See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323498315

Can climate mitigation help the poor? Measuring impacts of the CDM in rural
China

Preprint · February 2018

DOI: 10.13140/RG.2.2.31052.10888

CITATIONS READS

0 293

2 authors, including:

Kenji Takeuchi
Kyoto University
65 PUBLICATIONS 871 CITATIONS

SEE PROFILE

All content following this page was uploaded by Kenji Takeuchi on 01 March 2018.

The user has requested enhancement of the downloaded file.

 Can climate mitigation help the poor?
Measuring impacts of the CDM in rural China

Yimeng Du
Kenji Takeuchi

February 2018
Discussion Paper No.1808

GRADUATE SCHOOL OF ECONOMICS

KOBE UNIVERSITY

ROKKO, KOBE, JAPAN

 Can climate mitigation help the poor?
Measuring impacts of the CDM in rural China
Yimeng Du and Kenji Takeuchi∗
Graduate School of Economics, Kobe University
Kobe 657-8501 Japan

February 21, 2018

Abstract

This study aims to examine whether investment in climate change mitigation plays
a role in poverty alleviation. We investigate impacts of the renewable energy-based
clean development mechanism (RE-CDM) on rural communities in China. The im-
pacts of RE-CDM projects are estimated by combining propensity score matching with
the difference-in-differences approach. We found that the promotion of biomass-based
CDM projects significantly contribute to income improvement, employment generation,
and industrial transformation in rural communities in China. On the other hand, our
estimation results reveal that large-scale wind and solar energy-based CDM projects
have the potential to increase the labor force in the primary industry in rural areas.

Keywords: CDM; renewable energy; poverty alleviation; rural development; propensity

score matching; difference-in-differences

∗
E-mail: sissitoh@gmail.com (Du), takeuchi@econ.kobe-u.ac.jp (Takeuchi). Earlier versions of this
paper were presented at the Seventh Congress of the East Asian Association of Environmental and Re-
source Economics, 2017 annual meeting of the Society for Environmental Economics and Policy Studies, and
the Environmental Economics Seminar at Kyoto University. Authors acknowledge helpful comments from
Yushi Kunugi, Daisuke Ichinose, Makoto Hasegawa, Emiko Inoue and seminar participants. This work was
supported by JSPS KAKENHI Grant Numbers JP16H03006, JP26241033.
1 Introduction

More than 5.7% of Chinese population live below the poverty line as of 2015, mainly

in remote rural areas with limited energy access and job opportunities (Asian Development
Bank, 2017).1 As one of the policy packages to alleviate poverty in rural area, the Chinese

Government has adopted programs that promote renewable energy in remote areas, including
the Solar Energy for Poverty Alleviation Programme (SEPAP)2 and the 13th Five-year Plan
(FYP) for Rural Bioenergy Development.3 Implementation of the SEPAP, which commenced
in 2014, reveals the government’s aim to alleviate rural poverty through deploying distributed
solar photovoltaic (PV) systems in poor areas. On the other hand, one of ultimate targets of
the 13th FYP for Rural Bioenergy Development is to increase the income of rural residents

and improve living conditions of rural households by promoting utilization of agricultural

wastes. In addition, the Announcement on Accelerating Construction of Energy Projects in

Impoverished Areas for Promoting Poverty Alleviation, published by the National Energy

Administration (NEA) in 2017, emphasizes the contribution of renewable energy to poverty

reduction as well.4

In reality, can renewable energy play a key role in reducing the rural poverty? To explore

the answer to this question, we investigate the past Chinese experiences with clean devel-

opment mechanism (CDM) projects and examine their impacts on poverty reduction. The
CDM, as a part of the flexible mechanisms defined in the Kyoto Protocol, has opened a host
of possibilities to absorb foreign investment and enhance sustainable development (SD) in

1
The official national rural poverty line of China is 2,300 yuan per year at constant 2011 purchasing
power parity.
2
The National Energy Administration and Poverty Alleviation Office of the State Council decided to im-
plement a poverty alleviation program through installation of solar PV panels in poor households to increase
their incomes. The SEPAP is scheduled to run for a period of six years, starting from 2014. Information
on the SEPAP is available at <http://www.nea.gov.cn/2014-10/17/c 133723326.htm>, last viewed 17 July
2017.
3
Released by the National Development and Reform Commission (NDRC) on 25 January 2017. More
information is available at <http://www.gov.cn/xinwen/2017-02/16/content 5168559.htm#1>, last viewed
16 January 2018.
4
Informations on the announcement is available at <http://zfxxgk.nea.gov.cn/auto82/201711/t2017 1108
3046.htm>, last viewed 16 January 2018.

1
developing countries. According to the definition by the United Nations Framework Conven-
tion on Climate Change (UNFCCC), the SD co-benefits of CDM projects can be divided into
three categories: social benefits, economic benefits, and environmental benefits. Examples of

these benefits include poverty alleviation, employment generation and enhanced education
services (social benefits); new industrial activities, productivity growth, and technology in-
novation (economic benefits); and improvement of air, water and land quality (environmental
benefits).5
There are many studies that examine how far the CDM will achieve its SD goals. Studies
with positive findings suggest that the CDM could contributes to SD in host countries in

different ways. Olsen and Fenhann (2008) conclude, through a text analysis of 744 project
design documents, that small-scale renewable energy projects have comparatively higher

social benefits than large-scale projects. Wang et al. (2013) evaluate the employment impacts

through an input-output approach. Their results show that solar projects have the greatest

potential for indirect job creation, while hydro projects induce job losses. Weitzel et al.
(2015) indicate that larger CDM projects and more advanced technologies are more likely

to involve technology transfer.

However, several researchers provide contrasting results. Sirohi (2007) indicates that the

socio-economic development potential of CDM projects in India is ambiguous, and suggests

that for CDM to emerge as a “win-win” poverty alleviation strategy, its projects should
be implemented at the rural community level. By assessing 16 officially registered CDM

projects, Sutter and Parreno (2007) conclude that fewer than 1% of the CDM projects
are likely to contribute significantly to SD in the host country. Zhang and Wang (2011)
use an econometric approach to estimate the CDM effect on reducing local air pollution
and conclude that the CDM does not have a statistically significant effect in lowering SO2
emissions.

5
The SD tool provided by the UNFCCC enables the project owners to show the value behind the certified
emission reductions their CDM projects offer, by describing the SD co-benefits of projects. Available at:
http://cdmcobenefits.unfccc.int/Pages/SD-Tool.aspx.

2
 Previous studies show inconclusive results on whether or not CDM activities actually
contribute to the SD in host countries. Thus, the primary concern of our study is to evaluate
the SD benefits of the CDM on rural communities of the host country. As for the SD

benefits, our main focus is on social benefits, which include income generation, creation
of job opportunities, and changes in the industrial structure. Poverty alleviation through
income and employment generation is considered as one of the most important indicators
in CDM project evaluation. Moreover, the eradication of poverty is also regarded as an
indispensable requirement for SD (United Nations, 2012).
The contributions of this study can be summarized as follows. Most of above-mentioned

studies on the local impacts of the CDM adopt descriptive or the input-output analysis,
and are not based on rigorous econometric approach. In order to fill this research gap, we

use a fixed effect difference-in-differences (DID) model to investigate the social benefits of

the RE-CDM projects at the rural community level. In addition, this research applies the

propensity score matching (PSM) in conjunction with the DID model to adequately deal
with several issues of the simple DID approach, such as selection bias and omitted variable

bias. Besides, we check the robustness of our estimation results, obtained through the PSM-

DID approach by adopting the Mahalanobis distance matching (MDM) method. Finally,

our findings provide policy implications on the possibility of simultaneously achieving the

goal of climate change mitigation and poverty alleviation. It is of critical importance that
countries achieve their targets of poverty reduction under the Sustainable Development Goals

(SDGs)6 , while meeting their commitments of greenhouse gas emission reductions under the
Paris Agreement.7 With this respect, our study relates to the literature on poverty and the
environment (Sims, 2010; Sims and Alix-Garcia, 2017), but differs from these studies in that
we examine the effect of projects that require substantial investment and technology.
6
On 1 January 2016, the United Nations SDGs officially came into force. The first of the seventeen
proposed SDGs is “End poverty in all its forms everywhere.” More information on the SDGs is available at:
<http://www.un.org/sustainabledevelopment/>, last viewed 17 January 2018.
7
The Paris Agreement on climate change came into force in 2016 to limit the rise in global temperatures.
More information on the agreement is available at: <http://unfccc.int/paris agreement/items/9485.php>,
last viewed 17 January 2018.

3
 The main result of this study is that the RE-CDM contributes significantly to rural devel-
opment in China. Our findings suggest that biomass-based CDM projects can bring about
income growth, job creation, and industrial transition in rural communities in China. For

example, annual income of rural residents can be increased approximately 15.5% by adopt-
ing the biomass CDM projects. In addition, we find that large-scale wind and solar energy
projects can help to increase the labor force in the primary industry in rural communities.
These findings imply that investment in climate change mitigation can play a simultaneous
role in poverty alleviation.
This paper is organized as follows: Section 2 provides the current status of income in-

equality and promotion of the renewable energy in China. In Section 3 we introduce the data
for estimation and the measures of social benefits. Section 4 follows with an analysis frame-

work, including a description of the empirical model and matching techniques. Estimation

results and discussions are provided in Section 5. Finally, Section 6 presents our conclusions

and discuss the policy implication of this research.

2 Background

2.1 Income inequality in China

Beginning in 1978, China’s economic reform has led not only to rapid economic growth

but also to serious income inequality. Figure 1 shows the income trends of rural and urban
residents in China from 1985 to 2015.8 The rural population of China comprised 618 million
in 2014, accounting for about 45.2% of the total populations in 2014 (NBSC, 2014). At
the end of 2015, the net income of urban residents was nearly 3.5 times as much as that
of rural residents. The Gini coefficient, a measure of income inequality, has soared to 0.47

8
Individuals are categorized as either “rural” or “urban” residents by the hukou system, a household
registration system that serves as a domestic passport regime in China. Residents are required to stay
and work within their designated geographic areas. Individuals living in rural areas depend on agriculture
to make a living and are commonly known as rural residents. On the contrary, urban residents usually
dependent on non-agricultural sources of income.

4
from 0.25 in the middle 1980s (China Digital Times, 2013). Xie and Zhou (2014) argue
that China’s current high income inequality is significantly driven by the rural-urban divide
and the regional variation in economic well-being. Differences in economic structure play a

critical role in creating the overall income inequality between rural and urban residents.

[Figure 1]

Simultaneously, the income structure of the rural population has transformed over the
past two decades. As of 2015, the wage income9 has increased to around 43% of total
income of rural residents, while the proportion of rural residential income from the primary
sector has decreased to about 29%.10 The change reflects the fact that the source of income

of rural residents has been shifted from the primary sector to the secondary and tertiary

sectors. Rural areas tend to have a relatively smaller range of job opportunities, lower
payment, and thus higher unemployment. This has caused a large number of rural laborers

to move out from their registered places of residence and migrate to urban cities in search

of job opportunities. The total stock of rural migrant labor, estimated to be around 282

million as of 2016, constitutes more than one-third of the entire working population of China

(Walsh, 2017). The large rural-to-urban migration not only increases the burden on urban
cities but also creates many social problems in rural areas, such as the mental health and
education of the left-behind children, aging of the rural population, and decline in agricultural

productivity (China Labour Bulletin, 2016). In order to alleviate these issues of rural China,

policy makers focus on the way to improve the employment environment by providing high
quality and sustainable job opportunities to the rural community.

2.2 Rural poverty and renewable energy

Recently, the Chinese government has promoted investment in renewable energy in rural
areas. With the formulation of several national promotion policies for renewable energy,
9
The income earned by an individual working as an employee.
10
Authors’ own calculations. Data collected from the China Statistical Yearbook in 1996 and 2016.

5
such as the SEPAP and the 13th FYP for Rural Bioenergy Development, new energy in-
dustries are ready to exploit the wide development space in rural areas. Development of
the renewable energy industry is expected to attract both domestic and foreign investment,

as well as the working-age population, into rural areas. In addition, access to cleaner and
affordable energy options can improve the livelihood of rural households by raising their
living conditions and transforming the production structure of local firms. Moreover, re-
newable energy industries can focus on retraining the low-skill and low-income workers. For
instance, by the end of December 2014, a total of 16,542 rural residents in Qingxiu County11
had received vocational training related to renewable energy, and 15,308 of them obtained

national vocational qualifications through an examination system.12

By 2020, China’s renewable energy industries are expected to provide employment oppor-

tunities for nearly a million people, including research and development, design, production,

construction, operation, service, transportation, management, education, training, consult-

ing, and other related jobs (Worldwatch Institute, 2011).

At the same time, with the aggravation of the energy crisis and increasing importance

of environmental problems, climate policies have been high on the agenda of Chinese gov-

ernment for about a decade. The necessity and urgency of promoting the renewable energy

sector in China have been providing entry points for the RE-CDM. Moreover, because adop-

tion of RE-CDM projects could bring additional foreign investment to the host community,
and ultimately drive the development of local renewable energy industries, local governments

encourage local firms to develop renewable energy resources with CDM. Consequently, China
has become the world’s largest host country for CDM projects. Between 2005 and 2012, a
total of 2,983 CDM projects were formally registered in China. Among the registered CDM
projects, renewable energy projects make up the largest share, at about 82.7%. Of these,
40.6% is wind power project while other projects, including bioenergy and solar energy, make

11
A county belonging to Nanning city, Guangxi Zhuang Autonomous Region, China.
12
The Office of Rural Energy, Guangxi Province. <http://www.gxncny.cn/gxnycms/pxjn/3175.jhtml>,
last viewed 15 February 2018.

6
up about 5.2% and 1.6%, respectively.13
Rural counties14 attract a large part of investment related to RE-CDM deployment be-
cause they tend to be sparsely populated, amply endowed with renewable sources of energy,

and spacious enough for land-intensive developments like wind farms. As of 2012, a total
of 461 rural counties had adopted RE-CDM activities in China, which installed capacity
accounts for about 86.8% of total installed capacity of the RE-CDM. Figure 2 depicts the
locational distribution of RE-CDM projects by the cumulative installed capacity at the pre-
fecture level. The RE-CDM projects are not evenly distributed among regions, but mainly
concentrated in regions endowed with large renewable energy resources, the northern, north-

eastern, and northwestern regions.

[Figure 2]

3 Data

3.1 Measures of the social benefits

There are three dimensions that compose SD in the local community. The first is the

social dimension, which includes welfare indicators such as household income, employment,

and spending on health and education. The second is the economic dimension, which is
often related to consumption and investment in productive capital. The last one is the en-
vironmental dimension, including environmental quality, pollution emissions, and material

consumption (IRENA, 2016). Many previous studies focus on investigating both the eco-
nomic and environmental benefits of the CDM while its social benefits have received less
attention. In order to estimate the social impacts of increased renewable energy deployment

13
Authors’ own calculations. Data collected from UNFCCC’s Database for Project Activities and Pro-
gramme of Activities.
14
County-level administrative areas in China consist of the county and county-level city and municipal
districts, where the county is usually considered as the backward region in each prefecture. Considering that
the objective of this study is to evaluate the impact of the CDM on rural development, we only adopt those
CDM projects located in the county, also known as rural area in our analysis.

7
under the CDM, this study employs three indicators: income generation, job creation, and
the transformation of industrial structure.
First, we adopt the per capita net income of rural households to measure the impact

of the RE-CDM activities on rural income. Rural communities can diversify, stabilize, or
increase the income of their residents in several ways by adopting RE-CDM activities. For
instance, the income level of rural residents can be increased through subsides from local
governments. In 2009, the Chinese central government’s subsidy standard for rural household
biogas was improved to 1,500 yuan for each rural household in the northeastern and western
areas, 1,200 yuan in the central region, and 1,000 yuan in the eastern region (Qiu et al.,

2013). Besides, RE-CDM projects can reduce the poverty that characterizes rural regions by
helping unskilled laborers in rural areas, such as farmers, unemployed persons, and women

with low education level in rural area, to serve as assembly line workers, equipment installers,

and maintenance or sales staff.

Second, the number of rural laborers is used to capture the working population in a rural
county. The working population of rural communities increases with more job opportuni-

ties for rural residents. Development and promotion of the renewable energy industry is an

important way to increase employment of local residents. Worldwide, the renewable energy

sector provided about 6.5 million direct and indirect jobs in 2013. Fuel supply from bioen-

ergy feedstock, installations, and equipment manufacturing will generate most jobs in the
renewable energy value chain (IRENA, 2014). Some argue that the decentralized nature of

renewables deployment will raise the overall number of jobs. However, others hold that the
relatively higher monetary costs of deploying renewables will reduce purchasing power and
consequently employment. These arguments underscore the need for more country-specific
empirical analysis and reliable approaches to estimate the potential social benefits, especially
employment creation from renewable energy deployment.
Lastly, we employ the number of rural laborers in the primary sector to capture the impact
of RE-CDM on industrial transformation. Renewable energy related industries can create

8
valuable job opportunities for people in regions with low employment. It provides both direct
jobs, such as operating and maintaining equipment, and indirect jobs along the supply chain,
such as fuel supply, manufacturing, construction, and other related specialized services. For

example, if the presence of renewable energy installations can revive construction activities
related to renewable energy power plants, then the main income sources of farm households
could switch from agricultural activities to the construction industry. According to the
Construction Plan of the National Rural Biogas Project (2006-2010) released by the Ministry
of Agriculture of China in 2007, construction and maintenance of every 10,000 biogas pools
can absorb about 800 rural laborers; thus, the whole country can provide about 368,000 jobs

per year for the rural labor (MOA, 2007).

3.2 Data sources

To assess the effect of the RE-CDM on rural development, we collect information on

the construction period and location of RE-CDM projects, rural residential income, number

of laborers in rural area, and other characteristics of each county.15 The panel data used

for analysis cover a total of 1,939 rural counties across China and consist of three types
of variables, namely, social benefits, county characteristics, and characteristics of RE-CDM

projects. The sample period for this study is between 2005 and 2011.

Table 1 contains descriptive statistics on the variables used in our analysis. Counties
that adopted RE-CDM projects between 2005 and 2011 are included in the treatment group
in this study. On the other hand, counties with no RE-CDM activities during the research
period are included in the control group. The average rural household income is about 5,027
yuan in the treatment group, and approximately 5,486 yuan in the control group. The
mean number of rural laborers in a county is around 0.231 million in the treatment group
and about 0.217 million in the control group. The average number of rural laborers in the
primary sector is about 0.122 million in the control group; the corresponding number in the

15
A county is an administrative unit ranking below a prefecture and above a township.

9
treatment group is 0.011 million lower. A two-tailed t-test shows that there is a statistically
significant differences in the mean value of social benefits and county characteristics. It
suggests the need to adopt matching techniques in order to avoid selection bias.

[Table 1]

Data related to social benefits are collected from the China Statistical Yearbook for
Regional Economy. Per capita net income of rural households, number of rural laborers, and
number of rural laborers in the primary sector are used as indicators of social benefits.
The county characteristic variables, including share of gross output of the primary sector,
area of agricultural land, rural population, total government revenue, share of students
accepting compulsory education, production of oil crop, and total capacity of agricultural

machinery, are based on the China Rural Statistical Yearbook. Both geographical and social
characteristics are considered since these factors may affect the existing energy infrastructure

and influence the promotion of renewable energy industries.

CDM data are obtained from the UNFCCC’s Database for Project Activities and Pro-

gramme of Activities, which includes basic information on every registered project. Only

RE-CDM projects registered between 2005 and 2011 are considered in our study because of
the limitations of the county-level economic data. Hydroelectric projects are excluded from

the sample because of their potential to generate social benefits and social problems for rural
communities at the same time.16 The geographic location of each project is collected from

the CDM location map provided by the NDRC. ArcGIS 10.1 is used to generate the location
data of RE-CDM projects.

16
The construction of reservoirs can improve water supply, increase farmland irrigation, produce electric-
ity, and produce other social and economic benefits. However, it also has its disadvantages. For example,
the resettlement of residents will lead to changes in the economic structure.

10
4 Empirical analysis

4.1 Model

In order to measure social benefits of RE-CDM in rural communities, we employ a DID

estimator combined with a mix of fixed effects by running a LSDV model. The DID estimator
compares the change in social benefits connected to RE-CDM projects in counties that
adopted the project to the change in social benefits in counties that did not. The fixed effect
estimation allows us to control for time invariant and time varying unobservable county
characteristics that may be correlated with a county’s RE-CDM project adoption decision.
This study uses unbalanced panel data on the social benefit indicators for 1,939 rural

counties in China from 2005 to 2011. The general form of the model adopted can be written
as follows:

yit = β0 + β1 Dit + β2 Xit + δi + γt + εit ,

where yit indicates the social benefits variables, which includes: (a) rural residential income;

(b) the number of rural laborers; and (c) the number of rural laborers in the primary sector
in the county i in year t.

Dit is the treatment indicator that takes on the value one in and after the year the

CDM renewable energy power plants have been constructed in county i, and zero otherwise.

We also interact the treatment indicator with different scales and types of renewable en-
ergy sources to capture their differences in social benefits. There are two major classes of
renewable energy-based project scales defined by the CDM Executive Board17 : large utility-

scale (>15MW) projects that sell wholesale electricity to energy providers, and small-scale
(≤15MW) projects that often include biomass fuel switches, small wind generators for civil-
ian use, farm or rooftop solar implementation, and solar cooker projects (UNFCCC, 2014).
Xit is a set of time-varying county characteristics. δi is the vector of the county dummy
17
The CDM Executive Board supervises the Kyoto Protocol’s clean development mechanism under the
authority and guidance of the Conference of the Parties serving as the Meeting of the Parties to the Kyoto
Protocol.

11
variable, which is used to control for unobserved county characteristics that shape the level
of development across counties. Year dummy γt is included to control for trends that shape
rural development over time such as changes in policies and regulations at the national level.

εit is the error term.

4.2 Matching techniques

There is a concern that the DID estimator may suffer from two sources of bias. The first
may arise if the levels and trends in social benefit indicators in treatment and control counties
differ before the CDM project adoption. Another bias could arise if the CDM project sites are
not randomly assigned but determined by various geographical, political, and socio-economic

factors. Therefore, in this study, we adopt two matching approaches to mitigate potential
bias by pairing treatment counties with counties that have similar observed attributes from

the control pool.

We adopt the PSM approach developed by Rosenbaum and Rubin (1983). The objective

of the PSM is to construct a control group by finding controls that have observed x similar

to those of the treatment group. To match treatment and control units on the basis of x is
equivalent to matching them using a propensity score p(x), which gives the probability of

receiving treatment given the pretreatment value of x, that is, p(x) = P r(D = 1|x). The

matching method assumes that, in a set of subjects all with the same propensity score, the
observed outcome distribution will be the same between the treatment and control groups. To
check the robustness of the PSM, we also use the simple MDM, which was first discussed by
Cochran and Rubin (1973). For the MDM, the variance-covariance matrix of x is estimated
by the pooled within group sample covariance matrix S. The distance between covariate x1
and x2 is M (x1 ,x2 ) = (x1 -x2 )T S−1 (x1 -x2 ).
In the first step, in order to estimate the propensity score, we use covariates to identify the
probability of a county adopting a RE-CDM project, which include: gross regional product
of the primary sector, agricultural land area, amount of oil crop production, and a dummy

12
variable for regions that have relatively higher wind power potential.18 We assume each
covariate affects the county’s decision on RE-CDM adoption and the social benefit outcomes
of the treatment and control groups in the pre-treatment period.

In the second step, we use the estimated propensity score to match treatment and control
groups in the baseline year. Here, we use 2005 as the baseline year since most of the RE-CDM
projects were implemented after this year.19 In order to ensure that all the rural counties did
not have RE-CDM activities in the baseline year, counties that adopted RE-CDM activities in
2005 were dropped from the sample. A one-to-one matching approach without replacement
was adopted while using the nearest-neighbor PSM and MDM algorithm. It means that

we choose only one county from the counties without RE-CDM activities as a match for a
treatment county in terms of their closest propensity score and Mahalanobis distance. An

untreated county cannot be used more than once as a match. The observations decrease

from 11,537 to 2,078 after the PSM, and to 2,031 after the MDM, since the observations out

of the common support have been dropped from the sample.

In the final step, in order to ensure that the matching procedure successfully balances the

two groups, we compare the treatment and control groups after matching. We present the

balancing test results for the PSM in Table 2, Panel A and that of the MDM in Panel B. The

results illustrate that there are statistically significant differences between the mean values

of the estimated propensity scores of the treatment and control groups before matching. For
instance, in the first row of Table 2, Panel A, we find that the difference of primary industry

output between the treatment and control groups is nearly 29.0%. Whereas, the second
row shows that the difference between these two groups drops to 7.10% when the sample
is matched. In addition, results of the t-test indicate that matched groups do not have

18
High wind potential regions are those regions with on-grid tariffs for wind power less than or equal to
0.54 CNY per kWh including tax. According to The Notice on Tariff Price of On-shore Wind Power, on-grid
tariffs for wind power generators are 0.47–0.60 CNY per kWh, with the lower tariffs applying in regions with
higher wind power potential (NDRC, 2016).
19
The first CDM project in China was the Huitengxile wind farm project, which was successfully regis-
tered in 2005. <https://cdm.unfccc.int/Projects/DB/TUEV-SUED1113481234.64/view>, last viewed: 21
December 2017.

13
statistically significant differences in the mean value of covariates. These results illustrate
that no statistical difference emerges after matching the treatment and the control groups.

[Table 2]

The balancing test results are also shown in Figure 3, which depicts the differences in

the distribution of the propensity scores by treatment and control groups. The figure shows
that selected observations of the control groups have extremely similar kernel density of
propensity score with observations in the treatment groups. It suggests that differences in

the distribution of the two groups have been significantly reduced after the PSM is applied
as well.

[Figure 3]

5 Results and discussion

5.1 Impact on rural residential income

The estimation results of the RE-CDM’s effect on rural residential income are reported
in Tables 3 (PSM-DID) and 4 (MDM-DID). The results in Table 3 suggest that a positive

relationship exists between RE-CDM activities and rural residential income. The coefficient
of the treatment indicator re cdm is positive and statistically significant at the 10% level, as

shown in column 1. The estimated effects correspond to an increase of approximately 276

yuan in annual income, which is about 5.03% of the average rural income of residents.20

[Table 3]

[Table 4]

20
This calculation is based on the assumption that annual average income of rural residents is 5,486 yuan.

14
 Table 3 also reports the impact of the RE-CDM by different energy sources. The largebio
dummy is positive and significant at the 5% level, as shown in column 2. This result indicates
that the biomass based CDM projects stimulated income growth substantially for rural

residents. Specifically, the adoption of large-scale biomass-CDM projects generated 851

yuan, about a 15.5% increase in annual income for the rural residents. In addition, we find
a similar result when utilizing the MDM-DID approach. Our result regarding the impact of
the RE-CDM on income improvement illustrates that only the utility-scale biomass energy-
based CDM projects were significant in stimulating income generation. According to Faaij
et al. (1998), bioenergy-based electricity production tends to have a greater impact on local

income than power generation using coal because of the use of locally produced feedstocks.
Moreover, Gan and Smith (2007) estimated the co-benefits associated with the utilization of

logging residues for bioenergy production in East Texas, USA. The input-output modeling

revealed that the most noticeable socio-benefits of bioenergy production were income and

job creation.

5.2 Impact on employment generation

In Tables 5 and 6, we assess the impact of the RE-CDM projects on employment gen-

eration using the PSM- and MDM-DID method, respectively. The results indicate that the

employment generation impact of the RE-CDM activities in rural areas differ by different
renewable energy sources. The coefficients of largebio are positive and significant at the 5%
level, as shown in columns 2 in Tables 5 and 6, respectively. These results suggest that the
number of labor in a rural county can be increased by approximately 13,000 workers, 5.99%
of the average number of rural laborers, through the adoption of large-scale biomass-CDM
projects.21

[Table 5]

21
The calculation is based on the assumption that the average number of labor is 0.217 million workers
in a rural community.

15
 [Table 6]

In line with the arguments of Thornley et al. (2008) and Openshaw (2010), our results
illustrate that biomass energy based projects show remarkable contributions to employment
generation in rural communities. Thornley et al. (2008) quantified the expected employment
impacts of individual bioenergy development and suggested that the larger bioenergy power
plants had a larger employment impact, which confirms our results on the employment
creation impact of large-scale biomass projects. Additionally, Openshaw (2010) highlighted
the importance of bioenergy systems as a means to poverty alleviation. Openshaw found
that in Malawi, Africa, the equivalent to 93,500 and 133,000 full-time workers were employed

in the biomass supply chain in 1996 and 2008, respectively. In contrast, about 3,400 and
4,600 people were employed in the supply chain of other conventional fuels, such as coal and

petroleum, in those years.

5.3 Impact on industrial transformation

In Tables 7 and 8, we report the estimated impact of the RE-CDM adoption on industrial

transformation in rural communities. The coefficient of smallbio, shown in column 3 of Table

7, indicates that the implementation of small-scale bioenergy projects under the CDM is asso-

ciated negatively with 20,000 workers annually in primary industry, which is about a 16.39%
decrease in the number of rural laborers in the primary sector.22 The results estimated by the

MDM-DID approach in Table 8 confirm the robustness of our PSM-DID estimation. These

results on smallbio suggest that the transformation of the economic structure from primary
industry to other industries is likely to be achieved by introducing small-scale biomass based
CDM projects into the rural communities. The finding implies that the presence of individual
biomass-CDM projects may promote production activities beyond agricultural production
in those rural areas with large numbers of unskilled laborers. This is based on the thought

22
This calculation is based on the assumption that the average number in primary industry rural labor
force is approximately 0.122 million workers.

16
that small-scale projects are often community based, and therefore, it is easier for unskilled
rural laborers to be involved in the production process after skill training. According to the
report released by the Ministry of Agriculture of China, the industrial scale of bioenergy

production has been growing continually, with over 20,000 small-scale biogas projects and
more than 4,700 large and medium-sized biogas projects approved for production as of the
end of 2010. More than 300,000 farmers can be transferred to such local jobs each year in
the stages of biogas power plant construction and service providing alone (MOA, 2010).

[Table 7]

[Table 8]

Interestingly, we found that both large-scale wind and solar energy-based CDM projects

have the potential to induce the reintegration of migrant rural labor. The coefficients of

largewind, shown in column 4 and largesolar in column 5 of Table 7, indicate positive and
statistically significant impacts of large-scale wind- and solar-CDM power plants on the

number of rural labor in the primary industry sector at the 1% level. Our results illustrate

that large-scale wind projects attract approximately 4,000 rural laborers into the primary

sector each year, which is about 1.01% of the total population of a county.23 Moreover, due
to the adoption of large-scale solar energy projects, the number of rural work force in the

primary industry can be increased by 19,000 annually, which is about 4.80% of the total
rural population.24 The promotion of photovoltaic agriculture25 is said to have improved

the efficiency of agricultural production, encouraged small villages to form an agricultural

town, and promoted the return of migrant rural workers (China Energy Net, 2011). For
example, this effect is observed in the case of a 20 MW agriculture-photovoltaic power station
23
In this study, mean value of the total population of a rural county is about 0.396 million people in
control groups, and 0.430 million people in treatment groups.
24
The calculation is based on the assumption that the mean value of the rural population is 0.396 million
people.
25
Photovoltaic agriculture is the combination of photovoltaic power generation and agricultural activities.
There are several main application modes of photovoltaic agriculture such as a photovoltaic agricultural
greenhouse, photovoltaic breeding, and a new type rural solar power station, among others (Xue, 2017).

17
located in Zhengyang county.26 The expected power generation of the Zhengyang ecological
agricultural farm is 20 GWh per year, with the power generated by solar panels used for
agricultural production inside the farm and the rest of the generated electricity transmitted

to the national grid. More than 120 migrant workers who returned to the county are said to
be employed by this photovoltaic agriculture farm (Farmer Daily, 2017).

6 Conclusions

By focusing on the social benefits brought by the renewable energy projects, we examine
whether the RE-CDM improved Chinese rural communities in terms of rural residential

income, job opportunities, and transforming the industrial structure. In addition, our study
investigates the impact generated by various renewable energy sources in order to understand

which energy source provides higher social benefits.

Our results indicate that the bioenergy-based CDM projects significantly contribute to

local sustainability of the host counties. The increase in annual income of rural residents by

adopting bioenergy CDM projects was calculated at about 851 yuan per year. The growth
in rural residential income caused by biomass-CDM projects was likely due to increased job

opportunities and the transformation of the labor structure. As described in 2009 annual

report by the China Association of Rural Energy Industry, the biogas projects provide a large

number of employment opportunities for the rural surplus labor force and migrant workers,
significantly increasing the income of the farmers, and promoting the social stability in the
project areas (CAAE, 2010). Moreover, we find that not all renewable energy technologies

contribute to the social benefits in the same manner. Small-scale biomass-CDM projects
had the largest potential in improving the labor structure in rural areas since their impact
on transferring primary sector rural labor to other sectors was around 13,000 workers per
year. This result indicates that small-scale bioenergy provides more job opportunities for
unskilled laborers than other types of energy sources. In contrast, large-scale wind and solar
26
Zhengyang county is a rural county belonging to Zhumadian city, Henan province, China.

18
energy-based CDM projects promote rural development by encouraging agricultural reform
and attracting labor force into the primary sector.
Climate change represents a direct and immediate threat to poverty alleviation (World

Bank, 2015). In this study, we assess whether activities for climate change mitigation can
alleviate poverty of rural communities in China. We conclude that the adoption of renew-
able energy projects under the CDM can offer an effective way to both reducing poverty
and addressing the global externality. By promoting the development of renewable energy,
particularly bioenergy in local communities, it might be possible to reduce poverty in ways
that support low-carbon growth. Providing clean electricity and access to modern energy

services may also contribute to other type of social benefits, by improving health and welfare,
access to education and jobs, and drive economic growth while reducing pollution (Climate

Advisers, 2014).

Although our study confirms the role of the RE-CDM in assisting host countries in

achieving SD, a further investigation is necessary to understand links between climate change
mitigation and poverty reduction strategies. For example, it is important to compare the

social benefits of domestic renewable energy projects and RE-CDM projects in order to

evaluate the effectiveness of different investment channels. Another limitation of our study

is that the long-run effect of the RE-CDM has not been considered. Future research should

be designed to capture the dynamics of the relationship between the RE-CDM and rural
development in the long-run.

19
References

Asian Development Bank, 2017. Poverty in the People’s Republic of China. Available at

<https://www.adb.org/countries/prc/poverty>, accessed on 23 January 2018.

Climate Advisers, 2014. Reducing Poverty Through Climate Action. Available at

<http://www.climateadvisers.com/wp-content/uploads/2014/05/2014-05-Climate-
Poverty.pdf>, accessed on 19 July 2017.

Chinese Academy of Agricultural Engineering (CAAE), 2010. Rural Energy Keyword 2009.
Available at <http://www.caae.com.cn/ywbm/ncnyyhbyjs/nyhb hyxx/201403/t20140313 3
813878.html>, accessed on 18 December 2017.

Chinese Academy of Social Sciences (CASS), 2016. Annual Report on China’s Economic
Growth 2016.

China Digital Times, 2013. How to Fix China’s Income Inequality. Available at

<http://chinadigitaltimes.net/2013/02/how-to-fix-chinas-income-inequality/>, accessed on

17 July 2017.

China Energy Net, 2011. Agricultural development strategy of amorphous silicon. Available

at <http://www.china5e.com/energy/news-333270-1.html>, accessed on 18 December 2017.

China Labour Bulletin, 2016. Migrant workers and their children. Available at

<http://www.clb.org.hk/content/migrant-workers-and-their-children>, accessed on 6 De-

cember 2017.

Cochran, W. G., Rubin, D. B., 1973. Controlling bias in observational studies: A review,
Sankhyā: The Indian Journal of Statistics 35(4), 417–446.

Faaij, A., Meuleman, B.,Turkenburg, W., Wijk, A., Bauen, A., Rosillo-Calle, F., Hall, D.,
1998. Externalities of biomass based electricity production compared with power generation
from coal in the Netherlands, Biomass and Bioenergy 14(2), 125–147.

20
Farmer Daily, 2017. The photovoltaic farm benefits the farmers. Available at
<http://www.farmer.com.cn/jjpd/cyzf/201705/t20170510 1294935.htm>, accessed on 18
December 2017, in Chinese.

Gan, J., Smith, C.T., 2007. Co-benefits of utilizing logging residues for bioenergy production:

The case for East Texas, USA, Biomass and Bioenergy 31(9), 623–630.

IRENA, 2014. The Socio-economic Benefits of Solar and Wind Energy. Available at <http://
www.irena.org/DocumentDownloads/Publications/Socioeconomic benefits solar wind.pdf>,
accessed on 19 July 2017.

IRENA, 2016. Renewable Energy Benefits: Measuring the Economics. Avail-

able at <http://www.irena.org/DocumentDownloads/Publications/IRENA Measuring-the-

Economics 2016.pdf>, accessed on 17 July 2017.

Ministry of Agriculture of China (MOA), 2007. Construction Plan of National Rural Biogas

Project (2006-2010). Available at <http://www.ndrc.gov.cn/fzgggz/fzgh/ghwb/gjjgh/200709

/P020150630514209853009.pdf>, accessed on 21 December 2017, in Chinese.

Ministry of Agriculture of China (MOA), 2010. The development of biogas in rural areas

is leapfrogging. Available at <http://www.moa.gov.cn/ztzl/zyncgzhy/201012/t20101224 1

797317.htm>, accessed on 18 December 2017, in Chinese.

National Bureau of Statistics of China (NBSC), 2014. China Statistics Yearbook 2015. Avail-
able at <http://www.stats.gov.cn/tjsj/ndsj/2015/indexch.htm>, accessed on 17 July 2017.

National Development and Reform Commission (NDRC), 2016. The Notice on Tariff Price
of On-shore Wind Power. Available at <www.spicxj.com/FileDownload.do?affixid=10624>,
accessed on 24 July 2017, in Chinese.

21
National Energy Administration (NEA), 2015. Work program on implementation of
photovoltaic poverty alleviation project. Available at <http://www.nea.gov.cn/2016-
04/05/c 135250679.htm>, accessed on 17 July 2017.

Olsen, K.H., Fenhann, J., 2008. Sustainable development benefits of clean development mech-

anism projects: A new methodology for sustainability assessment based on text analysis of
the project design documents submitted for validation, Energy Policy 36(8), 2819–2830.

Openshaw, K., 2010. Biomass energy: Employment generation and its contribution to
poverty alleviation, Biomass and Bioenergy 34(3), 365–378.

Qiu, H., Cai, Y., Bai, J., Sun, D., 2013. Research on the implementation effect of rural
household biogas subsidy policy in China, Issues in Agricultural Economy 2, 85–92, in Chi-

nese.

Rosenbaum, P.R., Rubin, D.B., 1983. The central role of the propensity score in observational

studies for causal effects, Biometrika 70(1), 41–55.

Sims, K.R.E., Alix-Garcia, J.M., 2017. Parks versus PES: Evaluating direct and incentive-

based land conservation in Mexico, Journal of Environmental Economics and Management

86, 8–28.

Sims, K.R.E., 2010. Conservation and development: Evidence from Thai protected areas,
Journal of Environmental Economics and Management 60(2), 94–114.

Sirohi, S., 2007. CDM: Is it a ‘win-win’ strategy for rural poverty alleviation in India?
Climatic Change 84, 91–110.

Sutter, C., Parreno, J.C., 2007. Does the current Clean Development Mechanism (CDM)
deliver its sustainable development claim? An analysis of officially registered CDM projects,
Climatic Change 84, 75–90.

22
Thornley, P., Rogers, J., Huang, Y., 2008. Quantification of employment from biomass power
plants, Renewable and Sustainable Energy Reviews 33(8), 1922–1927.

United Nations, 2012. The Future We Want. RIO+20 United Nations Conference on Sus-
tainable Development. Available at <https://sustainabledevelopment.un.org/content/docu
ments/733FutureWeWant.pdf>, last viewed 17 July 2017.

UNFCCC, 2014. Small scale methodology: renewable electricity generation. Available at

<https://cdm.unfccc.int/methodologies/DB/W3TINZ7KKWCK7L8WTXFQQOFQQH4SB
K>, last viewed 19 July 2017.

Walsh, M., China’s Migrant Worker Poetry, The Asia-Pacific Journal: Japan Focus 15(12)

(2017) 4.

Wang, C., Zhang, W., Cai, W., Xie, X., 2013. Employment impacts of CDM projects in

China’s power sector, Energy Policy 59, 481–491.

World Bank, 2015. Climate Change Complicates Efforts to End Poverty, The World

Bank. Available at <http://www.worldbank.org/en/news/feature/2015/02/06/climate-

change-complicates-efforts-end-poverty>, last viewed 19 July 2017.

Weitzel, M., Liu, W.H., Vaona, A., 2015. Determinants of technology transfer through CDM:
a within-country analysis for China, Climate Policy 15(5), 626–646.

Worldwatch Institute, 2011. Green Economy and Green Jobs in China. Available at
<http://www.worldwatch.org/system/files/185%20Green%20China.pdf>, last viewed 19
July 2017.

Xie, Y., Zhou, X., 2014. Income inequality in today’s China, Proceedings of the National
Academy of Sciences of the United States of America 111(19), 6928–6933.

Xue, J., 2017. Photovoltaic agriculture - New opportunity for photovoltaic applications in
China, Renewable and Sustainable Energy Reviews 73, 1–9.

23
Zhang, J., Wang, C., 2011. Co-benefits and additionality of the clean development mech-
anism: An empirical analysis, Journal of Environmental Economics and Management 62,
140–154.

Zhao, Z.Y., Li. Z.W., Xia, B., 2014. The impact of the CDM on the cost price of wind power

electricity: a China study, Energy 69, 179–185.

24
 Table 1: Descriptive statistics
(1) Control groups (2) Treatment groups
Unit Obs Mean Std. dev. Obs Mean Std. dev.
Treatment Indicators
re cdm dummy 8,719 0.000 0.000 2,818 0.310 0.463
largebio dummy 8,719 0.000 0.000 2,818 0.057 0.232
smallbio dummy 8,719 0.000 0.000 2,818 0.004 0.059
largewind dummy 8,719 0.000 0.000 2,818 0.241 0.428
smallwind dummy 8,719 0.000 0.000 2,818 0.008 0.086
largesolar dummy 8,719 0.000 0.000 2,818 0.011 0.104
smallsolar dummy 8,719 0.000 0.000 2,818 0.011 0.106

Social Benefit Variables

income rural 1,000 yuan 8,247 5.486* 8.748 2,756 5.027 2.837
rural labor million person 8,719 0.217* 0.169 2,818 0.231 0.176
rural labor primary million person 8,719 0.122* 0.094 2,818 0.133 0.095

County Characteristics
grp primary% % 8,719 0.247 0.130 2,818 0.249 0.135
land area agriculture 1,000 km2 6,876 0.387* 0.409 2,178 0.633 0.520
rural population million person 8,719 0.396* 0.299 2,818 0.430 0.327
government income billion yuan 8,719 0.416 0.930 2,817 0.434 0.627
student% % 8,719 0.139* 0.035 2,817 0.133 0.039
oil production million ton 8,450 0.014* 0.022 2,716 0.020 0.034
machinery power 1,000 kw 8,719 0.033* 0.036 2,818 0.045 0.049

Note: 1) * indicates that the means differ with statistical significance in a two-tailed t-test at the 1% level between the
treatment and control groups; 2) re cdm is a dummy variable used to indicate if county i contains RE-CDM projects in year t:
0 = no,1 = yes; income rural is the annual per capita net income of rural households; rural labor is amount of working
population in rural area; rural labor primary is the number of rural laborers in the primary sector; grp primary% is the share
of primary industry product in the gross regional product; government income is the total value of the government budget
revenue; student% is the share of students accepting compulsory education out of total residents; oil production is the amount
of oil crop production; machinery power is the total capacity of agriculture machinery.

25
 Table 2: Balancing test results
Panel A: Nearest-neighbor propensity score matching (PSM)
Unmatched/ Mean t-test
Outcome var: income rural Matched Treatment Control %bias %bias reduction t-value p-value
grp primary U 1.232 0.977 29.0 4.40 0.000
M 1.232 0.170 7.10 75.6 0.82 0.412
land area agriculture U 0.610 0.412 43.6 6.34 0.000
M 0.610 0.578 7.00 83.9 0.75 0.451
wind potential U 0.272 0.166 25.8 3.84 0.000
M 0.272 0.331 -14.3 44.5 -1.50 0.135
oil production U 0.021 0.014 23.7 3.75 0.000
M 0.021 0.018 9.70 59.1 1.05 0.294

Panel B: Mahalanobis distance matching (MDM)

Unmatched/ Mean t-test
Outcome var: income rural Matched Treatment Control %bias %bias reduction t-value p-value
grp primary U 1.232 0.977 29.0 4.40 0.000
M 1.232 1.189 5.00 82.8 0.54 0.586
land area agriculture U 0.610 0.412 43.6 6.34 0.000
M 0.610 0.581 6.50 85.0 0.75 0.453
wind potential U 0.272 0.166 25.8 3.84 0.000
M 0.272 0.272 0.00 100 0.00 1.000
oil production U 0.021 0.014 23.7 3.75 0.000
M 0.021 0.019 7.60 67.8 0.80 0.422

26
 Table 3: Regression results (Explained variable: income rural)
Estimation method: PSM-DID
(1) (2) (3) (4) (5) (6) (7)
re cdm 0.276∗
(0.151)

largebio 0.851∗∗
(0.315)

smallbio 1.916
(1.279)

largewind 0.085
(0.147)

smallwind -0.440
(0.719)

largesolar -1.164
(0.879)

smallsolar -0.299
(0.817)

grp primary% 9.328∗∗∗ 9.274∗∗∗ 9.091∗∗∗ 9.286∗∗∗ 9.260∗∗∗ 9.191∗∗∗ 9.258∗∗∗

(3.126) (3.122) (3.128) (3.132) (3.123) (3.091) (3.125)

land area agri 2.709∗∗∗ 2.793∗∗∗ 2.843∗∗∗ 2.742∗∗∗ 2.775∗∗∗ 2.896∗∗∗ 2.776∗∗∗
(0.672) (0.671) (0.676) (0.676) (0.677) (0.717) (0.682)

rural population -0.243 -0.201 -0.172 -0.211 -0.190 -0.180 -0.186

(1.210) (1.203) (1.209) (1.215) (1.207) (1.207) (1.208)

income gov 4.763∗∗∗ 4.769∗∗∗ 4.768∗∗∗ 4.765∗∗∗ 4.767∗∗∗ 4.767∗∗∗ 4.766∗∗∗

(1.485) (1.483) (1.484) (1.486) (1.486) (1.486) (1.485)

student% 6.920∗ 6.755∗ 6.560∗ 6.650∗ 6.559∗ 6.567∗ 6.533∗

(3.966) (3.958) (3.955) (3.969) (3.970) (3.945) (3.968)

oil production 15.01∗∗ 14.72∗∗ 14.67∗∗ 15.28∗∗ 15.58∗∗∗ 15.10∗∗ 15.42∗∗∗

(5.865) (5.791) (5.787) (5.934) (5.884) (5.965) (5.876)

machinery power 0.626 0.599 0.581 0.634 0.615 0.604 0.624

(0.830) (0.812) (0.797) (0.842) (0.831) (0.815) (0.834)

constant -2.629 -2.646 1.574 1.563 1.583 1.571 1.593

(2.169) (2.179) (1.046) (1.042) (1.043) (1.042) (1.040)
Year dummy YES YES YES YES YES YES YES
County dummy YES YES YES YES YES YES YES
N 2,078 2,078 2,078 2,078 2,078 2,078 2,078
adj. R2 0.758 0.758 0.758 0.758 0.758 0.758 0.758
Robust standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01

27
 Table 4: Regression results (explained variable: income rural)
Estimation method: MDM-DID
(1) (2) (3) (4) (5) (6) (7)
re cdm 0.423
(0.312)

largebio 1.646∗∗∗
(0.488)

smallbio 3.720
(2.279)

largewind -0.013
(0.308)

smallwind 0.479
(1.355)

largesolar -0.890
(1.413)

smallsolar 0.414
(1.627)

grp primary% 8.477∗∗ 8.390∗∗ 8.077∗∗ 8.391∗∗ 8.422∗∗ 8.337∗∗ 8.398∗∗

(3.817) (3.809) (3.842) (3.828) (3.817) (3.802) (3.821)

land area agri 4.119∗∗∗ 4.250∗∗∗ 4.349∗∗∗ 4.196∗∗∗ 4.190∗∗∗ 4.283∗∗∗ 4.182∗∗∗
(0.935) (0.931) (0.931) (0.945) (0.936) (0.976) (0.937)

rural population 3.063 3.131 3.183 3.154 3.133 3.166 3.143

(3.186) (3.155) (3.173) (3.182) (3.173) (3.178) (3.172)

income gov 11.33∗∗∗ 11.34∗∗∗ 11.34∗∗∗ 11.33∗∗∗ 11.33∗∗∗ 11.33∗∗∗ 11.33∗∗∗

(2.925) (2.920) (2.923) (2.927) (2.925) (2.926) (2.926)

student% 13.97∗∗ 13.88∗∗ 13.59∗∗ 13.56∗∗ 13.58∗∗ 13.62∗∗ 13.58∗∗

(5.950) (5.945) (5.926) (5.909) (5.937) (5.917) (5.938)

oil production 30.19∗∗∗ 29.61∗∗∗ 29.56∗∗∗ 30.93∗∗∗ 30.98∗∗∗ 30.81∗∗∗ 30.80∗∗∗

(11.38) (11.21) (11.17) (11.49) (11.45) (11.46) (11.41)

machinery power -0.410 -0.464 -0.495 -0.404 -0.424 -0.423 -0.394

(0.919) (0.907) (0.902) (0.929) (0.921) (0.916) (0.923)

constant -6.005∗∗∗ -2.613 -2.587 -6.085∗∗∗ -2.597 -2.621 -2.593

(2.307) (1.957) (1.956) (2.331) (1.959) (1.959) (1.959)
Year dummy YES YES YES YES YES YES YES
County dummy YES YES YES YES YES YES YES
N 2,031 2,031 2,031 2,031 2,031 2,031 2,031
adj. R2 0.741 0.742 0.741 0.741 0.741 0.741 0.741
Robust standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01

28
 Table 5: Regression results (explained variable: rural labor)
Estimation method: PSM-DID
(1) (2) (3) (4) (5) (6) (7)
re cdm -0.004
(0.002)

largebio 0.013∗∗
(0.007)

smallbio 0.050
(0.036)

largewind -0.007∗∗∗
(0.002)

smallwind -0.005
(0.023)

largesolar -0.005
(0.007)

smallsolar -0.009
(0.014)

grp primary% -0.054∗∗∗ -0.053∗∗∗ -0.057∗∗∗ -0.055∗∗∗ -0.053∗∗∗ -0.053∗∗∗ -0.053∗∗∗

(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)

landarea agri -0.019∗∗∗ -0.020∗∗∗ -0.018∗∗∗ -0.018∗∗∗ -0.020∗∗∗ -0.020∗∗∗ -0.020∗∗∗

(0.006) (0.006) (0.007) (0.007) (0.006) (0.006) (0.006)

rural population 0.066 0.065 0.065 0.067 0.065 0.065 0.065

(0.045) (0.045) (0.045) (0.044) (0.045) (0.045) (0.045)

income gov -0.001 -0.001 -0.001 -0.001 -0.001 -0.001 -0.001

(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)

student% -0.038 -0.030 -0.033 -0.042 -0.033 -0.033 -0.034

(0.036) (0.035) (0.035) (0.036) (0.036) (0.036) (0.036)

oil production 0.015 0.000 -0.001 0.017 0.011 0.009 0.012

(0.098) (0.097) (0.095) (0.098) (0.098) (0.098) (0.097)

machinery power 0.035 0.035 0.034 0.035 0.035 0.035 0.035

(0.027) (0.027) (0.026) (0.027) (0.027) (0.027) (0.027)

constant 0.122∗∗∗ 0.148∗∗∗ 0.149∗∗∗ 0.121∗∗∗ 0.149∗∗∗ 0.148∗∗∗ 0.149∗∗∗

(0.013) (0.015) (0.015) (0.013) (0.015) (0.015) (0.015)
Year dummy YES YES YES YES YES YES YES
County dummy YES YES YES YES YES YES YES
N 2,078 2,078 2,078 2,078 2,078 2,078 2,078
adj. R2 0.988 0.988 0.988 0.988 0.988 0.988 0.988
Robust standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01
Note: The values of adjusted R-square are approximately 0.99 since the LSDV estimation including a dummy
variable for each county, namely the county dummy, perfectly explains the between variance.

29
 Table 6: Regression results (explained variable: rural labor)
Estimation method: MDM-DID
(1) (2) (3) (4) (5) (6) (7)
[1em] re cdm -0.003
(0.002)

largebio 0.013∗∗
(0.007)

smallbio 0.050
(0.037)

largewind -0.007∗∗∗
(0.002)

smallwind -0.003
(0.023)

largesolar -0.004
(0.007)

smallsolar -0.009
(0.014)

grp primary -0.054∗∗∗ -0.053∗∗∗ -0.057∗∗∗ -0.055∗∗∗ -0.053∗∗∗ -0.053∗∗∗ -0.053∗∗∗

(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)

land area agri -0.021∗∗∗ -0.021∗∗∗ -0.019∗∗∗ -0.020∗∗∗ -0.021∗∗∗ -0.021∗∗∗ -0.021∗∗∗
(0.007) (0.007) (0.007) (0.007) (0.007) (0.007) (0.007)

ruralpopulation 0.063 0.062 0.063 0.064 0.063 0.063 0.063

(0.043) (0.044) (0.044) (0.043) (0.044) (0.044) (0.044)

income gov 0.000 0.000 0.000 0.000 0.000 0.000 0.000

(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)

student -0.031 -0.030 -0.030 -0.034 -0.028 -0.028 -0.028

(0.034) (0.034) (0.034) (0.034) (0.034) (0.034) (0.034)

oilproduction -0.012 -0.027 -0.035 -0.009 -0.015 -0.017 -0.014

(0.097) (0.096) (0.094) (0.097) (0.094) (0.097) (0.095)

machinerypower 0.032 0.032 0.031 0.032 0.032 0.032 0.032

(0.025) (0.025) (0.024) (0.025) (0.025) (0.025) (0.025)

constant 0.123∗∗∗ 0.148∗∗∗ 0.149∗∗∗ 0.122∗∗∗ 0.149∗∗∗ 0.149∗∗∗ 0.149∗∗∗

(0.013) (0.015) (0.015) (0.013) (0.015) (0.015) (0.015)
Year dummy YES YES YES YES YES YES YES
County dummy YES YES YES YES YES YES YES
N 2,031 2,031 2,031 2,031 2,031 2,031 2,031
adj. R2 0.989 0.989 0.989 0.989 0.989 0.989 0.989
Robust standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01
Note: The values of adjusted R-square are approximately 0.99 since the LSDV estimation including a
dummy variable for each county, namely the county dummy, perfectly explains the between variance.

30
 Table 7: Regression results (explained variable: rural labor primary)
Estimation method: PSM-DID
(1) (2) (3) (4) (5) (6) (7)
re cdm 0.002
(0.002)

largebio -0.005
(0.003)

smallbio -0.020∗∗
(0.009)

largewind 0.004∗∗∗
(0.002)

smallwind 0.002
(0.011)

largesolar 0.019∗∗∗
(0.005)

smallsolar 0.012
(0.009)

grp primary% -0.056∗∗∗ -0.056∗∗∗ -0.054∗∗∗ -0.055∗∗∗ -0.056∗∗∗ -0.055∗∗∗ -0.056∗∗∗

(0.017) (0.017) (0.017) (0.017) (0.017) (0.017) (0.017)

landarea agri 0.000 0.000 -0.000 -0.001 0.000 -0.002 -0.000

(0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004)

rural population 0.037∗∗ 0.037∗∗ 0.037∗∗ 0.036∗∗ 0.037∗∗ 0.037∗∗ 0.037∗∗

(0.017) (0.017) (0.017) (0.017) (0.017) (0.017) (0.017)

income gov -0.001 -0.001 -0.001 -0.001 -0.001 -0.001 -0.001

(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)

student% -0.008 -0.010 -0.010 -0.004 -0.010 -0.010 -0.010

(0.030) (0.030) (0.030) (0.030) (0.030) (0.029) (0.030)

oil production 0.101 0.106 0.110 0.0989 0.102 0.107 0.100

(0.080) (0.080) (0.081) (0.080) (0.081) (0.081) (0.081)

machinery power -0.032∗∗ -0.032∗∗ -0.031∗∗ -0.032∗∗ -0.032∗∗ -0.031∗∗ -0.031∗∗

(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)

constant 0.100∗∗∗ 0.078∗∗∗ 0.078∗∗∗ 0.100∗∗∗ 0.078∗∗∗ 0.079∗∗∗ 0.078∗∗∗

(0.014) (0.007) (0.007) (0.014) (0.007) (0.007) (0.007)
Year dummy YES YES YES YES YES YES YES
County dummy YES YES YES YES YES YES YES
N 2,078 2,078 2,078 2,078 2,078 2,078 2,078
adj. R2 0.981 0.981 0.981 0.981 0.981 0.981 0.981
Robust standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01
Note: The values of adjusted R-square are approximately 0.98 since the LSDV estimation including a dummy
variable for each county, namely the county dummy, perfectly explains the between variance.

31
 Table 8: Regression results (explained variable: rural labor primary)
Estimation method: MDM-DID
(1) (2) (3) (4) (5) (6) (7)
re cdm 0.001
(0.002)

largebio -0.005
(0.003)

smallbio -0.019∗∗
(0.009)

largewind 0.004∗∗∗
(0.002)

smallwind -0.000
(0.011)

largesolar 0.020∗∗∗
(0.005)

smallsolar 0.013
(0.009)

grp primary% -0.071∗∗∗ -0.071∗∗∗ -0.070∗∗∗ -0.070∗∗∗ -0.071∗∗∗ -0.070∗∗∗ -0.070∗∗∗

(0.015) (0.015) (0.015) (0.015) (0.015) (0.014) (0.015)

land area agri 0.000 0.000 -0.000 -0.001 0.001 -0.000 0.000
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004)

ruralpopulation 0.038∗∗ 0.039∗∗ 0.039∗∗ 0.038∗∗ 0.039∗∗ 0.038∗∗ 0.038∗∗

(0.017) (0.017) (0.017) (0.018) (0.017) (0.017) (0.017)

income gov -0.002 -0.002 -0.002 -0.002 -0.002 -0.002 -0.002

(0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002)

student -0.002 -0.004 -0.003 0.001 -0.003 -0.004 -0.003

(0.028) (0.028) (0.028) (0.028) (0.028) (0.028) (0.028)

oilproduction 0.071 0.076 0.080 0.070 0.073 0.075 0.069

(0.079) (0.079) (0.079) (0.079) (0.079) (0.079) (0.079)

machinerypower -0.034∗∗ -0.033∗∗ -0.033∗∗ -0.033∗∗ -0.033∗∗ -0.033∗∗ -0.033∗∗

(0.015) (0.015) (0.015) (0.015) (0.015) (0.015) (0.015)

constant 0.105∗∗∗ 0.079∗∗∗ 0.079∗∗∗ 0.106∗∗∗ 0.079∗∗∗ 0.079∗∗∗ 0.079∗∗∗

(0.012) (0.007) (0.007) (0.012) (0.007) (0.007) (0.007)
Year dummy YES YES YES YES YES YES YES
County dummy YES YES YES YES YES YES YES
N 2,031 2,031 2,031 2,031 2,031 2,031 2,031
adj. R2 0.981 0.981 0.981 0.982 0.981 0.982 0.981
Robust standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01
Note: The values of adjusted R-square are approximately 0.98 since the LSDV estimation including a
dummy variable for each county, namely the county dummy, explains the between variance perfectly.

32
 30000
Household per capita income in yuan
25000
20000
15000
10000
5000
0

1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015

Urban Rural

Figure 1: Trends in growth of per capita income of urban and rural households in
China. Source: China Statistical Yearbook.

33
Figure 2: Locational distributions of RE-CDM projects by the cumulative installed
capacity (MW) of power plants in 2012

34
 6
kernel density of propensity score
2 0 4

0 .2 .4 .6 .8
propensity scores BEFORE matching

treatment control
4
kernel density of propensity score
1 2 0 3

0 .2 .4 .6 .8
propensity scores AFTER matching

treatment control

Figure 3: Distribution of propensity scores by treatment and control groups: before and
after the nearest-neighbor PSM

35

View publication stats