Competition and Green Signaling: The Case of LEED

Competition and Green Signaling: The Case of LEED Competition and Green Signaling: The Case of LEED
Daniel C. Matisoff*1, Mallory E. Flowers2, Douglas S. Noonan3
Acknowledgements: This research was supported by the National Science Foundation, grant
#1069138. We would also like to thank Dan Winters and the US Green Building Council, John
Maxwell, Tom Lyon, Dylan Minor, Karthik Ramachandran, Atalay Atasu, and participants at the
World Congress for Environmental and Resource Economists for helpful comments and
criticism. All errors are our own.
Keywords: non-market competition, signaling, green business, energy efficiency, green building,
green certification, corporate social responsibility, green labeling
1
Assistant Professor
School of Public Policy
Georgia Institute of Technology
Email: matisoff@gatech.edu
Phone: 404.385.2623
Fax: 404.385.0504
685 Cherry St NW
Atlanta, Georgia, 30332
2
Doctoral Student
School of Public Policy
Georgia Institute of Technology
Email: mflowers8@gatech.edu
Phone: 404.385.3082
685 Cherry St NW
Atlanta, Georgia, 30332
3
Director of Research
Indiana University Public Policy Institute
Associate Professor
Indiana-University-Purdue-University-Indianapolis
Email: noonand@iupui.edu
Phone: 317.278.2448
801 West Michigan Street, BS 3025
Indianapolis, IN 46202
1
Competition and Green Signaling: The Case of LEED Abstract
Firms have increasingly invested in green building certification to signal performance benefits
and non-performance reputational benefits associated with green building. Using Leadership in
Energy and Environmental Design (LEED) certification data, we demonstrate that firms invest
additional resources to attain a greater non-performance signal. Firms earn higher LEED scores
to achieve a higher certification and provide a greener signal to stakeholders, indicating the
presence of competition in green building. Over time, the market becomes more crowded and
signaling becomes less pronounced, particularly at higher certification levels. Further, while
buildings certified just above the highest thresholds cluster spatially, overall trends suggest
decreased clustering of non-performance signaling in markets subject to crowding. Together,
these findings provide a nuanced view of competitive pressures in green signaling.
Introduction
Firms engage in signaling to communicate characteristics to the competitive market. Green
building certification is a costly signal that firms have increasingly used to communicate
information about a firm’s environmental performance or construction practices that aim to
minimize environmental impacts. Much of the additional expense of building green comes from
costs associated with the verification process under third party certifiers (Mills, Friedman et al.
2004, D'Antonio 2007, Morris and Matthiessen 2007). Several studies have emphasized the
benefits to “being green” from the perspective of improved building market performance
(Eichholtz, Kok et al. 2010) and market signaling that goes beyond the productive improvements
of a building (Matisoff, Noonan et al. 2014). The signaling component of Leadership in Energy
and Environmental Design (LEED) benefits operates by reducing information asymmetry
2
Competition and Green Signaling: The Case of LEED between owners and possible customers, making the owners’ products more desirable, and may
be a core component of a firm’s “nonmarket” strategy (Delmas and Toffel 2008, Delmas and
Montes-Sancho 2010) in an increasingly competitive green marketplace. The performance
component of green certification is akin to Spence (1973)’s signaling model where certification
is a strategy that suppliers can use to reduce information asymmetries.
While some cases have demonstrated potential market advantages firms can receive from
environmental technologies and strategies (Shrivastava 1995, Sharma and Vredenburg 1998), the
impact of competition on sustainability initiatives has been difficult to identify. LEED, a wellknown environmental certification scheme for buildings, is of particular interest due to its broad
market uptake and interest it has received in the academic literature. Research that notes the
physical clustering of LEED buildings (Corbett and Muthulingam 2007, Cidell and Beata 2009,
Kahn and Vaughn 2009, Kok, McGraw et al. 2011) fails to control for spatial clustering of new
building starts and development opportunities for commercial office space, and of overall
building density patterns in a metropolitan area.1
In this study, we seek to identify the impact of the multitier LEED certification signal in
producing additional investment in LEED points by building owners. After identifying strategic
thresholding behavior by firms, non-profits, and governments, and our analysis examines relative
differences in signaling behavior across ownership type and building sector. By examining
strategic behavior around LEED certification thresholds, an important distinction between this
work and previous work in market signaling is that we focus on the non-performance component
of the signal communicated through certification. Certification not only communicates
1
Cidell 2009 examines LEED constructions by city and region; Cidell and Beata examine LEED
constructions and effort in certain point categories on a per capita basis. Kahn et al examine
LEED building in a Tiebout sorting context, and Kok et al. provide location controls.
3
Competition and Green Signaling: The Case of LEED information about the uncertain performance of a building, but it communicates characteristics
about the building owner and occupier as well that are independent of the building performance
characteristics.
Spatial and temporal trends around the certification thresholds help illustrate whether firms
compete by adopting greener practices to signal sustainable market practices in response to
competitive pressures. Alternatively, increased competition and a crowded market may decrease
the value of green signals and lead to less strategic behavior around the certification thresholds.
Theory and literature review
Performance and non-performance signaling
The competitive advantages associated with LEED certification go beyond that of building
energy efficiency and reducing the costs of production for a firm (Eichholtz, Kok et al. 2010,
Eichholtz, Kok et al. 2010, Fuerst and McAllister 2011). An uncertified building could have
identical qualities, but not credibly disclose its features. LEED seeks to address a classic
information asymmetry problem by providing a snapshot of the building’s qualities (Majumdar
and Zhang 2009, Fuerst and McAllister 2011, Mason 2013). It rates each building project’s
quality with a raw score that also determines its certification tier: Certified, Silver, Gold, or
Platinum. Due to information asymmetry regarding building quality, in the absence of LEED, we
expect underinvestment in green characteristics in buildings where builders, owners, and
occupiers do not share information about individual building improvements and performance.
LEED certification signals some of this information in an accessible and verifiable manner
(Eichholtz, Kok et al. 2010, Eichholtz, Kok et al. 2013).
In addition to signaling a building’s performance qualities, the LEED certification also signals
other qualities not directly related to building performance for tenants or owners. These non-
4
Competition and Green Signaling: The Case of LEED performance qualities include positive environmental externalities, management or owner
qualities, and qualities of output. LEED can certify a building’s positive external environmental
impacts, which investors (Saha and Darnton 2005), consumers (Sen and Bhattacharya 2001),
employees (Turban and Greening 1997), or other stakeholders (Wood 1991) may value above
and beyond the building’s performance gains internalized to its owner or operator. Further,
achieving LEED certification can effectively certify owner types, corporate social responsibility,
and other difficult-to-observe management qualities that stakeholders (e.g., investors,
employees) may value, above and beyond the building’s performance. While not the same as
certifying “green” products, LEED certification of a building may enhance the green image of its
tenants and that image may spill over to its goods and services. That LEED certification offers a
green signal about more than just the building itself can confer some market power or help the
builder or owner command a premium (Eichholtz, Kok et al. 2010, Eichholtz, Kok et al. 2013).2
Consequently, as more firms use this signal and enter the green building market, the advantages
arising from this non-performance signal may decline (Chegut, Eichholtz et al. 2013, Sexton and
Sexton 2014). This is a crucial distinction from the performance aspect of the LEED signal,
where certifying the internalized benefits from enhanced building performance does not dilute
with greater market penetration.
Thus the signal produced by LEED has two components and conveys information about both
building performance and other non-performance qualities (Shewmake and Viscusi 2014). The
LEED signal, measureable as the total number of LEED points and specific credits that each
2
Eichholtz, Kok, and Quigley (2013, p95) find that LEED commercial office space rents for a
higher rate compared with similar, matched class A office space, and that the rent premium is
monotonically and non-linearly related to the total number of LEED points. They find no
evidence for any price premium based on certification level independent from premium due to
the LEED points. This finding lends credence to the idea that the market interprets the
performance benefits to LEED based on the total and continuous point total.
5
Competition and Green Signaling: The Case of LEED building attains, conveys otherwise hidden information about quality (Shewmake and Viscusi
2014). Its value depends on the reliability of the signal’s form (i.e., the certifier, USGBC), the
importance of the characteristics measured (i.e., the hidden information, such as energy
efficiency), and higher costs of issuing false signals. A building’s raw LEED score, constructed
on a continuous scale from 26 to 69 for buildings that obtained certification under the LEEDNew Construction (NC) versions 2.0–2.2, can be construed as monotonically related to
unobserved quality (Corbett and Muthulingam 2007). As LEED’s design intends, we assume
higher scores imply higher performance qualities.3 Further, we assume higher certification tiers
imply stronger non-performance signals and more cachet to emphasize the firm’s social and
environmental responsibility. The relationship between LEED score and cachet or nonperformance signal, however, may be more weakly monotonic, with certification levels rather
than raw scores playing a prominent role (Corbett and Muthulingam 2007). Along the lines of
Shewmake and Viscusi (2014), Corbett and Muthulingam (2007) and Matisoff, Noonan et al.
(2014), we assume that the performance quality is continuous in raw LEED scores whereas the
non-performance quality is discontinuous around certification thresholds. The cachet segments
markets across levels of certification, promoting owners to upgrade their building’s certification
levels to obtain stronger signaling benefits.
If LEED were merely binary (i.e., certified or not), it would signal both performance and nonperformance qualities, and some of the LEED certified buildings would have been just as green
even without LEED. Making LEED a continuous score gives incentives to (some) buildings to
be even greener, as they can recoup those additional costs through LEED’s performance and
non-performance signaling (Eichholtz, Kok et al. 2013, Shewmake and Viscusi 2014). Putting
3
A separate analysis of LEED scorecards shows that the portion of improvements that fall into
the “performance” categories (e.g. energy efficiency) increase at higher LEED scores.
6
Competition and Green Signaling: The Case of LEED in the arbitrary tiers, we argue, does not affect the performance signals – but it does affect (and
arguably drives) the non-performance signal. 4
Strategic signaling in a competitive market
This interpretation of LEED signals allows insight into the strategic behavior of firms related
to green signaling. In monopolistic competition, firms aim to send stronger signals in order to
drive up demand for their product. If firms compete through signaling, as suggested in a wide
range of research (Fuerst and McAllister 2011, Kok and Kahn 2012, Chegut, Eichholtz et al.
2014), the proportion of firms certifying at higher levels will increase with time. In Sexton and
Sexton (2014), the utility derived by an individual owner of an environmental technology is
related to the uniqueness to the good and local environmental preferences. If firms build greener
to take advantage of a higher non-performance signal, LEED may facilitate a greening of the
building stock. This behavior may be conditioned by existing building stock and the behavior of
competitors. If the non-performance signal confers some market power, an “arms race” or “race
to the top” may occur as firms compete for market advantages by pursuing ever-higher
4
Let X measure a new building’s performance in the absence of LEED. With information
asymmetries present, X is suboptimal and a lower, “market for lemons” level of quality results –
although firms obviously still make some costly performance investments that they can recoup.
A credible signal would allow at least some firms to build better buildings. Let Y≥X be the same
building’s performance with an untiered LEED program in place. Some of the incentive to
upgrade building performance is in the “non-performance signal” is conveys. LEED’s tiered
system offers a convenient way to identify at least some of these non-performance signals.
Because the tiers’ thresholds are arbitrarily set with respect to performance, the observed
threshold discontinuities are a result of additional cachet or non-performance signal from
achieving higher tiers. Let Z be building performance under the observed, tiered LEED system.
As Matisoff et al. (2014) argue, the tiered system allows firms to upgrade further to capture this
cachet, so Z≥Y≥X. With LEED’s tiers, upgrading for non-performance signals does
coincidentally yield (otherwise unprofitable) performance gains.
While we observe only Z, the discontinuities at the thresholds can us identify the buildings that
upgraded from Y to Z to obtain non-performance signals. Of course, some non-performance
signal may be continuously rising in Y. But at least at the (arbitrary) thresholds we assume the
discontinuities result from cachet rather than performance.
7
Competition and Green Signaling: The Case of LEED certification levels. Improving performance to gain non-performance benefits is an indicator of a
classic race to the top (Auld, Bernstein et al. 2008).5 Upgrading-to-the-threshold and achieving
ever higher tiers of LEED certification would be consistent with this story.
Alternatively, as more green buildings certify, the monopoly rents from non-performance
signals are diluted with entry by new green-certified buildings (Auld, Bernstein et al. 2008,
Chegut, Eichholtz et al. 2014). In a race to the start, market leaders capture the benefits of nonperformance signaling, while laggards enter the market to capture the value of the nonperformance signal accrued by the early movers (Delmas and Montes-Sancho 2010). While firms
will still certify green to take advantage of the performance signal due to the reduction of
information asymmetry, with a diminished value of the non-performance signal due to a
competitive market, there would be little incentive to upgrade at the thresholds, leaving a smooth
distribution around the thresholds, consistent with a race to the start, as expected by others
(Eichholtz, Kok et al. 2010, Mason 2013, Chegut, Eichholtz et al. 2014).6
Building patterns and spatial competition
5
Note that, for new building construction, the “racing” may not occur among incumbent players
but rather among new entrants. Changes in certification and upgrading-to-a-threshold behaviors
over time indicate changes in the “entry points” of new firms, rather than changes in existing
LEED-certified buildings. (Other certification schemes exist for older buildings.)
6
Observing this is complicated by more market penetration of LEED not weakening the
performance signals of LEED and because costs and benefits of the performance benefits of
LEED may change over time. If the cachet loses its luster, non-performance signals suffer but
not the performance signal aspect of LEED. Further, the increasing uptake of LEED in a market
leads to its reduced utility. Increased upgrading behavior drives the premium for upgrading
down, although other trends in the value of greener performance may rise or fall independently.
Buildings’ green performance can be expected to actually fall, but only back to the level justified
by performance benefits. The cachet’s incentive to upgrade fades, not the entire incentive of the
certification scheme (i.e., it is a race to the start rather than a race to the bottom).
8
Competition and Green Signaling: The Case of LEED Real estate markets provide one way of viewing spatial competition in green signaling
(Eichholtz, Kok et al. 2010, Eichholtz, Kok et al. 2010, Eichholtz, Kok et al. 2013). If
environmental preferences are spatially clustered (Sexton and Sexton 2014), real estate markets
may respond accordingly (Kahn and Vaughn 2009). To the extent that real estate markets are
local and that real estate developers compete for tenants and retail establishments use green
signaling to compete for consumers, spatially correlated upgrading behavior may provide
evidence of spatial competition in green signaling. Further, because some types of businesses are
more likely to compete spatially in a monopolistically competitive market (due to a smaller
market radius), these patterns may be more pronounced in industries likely to be spatially
competitive, such as commercial office. While some types of firms are engaging in nonperformance signaling to investors or other businesses, others are engaging in non-performance
signaling to employees (commercial office) or consumers (hotels, restaurants, and retail). If firms
compete to be green, we also ought to observe spatial clustering of green non-performance
signals, as firms aim to send stronger non-performance signals than their neighboring
competitors in a monopolistically competitive market (Capozza and Van Order 1978). This
pattern, expected in a race to the top, should not be observed in a race to the start.
It is important to note that this means of identifying competition may not capture non-spatial
competition between firms for consumers or employees (e.g. developing a green brand
reputation) or signaling to investors (e.g. greening a corporate headquarters). A spatial model of
green competition only holds when the underlying preferences of consumers, tenants, or
employees are spatially correlated.
Up to this point, the discussion of signaling via LEED has centered on for-profit firms, yet
much of LEED-certified new construction has other ownership types (i.e., governments,
9
Competition and Green Signaling: The Case of LEED nonprofits). While these organizations still pursue rents from green signals, they face different
sorts of competition (e.g., competition for donors, pressure in political markets) that likely
operate much differently than conventional market competition. They serve as interesting
“control groups” to identify how market pressures affect green certification. It may be that forprofit firms are less sensitive to non-performance signals than other owners. If clustering of
green signals is due to competition, we expect stronger clustering of spatial clustering for forprofit firms, rather than non-profit firms or government, due to the need for for-profit firms to
compete for consumers or market share in a spatially dependent context. Conversely, if
clustering is due to public building regulatory compliance or procurement processes, we expect
greater spatial clustering of government owned buildings. While government and nonprofit firms
may have even stronger motivations to provide a non-performance signal, we do not expect that
these motivations are spatially dependent.
Identification of non-performance signaling
Following a similar methodology to Kleven and Waseem (2013) and Matisoff et al. (2013), we
establish a counterfactual distribution of LEED buildings in the absence of non-performance
signaling due to the thresholds. Because performance benefits are similar on either side of the
threshold, we interpret the excess number of buildings just above the certification threshold and
the lack of buildings just below the certification threshold as the non-performance signaling
component—an indicator of the signal value of the certification threshold. We identify the nonperformance signal by looking at these threshold effects. The performance signal exists but
cannot be identified with these data.
Data
10
Competition and Green Signaling: The Case of LEED Data were obtained from the USGBC. The available data includes LEED point total, certification
level, project name and address, LEED scoring system used, project type, buildings size, and site
context. We restrict our analysis to the buildings within the LEED New Construction (LEED
NC) versions 2.0–2.2 in order to maintain consistent methods and simplify results to comparable
metrics. Because we expect building owners to behave differently in very rural areas, and for
further reasons discussed in the next section, we have also excluded rural observations. Likewise,
we only assess buildings in the contiguous United States (there are several LEED certified
buildings in Hawaii, Alaska, and other U.S. territories which are too far to be considered in
spatial competition with the buildings in the other adjoining 48 states). Under LEED NC,
buildings must achieve at least 26 points to become Certified, at least 33 points for a Silver
certification, 39 points for Gold certification, and 52 points or more (at most 69) to earn Platinum
certification. These point thresholds represent 40, 50, 60 and 80 percent of the total number of
“base” points (65). 7 According to a long time employee of USGBC, these thresholds were
determined “arbitrarily, and not scientifically… when there were only a couple of employees of
USGBC.” The thresholds were meant to provide even spacing amongst the lower tiers, with
platinum requiring a significant extra investment. Points for sustainable site planning, water
safeguarding and efficiency, energy efficiency and renewable energy, resource and material
conservation, and indoor environmental quality are awarded after a documentation and
certification process, and all points are awarded independently. According to USGBC officials,
for the lower tiers, there are many different ways of achieving Certified, Silver, or Gold
7
There are 65 base points, with 4 “extra credit” points awardable making a total of 69 points.
The “arbitrary” thresholds highlight the idea that the thresholds are not natural cut points in
building technology investment. A long time employee noted that there were only two
employees at the time working on the project, and that these cut points were “not scientific in
any way.” As the scoring system was revised, minimums in certain categories were required to
prevent firms from avoiding substantive investments in energy efficiency improvements.
11
Competition and Green Signaling: The Case of LEED certifications. For Platinum, firms must get most points available in some costly categories.
Notably, most buildings earn point totals that are either at or just above these certification
thresholds. Building scores that are just below each threshold are rare. The frequency of LEED
buildings at all certified scores is displayed in Figure 1.
<<insert figure 1 about here>>
High frequencies of buildings at and just above the thresholds suggest several possible data
generating processes. First, building owners may complete cost-effective credits, and then work
for extra points to move up to the next level of certification. Second, firms set a certification
target, building in a point cushion to ensure the goal certification is obtained. Some firms may
fall short of their goal during the review and approval process and be unwilling to make costly
adjustments to ratchet up, explaining why buildings periodically obtain a score just below a
threshold value. Finally, some firms may seek to build ‘the greenest building possible,’
irrespective of certification thresholds. Discussions with LEED builders and consultants confirm
a mixture of these processes.
We divide our data into categories by owner type and building use. We examine three owner
types: government, non-profit, and for-profit firms. Of the 3,437 total projects, 1,588 are owned
by government agencies, 1,048 by for-profit entities, and non-profits own 801. Individual
owners make up only a small portion of the total dataset, and are excluded from this study.
Buildings are also separated by use into commercial office (N = 1,629), retail (N = 214),
healthcare (149), restaurant (N = 85), hotel (N = 37), and education (N = 1,324) buildings.
On average, buildings score 37.3 LEED points, though non-profits score higher (38 points on
average) compared to government and for-profit entities (averaging 37.5 and 36.6 points,
12
Competition and Green Signaling: The Case of LEED respectively). In the LEED new construction sample, 19.6 percent are Certified, 33.3 percent are
Silver certified, 41 percent are Gold certified, and six percent achieve Platinum certification.
Methodology
Summary of approach
We begin, following Kleven and Waseem (2013) by constructing a counterfactual distribution in
absence of the multitier thresholds. This counterfactual allows us to compare the observed
distribution to the counterfactual distribution by approximating the strength of the nonperformance signal for each building. After characterizing the quantity of non-performance
signaling that exists across building types, we observe how the strength of this signaling changes
with time. As described above, we expect non-performance signaling to ‘wash away’ in purely
competitive markets, and to persist or strengthen under monopolistic competition. Nonperformance signaling can be understood to increase when the excess portion of certified
buildings achieving scores at or just above the threshold increases. We test for these trends
across owner types. Finally, we assess evidence of clustering of these signals across ownership
types and building uses.
Measuring non-performance signaling propensities
First, this study identifies the proportions of upgrading within the LEED certification due to
benefits from the non-performance signal. To determine the propensity for non-performance
signaling at each point value within the LEED certification program, a counterfactual
distribution of LEED buildings describes a hypothetical distribution of buildings were no
certification thresholds exist and building attributes are described by the LEED certification
point total.
13
Competition and Green Signaling: The Case of LEED While other factors (e.g. energy prices, contractor learning and behavior, locally appropriate
building attributes) may impact the types of investments or improvements made by building
owners, a key assumption in this analysis is that the performance benefits are locally smooth
around certification thresholds. Because some of the points are accrued for positive externalities
of a firm that provide no operational benefits (e.g. reducing construction waste), firms could
forgo just one or two of these LEED points and have the same level of performance benefits,
with a lower level of non-performance signaling. We test for spatial clustering in nonperformance signaling, independent of spatial clustering in performance signaling.
Following Kleven and Waseem (2013) a locally smooth polynomial function is generated
based on the distribution of buildings with LEED scores not immediately impacted by the
thresholds. The buildings just above (“bunchers”) and below (“dominated”) each threshold are
dropped to approximate a distribution unaffected by strategic behavior around the thresholds.
The smooth polynomial function is drawn based on the remaining data. This eliminates the
pronounced discontinuities visible in the observed distribution (Fig. 1), creating a feasible
empirically determined counterfactual distribution. This counterfactual illustrates the LEED
building distribution in the absence of discontinuous signaling due to the certification thresholds.
If LEED buildings reflect only production-related benefits to LEED attainment rather than
discontinuities at the threshold, the observed distribution would be smooth (Corbett and
Muthulingam 2007), peaking in density at the average cost-effective LEED score due to the
performance benefits and related performance signaling alone (Corbett and Muthulingam 2007).
The range of LEED scores around each threshold dropped from the distribution may impact
the resulting counterfactual. A tradeoff exists between information retention and reducing the
impact of the signaling thresholds. To keep more information in the distribution to generate the
14
Competition and Green Signaling: The Case of LEED kernel density, dropping fewer LEED scores is ideal. However this may cause some buildings
that are influenced by signaling effects to remain in the counterfactual distribution, skewing
results. To avoid either extreme, in which the counterfactual density would be either over- or
underestimated, we drop four scores around each threshold: the point value of the threshold, the
one above it, and the two below it. This creates a smooth counterfactual of LEED building
patterns.8 Following similar conservative logic, we estimate the polynomial counterfactual using
a smoothing bandwidth of two LEED points.
The observed and counterfactual distributions are graphed in Figure 2. The counterfactual
distribution is compared to the observed distributions to assign a non-performance signal value
for each building project based on the number of points achieved in the LEED scoring system.
This value represents the portion of the observed frequency that exceeds the counterfactual
expectation divided by the individual observations composing the observed frequencies. To
avoid double-counting, and because we are interested only in the excess of buildings above
thresholds, the signaling value is censored to preserve only positive values.
<<<insert figure 2 about here>>>
The signaling value at each LEED score i can be defined as:
xi = (observed frequencyi – counterfactual frequencyi) / observed frequencyi
(1)
Thus, a building scoring at or just above a certification threshold, where the observed
frequency is much higher than expected by the counterfactual, will have a very high signaling
value. If the frequencies observed in the data and predicted by the counterfactual are similar at a
8
Corbett et al. (2007) contend that a performance-based distribution must be unimodal and
imposed various distributions on the empirical distribution of LEED buildings to determine the
most appropriate explanation for the empirical multi-modal LEED distribution. We remain
atheoretical about the shape of the expected distribution, and simply estimate a smooth
distribution based on the empirical distribution, excluding the observations around the
thresholds.
15
Competition and Green Signaling: The Case of LEED given building’s score, that building will have a signaling value close to zero. For signaling
values below zero, xi is censored. While other methods to construct a counterfactual are possible,
across all potential counterfactuals, the high points in the observed distribution will correspond
to high signaling values, and the low points to low signaling values. Results remain consistent
across a wide range of potential counterfactual functions.
Temporal trends in non-performance signaling
After identifying the intensity of the signaling effect for each LEED building score, we
calculate average non-performance signaling by building and owner type, observe changes in the
prevalence of non-performance signaling over time, and whether the relative importance of
signaling tiers increases or decreases over time. We observe the changes in the portions of LEED
buildings certifying annually in each of the four tiers and the excess portions of new LEED
buildings bunching just above the certification thresholds for each year by owner type and
certification level, to determine whether the sharp discontinuities in the observed distribution are
increasing or decreasing over time, and whether trends are consistent for each level of
certification. This is done by calculating xi for each year T, based on only on observations in year
t ≤ T and summing the number of signaling buildings (xini, for observed frequency as score i = ni)
across all scores in a particular certification level, then dividing by the total number of
observations those same years. Thus, the values reported in Table 2 follow from:
Share SignalingcT = [Σ(observed frequencyiT – counterfactual frequencyiT)]/Ni
(2)
as summed over all scores i in certification level c and calculated using only observations in year
t ≤ T, the count of which is Ni. We also compute this for subsamples of owner types (i.e.,
government, for-profit, non-profit), in which case the counterfactual for xi is computed based on
16
Competition and Green Signaling: The Case of LEED observations for that year or prior. For Table 1, this share of signaling is calculated using all
years (T=2013).
Calculating spatial clustering of non-performance signaling
The second stage of the study identifies the spatial dependence of signaling effects by
calculating the global Moran’s Index (Moran’s I), which reflects the level of spatial
autocorrelation (Moran, 1950) for each sample. Moran’s I ranges from negative to positive one.
A value of negative one indicates perfect dispersion of the sample; a positive one illustrates
perfectly clustered observations, and a zero indicates random spatial distribution. Moran’s I is
defined as:
I=
N
i
∑∑w
i
∑ ∑ w (X − X)(X − X)
∑ (X − X)
j
ij
j
ij
i
i
j
i
2
(3)
where N is the number of observations which are indexed by i and j, and X is the variable of
€
interest (in this case, either the identity of the building or the signaling factor of the building’s
score). The spatial weights matrix, w, is constructed to define which neighboring observations
may influence one another. This can be constructed in several ways including distance within a
given radius, number of nearest observations, or inverse distance. Distance or nearest-neighbor
threshold comparisons are highly sensitive to the selected threshold values, which may not be
representative of each building’s competitive market. We utilize the inverse-distance weight
construction to compare LEED building decisions across variable market sizes, which assumes
that observations are more dependent on other nearby observations compared to those that are far
away. To eliminate buildings that may skew the results because they do not have any neighbors
within a reasonable distance, we remove the very rural observations from this data set.
17
Competition and Green Signaling: The Case of LEED For the full sample (i.e., all new construction) and each subsample (i.e., building owner types
and end uses), we calculate the Moran’s I for the signaling values calculated in equation (1). We
call this “signal clustering” or dispersion.
Results
Non-performance signaling in LEED buildings
Signaling values are assigned to each building based on the counterfactual distribution obtained
from the locally smooth polynomial. Results demonstrate that many LEED buildings bunch at
and above each certification threshold (seen in Fig. 1), producing strong positive nonperformance signaling at these LEED point values. Few buildings attain just below the
threshold, producing signaling values equal to zero at those LEED scores, as seen in Figure 2.
Figure 2 also displays the signaling value (xi) assigned to each LEED score. The locally
smoothed density for the counterfactual, depicted in Figure 2, is a rather conservative approach.
Note how, even with its estimation based on data that omits all observations at or just above
thresholds, it still shows local peaks at each threshold. This is a result from selecting a relatively
narrow window around each threshold; widening that window would smooth out the
counterfactual distribution even further than quickly result in a more unimodal distribution.
Alternative counterfactual distributions, such as χ2 or normal, tend to be even less data-driven
and even less conservative. The current approach shown in Figure 2 yields conservative
signaling values that likely substantially understate the extent of signaling in practice.
Variation by building owner and use
Table 1 displays the share of signalers for all observation and for various subsamples defined
based on the building’s owner type, certification level, or end use. For all 3,036 observations
LEED buildings, almost 4% are Silver signalers (i.e., at or just above the Silver threshold),
18
Competition and Green Signaling: The Case of LEED almost 6% are Gold signalers, and 1% are Platinum signalers. Put another away, at least one in
ten LEED-certified buildings have upgraded to the next higher threshold for some nonperformance signaling gain. The rest of Table 1 indicates the prevalence of this signaling
behavior, overall and across the different tiers of LEED certification, for several key subsamples.
All subsamples demonstrate some signaling, relative to what would have been expected by the
counterfactual distribution. Results show more non-performance signaling behavior at the Gold
tier than other tiers for all ownership types. This partly reflects the overall greater frequency of
Gold buildings in the data. Hotels and restaurants exhibit above-average signaling shares (16%
and 14%, respectively). This is driven by a preponderance of signaling behavior at the Silver tier
for hotels and at the Gold tier for restaurants. Community development, interpretive centers,
laboratories, and stadiums also have above-average signaling shares.
<<insert Table 1 about here>>
Temporal trends in non-performance signaling
Each year, the number of new buildings added under the LEED-NC 2.0 to 2.2 certification
standards grows, peaking just after 2010, when a new version of LEED-NC (v2009) standards
was initiated. Notably, 2008 seems to be an outlier year, where Gold non-performance signaling
dropped and Silver non-performance signaling spiked. This coincides with the financial crisis
and perhaps a temporary shift in decision-making as capital became scarce.
<<<insert Figure 3 about here>>>
<<<insert Figure 4 about here>>>
Figure 3 demonstrates that roughly one-third of buildings certify at the Silver level during each
of those years. The annual portion of buildings certifying Gold rose from less that 25 percent to
over 40 percent of new buildings over the period from 2005 and 2009. The portion of buildings
19
Competition and Green Signaling: The Case of LEED certifying Platinum remains between about five and seven percent each year, although the
number of Platinum buildings added annually grows almost every year. The difference between
the top and bottom panels of Figure 4 shows that the spikes and sharp drop-offs in density as
points pass thresholds are becoming more muted over time. This suggests a shift in nonperformance signaling over time as well as shifts to higher levels of certification overall.
Table 2 displays changes in signaling shares over time. For the full sample, the signaling share
is erratically falling over time, from 14.2% at the start to 11.7% when the most current
observations are included. This trend is reflected within the Silver and Gold certification levels
as well. Table 2 also shows these trends for subsamples based on owner types, and the results
are largely consistent regardless of owner types. There are generally declining shares of nonperformance signalers in these subsamples from 2002 to 2010.
<<<insert Table 2 about here>>>
Spatial clustering of non-performance signaling
LEED buildings spatially cluster, especially in metropolitan areas (Cidell 2009). A result
showing Moran’s I of 0 would indicate random dispersion of the signaling values across these
locations and would suggest that there is no clustering of new buildings associated with nonperformance signaling behavior. In Table 3, values of Moran’s I demonstrate that LEED
buildings with high signaling factors (i.e., tended to be upgrading buildings) tend to locate
nearby other LEED buildings that also had higher signaling factors. In the terminology provided
by Kleven and Waseem (2013), the “bunchers” are spatially correlated (I = .12). It is important
to recall that the spatial correlation we observe is for non-performance signaling and thus
controls for any performance reasons to bump up as well as for the physical clustering of LEED
buildings due to patterns in new construction. In other words, the positive spatial correlation
20
Competition and Green Signaling: The Case of LEED reported in Table 3 is above and beyond colocation driven by spatial clustering of buildings
generally or spatially correlated performance costs and benefits. It is, given where all the LEED
certified buildings locate, a measure of how spatially clustered their non-performance signaling
intensities (xi) are.
The statistically significant spatial autocorrelation index for the full sample is driven by several
subsets of buildings. In particular, government buildings and buildings constructed just above the
Silver and Gold and Platinum thresholds demonstrate significant spatial clustering. Note too, that
even fairly small samples achieve statistical significance (as in retail buildings), and that some
larger samples do not (as in commercial office buildings). This confirms that the clustering
finding is not simply the result of a large sample size driving down standard errors.
<< Insert Table 3 about here >>
The strong statistical significance for the clustering of government buildings suggests that
clustering may be driven by public building procurement and policies, rather than by
competition. The strong result of clustering in the highest tiers, however, also suggests that
green competition may be playing a large role in the higher tiers.
Discussion
LEED encourages higher non-performance signaling
Our analysis reveals a nuanced relationship between certification, green building, and nonperformance signaling trends. Several pieces of evidence demonstrate that the multitier labeling
context of LEED encourages building owners to build “greener” to send a stronger nonperformance market signal by ratcheting up at the tiers to capture additional cachet associated
with a firm or organization’s positive externality. Across all subsamples, buildings bunch just
above the certification thresholds, demonstrating that building owners have invested additional
21
Competition and Green Signaling: The Case of LEED resources to achieve a higher level of certification. However, over time, we observe that firms
and other organizations are less likely to construct buildings that achieve a point total just above
higher levels of certification. This trend suggests that a crowding market has led to the nonperformance signal associated with LEED to fade over time, even though overall market trends
have pushed towards higher levels of certification.
The multitier and continuously scored label, in contrast to a binary certification, allows firms
to continuously differentiate themselves in the market. Because USGBC revises standards every
few years, the tightening standards force firms to repeatedly become “greener” in order to send a
non-performance signal. Notably, several shifts occurred in green building patterns in
conjunction with the introduction of the v2009 certification scheme. First, 2009 and 2010
represented a local peak for non-performance signaling with a larger share of buildings achieving
Gold certification in those years, just as the new standards were introduced. This suggests that
firms rushed to gain Gold certification before the standards were changed. Second, 2009
represents the peak of Gold certification. While market trends had shifted from Certified and
Silver and towards Gold, this trend stopped with the introduction of the new v2009 standards.
These results suggest that the new standards represented an opportunity to send a stronger nonperformance signal in a newer less crowded market, but also a last opportunity to get projects in
under the old standards.
Non-performance signaling is spatially clustered
Supporting previous research (Cidell 2009, Kahn and Vaughn 2009), we find evidence of
spatial clustering of non-performance signaling in green building behavior, once building
patterns are accounted for, suggesting that spatial competition may be a driver of ratcheting up
behavior at the highest tiers. However, these results are nuanced, based on varying findings
22
Competition and Green Signaling: The Case of LEED across the thresholds and within specific building sectors. Buildings at the Silver or Gold
certification standards are especially more likely to be built just above the threshold if nearby
LEED buildings have also been built just above the threshold. This is consistent with firms that
upgrade for non-performance signaling benefits to do so when nearby buildings do likewise.
This suggests a role for spatial competition in signaling akin to an arms race over green
signaling. That government buildings demonstrate the strongest effects of spatial autocorrelation
confirms the role of procurement policies in driving green building (Simcoe and Toffel 2011).
Unfortunately, due to spatial dispersion of new LEED construction at large, it is likely that the
data are simply not dense enough to consistently identify these sorts of patterns within many of
the particular building uses.
A case for competition in non-performance signaling?
Taken together, our findings suggest that the multitier context, combined with increasing
demand for green building, allows firms to engage in competition to send a stronger nonperformance signal. As opposed to a wide range of other voluntary programs where firms receive
the same marketing benefit for participation regardless of participation intensity, LEED uses a
tiered grading system that adds transparency and allows firms to differentiate performance based
on LEED labels. The success of LEED may hinge on allowing firms to distinguish themselves by
sending a strong green signal by certifying at the highest tiers, while still allowing for other firms
and organizations to send a weaker green signal by complying with the lower tiers and certify
performance. This dynamic invites further investigation.
Discussions with LEED officials, real estate management companies, and other professionals
in the industry suggest several behavioral and economic trends that help elucidate these results.
Real estate investment professionals note continued pressure to certify green by real estate
23
Competition and Green Signaling: The Case of LEED investment trusts and to attempt to differentiate a building by achieving higher levels of
certification. Other stakeholders, such as university campus architects, note competitive pressure
due to the behavior of other universities, sustainability rankings, and the desire to market a
“green” campus. These observations are consistent with the overall findings of signaling, paired
with a lack of significance in the for-profit sector of the spatial analysis, suggesting that firms
may be signaling primarily to investors or other stakeholders that are not spatially clustered. The
non-performance portion of the LEED signal likely targets investors, donors, and other
stakeholders more than it does to employees, tenants, or customers (who may be targeted by the
performance portion of the signal).
As green building practices have become standard, certifying basic levels of building
performance may have reduced value, and firms seeking to differentiate themselves have
pursued higher levels of certification. Nevertheless, some building owners have continued to
pursue LEED at the lower levels, presumably to pursue a performance signal.
Second, real estate investment professionals point to dynamic competition in the real estate
market based on a combination of types of actors and increasing demand for “green” real estate.
They point to pioneers that sought to build and certify green for ideological reasons or to be able
to market themselves as leaders in environmental sustainability. As the market shifted, and
LEED certification has become increasingly common, higher levels of certification were
required to be pioneers.9 This suggests that the additional marketing benefits of going above
Certified or Silver become more valuable after competition crowds out the marketable benefit of
the lowest levels, consistent with results from Kok et al. (2011). Buildings must become greener
9
Interestingly, because USGBC has revised the standards (e.g., 2009 and 4.0), real estate
professionals note that the new standards provide opportunity to be pioneers within the new
standard, “Firms want to be the first platinum building under the 4.0 standard.”
24
Competition and Green Signaling: The Case of LEED if they want to get the same benefits as their early-adopting counterparts. The non-increasing
signaling share over time (see Table 2) contrasts with this common perception expressed in
discussions with stakeholder. The inconsistency could result from “pioneers” being sufficiently
rare that they do not drive the estimates here. The signaling factor at the higher levels of
certification might not rise over time if the marginal point cost falls fast enough or builders’
aversion to the risk of inadvertently falling just below a threshold grows strong enough. Either
of those trends could lead to pioneers increasingly “overshooting” thresholds when their upgrade
for non-performance signaling benefits.
Others, in contrast, are responding to investor preferences – and in particular pressures from
European real estate investment trusts (REITs) and private equity funds, which have increasingly
demanded LEED certification. A wide range of institutional investors now frequently requires
socially responsible designations. For example, GRESB (Global Real Estate Sustainability
Benchmark), a sustainable real estate rating system, backed by trillions of dollars of institutional
investors, rates REITs and other investment groups to generate sustainability reports and awards
points for LEED certification. Other disclosure organizations, such as TruCost, Global Reporting
Initiative, and the Carbon Disclosure Project, play similar roles.
Finally, professionals in the field note a third type of participant – those who “are just keeping
up with the Joneses.” According to discussions with real estate managers, this class of
participants simply certifies green because the market has moved in this direction.
Several market trends may also help explain some of this behavior. As market penetration of
green building increases, the costs of green building are likely to drop and it may be easier to
achieve higher levels of certification. Decreasing costs may explain the overall shift to higher
certification levels in the market. While overall demand and decreased costs for green building
25
Competition and Green Signaling: The Case of LEED may have shifted the market towards higher tiers of certification, it is apparent that an
increasingly competitive market has diminished the ability of a building to secure rents from
certification. That spatial clustering of non-performance signaling suggests a mechanism where
colocating buildings apply competitive pressures in seeking and diluting the non-performance
signaling rents.
According to discussions with engineers, LEED has reduced costs to higher levels of
certification by driving the market, changing norms, and making it easier to pursue certain
credits. Early in the LEED program, it was difficult to achieve credits related to the procurement
and handling of waste because recycled content and recycling construction waste were not
readily available in the market. As market penetration increased (Eichholtz et al. (2013) note an
increase from less than two to near 30 percent of the commercial office market), the availability
of these options has greatly increased. Many of these options are now considered standard best
practices. Many of these basic practices have been adopted as local building codes. For example,
Washington D.C. now requires new buildings over 50,000 square feet to be built at the LEED
Silver level, though does not require certification, and the ASHRAE 90.1-2013 building codes
require basic LEED practices.
These findings highlight the role of the USGBC and conditions present in the real estate
market that at times foster a race to the top, and at times foster a race to the start. Increasing
demand for green real estate, due to pressures from investors, employees, and consumers has
sustained demand for the non-performance signal associated with green building. If demand
were to drop as the market would become crowded due to new entrants, the value of the nonperformance signal would disappear. While the value of the performance signal would remain,
26
Competition and Green Signaling: The Case of LEED the distribution of LEED building points would increasingly reflect the value of the performance
signal, which would be continuous rather than discontinuous at the thresholds.
Conclusion
The multitier labeling structure of LEED encourages firms to invest more to become “greener” in
order to obtain a non-performance signal associated with a higher LEED certification level. Over
time, a greater portion of buildings have been built at higher tiers, while a smaller portion have
been built just above the certification thresholds. While the market has shifted to be “greener,”
the non-performance portion of the certification signal has diminished. Spatial correlation of the
signaling behavior exists at the higher levels of certification, but not at the lowest level of
certification. While the competitive pressures for green building exist across all owner types,
these trends are most pronounced in the government building sector, suggesting a strong role for
politics and policy in driving green building patterns.
27
Competition and Green Signaling: The Case of LEED References:
Auld, G., S. Bernstein and B. Cashore (2008). "The new corporate social responsibility." Annual
Review of Environment and Resources 33: 413-435.
Capozza, D. R. and R. Van Order (1978). "A generalized model of spatial competition." The
American Economic Review: 896-908.
Chegut, A., P. Eichholtz and N. Kok (2011). The Value of Green Buildings: New Evidence from
the United Kingdom, European Center for Corporate Engagement.
Chegut, A., P. Eichholtz and N. Kok (2013). "Supply, Demand and the Value of Green
Buildings." Urban Studies.
Chegut, A., P. Eichholtz and N. Kok (2014). "Supply, Demand and the Value of Green
Buildings." Urban Studies 51(1): 22-43.
Cidell, J. (2009). "Building Green: The Emerging Geography of LEED-Certified Buildings and
Professionals." Professional Geographer 61(2): 200-215.
Cidell, J. and A. Beata (2009). "Spatial Variation Among Green Building Certification
Categories: Does place matter?" Landscape and Urban Planning 91(3): 142-151.
Corbett, C. J. and S. Muthulingam (2007). Adoption of Voluntary Environmental Standards: The
Role of Signaling and Intrinsic Benefits in the Diffusion of the LEED Green Builidng Standards,
UCLA Anderson School of Management.
D'Antonio, P. (2007). Costs and Benefits of Commissioning LEED-NC Buildings. National
Conference on Building Commissioning, Chicago, IL.
Delmas, M. A. and M. J. Montes-Sancho (2010). "Voluntary agreements to improve
environmental quality: symbolic and substantive cooperation." Strategic Management Journal:
575-601.
Delmas, M. A. and M. W. Toffel (2008). "Organizational responses to environmental demands:
Opening the black box." Strategic Management Journal 29(10): 1027-1055.
Eichholtz, P., N. Kok and J. M. Quigley (2010). "Doing Well by Doing Good? Green Office
Buildings." American Economic Review 100(5): 2492-2509.
Eichholtz, P., N. Kok and J. M. Quigley (2010). "Doing well by doing good? Green office
buildings." The American Economic Review 100(5): 2492-2509.
Eichholtz, P., N. Kok and J. M. Quigley (2010). Why Do Companies Rent Green? Real Property
and Corporate Social Responsibility. Berkeley Program on Housing and Urban Policy,
University of California Berkeley.
Eichholtz, P., N. Kok and J. M. Quigley (2013). "The economics of green building." Review of
Economics and Statistics 95(1): 50-63.
Fuerst, F. and P. McAllister (2011). "Eco-labeling in commercial office markets: Do LEED and
Energy Star offices obtain multiple premiums?" Ecological Economics 70(6): 1220-1230.
Kahn, M. E. and R. K. Vaughn (2009). "Green Market Geography: The Spatial Clustering of
Hybrid Vehicles and LEED Registered Buildings." The B.E. Journal of Economic Analysis &
Policy 9(2).
Kahn, M. E. and R. K. Vaughn (2009). "Green Market Geography: The Spatial Clustering of
Hybrid Vehicles and LEED Registered Buildings." B.E. Journal of Economic Analysis & Policy
9(2).
Kleven, H. J. and M. Waseem (2013). "Using notches to uncover optimization frictions and
structural elasticities: Theory and evidence from Pakistan*." The Quarterly Journal of
Economics: qjt004.
28
Competition and Green Signaling: The Case of LEED Kok, N. and M. E. Kahn (2012). The Value of Green Labels in the California Housing Market:
An Economic Analysis of the Impact of Green Labeling on the Sales Price of a Home. San
Fransisco, CA.
Kok, N., M. McGraw and J. M. Quigley (2011). The Diffusion of Energy Efficiency in Building.
UCE3 Working Paper Series. Berkeley, CA, UC Center for Energy and Environmental
Economics.
Majumdar, S. and Y. Zhang (2009). "Market for Green Signaling." The Business Review
Cambridge 13(2): 87-92.
Mason, C. F. (2013). "The Economics of Eco-Labeling: Theory and Empirical Implications."
International Review of Environmental and Resource Economics 6(4): 341-372.
Matisoff, D. C., D. S. Noonan and A. M. Mazzolini (2014). "Performance or Marketing
Benefits? The CASE of LEED Certification." Environmental Science & Technology 48(3):
2001-2007.
Matisoff, D. C., D. S. Noonan and A. M. Mazzolini (2014). "Performance or Marketing
Benefits? The Case of LEED Certification." Environmental Science & Technology.
Mills, E., H. Friedman, T. Powell, N. Bourassa, D. Claridge, T. Haasl and M. A. Piette (2004).
The Cost-Effectiveness of Commercial-Buildings Commissioning, A Meta-Analysis of Existing
Buildings and New Construction in the United States. Lawrence Berkeley National Laboratory,
Portland Energy Conservation Inc., Energy Systems Laboratory, Texas A&M University.
Morris, P. and L. F. Matthiessen (2007). Cost of Green Revisited: Reexamining the Feasibility
and Cost Impact of Sustainable Design in the Light of Increased Market Adoption.
Saha, M. and G. Darnton (2005). "Green Companies or Green Con-panies: Are Companies
Really Green, or Are They Pretending to Be?" Business and Society Review 110(2): 117-157.
Sen, S. and C. B. Bhattacharya (2001). "Does Doing Good Always Lead to Doing Better?
Consumer Reactions to Corporate Social Responsibility." Journal of Marketing Research 38(1):
225-243.
Sexton, S. E. and A. L. Sexton (2014). "Conspicuous conservation: The Prius halo and
willingness to pay for environmental bona fides." Journal of Environmental Economics and
Management 67(3): 303-317.
Sharma, S. and H. Vredenburg (1998). "Proactive Corporate Environmental Strategy and the
Development of Competitively Valuable Organizations Capabilities." Strategic Management
Journal 19(1): 729-753.
Shewmake, S. and W. K. Viscusi (2014). "Producer and Consumer Responses to Green Housing
Labels." Economic Inquiry, Forthcoming.
Shrivastava, P. (1995). "Environmental Technologies and Competitive Advantage." Strategic
Management Journal 16(1): 183-200.
Simcoe, T. and M. W. Toffel (2011). LEED Adopters: Public Procurement and Private
Certification. Boston, Boston University School of Management and Harvard Business School.
Spence, M. (1973). "JOB MARKET SIGNALING." Quarterly Journal of Economics 87(3): 355374.
Turban, D. B. and D. W. Greening (1997). "Corporate Social Performance and Organizational
Attractiveness to Prospective Employees." Academy of Management Journal 40(3): 658-672.
Wood, D. J. (1991). "Corporate Social Performance Revisited." The Academy of Management
Review 16(4): 691-718.
29
Competition and Green Signaling: The Case of LEED Figures and Tables
0.1 Gold Silver Frequency 0.08 0.06 0.04 0.02 Certi&ied Platinum 0 20 24 28 32 36 40 44 48 52 56 60 64 68 LEED building score Figure 1: Observed distribution of LEED building scores.
30
Competition and Green Signaling: The Case of LEED Figure 2: Observed and counterfactual building distributions. Counterfactual distribution constructed with
locally smoothed polynomial. Data labels indicate signal values associated with each LEED building score.
31
Competition and Green Signaling: The Case of LEED CertiGied Silver Gold Platinum Percent of New LEED Buildings 100% 80% 60% 40% 20% 0% ( 35) 2004 ( 74) 2005 ( 97) (140) (236) (412) (589) 2006 2007 2008 2009 2010 (506) 2011 (304) 2012 Year (# New LEED Buildings) Figure 3: Changes in certification level over time.
32
Competition and Green Signaling: The Case of LEED Figure 4: Changes in LEED NC score distribution over time. A locally-smooth polynomial, calculated in the
same fashion as the counterfactual used throughout this work, is superimposed over each histogram.
33
Competition and Green Signaling: The Case of LEED Table 1. Non-performance signaling by subsample (percentage represents excess proportion of buildings built
just above threshold).
Group All Gov Profit Non-­‐Profit Certified At Least Silver At Least Gold Commercial Office Community Dev. Education Health Hotel Industrial Interpretive Center Laboratory Military Base Multi-­‐unit Residence Parks & Recreation Restaurant Retail Stadium Public Order/Safety Other Size Certified 3036 0.08 1472 0.07 860 0.09 704 0.08 590 0.39 2446 1435 962 0.06 50 0.04 694 0.06 136 0.14 28 0.00 92 0.08 55 0.07 119 0.08 83 0.07 149 0.06 55 0.14 58 0.03 158 0.10 19 0.20 205 0.07 147 0.11 % Signaling Silver Gold Platinum Cumulative 3.89 5.83 1.11 10.91 4.08 5.96 1.07 11.17 3.89 5.56 1.11 10.65 3.51 5.90 1.33 10.82 0.39 4.83 7.24 1.38 13.44 12.34 2.35 14.69 3.83 5.40 1.24 10.53 3.17 7.85 2.09 13.15 3.77 6.06 1.33 11.22 2.89 7.11 0.29 10.43 8.93 6.28 1.16 16.37 2.20 5.49 0.00 7.77 3.01 5.30 3.35 11.72 3.88 8.00 1.82 13.78 2.88 6.43 0.48 9.86 3.70 5.36 0.75 9.88 6.64 4.14 0.72 11.64 4.36 8.90 0.69 13.98 3.71 4.35 0.46 8.61 4.11 7.45 1.71 13.46 4.98 5.84 0.47 11.36 4.61 5.71 1.42 11.86 34
Competition and Green Signaling: The Case of LEED Table 2: Non-performance signaling over time. Percentage represents excess portion of buildings certified just above a threshold,
compared to a counterfactual uniquely constructed for each time period.
All Buildings
% Signaling by Certification Level
Time Period
2002-2005
2002-2006
2002-2007
2002-2008
2002-2009
2002-2010
2002-2011
2002-2012
2002-2013
Size
109
206
346
582
994
1583
2089
2393
2492
Certified
2.24
0.10
0.07
0.36
0.20
0.16
0.19
0.12
0.15
Silver
2.94
6.32
5.08
7.95
3.92
2.39
3.46
3.72
4.24
Gold
9.05
7.72
7.32
4.45
7.10
7.17
6.30
6.59
6.37
Platinum
0.00
0.00
1.18
1.93
0.82
0.27
0.84
0.90
0.93
For-Profit Buildings
% Signaling by Certification Level
Time Period
2002-2005
2002-2006
2002-2007
2002-2008
2002-2009
2002-2010
2002-2011
2002-2012
2002-2013
Size
48
77
121
199
369
578
702
758
786
Certified
1.03
0.05
0.03
0.14
0.08
0.06
0.07
0.04
0.06
Silver
1.76
2.92
2.06
2.61
1.29
0.86
1.22
1.20
1.34
Gold
5.64
4.07
3.59
1.87
3.04
2.74
2.09
2.09
2.00
Platinum
0.00
0.00
0.21
0.42
0.22
0.08
0.24
0.24
0.25
Government Buildings
% Signaling by Certification Level
% Signaling
All Thresholds
14.23
14.14
13.65
14.68
12.03
9.99
10.80
11.33
11.69
Size
60
101
163
259
405
634
901
1071
1114
Silver
1.57
3.16
2.06
3.64
1.85
1.05
1.59
1.76
2.00
Gold
6.14
4.27
3.64
1.96
2.65
2.90
2.75
2.92
2.82
Platinum
0.00
0.00
0.40
0.57
0.29
0.10
0.37
0.42
0.42
Non-Profit Buildings
% Signaling by Certification Level
% Signaling
All Thresholds
8.44
7.04
5.90
5.04
4.63
3.74
3.62
3.57
3.65
Certified
0.69
0.04
0.03
0.12
0.08
0.07
0.09
0.05
0.06
Size
29
56
90
152
248
399
514
592
620
Certified
0.69
0.02
0.01
0.10
0.05
0.03
0.04
0.03
0.03
Silver
0.39
1.22
1.51
2.09
0.95
0.54
0.73
0.83
0.95
Gold
0.69
0.99
1.02
0.96
1.66
1.68
1.55
1.67
1.62
Platinum
0.00
0.00
0.56
0.93
0.31
0.09
0.23
0.24
0.26
% Signaling
All Thresholds
8.40
7.48
6.14
6.30
4.87
4.12
4.79
5.14
5.31
% Signaling
All Thresholds
1.77
2.23
3.10
4.09
2.96
2.35
2.55
2.76
2.87
Competition and Green Signaling: The Case of LEED Table 3: Clustering of signaling values in LEED-NC buildings. Building use
Level
Owner
Category
Sample size
All
Government
Profit
Non-profit
Certified
At Least Silver
At Least Gold
Education
Hotel
Restaurant
Retail
Office
Health
Other
* p < 0.1, ** p < 0.05, *** p < 0.01
3,036
1,472
826
738
590
2,446
1,435
693
28
58
158
962
136
1,000
Moran's I
0.1185
0.1325
0.1416
0.2855
0.0266
0.1961
0.4170
0.1579
–0.0290
0.0193
1.2328
–0.0981
0.2283
0.1471
**
***
*
***
***
*
*