University of California Transportation Center
UCTC- FR- 2010- 22
Leaders, Followers and Laggards: Adoption of the U. S. Conference of
Mayors Climate Protection Agreement in California
Rui Wang
University of California, Los Angeles
August 2010
Leaders, Followers and Laggards: Adoption of the U. S. Conference of Mayors
Climate Protection Agreement in California
Rui Wang
Assistant Professor
UCLA School of Public Affairs
Abstract: Little quantitative research has been devoted to voluntary climate actions at the
local level in comparison to those at federal and state levels. It is unclear why some cities
act as leaders in the fight against climate change, some act as followers, while others
remain laggards. This study empirically tests some hypotheses about local political will
to mitigate climate change. Applying a survival analysis to California cities’ adoption of
the U. S. Conference of Mayors Climate Protection Agreement, this study examines the
association between cities’ adoption of the Mayors Agreement and a broad range of
characteristics, such as: local demographics, jurisdiction size, government structure,
political preference and environmentalism, local air quality and congestion level, and
behavior of neighboring jurisdictions. Results support the importance of income level,
political preference and environmentalism of the local communities, as well as a city’s
administrative capacity and autonomy. Congestion relief seems to be an important co-benefit
motivating cities to reduce greenhouse gas emissions.
Keywords: local climate action; political will; U. S. Conference of Mayors Climate
Protection Agreement; California
The role of local government in greenhouse gas ( GHG) emissions reduction has
drawn an increased amount of attention for several reasons. First, the projected failure to
achieve the Kyoto Protocol targets by the major industrialized nations participating in the
treaty signals that focusing on national and industrial policies, e. g., carbon credit trading,
may be insufficient to meet our goals in a timely fashion. Second, the potential for carbon
reduction and ancillary benefits of local climate actions have been recognized by more
researchers and policy makers ( Lindseth, 2004; Climate Action Team, 2006; Schreurs,
2008; Gore and Robinson, 2009). Local governments’ own distinctive policies, such as
zoning, building codes and municipal services, have significant effects on carbon
intensities of major GHG- emitting sectors such as transportation, energy, water and solid
waste, often argued as independent from state policies ( Selin and Vandeveer, 2007). The
proven effectiveness of state and local voluntary measures strongly argues for the
position of these actions as an integrated part of, rather than being substituted by, the
emerging federal climate policies under President Obama’s administration ( Lutsey and
Sperling, 2008). In addition, local governments’ willingness to cooperate with their
neighbors is crucial to the success of regional policy measures, an emerging climate
policy area pioneered by the state of California, through its Senate Bill 375, to link
regional development patterns with GHG emissions. Finally, compared to top- down
policies and programs, localized measures can be designed to match unique local
circumstances and be implemented with fewer bureaucratic impediments. They can also
be monitored more directly and adjusted more quickly.
There is, unfortunately, a gap in academic literature on voluntary climate actions
at the local government level. Although a large literature exists on voluntary actions
taken by industries ( Dietz and Stern, 2002; OECD, 1999, 2003) and households ( Rege
and Telle 2004; Kahn, 2007; Kotchen and Moore, 2007, 2008), their findings may not
apply to cities, where voluntary carbon reduction has different aims and is implemented
in a much broader social context. Among the handful of studies on local climate actions,
almost all are case analyses that provide detailed information on the processes of local
climate actions ( Betsill, 2001; Lindseth, 2004; Betsill and Bulkeley, 2006; Selin and
Vandeveer, 2007; Engel and Orbach, 2008; Schreurs, 2008; Gore and Robinson, 2009).
They generally rely on qualitative case analysis about motivations and mechanisms
instead of robust hypothesis testing. For example, Betsill ( 2001) looks at the ICLEI- CCP
experience and concludes that localizing the global climate issue is key to political
support. Through interviewing local officials, Kousky and Schneider ( 2003) highlight the
perceived importance of local co- benefits of controlling GHG emissions. Very few
published studies quantify the relationship between city characteristics and carbon
reduction commitment. Zahran et al. ( 2009) analyze the correlates of climate action
commitment by U. S. metropolitan statistical areas ( MSAs) by examining some aggregate
measures of local climate change risks, emission intensities and local socio- demographic
characteristics. However, their analysis may only help explain the differences between
two groups of MSAs depending on whether they have committed to a climate action
program at a single point in time. Such treatment loses important information carried by
the diffusion of local policy innovations, given that the number of cities taking climate
actions has been continuously growing in recent years. Overall, there has been little
quantitative analysis on the relationship between local climate actions and local
characteristics such as co- benefits, capacity, and governance structure.
On February 16, 2005, when the Kyoto Protocol became legally binding for the
countries that had ratified it, Seattle Mayor Greg Nickels launched an initiative to
advance the goals of the Kyoto Protocol through leadership and action by American cities
willing to participate. By June 2005, 141 mayors had signed the U. S. Conference of
Mayors Climate Protection Agreement ( MCPA), vowing to reduce carbon emissions in
their cities below 1990 levels, in line with the Kyoto Protocol. The number of signees
quickly rose to 967 as of August 21, 2009. 1
The climate action commitments of these cities raise two questions. What
determines a city’s participation in the MCPA? Why did cities act differently – some as
leaders, some as followers, while some as laggards? Using the time sequence of cities’
participation in the MCPA, this study addresses the gap left by previous studies on local
climate actions. The remainder of this paper presents major hypotheses explaining local
voluntary climate actions, followed by descriptions of data and methodology. After
presenting the results, the paper concludes with policy implications, limitations, and
future research possibilities.
Hypotheses about Local Voluntary Climate Actions
As a standard example of free riding, voluntary GHG emissions mitigation at the
local level seems difficult to be interpreted as a rational choice. In reality, however, the
driving forces of local political will to take climate actions can come from a mixture of
local collective or individual self- interests, behavioral biases and true environmental
altruism, as postulated by Engel and Orbach ( 2008). This paper specifically addresses the
range of theories and/ or hypotheses potentially applicable to local voluntary climate
actions described below.
Income effect is also known as the environmental Kuznets curve ( e. g., Kahn,
2006). Indicated by Kahn’s comparison between smart growth cities and “ brown” cities,
people with higher income levels tend to care more about quality of life issues, and
wealthier communities with more resources and expertise are more capable of creating
strategies and implementing them. Whether the environmental Kuznets curve extends to
the realm of climate change remains a question. The hypothesis that wealthier
communities are more concerned with climate change and willing to act faster may prove
true, given that as local communities aim to lessen their environmental impact, but
continue wanting the same goods, wealthier communities can afford to substitute
environmentally harmful industries and behaviors with those less so. Meanwhile, poorer
communities are often constrained by their budgets, and have fewer choices ( Betsill,
2001).
Administrative capacity refers to a city’s ability to motivate and coordinate its
resources, particularly the necessary human skills to address the complex issue of local
GHG emissions. All else being equal, a more resourceful city government is certainly
more capable of addressing this issue. For example, a richer local government has more
funds available to enhance their technical capacity by consulting outside organizations.
However, given the same level of per capita government expenditure, a city with more
resources allocated toward planning or sustainability- related staff is likely to have higher
administrative capacity to design and implement climate policies. Finally, as economies
of scale may exist in the administrative capacity cities rely on to address specialized
issues such as climate change, a larger city may have dedicated personnel for climate and
sustainability issues, while a smaller but wealthier city may not.
Vulnerability perception stresses that people will react when they feel threatened.
The local effects of climate change can sometimes be a part of daily life for a community,
rather than being uncertain and intangible. Weather fluctuations, smoggy air, or wildfires
can compromise local economies and force traditional forms of economic activity to
adapt if the hazards are sufficiently severe and/ or frequent. Subsequently, the affected
communities are more motivated to mitigate future upheaval than those not directly
affected by climate change. For example, one of the deciding factors that influenced the
City of Durban, South Africa to pursue climate actions included “ a series of extreme
weather events” that threatened the city ( Roberts, 2008). One way to test the vulnerability
hypothesis might be examining the association between local climate actions and the
extent to which the local economy may be affected by climate change, such as the portion
of economic activities based on the weather and/ or the ecosystem. However, due to the
many unknowns in climate vulnerabilities of localities, what matters most is perhaps not
the scientifically measured vulnerability of each community, but the perceived
vulnerability. The education attainment of the citizens may play a role in this perception.
To most local communities, anthropogenic climate change is an abstract theory not
experienced in daily life. Education provides people with the ability to better understand
complex issues such as the risks associated with global climate change ( Kahn, 2006).
Thus, cities populated by citizens with higher educational attainment are expected to be
more responsive to climate change, as these individuals can more readily grasp the
complexities of the issue.
Neighboring effect, also known as peer influence, hypothesizes the positive
correlation between a city and its surrounding jurisdictions with respect to climate actions.
This is similar to the hypothesis that individuals are more likely to adopt a more
environmentally friendly lifestyle if they observe their neighbors doing so. The
underlying mechanism of neighboring effect, however, is unclear. It may come from a
broad range of connections among peer jurisdictions including the simple transfer of
climate change knowledge between local political elites, the diffusion of
environmentalism among neighboring communities, the influence of regional top- down
policies, and the economies of scale that occur when neighboring cities undertake joint
efforts.
Local co- benefits exist mainly because GHGs are often emitted by activities that
produce other non- market costs to the society. Even if climate policies cannot be
implemented at a net savings or zero cost, they often generate local co- benefits that help
localize climate issues, which may convince the public to support climate policies. For
example, the sheer amount of vehicle- miles traveled in California produces not only
GHGs, but also congestion and conventional air pollution. However, the significance of
the effects of local co- benefits such as air quality improvement and congestion relief have
to be carefully tested, as poor air quality and congestion may be results of higher
concentrations of population, economic activity, and income level.
Local interest groups may also explain why cities act as leaders, followers, or
laggards on climate actions. The net influence on local collective climate actions from
local interest groups depends on the relative strength of support and opposition groups.
Vocal individual environmentalists, organized local environmental groups, and those in
green industries may act to increase the social awareness of climate change and influence
political decisions. Conversely, local interest groups that view a certain climate policy as
not being of interest to them may oppose the policy and even impede the overall idea of
climate change mitigation.
Local governance structure is hypothesized to affect local willingness to act on
climate change because structure is often considered to affect function. The structure of
local government includes multiple aspects, such as ( 1) the extent to which an individual
local political leader is allowed to influence public policy, and ( 2) a local jurisdiction’s
degree of autonomy with respect to upper- level governments. As some evidence suggests,
“[ L] ocal mitigation policy is predominantly a top- down decision based on what officials
or staff members believe to be ‘ good business’ or rational economic and political
choices” ( Kousky and Schneider, 2003, p. 361). A stronger mayoral governance system
can better enable a mayor to break the traditional bureaucratic structure of the city
government so that climate policies can be coordinated across various government
departments. However, even with strong local leadership, some administratively less
autonomous cities are constrained by state governments when they attempt to further land
use or economic development policies ( Portney, 2003). In this case, cities with more
administrative autonomy are more likely to lead policy innovations.
There are other explanations of local voluntary climate actions, such as ( 1) local
elected or appointed officials’ environmental preference and political entrepreneurship
( Engel and Orbach, 2008) and ( 2) the local potential to save costs through “ doing better
by doing good”, as suggested by the literature on the energy efficiency gap ( e. g., Jeffe
and Stavins, 1994; Levine et al., 1995). However, relevant data are very difficult to
obtain. As will be shown in the data section, this study can only provide suggestive
evidence by testing a small number of variables, such as local climate and governance
structure, that may be correlated or interact with the above factors.
Data
California is famous for its leading position in fighting against climate change
( Rabe, 2007). The state contains a wide range of local jurisdictions varying in their
commitment to climate change mitigations, ranging from some of the early signatories of
the MCPA, such as Berkeley, Palo Alto, San Francisco, Santa Monica, and West
Hollywood to the long list of non- adopters. This study uses California cities’ adoption of
the MCPA by the end of 20082 as a proxy of local political will to take climate actions.
Focusing on cities in a single state avoids potential effects of variations in state policies
on local actions. A discussion on the limitation of the MCPA as a proxy is discussed in
the conclusion.
The U. S. Conference of Mayors is a nonpartisan organization representing cities
with populations of 30,000 or more. The MCPA has 969 members, the largest number of
city members nationally in comparison to the other major voluntary climate programs
involving local governments -- the International Council for Local Environmental
Initiatives ( ICLEI)’ Cities for Climate Protection ( CCP) and the California Climate
Action Registry ( CCAR). ICLEI boasts 545 members and CCAR 23, but both programs
include significant numbers of other types of local governments, such as counties and
utility districts. In addition, CCAR focuses specifically on GHG emissions inventory
rather than a broader range of local climate actions promoted by the MCPA and the
ICLEI- CCP. The ICLEI- CCP strategically targets specific cities and charges annual
membership fees, while the MCPA is open to all city mayors willing to proclaim their
climate policy position. Cities participating in the MCPA commit to taking three actions:
( 1) strive to meet or exceed the Kyoto Protocol targets ( seven percent below 1990 GHG
emission levels by 2012) in their own communities through various local policies,
projects and campaigns; ( 2) urge their state governments and the federal government to
enact policies and programs to meet or exceed the Kyoto Protocol targets; and ( 3) urge
the U. S. Congress to pass comprehensive bipartisan GHG reduction legislation.
Membership of the MCPA has grown steadily since its founding on February 16, 2005.
By the end of 2008, 147 out of the 230 California cities with populations greater than
30,000, had not yet signed the MCPA ( Population count based on 2005 Census estimates).
Among the 83 signatories of the agreement, three are known to be among the nationally
earliest adopters of the agreement, and 72 have made their signatory date available.
The data set used in this analysis was constructed from a set of characteristics that
may be associated with a city’s voluntary climate action, as described below. Population
size indicates a city’s overall administrative capacity, as it is highly correlated with total
city government expenditure. Population size is expected to positively affect a city’s
likelihood of adopting the MCPA, although the effect of the overall administrative
capacity on the likelihood of adopting the MCPA may not be linear. Two variables are
used to measure a city government’s capacity to implement climate actions in addition to
population size. Per capita government expenditure represents the amount of resources a
city government may spend at the per capita level. A city’s number of planning
professionals, who are often the key technical staff involved in land use, transportation,
and environmental decisions, serves as a proxy of the level of technical capacity available
to plan and implement climate policy and actions.
Average household annual income and percentage of population with bachelor
degrees are obtained by matching the 2000 Census data at the place- level to cities. We
also include population racial compositions as they often reflect the socio- economic
status of a population, as well as potential cultural characteristics. We use percentage of
registered Democratic voters to represent the general political preference of the local
population, and use the percentage of registered Green Party voters as a partial proxy of
the preference for environmental protection.
Three aspects measure the political and administrative structures of the local
government. Local administrative autonomy is measured by a city’s status as a charter
city ( i. e., a city governed by its own charter) or a city governed by the California general
law. The establishment of a mayor- council or a council- manager leadership structure
represents the strength of a mayor’s ability to make decisions based on his/ her personal
political agenda. For example, a majority council vote may not be necessary to approve a
policy decision if an influential mayor governs a city. We measure how closely a mayor’s
political decisions are aligned with the majority of voters by whether a city’s mayor is
elected through direct election or some other method.
We measure two social co- benefits of climate policies: local air quality and traffic
congestion. We document whether a city is located within a nonattainment county/ air
basin, as designated by the U. S. EPA. The number of injuries in traffic collisions,
standardized by population, is used as a proxy, due to the lack of more accurate indicators
of traffic congestion. 3
Local climate variables reflect multiple characteristics of a city. Climate directly
affects the energy use pattern of a city, mainly through indoor climate control and water
use. Local climate characteristics also reflect local vulnerabilities to climate change risks
such as sea level rise and wildfires in California. The desirability to live in a specific
climate can be correlated with population groups who differ in preference, as suggested
by Glaeser and Kahn ( 2008).
Methodology
We use a survival model ( also known as duration model) to analyze the sequence
of adopting the MCPA by California cities. We estimate a hazard rate, h, which is the
conditional probability that a city adopts the MCPA at time t, given that the city has not
already joined, and given the characteristics of the city at time t. A survival analysis is
appropriate for analyzing the effects of a city’s characteristics on its participation in the
MCPA because in addition to accounting for the differences between the adopters and the
non- adopters, it also takes into account the sequence of adoption among cities. This is
preferred to a simple cross- sectional choice model because cities not participating at the
time when data were collected may have joined later. A survival model circumvents this
problem by estimating the conditional probability of participation at each point in time
when a new city ( or new cities, if there is a tie) joined the MCPA.
There are two broad approaches to specifying survival models. The parametric
models assume specific forms of time- dependence of the probability density function.
Common assumptions include exponential, Weibull, and log- logistic distributions.
However, parametric models are difficult to specify in this study for many reasons. For
example, one may expect that the longer a city waits to sign the MCPA, the more likely it
will sign in the future. This may be due to three reasons: ( 1) more knowledge of and
information about climate change will have accumulated and become available; ( 2) the
city will more likely become a target of national or international environmentalists; and
( 3) the city may feel peer pressure as more neighboring jurisdictions become signees.
Conversely, one may also expect the reverse to happen because the longer a city has not
signed the MCPA, the less likely local supporters may be passionate about climate
change, as pointed out by Downs’ ( 1972) “ issue attention cycle”, and the fewer pro-environment
voters will remain in the community. Together these factors indicate that
commitment dynamics are likely to produce non- monotonic hazards, while the exact
trends are unclear.
The semi- parametric models do not require a parametric assumption about the
density function. Instead, this method breaks the hazard rate down into two components:
( 1) a baseline hazard that is a function solely of time and is assumed to be constant across
all cities, and ( 2) a component that is a function of the explanatory variables. We chose
the most commonly used Cox ( 1975) proportional hazard model for this study. The Cox
proportional hazard model further assumes that a city’s hazard rate is proportional to the
baseline hazard, and the ratio is represented as an exponential function of the explanatory
variables that differentiate cities. The Cox proportional hazard model provides estimates
of coefficients ( frequently given directly in exponential form and referred to as “ hazard
ratios”), which show how each of the covariates may affect the hazard rate relative to a
common baseline. The coefficients are estimated using maximum likelihood. Days serve
as the temporal unit of analysis and the Breslow ( 1974) method is used to treat ties.
Results
Table 1 summarizes the maximum likelihood estimates from survival analyses in
five alternative specifications. The estimated coefficients are hazard ratios, with a
greater/ smaller- than- one ratio representing an increase/ decrease in likelihood to adopt the
MCPA. Standard errors reported are adjusted to county- level clusters. Overall, results
across the five alternative model specifications are quite consistent. We use results from
Model 3 when estimating the impacts of variations in independent variables.
The probability of a city signing the MCPA increases as population increases ( or
as the size of the local government expenditure expands), although at a slowly decreasing
pace, as can be seen from the coefficient of the secondary term. For cities with small to
medium- sized populations ( e. g., under half million), a 300,000 increase in population size
roughly doubles the likelihood of signing the MCPA. Such positive effects diminish
gradually until the city population reaches about 3 million. If per capita government
resources do not positively correlate with the hazard ( as suggested by the small and
significant coefficient estimate of per capita government expenditure), this total size
effect may indicate some kind of economies of scale in local climate policies for cities
with fewer than three million residents.
As expected, average income and education levels are highly correlated, and our
models show both coefficients have a positive correlation with cities’ participation in the
MCPA. However, the effect of the percentage of population with bachelor degrees
becomes statistically insignificant once average household income is controlled for in the
regression. A standard deviation ( about $ 21,400 per annual household) higher in average
household income reflects a 47% increase in the likelihood a city signs the MCPA.
Table 1: Hazard ratio coefficients from survival analyses a
Variables Model 1 Model 2 Model 3 Model 4 Model 5
population 1.003261*
( 0.054)
1.003364**
( 0.041)
1.003203**
( 0.038)
1.003208**
( 0.032)
1.003699**
( 0.024)
( population) 2 0.9999995*
( 0.08)
0.9999994**
( 0.047)
0.9999995**
( 0.039)
0.9999995**
( 0.039)
0.9999994**
( 0.02)
income 1.000022***
( 0.002)
1.000026***
( 0.001)
1.000019**
( 0.013)
pct. college
grad
1.020159*
( 0.071)
1.030509**
( 0.012)
1.004568
( 0.768)
0.9802583
( 0.246)
1.006246
( 0.688)
pct. Democrat 1.035775**
( 0.026)
1.038263**
( 0.019)
1.056953***
( 0.006)
1.033086*
( 0.07)
pct. Green 1.953259**
( 0.042)
1.350509
( 0.392)
1.91064*
( 0.073)
1.760096
( 0.124)
1.560368
( 0.173)
per cap gov’t
expenditure
1.000307
( 0.172)
1.000193
( 0.32)
1.000074
( 0.707)
0.9999623
( 0.853)
1.00008
( 0.693)
per cap no. of
planners
1.192522
( 0.568)
1.298821
( 0.322)
1.436643
( 0.122)
1.697857*
( 0.062)
1.418389
( 0.132)
charter city 2.083095***
( 0.01)
2.212598***
( 0.008)
2.378397***
( 0.009)
2.497109**
( 0.011)
2.250066**
( 0.013)
strong mayor 0.1196552
( 0.265)
0.1637117
( 0.301)
0.230828
( 0.39)
0.2790807
( 0.437)
0.1975593
( 0.362)
directly elected
mayor
1.310246
( 0.475)
1.392227
( 0.391)
1.60546
( 0.241)
1.646525
( 0.198)
1.509553
( 0.33)
air quality non-attainment
2.171009*
( 0.091)
2.002178*
( 0.096)
1.602376
( 0.259)
1.456421
( 0.36)
1.297692
( 0.544)
per cap. no. of
traffic injuries
1.080804**
( 0.014)
1.090894***
( 0.006)
1.068132*
( 0.087)
1.064589*
( 0.079)
1.071409*
( 0.059)
pct. peer cities
signed
0.9861128
( 0.124)
0.9832421*
( 0.084)
0.9810453*
( 0.058)
0.9786026**
( 0.048)
0.9819463*
( 0.07)
pct. Hispanic 0.9777983
( 0.107)
pct. black 0.9810766
( 0.292)
pct. Asian 1.014922
( 0.27)
avg.
precipitation
1.027977
( 0.205)
avg. cooling
degree days
0.9996685
( 0.342)
a. P>| z| in parentheses; ***, ** and * represent significance at levels of 1%, 5 % and
10%, respectively.
Both percentages of registered Democrats and Green Party members are
positively associated with cities’ participation in the MCPA. A city with a share of
Democratic voters that is one standard deviation ( about 12%) higher is expected to reflect
a 45% increase in the likelihood that the city signs the Agreement. One standard
deviation ( about 0.44%) increase in the share of Green Party voters in a city is associated
with an 84% increase in the likelihood of signing. However, the magnitude and level of
statistical significance varies across our models. When both percentages of Democratic
and Green Party voters are included in the regression, the latter variable’s influence
becomes less stable and less statistically significant. Overall, the evidence still seems to
support positive associations between local environmentalism and local voluntary climate
actions. The coefficient estimates of Green Party voter share is less stable across models,
indicating that this variable might be an imperfect measure of local environmentalism.
Although the total size of the local government matters, and is almost
proportional to population size, two other measures of local government capacity do not
strongly support the positive relationship between government capacity and voluntary
climate actions. Both per capita governmental expenditure and per capita number of
planners seem to be positively associated with cities’ participation in the MCPA, but only
the per capita number of planners shows some marginal statistical significance in our
models.
Among all the measures of local government structure, the charter city dummy
variable stands out as a powerful predictor of voluntary climate actions of the cities.
Other things equal, charter cities are more than twice as likely to participate in the MCPA
as general- law cities. This may indicate that administrative autonomy – the ability to
create a city’s “ own rules” – may free the cities from governing in the conventional ways
specified by general law at the state level, and allow them to adopt progressive policies.
On the other hand, whether a strong mayoral government exists or whether the mayor is
directly elected does not seem to make a difference ( if not in the negative way). This
poses a question on the important role of individual political leadership suggested by
Kousky and Schneider’s ( 2003) survey.
Estimated impacts of air quality and per capita traffic injuries partially support the
co- benefits hypothesis. One standard deviation ( about four cases per thousand people) in
traffic injury per thousand people roughly increases the hazard by more than 26%. Cities
within the nonattainment area seem to be more likely to participate in the MCPA, but this
result lacks statistical significance. These results suggest that people may be more aware
of or concerned with the interconnection between driving, congestion and climate change,
but less aware of conventional air pollution.
Perhaps the biggest surprise of our analysis is the strong negative association
between a city’s likelihood to participate in the MCPA and the percentage of
participation among its peer cities within the same county. All else equal, one mayor will
be almost 50% less likely to sign the MCPA than the other mayor if her peer cities’
participation rate is 25% higher. This seemingly erroneous effect is strong and consistent
across our models. A possible explanation for this result is political opportunism of the
mayors, which tends to happen when an official makes policy decisions based solely on
whether the action will help advance his/ her career. Namely, a mayor is more likely to
make a political commitment when he/ she finds that by making the commitment they will
be a leader among peers instead of a follower, as long as such a commitment remains
voluntary and no penalty will be applied to the laggards.
Models 4 and 5 test the effects of race and climate of local communities,
respectively. As one would expect, after controlling for other variables, none of the race
and climate variables show statistically significant impacts on local voluntary climate
actions. In California, precipitation and temperature often reflect the potential
vulnerability of a city to climate change hazards, such as sea level rise and wildfires. The
insignificance of the two climate variables seems to indicate that such vulnerabilities
were not well perceived, and/ or perceived vulnerabilities did not translated into local
willingness to mitigate GHG emissions.
Conclusion and Discussion
Understanding what truly drives the voluntary actions at the local level has the
potential to help federal and state policy makers design policies that are more compatible
with local incentives and more cost- effective to implement. Although both the
independent and dependent variables in this analysis may be perceived as incomplete
measurements of the hypothesized factors and the outcome, they provide an early set of
quantitative evidence of voluntary local climate actions.
The overall capacity of a local jurisdiction, measured by population size ( or total
local government budget), affects the likelihood of joining the MCPA. Local
communities with higher average household incomes are more likely to be early adopters.
This seems supportive of the carbon emissions Kuznets curve and previous observations
on local awareness of conventional environmental issues, such as those by Kahn ( 2006).
Both general political preference and local environmentalism exhibit significant impacts
on local willingness to take climate actions. Charter cities’ significantly higher likelihood
of joining the MCPA indicates that administrative autonomy or “ home rule” probably
matters. An increase in the per capita number of traffic injuries significantly improves the
likelihood to act on climate change, showing that traffic congestion alleviation and/ or
safety improvements might be perceived as an important co- benefit of climate actions.
This confirms that policy makers should link co- benefits when trying to build political
will for voluntary climate actions.
Some hypotheses are not strongly supported by the results. Education level of the
citizens may not be important if average household income is held constant. Capacity of a
local government measured by per capita government expenditure and per capita number
of planners do not seem to be crucial in forming the local political will on climate issues.
The results also fail to support the importance of individual political leadership in climate
policy because neither a strong mayoral government nor a directly elected mayor
demonstrates significant impact. The insignificance of the climate variables may indicate
that climate vulnerabilities are either not well perceived or considered to be less
compelling. This is consistent with the findings of Zahran et al. ( 2009) regarding the
effect of climate change risks on MSAs’ climate commitments. Lastly, our result clearly
rejects the hypothesis of positive peer influence among cities. In fact, they even indicate
the existence of political opportunism. This may be a warning to those who are optimistic
about the diffusion of voluntary climate actions across jurisdictions.
It is important to acknowledge that the validity of the results is constrained by
what the data truly measure. Above all, one could argue that political commitments by
the mayors sometimes do not translate into meaningful actions and may be largely
irrelevant. Indeed, it would be ideal to analyze substantial climate actions instead of just
political commitments if reliable data were available. However, political commitments by
a mayor and his/ her council are usually backed by a significant number of constituents. In
addition, they are often important political strategies necessary to catalyze a series of
local climate actions. The experience of cities’ participation in the Mayors Climate
Protection Agreement in California suggests that the leaders and the laggards do differ in
multiple aspects, including the characteristics of local communities and how they are
governed. Nonetheless, it is essential in further studies to analyze substantial climate
actions taken by cities to see why some have been able to move beyond political rhetoric
to substantial actions.
Notes
1. Data are from the U. S. Conference of Mayors Climate Protection Center
( http:// www. usmayors. org/ climateprotection/ revised/), retrieved on August 21, 2009.
2. We chose the “ right censoring” date to avoid possible impacts on cities’ primary
motives given the change of national environmental politics following the 2008
presidential election.
3. The number of fatalities, perhaps surprisingly, is uncorrelated with the number of
injuries. This may indicate that the number of injuries better reflects congestion levels.
Traffic fatalities more strongly reflect high driving speeds, which are negatively
correlated with congestion level.
References
Betsill, M. M., 2001, Mitigating Climate Change in U. S. Cities: Opportunities and
Obstacles. Local Environment 6( 4), 393- 406.
Betsill, M. M. and Bulkeley, H., 2006, Cities and the Multilevel Governance of Global
Climate Change. Global Governance 12, 141- 59.
Breslow, N., 1974, Covariance Analysis of Censored Survival Data. Biometrics 30, 89.
Climate Action Team, 2006, Report to Governor Schwarzenegger and the Legislature.
California Environmental Protection Agency, Sacramento.
Cox, D. R., 1975, Partial Likelihood. Biometrika 62, 269.
Dietz, T. and Stern, P. C. ( eds), 2002, New Tools for Environmental Protection:
Education, Information, and Voluntary Measures. Committee on the Human
Dimensions of Global Change, National Research Council.
Downs, A., 1972, Up and Down with Ecology: the “ Issue- Attention Cycle”. The Public
Interest 28, 38- 50.
Engel, K. H. and Orbach, B. Y., 2008, Micro- Motives for State and Local Climate Change
Initiatives. Harvard Law and Policy Review 2, 119- 137.
Glaeser, E. L. and Kahn, M. E., 2008, The Greenness of Cities. Rappaport
Institute/ Taubman Center Policy Briefs, Harvard Kennedy School.
Gore, C. and Robinson, P. 2009, Local Government Response to Climate Change: Our
Last, Best Hope? In: Selin, H. and VanDeveer, S. D. ( eds), Changing Climates in
North American Politics: Institutions, Policymaking, and Multilevel Governance.
MIT Press, Cambridge, 137- 158.
Jaffe, A. B. and Stavins, R. N., 1994, Energy Efficiency: What Does It Mean? Energy
Policy 22( 10), 804- 810.
Kahn, M. E., 2006, Green Cities: Urban Growth and the Environmen. Brookings,
Washington, DC.
Kahn, M. E., 2007, Do Greens Drive Hummers or Hybrids? Environmental Ideology as A
Determinant of Consumer Choice. Journal of Environmental Economics and
Management 54( 2), 129- 145.
Kotchen, M. J. and Moore, M. R., 2007, Private Provision of Environmental Public Goods:
Household Participation in Green- Electricity Programs. Journal of Environmental
Economics Management 53, 1– 16.
Kotchen, M. J. and Moore, M. R., 2008, Conservation: from Voluntary Restraint to a
Voluntary Price Premium. Environmental and Resource Economics 40( 2), 195-
215.
Kousky, C. and Schneider, S. H., 2003, Global Climate Policy: Will Cities Lead the Way?
Climate Policy 3, 359- 372.
Levine, M. D., Koomey, J. G., McMahon, J. E., Sanstad, A. H. and Hirst, E., 1995, Energy
Efficiency Policy and Market Failures. Annual Review of Energy and the
Environment 20, 535- 55.
Lindseth, G. 2004, The Cities for Climate Protection Campaign ( CCPC) and the Framing
of Local Climate Policy. Local Environment 9( 4), 325- 336.
Lutsey, N. and Sperling, D., 2008, America’s Bottom- Up Climate Change Mitigation
Policy. Energy Policy 36, 673- 85.
Organization for Economic Cooperation and Development ( OECD), 1999, Voluntary
Approaches for Environmental Policy: An Assessment. OECD Environment
Directorate, Paris.
Organization for Economic Cooperation and Development ( OECD), 2003, Voluntary
Approaches for Environmental Policy: Effectiveness, Efficiency and Usage in
Policy Mixes. OECD Environment Directorate, Paris.
Portney, K. E., 2003, Taking Sustainable Cities Seriously: Economic Development, the
Environment, and Quality of Life in American Cities. MIT Press, Cambridge.
Rabe, B. G., 2007, Beyond Kyoto: Climate Change Policy in Multilevel Governance
Systems. Governance: An International Journal of Policy, Administration, and
Institutions 20( 3), 423- 444.
Rege, M. and Telle, K., 2004, The Impact of Social Approval and Framing on
Cooperation in Public Good Situations. Journal of Environmental Economics and
Management 88, 1625- 1644.
Roberts, D., 2008, Thinking Globally, Acting Locally: Institutionalizing Climate Change
at the Local Government Level in Durban, South Africa. Environment and
Urbanization 20, 536.
Selin, H. and VanDeveer, S. D., 2007, Political Science and Prediction: What’s Next for
U. S. Climate Change Policy? Review of Policy Research 24( 1): 1- 27.
Schreurs, M. A. 2008, From the Bottom Up: Local and Subnational Climate Change
Politics. The Journal of Environment & Development 17( 4), 343- 355.
Zahran, S., Grover, H., Brody, S. D. and Vedlitz, A., 2008, Risk, Stress, and Capacity:
Explaining Metropolitan Commitment to Climate Protection. Urban Affairs
Review 43( 4), 447- 474.
Table 1: Hazard ratio coefficients from survival analyses a
Variables Model 1 Model 2 Model 3 Model 4 Model 5
population 1.003261*
( 0.054)
1.003364**
( 0.041)
1.003203**
( 0.038)
1.003208**
( 0.032)
1.003699**
( 0.024)
( population) 2 0.9999995*
( 0.08)
0.9999994**
( 0.047)
0.9999995**
( 0.039)
0.9999995**
( 0.039)
0.9999994**
( 0.02)
income 1.000022***
( 0.002)
1.000026***
( 0.001)
1.000019**
( 0.013)
pct. college
grad
1.020159*
( 0.071)
1.030509**
( 0.012)
1.004568
( 0.768)
0.9802583
( 0.246)
1.006246
( 0.688)
pct. Democrat 1.035775**
( 0.026)
1.038263**
( 0.019)
1.056953***
( 0.006)
1.033086*
( 0.07)
pct. Green 1.953259**
( 0.042)
1.350509
( 0.392)
1.91064*
( 0.073)
1.760096
( 0.124)
1.560368
( 0.173)
per cap gov’t
expenditure
1.000307
( 0.172)
1.000193
( 0.32)
1.000074
( 0.707)
0.9999623
( 0.853)
1.00008
( 0.693)
per cap no. of
planners
1.192522
( 0.568)
1.298821
( 0.322)
1.436643
( 0.122)
1.697857*
( 0.062)
1.418389
( 0.132)
charter city 2.083095***
( 0.01)
2.212598***
( 0.008)
2.378397***
( 0.009)
2.497109**
( 0.011)
2.250066**
( 0.013)
strong mayor 0.1196552
( 0.265)
0.1637117
( 0.301)
0.230828
( 0.39)
0.2790807
( 0.437)
0.1975593
( 0.362)
directly elected
mayor
1.310246
( 0.475)
1.392227
( 0.391)
1.60546
( 0.241)
1.646525
( 0.198)
1.509553
( 0.33)
air quality non-attainment
2.171009*
( 0.091)
2.002178*
( 0.096)
1.602376
( 0.259)
1.456421
( 0.36)
1.297692
( 0.544)
per cap. no. of
traffic injuries
1.080804**
( 0.014)
1.090894***
( 0.006)
1.068132*
( 0.087)
1.064589*
( 0.079)
1.071409*
( 0.059)
pct. peer cities
signed
0.9861128
( 0.124)
0.9832421*
( 0.084)
0.9810453*
( 0.058)
0.9786026**
( 0.048)
0.9819463*
( 0.07)
pct. Hispanic 0.9777983
( 0.107)
pct. black 0.9810766
( 0.292)
pct. Asian 1.014922
( 0.27)
avg.
precipitation
1.027977
( 0.205)
avg. cooling
degree days
0.9996685
( 0.342)
a. P>| z| in parentheses; ***, ** and * represent significance at levels of 1%, 5 % and
10%, respectively.