China Policy Journal Volume 1, Number 1, Fall 2018 | Page 76

Payment for Ecological Services and River Transboundary Pollution
Table 2. Statistic Descriptives of the Respondents
Variables Definition Mean S.D.
Age Years 34.09 10.94
Years of education Years 14.56 3.13
Income level Income (1,000 Yuan/month) 4.44 4.1
Income higher than
need
Male
Respondents’ income can meet
the needs of their daily life?
(1=yes,0=no) 0.25 0.43
Dummy for male
(0=female,1=male) 0.51 0.5
The database illustrated here is
the sub-sample (about the half of the
total sample) of the survey using the
Multiple-Bound Discret Choice format
WTP question. The other part of the
data that we did not use in this paper is
based on a dichotomous choice format
WTP question.
In Table 3 we reproduced the part
of the estimation results of He, Huang,
and Xu (2015a) based on the subsample
using the MBDC (Multiple Bound Discrete
Choice, Wang and He 2011; Welsh
and Poe 1998) format WTP questions. 2
The last two estimation functions (hypotheses
2 and 3) illustrated in Table
1 used both individual- and city-level
characteristics to explain the monthly
WTP. The key variables of our interest
are those included in the section called
water quality-related variables, where
both the water quality of the river flowing
through the city of a respondent’s
residence (degree) and the water quality
of the section of river flowing through
the direct upstream cities (degree_up-
per) were included in the explanation
of monthly WTP. As can be seen from
the estimation called “hypothesis 2”, the
monthly WTP of a respondent was positively
and significantly affected by the
water quality of the river crossing his/
her resident cities but was negatively
related to the water quality of the direct
upstream city. The LR test reported
at the bottom of Table 1 compared
the model, including the variables of
the water quality of the direct upstream
city, with that excluding such variables.
Including the water quality of the upstream
cities significantly increased the
explanative power of the estimation
models and thus supported the relevance
of including those variables.
However, the results associated
with the variable degree_upper in the
model called “hypothesis 2” do not exactly
correspond to the new framework
that we proposed. This is because the
upstream city’s water quality was used
as the determinant of WTP, not the exact
information of the water quality at
2 The other half split questionnaires used the double-bound dichotomous choice (DBDC) elicitation
strategies, whose results illustrated obvious bias related to the starting price anchor effect.
73