the PC technique can
still have some drawbacks associated with the
provision of bids, anchoring effects and the size of
intervals (Cameron and Huppert, 1989).
b. Econometric model:
For PC method, the monetary value of WTP
that respondents choose is treated as an ordinal
variable and analyzed with an ordered regression
model. The ordered probit model builds around a
latent regression in the same manner as the binominal probit model with attitudinal, behavioral
and demographic information as explanatory variables of WTP (Cameron and Huppert, 1989). We
assume a standard normal distribution with linearity in WTP as follows:
(1)
where
and
denotes the unobserved latent variable of willingness to pay for ob-

servation i which lies between cut-points tUi and
tLi in the distribution of . Let Y be the observed
ordinal variable, that is:
Y = j if
or
(j = 0,1,..J)
where both tUi and tLi are unknown parameters to be estimated with b (Greene, 2003).
Respondents have their own WTP intensity but
cannot express these given the limited number of
possible answers and will choose the answer that
most closely represents their own WTP intensity.
Then the probability of WTP that lies within
the interval is:
(2)
With the assumption that the error term is
normally distributed between zero and standard
deviation s, equation (2) can be re-written as
(Cameron and Huppert, 1989; Haab and McConnell, 2002):
(3)
where the function is the cumulative standard
normal density function and equation (3) is called
the ordered probit model (Greene, 2003). With
number of observation n, the log-likelihood function for the responses can be written as:
(4)
The parameters of coefficients b are estimated
using maximum likelihood estimation of the ordered probit model. Then, the interval bounds i.e.
tL and tU are derived (these are the “cuts” values
in STATA).
Since both parameters b and standard deviation s need to be estimated, the log-likelihood
function (4) will not result in a unique solution to

fit the data well (Jackman, 2000). Because the intercept term is dropped from the maximum likelihood estimation, it is necessary to assume the
constant, and standard deviation s.
The coefficients derived from an ordered probit
regression have the form b/s and constant 1/s
and the variance of error term is fixed at 1 from
ordered probit model in most of the standard computer programs (Winship and Mare, 1984). Thus,
it is required to recalculate the original b by
rescaling the estimated coefficients with standard
deviation s. This process is called re-calibrating
the b terms once we set the thresholds to cut

Economic Development Review - June 2011

53


RESEARCHES & DISCUSSIONS

points in monetary amounts. Hence, it is possible
to interpret the effects of explanatory variables in
dollar metric, rather than in probit metric (Jackman, 2000). We obtain a rescaling constant (or
standard deviation s) from a linear transformation as:
z* = mz + c
(5)
where z is a location estimate from the probit
model, z* is the midpoints of thresholds in dollar
amount given as bid levels in the payment card
determined by (tL + tU)/2 , m is the re-scaling constant (called the standard deviation s), and c is
the location shift. It is also assumed that the location shift c is the intercept term which was
dropped from the ordered probit model (Jackman,

2000).
The re-calibrating process results in a new set
of b, denoted as brescaled. Given a set of brescaled, the
expected willingness to pay (EWTP) is derived by
reconstructing the original form of WTP from
equation (1).
EWTPoprobit = Xibrescaled
(6)
Then, the mean of expected willingness to pay
from the population are estimated by:
(7)
The marginal effects of changes in the regressors of the ordered probit model can be evaluated
at sample means or at other relevant values of the
regressors. The marginal effect is calculated as:
(8)
where k denotes a single explanatory variable
and change in probabilities for the WTP categories must sum to zero (Cranfield and Magnusson, 2003).

3. Methodology
a. Value to be measured: consumers’ willingness to pay for plastic recycling:
Depending on available collection scheme, consumers will have different choices. In general,
there are three main schemes of plastics waste
collection in Vietnam: vehicles used for collection
and residents paying for a monthly waste collection charge; the plastics waste re-purchase scheme
in which household would receive an amount for a
quantity of solid waste from a collector; and the
environmental promotion programs of some institutions going ‘green’. The basic difference in the

54

Economic Development Review - June 2011

three schemes above is the consumer utility optimization problem. With the first scheme where
consumers are asked for their WTP, they will try
to minimize their pay-out consistently with their
utility, so the bid level would be distributed between zero and a relative low upper value. In contrast, the consumer may seek to maximize the
amount to be paid by the collectors then the willingness to accept (WTA) would be relative high in
the second scheme or they would even have zero
WTP or WTA in the third scheme. This study is
focused to the first scheme only as people will be
asked for their WTP. The proposed payment vehicle would be an addition to the monthly waste
collection charge.
b. Determinants of consumers’ willingness
to pay for recycling:
The empirical model regresses consumers’ willingness to pay for recycling on a number of socioeconomic factors and behavioral explanatory
variables.
- Socioeconomic variables:
Socioeconomic determinants are factors that
reflect demographic characteristics of a consumer
such as sex, age, marital status, income, employment, education level, and household size. These
factors are widely used in most CV studies. Except household size and age which are continuous
values, all the socioeconomic variables are defined
as either dummies (i.e. sex, marital status, and
employment) or as categories (i.e. income, and educational level). Income is categorized into six segments rather than as continuous numbers (Haab
and McConnell, 2002). For WTP measurement,
both income and education are employed in the regression model with their ordinal values.
- Attitudinal variables:
Behavioral factors included in the model are
based on consumers’ moral norm in plastic recycling and represent consumers’ perception regarding environmental problem (Hage et al. 2009). In
this study, we hypothesize that behavioral factors

might affect WTP for recycling, which are: (i) Perception of the costs/threats of plastic wastes; (ii)
Perception of the benefit of plastic recycling; (iii)
Perception of the needs for recycling; and (iv) The
habits of dealing with waste. There were 10 factors measured in terms of a Likert score value
ranging from 1 to 5 in the questionnaire. Regarding those answers regarding the consumers’ habit


RESEARCHES & DISCUSSIONS

in dealing with plastic wastes which have 4 values
ranging from 1-4, Likert scores 3 and 4 are converted to “Good habit” and scores 1 and 2 are converted to “Bad habit”. The remaining answers
regarding the consumers’ consideration towards
the agreement of recycling benefits and towards
the need, the costs of plastic wastes were converted to “High level of agreement” or “High concern” for scores 4 and 5 and “Neutral to low level
of agreement” or “Neutral to low concern” for
scores 1, 2 and 3, respectively.
c. Survey administration and data collection:
A survey of residents currently living in HCMC
in the 18-60 age bracket and having ability to access internet and respond to the online questionnaire was conducted in August 2010 via web-mail.
A total of 48cores 4 and 5)
0 = Low level of agreement (Likert scores 1,2, and 3)

4. Empirical results
It is remarkable that there are only 36 no-responses which account for the smallest percentage

56

Economic Development Review - June 2011

in total responses. This means that most people

are willing to pay an additional charge. Meanwhile, 30.53% of the respondents voted for the
highest bid level (VND72,000 per year) and thus


RESEARCHES & DISCUSSIONS

the distribution of stated values is skewed towards
the highest bid.
a. Interpretation of ordered probit regression estimates:
The parameter estimates are presented in
Table 3 for both the full (unrestricted) model with
all independent variables and the final (restricted)
model with only significant explanatory factors.
The results from the full model reveal that there
are seven significant variables: four socioeconomic
factors (MARRIED, EDU, EMPLOY, and INC)
and three behavioral factors (ENVQUAL, ENVTHREAT and BEN3). The Likelihood Ratio (LR)
Chi square goodness of fit statistics of the two
models are 50.82 and 43.81, respectively, and significant at 1% level. These test results indicate
that the H0 hypothesis of all estimated parameters equal to zero is rejected, and that the model
specification is appropriate and has a power to explain for the variation of WTP choice.

b. Mean of expected willingness to pay:
To eliciting the WTP in monetary values, it is
necessary to rescale coefficients as indicated in
the previous section. The re-calibrating process is
described in Table 4 (see next page).
Table 4 shows that the coefficient m has a positive sign and is significant at 1% level and the
intercept term derived from this regression is also
significant at 5%. The model has a well fit at Rsquared = 0.96. Then, the expected WTP and

mean WTP for plastics recycling from the final
WTP model are calculated by using equations (5)
and (6) as follows:

= 43.193 or VND 43,190 per year
Since there are no previous studies on WTP for
plastics recycling in HCMC, it is impossible to
compare the empirical findings with others and

Table 3: Ordered probit model estimates for parameters explaining WTP
N = 452
Variable
SEX

WTP Full model
Coefficient

WTP Final model
P>|z|

0.075 (0.108)

0.489

MARRIED

-0.202 (0.127)

0.111*

AGE

-0.010 (0.011)

0.334

EDU

0.230 (0.106)

EMPLOY

Coefficient

P>|z|

-0.266 (0.112)

0.017**

0.030**

0.225 (0.103)

0.028**

-0.383 0.181)

0.034**

-0.415 (0.176)

0.019**

INC

0.199 (0.045)

0.000***

0.173 (0.043)

0.000***

MEM

0.012 (0.021)

0.570

INFO

0.126 (0.106)

0.236

-0.404 (0.156)

0.010***

-0.365 (0.154)

0.018**

ENVTHREAT

0.505 (0.242)

0.037**

0.558 (0.215)

0.010***

HEALTH

0.008 (0.146)

0.959

NEED

0.285 (0.284)

0.314

ACT1

-0.016 (0.152)

0.918

ACT2

0.012 (0.112)

0.915

ACT3

0.039 (0.161)

0.808

BEN1

-0.342 (0.269)

0.203

BEN2

-0.289 (0.365)

0.430

BEN3

0.682 (0.292)

0.019**

0.423 (0.227)

0.063*

50.82

LR chi-square (7)

43.81

0.0001

Prob. > chi-square

0.000

ENVQUAL

LR chi-square (18)
Prob. > chi2

Note: figures in parentheses are standard errors of the estimates
*, **, *** denoted level of significance at 10%, 5% and 1%, respectively.

Economic Development Review - June 2011

57

RESEARCHES & DISCUSSIONS

Table 4: Defining standard deviations and intercept

/cut1 = -0.3968087

Midpoints of thresholds in monetary metric
z* = (tL + tU)/2
(0 + 12)/2 = 6

/cut2 = 0.3892714

(12 + 24)/2 = 18

z

/cut3 = 0.7156811

(24 + 36)/2 = 30

Const.

/cut4 = 0.9656855

(36 + 50)/2 = 43

Prob. > F

= 0.0035

/cut5 = 1.624124

(50 + 70)/2 = 60

R-squared

= 0.9596

Cut points from STATA
oprobit

discuss the performance of the analysis overall.
From 452 observations including true zero bids,
we obtain a mean willingness to pay an additional
charge in waste collection fee around VND43,200
per year or VND3,600 per month. This is a reasonable price in comparison with the average
monthly waste collection charge of VND13,000 obtained from the survey.
c. Marginal change of WTP:
In order to examine the change in the predicted probability of WTP by a marginal change
in one explanatory variable, others remain unchanged, the MEOPROBIT module in STATA was
used (Cornelissen, 2006).
It is seen that being married decreases the possibility of paying for high bids VND50,000 (1.6%)
and VND72,000 (8.9%). The marginal effect of
dummy MARRIED is significant at 5% for most

Linear regression
z* = mz + c (m = s)
Coef.

p-value

27.756

0.003***

13.093

0.024**

bids. Unemployed respondents have the negative
marginal effects on the last two WTP categories,
but positive effects on all the remaining bids.
However, EMPLOY is insignificant for the bid of
VND50,000. Marginal effects on WTP are also
stronger for EMPLOY than for the MARRIED
(Table 5).
For the two categorical variables INCOME and
EDU, the pattern is reverse to the socioeconomic
dummies. Higher education level has the highest
positive marginal effect on the highest bid
(VND72,000) by 7.8% and decreases the possibility of being willing to pay for the lower yea-saying
bid (VND12,000) by 4.1%. A marginal increase in
income will increase the probability of willingness
to pay the highest bid (VND72,000) by 6%. INCOME is the only variable having marginal effects significant at 1% for every bid.

Table 5: Marginal effects of ordered probit model
Y=0

Y = 24

Y = 36

Y = 50
0.239

Y = 72

0.080

0.177

0.108

0.091

Prob. At mean

0.069

0.174

0.113

0.097

0.253

0.295

MARRIED

0.038

0.048

0.015

0.005

-0.016

-0.089

p-value

0.030

0.019

0.019

0.030

0.061

0.014

EDU

0.305

-0.030

-0.041

-0.013

-0.005

0.012

0.078

p-value

0.032

0.032

0.040

0.061

0.061

0.029

EMPLOY

0.044

0.071

0.028

0.015

-0.005

-0.153

p-value

0.005

0.013

0.040

0.089

0.435

0.024

-0.023

-0.031

-0.010

-0.004

0.009

0.060

p-value

0.000

0.000

0.001

0.009

0.006

0.000

ENVQUAL

0.040

0.063

0.025

0.013

-0.007

-0.134

p-value

0.005

0.014

0.039

0.084

0.211

0.023

-0.104

-0.095

-0.019

0.000

0.055

0.163

0.050

0.004

0.000

1.000

0.067

0.001

-0.073

-0.074

-0.017

-0.002

0.038

0.129

0.139

0.049

0.003

0.555

0.180

0.031

INC

ENVTHREAT
p-value
BEN3
p-value

58

Y = 12

Sample frequency

Economic Development Review - June 2011


RESEARCHES & DISCUSSIONS

Marginal effects of ‘High concern about the environmental quality’ ENVQUAL dummy indicate
that consumers are more likely to pay lower prices
(1.3% to 6.3%) and less likely to pay the highest
price (13.4%), if they are ‘seriously’ or ‘somewhat

seriously’ concerned on the current ambient environmental quality.
Relative to those who are ‘seriously’ or ‘somewhat seriously’ concerned about the threats from
plastic wastes to the environment, the marginal
effects of ENVTHREAT are positive on the two
highest bids (5.5% to 16.3%). Similarly, ‘High
level of agreement’ in the benefit of plastics recycling, i.e. ‘help keeping surrounding clean,’ BEN3
dummy variable, has positive effect on the highest
bid (12.9%). This implies that consumers who
‘strongly agree’ or ‘somewhat agree’ with this benefit are more likely willing to pay VND72,000
(12.9%). In contrast, those who with ‘neutral to
low level of agreement’ are willing to pay lower
bids, i.e. VND12,000 (7.4%) and VND24,000
(1.7%).
Among the three behavioral factors, i.e. ENVQUAL, ENVTHREAT, and BEN3, all other
things being equal, the ENVTHREAT dummy
tends to have strongest marginal effects over WTP
categories than the other two variables. The result
suggests that consumers with higher concerns
about the threats from plastics waste are willing
to pay more than those who are concerned about
the environment quality and the surrounding
cleanliness as a benefit from plastic wastes recycling.

makers, the functional environmental agencies
and the plastics producers. Thus, there should be
parallel action programs from all entities so that
the environmental quality enhancement plan
would be implemented simultaneously.
According to the individuals’ opinions, the best
solutions to handling of plastic wastes were a

higher monthly solid waste collection fee and a
surcharge/environmental tax imposed on products
containing plastics as perceived by 46% and 35%
of the respondents, respectively. However, in
order to implement these two solutions, more consideration and measurements should be carried
out so that the charges would be well affordable
for both consumers and producers. Besides, stopping supply of free plastic bags in supermarkets
and improving the deposit-refund system on plastic items were also possible solutions stated by
32% and 24% of the respondents, respectively.
These two later solutions have a same characteristic in which consumers do not have financial responsibility, but environmental awareness
instead.
Hygiene quality and safety standards of recycled products are highly concerned. Because the
products are indeed made from disposals which

5. Policy implication and recommendation
Recycling plastics would result in costs to the
waste management agencies and recyclers, yet enhance an eco-friendly environment. The extra
amount paid by consumers aims to cover the cost
of recycling plastics. As presented above, the average expected WTP of consumers is of
VND43,200 per year. Thus, the policy makers
should consider whether the charge policy applies
on household-based unit or on adult individuals in
the coming development plans. Moreover, recycling is not only the responsibility of the consumers but also of the Government – policy

Economic Development Review - June 2011

59


RESEARCHES & DISCUSSIONS

could have been mixed with other wastes, such a
low quality treatment process could even endanger
human health. Thus, in order to implement any
plan, one should introduce guarantees in quality
standards to ensure that the recycled products
would not have negative effects on human health.
A well-managed process for the collected funds
should be considered seriously and efficiently by
policy makers. This matter in fact was mentioned
in most of respondents’ opinions collected from the
survey. Respondents indicate that a more transparent and efficient program management is
strongly necessary. It is also the main reason of
most protest zero responses.

of solutions, especially awareness inspiration and
education, before taking into account financial responsibilities such as surcharges, taxes or increment in wastes collection charges. The same
characteristic found in solutions of stopping the
supply of free plastic bags and improving the deposit-refund system on plastic items is that consumers do not have financial responsibility, but
environmental awareness instead, while policies
that emphasize the role of Government and functional agencies are highly voted by the respondents and worth implementation. Encouraging
re-using plastics and finding alternatives to plastics products are possible solution as well.

Last but not least, because a large proportion
of consumers are still new to and not very familiar
with environment protection programs, especially
those asking for their financial responsibility, it
is necessary to ensure a wide propagation about
such programs and their necessity via a range of
possible approaches. Regarding the surveyed respondents’ preferences, telecommunication/ radio

programs and conferences, and TV news programs
are mostly preferred for information channels.
The next important channels are websites, while
environmental programs run by institutions, functional agencies, and local authorities also play an
important role in which people will have more
chances to shift from perception to action.

6. Conclusion

Possible solutions to plastic wastes include: imposing some surcharge/tax on plastic items, stopping the supply of free plastic bags in
supermarkets, improving the deposit-refund system on plastic items (e.g. plastic bottles), and increasing the monthly solid waste collection fee.
However, whether consumers have to pay an increment in monthly waste collection charge or
they have to pay some surcharge or tax on plastic
items together with manufacturers, these policies
take into account the financial obligation from the
users’ side, yet do not reflect the attitudinal responsibility of consumers and role of Government.
Thus, in order to implement these two solutions,
more consideration and measurements should be
carried out so that the charges would be well affordable for both consumers and producers. Also,
it is necessary to apply simultaneously a variety

60

Economic Development Review - June 2011

The study presents a CV approach with an anchored payment card technique to measure consumers’ WTP for plastics recycling in HCMC,
Vietnam. The first significant finding in this study
is that most consumers (90%) are highly concerned about the current ambient environment
quality and the threats caused by plastics wastes
to the environment. The results from the ordered

probit regression show that the mean expected
willingness to pay an additional charge for plastic
recycling is VND43,200 per year. Secondly, the results show that behavioral factors have more influences on the consumers’ WTP. A marginal
increase in the consumers’ perception towards the
threats from plastic wastes to the environment
has the strongest effect on the probability of WTP
in comparison to the other two behavioral variables, which are the concerns about the ambient
environment quality. Notably, income plays an important role in determining consumers’ WTP.
Higher income and higher educated consumers are
likely willing to pay higher bids. Marital status
and employment are also significant factors but
have opposite signs for the marginal effects on the
predicted probability of WTP. Moreover, a number
of possible solutions to plastic wastes problem
were also investigated via voting of the respondents in which two solutions suggest financial responsibilities and two others take into account
consumers’ awareness of disposing wastes and
habit of using plastics. Media and telecommunica-


RESEARCHES & DISCUSSIONS

tion are the most potential channels to propagate
and disseminate information among plastics endusers regarding threats from plastic wastes, the
need for recycling and any available plans/policies
relating to the problemn
Notes:
(1) The Saigon Times Online, Accessible on Oct. 7,
2010; Available at:
/>
dered Probit, available at:

/>4. Cranfield, J. & E. Magnusson (2003) “Canadian
Consumer’s Willingness-To-Pay for Pesticide-Free Food
Products: An Ordered Probit Analysis”, International Food
and Agribusiness Management Association (IAMA).
5. Greene, W. (2003), Econometric Analysis, 5th Edition, Prentice Hall.
6. Hage, O., P. Soderholm & C. Berglund (2009)
“Norms and Economic Motivation in Household Recy-

(2) True zero responses reflect the valueless of

cling: Empirical Evidence from Sweden”, Resources,

amenity, where as protest zero responses are placed

Conservation and Recycling, Vol. 53, Issue 3, Pp 155-

when respondents provide nay-saying due to some as-

165.

pects of the scheme though they find the positive value

7. Hanemann, W. (1994) “Valuing the Environment

of the amenity. The reason of a respondent placing

through Contingent Valuation”, Journal of Economic Per-

protest vote may be because he/she does not fully trust

spectives, Vol. 8, No.4, Pp 19-43.

the proposed service, or he/she may think that the project
is unreliable (Fonta et al., 2010).
References
1. Boyle, K. J. & R.C. Bishop (1988) “Welfare Meas-

8. Jackman, S. (2000) “Models for Ordered Outcomes”,

Political

Science

200C,

available

at:

/>9. Winship, C. & R. Mare (1984) “Regression Models

urements Using Contingent Valuation: A Comparison of

with Ordinal Variables”, American Sociological Review,

Techniques”, American Journals of Agricultural Econom-

Vol. 49, pp 512-525.

ics, Vol. 70, No. 1, pp 20-28.

2. Cameron, T. & D. Huppert (1989) “OLS versus ML
Estimation of Non-market Resource Value with Payment

10. Wooldridge, J.M. (2009), Introductory Econometric: A Modern Approach, 4th Edition, South Western College.

Card Interval Data”, Journal of Environmental Economics
and Management, Vol.17, Issue 3, pp 230-246.
3. Cornelissen, T. (2006), MEOPROBIT, Stata Module to Compute Marginal Effects after Estimation of Or-

Economic Development Review - June 2011

61