El siguiente material se reproduce con fines estrictamente académicos y es para

uso exclusivo de los estudiantes de la Universidad ICESI, de acuerdo con el Artículo

32 de la Ley 23 de 1982. Y con el Artículo 22 de la Decisión 351 de la Comisión del
Acuerdo de Cartagena.

ARTÍCULO 32:

“Es permitido utilizar obras literarias o artísticas o parte de ellas, a título de

ilustración en obras destinadas a la enseñanza, por medio de publicaciones,
emisiones o radiodifusiones o grabaciones sonoras o visuales, dentro de los límites
justificados por el fin propuesto o comunicar con propósito de enseñanza la obra
radiodifundida para fines escolares educativos, universitarios y de formación
personal sin fines de lucro, con la obligación de mencionar el nombre del autor y el
título de las así utilizadas”.

Artículo 22 de la Decisión 351 de la Comisión del Acuerdo Cartagena.

ARTÍCULO 22:

Sin prejuicio de lo dispuesto en el Capítulo V y en el Artículo anterior, será lícito

realizar, sin la autorización del autor y sin el pago de remuneración alguna, los
siguientes actos:

b) Reproducir por medio reprográficos para la enseñanza o para la realización de

exámenes en instituciones educativas, en la medida justificada por el fin que se
persiga, artículos lícitamente publicados en periódicos o colecciones periódicas, o
breves extractos de obras lícitamente publicadas, a condición que tal utilización se
haga conforme a los usos honrados y que la misma no sea objeto de venta o
transacción a título oneroso, ni tenga directa o indirectamente fines de lucro;...”.

• Hyun, S., Park, D., & Tian, S. (2019). Differences between green bonds
versus conventional bonds: An empirical exploration. Handbook of green
finance, sustainable development. Springer, Singapore, 91-134.
Differences Between Green Bonds Versus
Conventional Bonds 7
An Empirical Exploration

Suk Hyun, Donghyun Park, and Shu Tian

Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Green Bonds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Issuers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Investors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Other Players Involved in Green Bonds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Differences Between Green Bonds and Conventional Bonds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
A Green Bond Premium Does Exist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
A Green Bond Premium Does Not Exist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Empirical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Sample Construction Using the Matching Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Empirical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Summary and Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

Abstract
This chapter empirically investigates how the green bond markets price green-
ness. Using the liquidity-adjusted yield premium of green bonds over their
synthetic conventional bonds, this study explores the possible determinants that
drive green bond premiums. Evidence shows that, on average, there is no
significant yield premium or discount on green bonds compared with their paired
conventional bonds. Interestingly, in the case of a subsample, such as the euro and
S. Hyun (*)
East Asia International College, Yonsei University, Wonju, Republic of Korea
e-mail: bigsuk@gmail.com
D. Park · S. Tian
Economic Research and Regional Cooperation Department, Asian Development Bank,
Manila, Philippines
e-mail: dpark@adb.org; stian@adb.org

© Asian Development Bank Institute 2019 127

J. D. Sachs et al. (eds.), Handbook of Green Finance, Sustainable Development,
https://doi.org/10.1007/978-981-13-0227-5_34
128 S. Hyun et al.

the US dollar, to control for the currency effect, most explanatory variables are
not statistically significant, which suggests that a green bond premium does not
exist in dollar-denominated and euro-denominated bonds, while a green bond
premium does exist in other currency-denominated bonds. However, green bonds
are also not standardized instruments. Certain factors, like “greenness,” are
necessary to match the needs of issuers and investors. These factors might have
an impact on the price, liquidity, and volatility of green bonds.

Keywords
Green bond · Green premium · Market development

JEL Classification
G12 · G14 · G19

Introduction

The green bond market has grown rapidly over the past decade. It started with the
“climate awareness bond” issued by the European Investment Bank in 2007 as the
first green bond. A key catalyst for the subsequent market development was the
International Capital Market Association’s (ICMA) introduction of the Green Bond
Principles (GBPs) in January 2014. Although the green bond market has expanded
substantially since then, Ehlers and Packer (2017) showed that it is nevertheless still
very small compared to the wider global bond market, with a share of less than 1.6%
of the global debt issuance in 2016.
While the green bond market began to form in 2007, due to the lack of a globally
accepted definition and standards for green bonds, the nascent market faced two
issues. First, as Ehlers and Packer (2017) pointed out, while the GBPs and the
Climate Bonds Standard (CBS) serve as the general guidelines to distinguish
between green bonds and conventional bonds, due to the lack of enforcement
mechanisms, some countries have produced their own standards for regulating
green bonds. The European Union (2016) stated that the central bank of the People’s
Republic of China (PBoC) suggested its own version of what constitutes green
bonds in 2015 in lieu of the Green Bond Principles and Climate Bonds Standard.
This has led to the issue of heterogeneous data sets on green bonds and the problem
of the generation of green bonds that do not necessarily improve the environment.
Second, the green bond premium, which Zerbib (2017) defined as the difference in
yield between a green bond and an equivalent conventional bond, is not associated
with the impact that green bonds exert on the environment but the sizeable excess
demand. As VanEck (2017) asserted, whether projects are successful or not does not
affect the principal and the interest payment of green bonds.
Therefore, this chapter addresses the above issues as follows. Section “Green
Bonds” is devoted to understanding issuers and investors and the other players
involved in green bonds, since a thorough understanding of them could facilitate
further development of green bond markets and improve policy measurements.
7 Differences Between Green Bonds Versus Conventional Bonds 129

Section “Differences Between Green Bonds and Conventional Bonds” uncovers the
differences between green bonds and conventional bonds and addresses the deter-
minants of a green bond premium. Section “Empirical Analysis” focuses on the
measures and building blocks to develop further green bond markets.

Green Bonds

Green bonds, in short, are fixed-income securities that fund exclusively green pro-
jects with environmental or climate-related benefits. In most countries, any issuer
can label its bonds “green,” because only the People’s Republic of China (PRC) and
India have defined the term legally. In practice, issuers can label bonds “green” only
if they comply with the GBPs, the CBS, or the national guidelines and regulations
governing green bond issuance. Each national standard incorporates many or most
elements of the GBPs; however, there are some differences between them. There-
fore, more coordination is necessary to facilitate further development of green bond
markets. ICMA’s GBPs, which are a voluntary process and guidelines, best describe
the specific criteria and requirements underpinning the concept of green bonds.
Major private financial institutions compiled these under the aegis of ICMA
(2017). The GBPs guide prospective issuers on the four key components of green
bond issuance: (a) the use of proceeds, (b) the process for project evaluation and
selection, (c) the management of proceeds, and (d) reporting. Though external
review is not one of the main four GBPs, the 2015 edition of the GBPs recommended
that green bond issuers “use external assurance to confirm alignment with the key
features of green bonds”. This can include second opinions and verifications. From
2016, the Principles referred to “external reviews” rather than “external assurance,”
while the list of recommended external reviews expanded to include those that rating
agencies provide (ICMA 2017).
Once issuers have issued green bonds in accordance with the GBPs, certification
and scheduled reporting are necessary. If the proceeds from the green bonds do not
fund green projects, as the pre-issuance reports suggest, within 24 months of
issuance, then the issued bonds will lose their green status, according to Petrova
(2016). As investors do not have reliable data and analyses of the impact of green
bonds on the environment, all that investors may use as a yardstick with which to
discern what is green from what is not green is the green bond label.
According to the Asian Development Bank (2018), issuers in 38 jurisdictions had
brought green bonds to the market by the end of June 2017. In many countries, the
green bond issues are very limited in unlocking private investments, except for those
in the most established markets, such as the PRC, the US, and Europe. Although the
green bond markets have rapidly grown over the last 10 years since the European
Investment Bank’s first green bond in 2007, the Climate Bonds Initiative estimated
that $895 billion of climate-aligned bonds were outstanding in 2017, comprising
both labeled and unlabeled green bonds and amounting to $221 billion and $674
billion, respectively, in 2017. The labeled green bonds account for only 24.7% of the
130 S. Hyun et al.

climate-aligned bond universe, while the remaining 75.3% are unlabeled and gen-
erally non-investable, although they also contribute to a low-carbon economy.
Therefore, the lack of a common practice and definition might deter the further
development of green bonds enough to unlock the private sector’s issuance and
investment in green projects. For the green bond market to channel a significant
amount of funds into environment-friendly projects, green bonds should also meet
the needs of both issuers and investors at the same time.

Issuers

Green bonds are no different from conventional bonds, except that their proceeds are
earmarked for investment solely in certain green projects with environmental ben-
efits. Green bonds are also in general less liquid than conventional bonds, since
many investors in green bonds are long-term investors who seek incentives to protect
themselves against inflation risk, default risk, and market volatility. Therefore, most
green bonds offer tax privileges, guarantees, and letters of comfort to attract more
investors (Veys 2010). Although maintaining the green label helps reduce the
environmental risks for the issuer, the costs of providing scheduled reports and
recurring research and development (R&D) expenditure to become green may be
burdensome to small-sized green bond issuers. As the concerns in addressing
environmental risks associated with corporate finance increase, global reporting on
greenhouse gas emissions has become a requirement rather than an option. Standard
& Poor’s Ratings Services (2016) introduced the Carbon Disclosure Project on
behalf of 827 institutional investors, managing $100 trillion in assets, and the United
Kingdom (UK) government’s 2013 requirement for all UK-listed companies for
reporting greenhouse gas emissions.
Considering the same issuer, the risk characteristics of a green bond are essentially
identical to those of a conventional bond: while the proceeds from the issuance of a
green bond are earmarked for environment-friendly projects, the cash flows of the
entire operations of the issuer—not just the green project—service the green bonds.

Investors

Investors in green bonds do not purchase them only because they can hedge against
environmental risks for their portfolios. Investors check the indexes and reports that
the issuers provide to make their investment decision regarding green bonds. Investors
in green bonds are usually institutional investors, such as pension funds, insurance
companies, and so on. Treasurers like to offer green bonds since doing so diversifies
their investor base. When the market is volatile, a diverse portfolio deals better with the
volatility of the market. In many cases, fans of green bonds have no alternative
investment, since green bonds are insufficient to meet the increasing demand. Thanks
to oversubscription in the green bond market along with solid supporters of green
bonds, the pricing of green bonds is tighter than that of conventional bonds.
7 Differences Between Green Bonds Versus Conventional Bonds 131

Furthermore, because there is no widely recognized mechanism to monetize the

environmental impact of green bonds, green bond investors currently take advantage
of the same yields as conventional bonds, with the bonus of the reputation of
investing in green projects that resolve environmental issues. In other words, if
investors are not in need to withdraw cash from green bonds, in the current market
with imperfect information, green bonds compensate investors for their patience.
Investigations into whether a green bond premium exists have produced mixed
empirical results. Investors who purchase green bonds clearly do not do so because
they have a special preference for green bonds over conventional bonds. However,
Östlund (2015) found no evidence of green bond preference. An empirical analysis
with 28 matching pairs would not be highly random, but it is hard to be confident
about the existence of a green bond premium. Zerbib (2017) reported evidence of the
existence of a green bond premium, but the amount of data is limited for green
bonds. In addition, depending on the framing of the data, the results lead to different
conclusions. If investors have green preferences, it would mean that they would
accept a lower return on their investment at the identical risk level as the alternative
to conventional bond if and only if the bond they are investing in is green.
The distinction between conventional investors and responsible investors is an
important factor for green bond markets. In the case of conventional investors, their
investment decision making depends mainly on financial factors, such as profitability
and cash flows. The financial statements of securities issuers provide this information.
On the other hand, the decision-making processes of responsible investors include both
financial and nonfinancial factors, such as ESG (environmental, social, and gover-
nance) factors. Both conventional and responsible investors hold green bonds. How-
ever, green bonds are especially attractive to responsible investors because their
proceeds are limited to environmentally beneficial projects, although the division
between conventional investors and responsible investors is not clear-cut. Given the
absence of institutional investors with responsible investment mandates to guide issuers
and underwriters in structuring green bonds, a comprehensive policy framework for
green bonds is necessary to support the further development of green bond markets.

Other Players Involved in Green Bonds

For green bond issuance, there are four players who play different roles in the market
in addition to issuers and investors. They are the underwriters, the external
reviewers, the index providers, and other market intermediaries. Underwriters are
the financial institutions that set up the public issuance and distribution of the bonds.
Underwriters specify the terms, definitions, and obligations of the bonds. External
reviewers verify the greenness of the underlying projects. Index providers are not
directly involved in the issuance and distribution of the bond; however, they create
green bond indexes according to their own standards. In practice, a widely recog-
nized index provider’s inclusion of a green bond may add an extra inch of reliability
for investors. Lastly, other market intermediaries are optional in a way, since most
green bonds are traded over the counter (Östlund 2015).
132 S. Hyun et al.

Rating agencies and third-party auditors generally check the financial soundness
of bond issuers or projects. However, when it comes to green bonds, their main
objective is to check whether green bonds are environmentally responsible and
observe the eligibility criteria that they pledged to abide by at the time of issuance
(ADB 2018). Rating agencies, such as Moody’s, investigate green bonds and
climate-related bonds and assign quantitative scores to them based on their green-
ness. Both Moody’s and S&P Global Ratings use a transparent scoreboard approach
to provide forward-looking opinions about the issuer’s effectiveness in meeting the
green criteria. In both cases, green bonds receive a grade, which assesses their
greenness. The green bond ratings are not credit ratings, which the agencies issue
separately. In the PRC, the China Chengxin International Credit Rating Company
has also developed a green bond methodology, which follows the GBP approach and
assigns one of five ratings to green bonds. While only a handful of issuers have
sought green ratings, both Moody’s and S&P Global Ratings have made efforts to
develop their green bond assessment capabilities expecting that all issuers will look
for both a credit and a green rating soon.
The following section will analyze the differences between green bonds and
conventional bonds more thoroughly. These characteristics have implications for
the pricing of green bonds and their attractiveness to investors. A premium at
issuance over comparable bonds without a green label would indicate that a signif-
icant number of investors value the label, enough to give issuers an extra incentive to
issue bonds that have it. At the same time, these investors will still be interested in an
acceptable financial performance of green bonds over time.

Differences Between Green Bonds and Conventional Bonds

Apart from the fact that green bonds require scrutiny by observing the GBPs and
strive to meet the requirements that the Climate Bonds Initiative suggested, VanEck
(2017) and Östlund (2015) insisted that they are not particularly different from
conventional bonds, especially in primary markets, where brokers must sell green
bonds to a large pool of investors buying both green bonds and conventional bonds.
Standard & Poor’s Ratings Services (2016) showed that green bonds’ trading occurs
near the yield spreads of conventional bonds. However, in secondary markets,
investors also consider the green bond label as a criterion for selecting bonds.
Therefore, the only way to assure investors that the bonds they are buying are
truly green bonds is for them to check the binding guidelines regulating green bonds’
issuance and maintenance. The PRC, for example, strongly limits the use of the
proceeds of green bonds to the environmental projects of the issuers only. In fact, the
country has two guidelines for overseeing green bonds: China’s Green Bond Guide-
lines and the PBoC’s Green Bond Endorsed Project Catalogue, according to the
European Union (2016). In Norway, on the other hand, for green bonds to maintain
their label, their issuers need to list them on Oslo’s Stock Exchange. The European
Union (2016) introduced Norway’s example of how to run green bonds. Norway
also requires green bonds to seek a second opinion for listing on the stock exchange.
7 Differences Between Green Bonds Versus Conventional Bonds 133

The issuer must make the result of the second opinion available to the public. Lastly,
the only certified green bond in Mexico underwent a second review in accordance
with the CBS. Whether a green bond premium exists or not has been a topic of fierce
discussion among economists. The chapter introduces the existing analyses of the
green bond premium to scrutinize the issue further.

A Green Bond Premium Does Exist

To investigate the existence of a green bond premium, Zerbib (2017) used a matching
method, comparing each eligible green bond with two similar conventional bonds with
identical conditions, such as currency, rating, bond structure, seniority, collateral, and
coupon type. He started out with 681 green bonds complying with the GBPs on
30 December 2016; however, after filtering out the outliers and incomplete data, he selected
only 135 investment grade senior bullet fixed-coupon bonds for analysis. Zerbib (2017)
aimed to estimate the green bond premium and to identify its determinants. He found that
the average green bond premium was –8bps against conventional bonds within the whole
sample of investment grade bonds, –5bs in US dollar bonds and –2bps in euro bonds. He
drew attention to the presence of an excess demand for green bonds in the market.
Ehlers and Packer (2017) compared the credit spreads at issuance of a cross-
section of 21 green bonds issued between 2014 and 2017 with the credit spreads at
issuance of conventional bonds of the same issuers with the closest possible issue
date. They showed that green bond issuers on average have borrowed at lower
spreads than they have through conventional bonds. Their findings confirmed the
results from other recent studies, such as Barclays (2015) and Zerbib (2017).
Wulandari et al. (2018) investigated the relationship between the liquidity risk and the
yield spread for both green and conventional bonds. The evidence showed a positive
relationship between the liquidity risk and the yield spread. However, for green bonds,
the impact of liquidity risk on the yield spread has become negligible over time. Barclays
(2015) indicated in across-sectional analysis a–17 bps premium as of mid-2015. How-
ever, the historical returns of green bonds are like those of conventional bonds.

A Green Bond Premium Does Not Exist

Östlund (2015) used a data set of 28 matching pairs of bonds from Bloomberg on
17 March 2015, focusing on the essence of a green bond premium defined as the
spread differentials between green bonds and conventional bonds of the same issuer.
The analysis tests the null hypothesis that there are no differences between green
bonds and conventional bonds. The results showed that indeed there is no evidence
of a green bond discount in the overall dataset. However, using the data for the year
2015, the mean difference was statistically significant at the level of 10%, and the
yield spread was 0.0758, indicating that conventional bonds have lower yields than
green bonds. The only case in which a green bond premium was apparent involved
the six green bonds related to the real estate industry.
134 S. Hyun et al.

Petrova (2016) conducted both panel regression analysis and timeseries analysis
to evaluate and compare the performance of green bonds and conventional bonds
during a sample period covering 2008–2016. After controlling for various possible
factors, such as default risk and term premium, and different time-series parameters,
no statistically significant difference between green bond and conventional bond
yields was documented.

Empirical Analysis

Sample Construction Using the Matching Method

The matching method is a useful and simple technique as a model-free approach to

analyzing the specificity of a financial instrument by matching a pair of instruments
with the same characteristics except for the one characteristic of interest from the
point of view of its effects. In line with Zerbib (2017), the authors construct a
database to evaluate the yield spread between a green bond and an equivalent
conventional bond of the same issuer. For this purpose, for each pair, we match a
green bond to a conventional bond with identical characteristics except for its
liquidity. We, therefore, define the difference between the green bond yield and the
equivalent conventional bond yield as the so-called “green bond premium.”
We collect conventional bonds with a maturity that is neither 2 years shorter nor
2 years longer than the green bond’s maturity, because the maturities cannot be
exactly equal. The difference between the two categories of bonds is their liquidity,
which is possible to measure from either the issued amount or the issuance date of
the benchmark. A substantial difference in liquidity can considerably affect the yield
level; therefore, we must control for it. We restrict the eligible conventional bonds to
those with an issued amount of less than four times the issued amount of green bonds
and greater than one-quarter of this amount to ensure a fair approximation. We also
restrict the range of conventional bonds to those with an issue date that is 6 years
earlier or 6 years later than the green bond’s issue date.
We examine the entire sample of 990 green bonds complying with the GBPs on
10 November 2017. This set includes bonds of various kinds: supranational,
sub-sovereign, and agency, municipal, corporate, financial, and utilities. In the first
step in building the equivalent conventional bond, for each green bond, we search for
the conventional bond with the closest maturity from the same issuer, with the same
characteristics: they all have the same currency, rating, bond structure, seniority,
collateral, and coupon type. Among the 158 green bonds stemming from the
selection process following Zerbib (2017), we focus on 123 remaining green
bonds with an investment grade senior bullet fixed coupon, while we select the
123 equivalent conventional bonds from the same issuers among 121,777 observa-
tions. The 123 green bonds of the sample are the ones for which we managed to find
7 Differences Between Green Bonds Versus Conventional Bonds 135

Figure 1: Issuers by Sector Utilities, Others, 2,262.8

(Unit: $ Million). (Source: 8,755.0 , 11% , 3%
Bloomberg)
Financial,
16,065.3 , 20%
Government,
51,564.4 , 66%

Government Financial Utilities Others

an equivalent conventional bond. The issuers are mainly among the most active bond
issuers in the market.
In the second step, we eliminate the maturity bias by building a data set consisting
of pairs of bonds: we assign one conventional bond with the closest characteristics to
each green bond. We retrieve the bid and ask yields of the green bonds and the
corresponding conventional bonds from Bloomberg on 10 November 2017. Since
this study focuses on the investors’ demand and the issuers’ supply of green bonds,
we focus on the ask yields of each bond to ensure a more precise analysis.

Descriptive Statistics

Considering the green bond issuance by sector in our data set by using the BICS
level 1 of the Bloomberg classification, Figure 1 shows that the government and the
finance sectors are the major issuers of green bonds. Although corporate (others)
activity has grown recently, it is still limited.
Table 1 presents the descriptive statistics of all the green bonds and conventional
bonds in our data set and definitions of variables are listed in Table A1. All the
statistics show no statistically significant differences between the 123 pairs of green
bonds and conventional bonds except for the amount issued. For example, while the
mean of the issue amount of green bonds is $639 million, the mean of the issue
amount of conventional bonds is $1.18 billion. Considering the issue amount by
sector, significant variations are observable in the issue amount of the government
and the finance sectors. However, statistically significant differences are not apparent
in the case of utilities and other sectors in Tables A2, A3, A4 and A5 in the appendix.
Statistically significant differences in amount issued are apparent in the case of the
US dollar and the euro as seen in Tables A6–A7 in the appendix. The difference in
the issue amounts can affect the liquidity of green bonds and conventional bonds.
Therefore, it is necessary to consider the impact of the difference between them on
the liquidity to calculate the green bond premium.
 136

Table 1: Descriptive Statistics

Variables N Mean Median Std Dev. Min. Max. Mean difference (P-value)
AskYLD (%) 123 1.30 0.85 1.72 –1.22 12.60 0.9422
AskYLD_M (%) 123 1.32 0.89 1.72 –0.71 12.19
BidYLD (%) 123 1.37 0.90 1.74 –0.56 13.07 0.9757
BidYLD_M (%) 123 1.38 0.94 1.73 –0.50 12.42
BidAskSP (%) 123 –0.07 –0.05 0.11 –1.14 0.00 0.3974
BidAskSP_M (%) 123 –0.06 –0.05 0.04 –0.24 0.00
TimeToMat (days) 123 1,535 1,328 926 18 4,944 0.8115
TimeToMat_M (days) 123 1,506 1,299 965 75 5,176
CPN (%) 123 1.72 1.63 1.40 0.00 8.50 0.9798
CPN_M (%) 123 1.73 1.50 1.30 0.00 8.00
AmtIssued ($) 123 639,410,207 500,000,000 529,038,534 27,531,000 3,499,500,000 <.0001
AmtIssued_M ($) 123 1,180,293,338 816,550,000 1,331,281,622 23,940,000 7,353,425,000
Source: Authors.
S. Hyun et al.
7 Differences Between Green Bonds Versus Conventional Bonds 137

Table 2: Average Issue Amount by Currency

Average issue amounts
(in $ million)
Currency Green bonds Conventional bonds
AUD 402.15 631.95
CAD 1222.18 3942.50
CHF 351.40 306.22
EUR 937.47 1589.21
GBP 1846.60 4468.11
INR 191.25 260.10
MXN 109.83 277.19
SEK 168.18 216.94
TRY 71.17 284.68
USD 624.44 1266.00
Mean 592.47 1324.29
Median 376.78 469.09
AUD Australian dollar, CAD Canadian dollar, CHF Swiss franc, EUR euro, GBP British pound,
INR Indian rupee, MXN Mexican peso, SEK Swedish krona, TRY Turkish lira, USD US dollar.
Source: Authors.

Table 2 indicates the average amount of green bonds and conventional bonds
issued in each currency. It shows that the average amounts of green bonds and
conventional bonds issued are statistically different, which suggests the existence of
different levels of liquidity. For example, while an average issue amount for green
bonds is $624 million, that of conventional bonds is $1.26 billion. In the case of the
euro, while the average amount issued for green bonds is $937 million, that of
conventional bonds is $1.59 billion. Therefore, when we specify the empirical
model, we also consider the impact of currency.

Empirical Analysis

Table 3 shows each green bond’s average yield with time, depending on the rating
and the currency of issuance. The average euro green bond yields are much lower
than those of emerging countries, such as Turkish lira(TRY), Mexican pesos(MXN),
or Indian rupees (INR). Significant variations are observable in the yield levels,
mainly between the various issue currencies.
The first step in our calculation of the green bond premium is to determine
whether a green bond is cheaper (discount) or more expensive (premium) than an
equivalent conventional bond. Based on the above statistics, the difference in yield
between a green bond and the equivalent conventional bond might have a liquidity
bias that results from different amounts issued in comparison with other character-
istics. To correct the impact of the different amounts issued of green bonds and
conventional bonds, we design a proxy variable as the liquidity difference between a
138 S. Hyun et al.

Table 3: Rating and Yield by Currency

Average ask yield to conventional (%)
Green bonds Conventional bonds
Currency AAA AA A BBB AAA AA A BBB
AUD 2.01 2.69 1.95 2.72
CAD 2.13 2.04
CHF –0.25 –0.14
EUR –0.14 0.21 0.11 0.69 –0.10 0.30 0.20 0.72
GBP 0.66 0.72
INR 5.69 6.71 5.84 6.72
MXN 6.76 7.33
SEK 0.01 0.20 0.47 0.94 0.06 –0.02 0.41 0.42
TRY 12.60 12.19
USD 2.04 2.12 2.61 2.44 2.05 2.13 2.63 2.51
Mean 3.26 1.30 1.33 2.69 3.32 1.28 1.32 2.59
Median 2.01 1.16 1.30 1.69 1.95 1.21 1.23 1.62
AUD Australian dollar, CAD Canadian dollar, CHF Swiss franc, EUR euro, GBP British pound,
INR Indian rupee, MXN Mexican peso, SEK Swedish krona, TRY Turkish lira, USD US dollar.
Source: Authors.

Table 4: Results of the Step 1 Regression (OLS)

Dependent variable: Difference_AskYLD
Liquidity 0.496**
(2.432)
Intercept –0.011
(–0.631)
N 123
Adjusted R2 0.0387
T statistics in parenthesis calculated from robust standard errors. ***, **, and * denote significance
at the 1%, 5%, and 10% levels, respectively.
Source: Authors.

green bond and a conventional bond in line with Zerbib (2017), who used the spread
between the ask yield and the bid yield as a liquidity indicator. We, therefore, define
the green bond premium as the regression residual unexplained by the liquidity
difference between green bonds and conventional bonds. The bid–ask spread proxy
that we use to control for the difference in liquidity proves to be statistically
significant as seen in Table 4.
Figure 2 shows that the green bond premium is concentrated around zero, and a
low standard deviation is observable. This indicates that the first liquidity control on
the amounts issued and the date of issuance in the data construction procedure yields
satisfactory results. The distribution of the green bond premium ranges from –0.49%
to +0.54%, with a mean and a median value of near 0 bps (basis point) and –73.17
bps, respectively, and 51% of the premiums in our sample are negative.
 7

Premium
18
16
16 16
14
12
12
10 11 11
8
8 8
6 7
4 5 5
4 4 4
2 3 3
0 2 0 2 0 0 2 0
0
Differences Between Green Bonds Versus Conventional Bonds

0
0.1
0.2
0.3
0.4
0.5
0.6

-0.5
-0.4
-0.3
-0.2
-0.1
0.05
0.15
0.25
0.35
0.45
0.55

-0.45
-0.35
-0.25
-0.15
-0.05
Figure 2: Distribution of Green Bond Premiums. (Source: Authors)
139
140 S. Hyun et al.

In the second step, we attempt to identify the determinants of the green bond
premium. Table 5 shows each green bond’s average yield with time, depending on
the rating and the currency of issuance and the amount issued. We base the
estimation on the structural part of the curve (yield) and the specific characteristics
of each bond. We construct the following specification in which the impacts of the
rating, the maturity, the issued amount, and the group to which the bond belongs are
the same in all currencies. Specifically, we express the maturity in years and the
issued amount in US dollar with the reference date of 10 November 2017 and the
exchange rate that applied on that date. We use the level 1 Bloomberg classification
(BICS level 1) for the sector group. Government includes municipalities; regional
and sovereign agencies; and national, supranational, and multilateral development
banks. In addition, Utilities and Financial encompass nonpublic banks and financial
services. The specification considers nonlinearities, such as Issued Amount2i and
Maturity2i , based on Zerbib (2017).
The empirical results that we obtain for the whole sample show that the structural
part of the equation (yield) does not play a significant role; the yield is not statisti-
cally significant, while the issued amount is statistically significant in all the spec-
ifications. The issued amount negatively impacts the green bond premium, in
contrast to the results of Zerbib (2017). The larger the issued amount is, the lower
the green bond premium becomes. Financial bonds show a 6.9–15.5 bps premium
below the reference level (government group). However, in the case of a subsample,
such as the euro and US dollar in Tables A8 and A9, to limit the currency effect, most
explanatory variables are not statistically significant, which suggests that a green
bond premium does not exist in US dollar-denominated and euro-denominated
bonds, respectively, in Tables 6 and 7 while a green bond premium does exist in
other currency-denominated bonds.
The negative green bond premium highlights the fact that the buying pressure
relative to the supply capacity is greater in the case of green bonds than in the case of
conventional bonds. These market phenomena result from an excessive investment
demand due to the characteristic of green bonds and an insufficient volume of bond
issuances. The negative green bond premium leads to a favorable situation for green
bond issuers, which prioritizes a green project instead of a conventional project, due
to a lower cost of capital. Nevertheless, this situation poses two problems. First, it
reveals the structural lack of green projects. Second, a negative green bond premium
subdues the appetite of institutional investors that have not set up a binding floor for
green assets in their strategic asset allocation.

Summary and Policy Implications

Green bonds significantly contribute to enhancing green investments, which aim to

mitigate and adapt to climate change (CICERO 2015). The green bond markets have
substantially grown over the past decade, although the market size is still insignif-
icant compared with that of conventional bonds. This paper attempts to analyze the
 7

Table 5: Results of the Step 2 Regression (OLS): Whole Sample

Dependent variable: green bond premium (residual terms of the step 1 regression)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
AskYLD 0.010 0.009
(0.927) (0.860)
TimeToMaturity (years) 0.032 0.046**
(1.404) (2.061)
TimeToMaturity_Square (years) –0.003 –0.004**
(–1.393) (–2.042)
AmtIssued ($ billion) –0.203** –0.188** –0.188** –0.170** –0.170*
(–2.477) (–2.212) (–2.228) (–2.000) (–1.954)
AmtIssued_Square 0.055* 0.052* 0.047 0.047 0.050
(1.793) (1.665) (1.504) (1.509) (1.583)
Rating_AA 0.075 0.111** 0.050 0.087* 0.104**
(1.634) (2.357) (1.063) (1.811) (2.090)
Rating_A –0.011 0.123* –0.011 0.119* 0.119*
(–0.241) (1.776) (–0.226) (1.720) (1.719)
Rating_BBB 0.026 0.142* 0.034 0.129* 0.134*
(0.452) (1.931) (0.595) (1.751) (1.830)
Group_Utilities –0.089 –0.184** –0.045 –0.142 –0.146
Differences Between Green Bonds Versus Conventional Bonds

(–1.284) (–2.089) (–0.637) (–1.585) (–1.638)

Group_Financials –0.069* –0.149** –0.072* –0.147** –0.155**
(–1.680) (–2.462) (–1.755) (–2.450) (–2.580)
Intercept –0.013 –0.066 0.092** –0.018 0.024 –0.018 0.071 0.109*** 0.063 –0.048
(–0.561) (–1.215) (2.398) (–0.676) (1.112) (–0.692) (1.537) (2.737) (1.362) (–0.697)
N 123 123 123 123 123 123 123 123 123 123
R2 0.0071 0.0164 0.0589 0.0279 0.0312 0.0876 0.0727 0.0838 0.1227 0.1595
Adjusted R2 –0.0012 0.0000 0.0432 0.0034 0.0151 0.0486 0.0331 0.0528 0.0693 0.0845
T statistics in parenthesis calculated from robust standard errors. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
141

Source: Authors.
Table 6: Results of the Step 2 Regression (OLS): US Dollar Sample
142

Dependent variable: green bond premium (residual terms of the step 1 regression)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
AskYLD 0.046 0.005
(0.921) (0.030)
TimeToMaturity (years) 0.058 0.053
(1.629) (1.038)
TimeToMaturity_Square (years) –0.005 –0.005
(–1.524) (–1.220)
AmtIssued ($ billion) –0.285 –0.259 –0.176 –0.130 –0.077
(–0.605) (–0.534) (–0.370) (–0.263) (–0.142)
AmtIssued_Square 0.080 0.057 0.014 –0.003 –0.038
(0.302) (0.207) (0.051) (–0.010) (–0.125)
Rating_AA 0.011 0.038 –0.007 0.019 0.009
(0.136) (0.485) (–0.095) (0.231) (0.093)
Rating_A 0.002 0.127 –0.018 0.091 0.069
(0.035) (1.288) (–0.256) (0.893) (0.376)
Rating_BBB –0.043 0.046 –0.066 0.013 0.040
(–0.549) (0.447) (–0.827) (0.121) (0.207)
Group_Utilities –0.071 –0.101 –0.083 –0.086 –0.066
(–0.586) (–0.657) (–0.687) (–0.556) (–0.360)
Group_Financials –0.082 –0.165* –0.085 –0.142 –0.121
(–1.314) (–1.776) (–1.353) (–1.483) (–0.993)
Intercept –0.112 –0.114 0.131 –0.006 0.010 –0.006 0.139 0.115 0.084 –0.036
(–0.984) (–1.637) (0.742) (–0.170) (0.345) (–0.172) (0.759) (0.649) (0.454) (–0.101)
N 45 45 45 45 45 45 45 45 45 45
R2 0.0193 0.0600 0.0600 0.0091 0.0438 0.0836 0.0765 0.1070 0.1285 0.1678
Adjusted R2 –0.0035 0.0153 0.0152 –0.0634 –0.0017 –0.0339 –0.0419 0.0177 –0.0364 –0.0770
S. Hyun et al.

T statistics in parenthesis calculated from robust standard errors. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Source: Authors.
 7

Table 7: Results of the Step 2 Regression (OLS): Euro Sample

Dependent variable: green bond premium (residual terms of the step 1 regression)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
AskYLD –0.005 –0.079
(–0.105) (–0.624)
Time To Maturity (years) –0.003 0.006
(–0.093) (0.112)
Time To Maturity_Square (years) 0.000 0.000
(0.137) (0.121)
AmtIssued ($ billion) 0.013 –0.037 0.024 0.025 0.040
(0.094) (–0.229) (0.149) (0.127) (0.174)
AmtIssued_Square –0.006 0.003 –0.011 –0.011 –0.015
(–0.140) (0.059) (–0.236) (–0.212) (–0.244)
Rating_AA –0.012 –0.012 –0.042 –0.018 –0.008
(–0.164) (–0.143) (–0.426) (–0.163) (–0.066)
Rating_A –0.047 –0.033 –0.073 –0.040 –0.007
(–0.674) (–0.267) (–0.824) (–0.285) (–0.048)
Rating_BBB 0.013 0.054 –0.007 0.046 0.120
(0.154) (0.467) (–0.070) (0.334) (0.659)
Group_Utilities –0.046 –0.070 –0.052 –0.072 –0.097
Differences Between Green Bonds Versus Conventional Bonds

(–0.593) (–0.599) (–0.607) (–0.537) (–0.672)

Group_Financials –0.030 –0.005 –0.034 –0.005 –0.001
(–0.523) (–0.044) (–0.523) (–0.043) (–0.011)
Intercept –0.045 –0.042 –0.050 –0.028 –0.028 –0.028 0.022 –0.034 –0.032 –0.098
(–1.626) (–0.466) (–0.588) (–0.531) (–0.737) (–0.520) (0.155) (–0.315) (–0.186) (–0.439)
N 41 41 41 41 41 41 41 41 41 41
R2 0.0003 0.0009 0.0010 0.0205 0.0124 0.0348 0.0275 0.0158 0.0377 0.0518
Adjusted R2 –0.0254 –0.0517 –0.0516 –0.0590 –0.0396 –0.1031 –0.1114 –0.0935 –0.1664 –0.2643
Source: Authors.
143
144 S. Hyun et al.

green bond premium, which is defined as the yield difference between a green bond
and an equivalent conventional bond after controlling for the difference in liquidity
by using a matching method in line with Zerbib (2017). The main objective of the
article is to determine the value of the green bond premium and to identify its major
determinants.
Our empirical results indicate that on average there is no significant premium for
green bonds relative to conventional bonds. The empirical results that we obtain for
the whole sample show that the yield is not statistically significant while the issued
amount is statistically significant in all the specifications. The issued amount nega-
tively impacts the green bond premium, in contrast to the results of Zerbib (2017).
The larger the issued amount is, the lower the green bond premium becomes.
Financial bonds show a 6.9–15.5 bps premium below the reference level (govern-
ment group). Interestingly, in the case of a subsample, such as the euro and US
dollar, to control for the currency effect, most explanatory variables are not statis-
tically significant, which suggests that a green bond premium does not exist in US
dollar-denominated and euro-denominated bonds while a green bond premium does
exist in other currency-denominated bonds.
Like conventional bonds, green bonds are not standardized instruments, and
certain factors, like “greenness,” are necessary to match issuers’ and investors’
needs. These factors might impact the price, liquidity, and volatility of green
bonds. Therefore, for further research to construe the pricing mechanism of green
bonds, it should incorporate the definition and measurement of “greenness” into the
empirical analysis with a larger data sample. One limitation of this study is also due
to the quality of the data; since green bonds are not frequently traded, a bond yield
sometimes does not accurately reflect its fair value.
Going green does not mean sacrificing the yield because some institutional
investors are not willing to pay a green bond discount. The excessive demand, rather
than the environmental impacts, has driven the current premium. Essentially, to
make green bonds more attractive and desirable than conventional bonds, issuers
should monetize externalities like the environmental benefits generated from green
projects for investors in green bonds. Yoshino and Abidhadjaev (2016) insist that it
is possible to measure the spillover effects from the tax increment that infrastructure
projects generate through the difference-in-difference method, which the issuer can
internalize as an additional return for investors and not by increasing the issuer’s
burden.
On the demand side, a sufficiently large, committed demand from institutional
investors would lower the cost of capital for green projects. There is a good
opportunity for issuers to expand their funding capacity for green projects, thanks
to ethical investors who are willing to sacrifice some of the yield to participate in
environment-friendly projects. The urgent need to finance the change to a
low-carbon economy makes green bonds more attractive and essential investments
based on ESG (environmental, social, and governance) criteria. More green bonds
7 Differences Between Green Bonds Versus Conventional Bonds 145

would help economies meet the huge green investment needs to cope with climate
change and mitigation.
The supply of green bonds should increase to provide the huge amount of green
investments required for climate change. The European Union (2016) pointed out
five main bottlenecks in the supply of green bonds that issuers need to address to
remove the hindrances to green bond market development: (a) the lack of green
bonds and green project pipelines, (b) the lack of aggregation mechanisms for
green projects, (c) the lack of green bond definitions and framework, (d) the lack of
information and market knowledge, and (e) the lack of a clear risk profile for green
investments.
The lack of commonly recognized standards in the green bond market especially
has limited the universe of institutional investors, who are ethically mandated to
invest in green bonds since the standardization of green bonds’ definition and
framework is ongoing. However, external reviewers and issuers are cooperating to
establish more up-to-date methods to track the proceeds from the green bonds and
their impact on the environment. The ASEAN Capital Market Forum (2017) has
recently developed the ASEAN Green Bond Standards based on ICMA’s GBPs,
which are internationally accepted and widely used for the development of national
green bond guidelines or standards issued globally. The expectation is that this
regional standardization and harmonization will facilitate the further development
of the green bond market.

Appendix

Table A1: Variable Definition

Variables Definition of variables
AskYLD (%) Ask yield of the GB (green bond)
AskYLD_M (%) Ask yield of the matched CB (conventional bond)
BidYLD (%) Bid yield of the GB (green bond)
BidYLD_M (%) Bid yield of the matched CB (conventional bond)
BidAskSP (%) AskYLD - BidYLD (bid–ask spread of GB)
BidAskSP_M (%) AskYLD_M - BidYLD_M (bid–ask spread of CB)
TimeToMat (days) Time to maturity of the GB (green bond)
TimeToMat_M (days) Time to maturity of the matched CB (conventional bond)
CPN (%) Coupon of the GB (green bond)
CPN_M (%) Coupon of the matched CB (conventional bond)
AmtIssued ($) Amount issued of the GB (green bond)
AmtIssued_M ($) Amount issued of the matched CB (conventional bond)
Source: Authors’ calculation from Bloomberg.
 146

Table A2: Descriptive Statistics for Government

Variables N Mean Median Std dev. Min. Max. Mean difference (P–value)
AskYLD (%) 78 1.27 0.74 1.87 –1.22 12.60 0.9954
AskYLD_M (%) 78 1.27 0.73 1.86 –0.71 12.19
BidYLD (%) 78 1.35 0.80 1.89 –0.56 13.07 0.9561
BidYLD_M (%) 78 1.33 0.77 1.88 –0.50 12.42
BidAskSP (%) 78 –0.08 –0.05 0.14 –1.14 0.00 0.3647
BidAskSP_M (%) 78 –0.06 –0.05 0.05 –0.24 0.00
TimeToMat (days) 78 1,546 1,297 1,002 18 4,944 0.7753
TimeToMat_M (days) 78 1,499 1,263 1,038 75 5,176
CPN (%) 78 1.63 1.48 1.42 0.00 8.50 0.7684
CPN_M (%) 78 1.57 1.44 1.27 0.00 8.00
AmtIssued ($) 78 661,081,865 500,000,000 617,546,321 27,531,000 3,499,500,000 0.0005
AmtIssued_M ($) 78 1,351,253,776 897,037,500 1,586,196,618 23,940,000 7,353,425,000
Source: Authors.
S. Hyun et al.
 7

Table A3: Descriptive Statistics for Financial

Variables N Mean Median Std dev. Min. Max. Mean difference (P-value)
AskYLD (%) 31 1.15 0.59 1.19 –0.13 3.20 0.8496
AskYLD_M (%) 31 1.20 0.63 1.19 –0.19 3.27
BidYLD (%) 31 1.19 0.65 1.17 –0.08 3.26 0.8510
BidYLD_M (%) 31 1.24 0.69 1.17 –0.13 3.29
BidAskSP (%) 31 –0.04 –0.05 0.03 –0.10 0.00 0.8573
BidAskSP_M (%) 31 –0.04 –0.05 0.03 –0.10 0.00
TimeToMat (days) 31 1,433 1,431 696 183 3,175 0.9131
TimeToMat_M (days) 31 1,452 1,383 685 280 3,139
CPN (%) 31 1.61 1.13 1.09 0.13 4.00 0.5812
CPN_M (%) 31 1.77 1.80 1.05 0.38 3.60
AmtIssued ($) 31 518,236,129 583,250,000 206,812,705 101,745,000 1,000,000,000 0.0011
Differences Between Green Bonds Versus Conventional Bonds

AmtIssued_M ($) 31 931,997,903 874,875,000 618,570,693 95,760,000 2,000,000,000

Source: Authors.
147
 148

Table A4: Descriptive Statistics for Utilities

Variables N Mean Median Std dev. Min. Max. Mean difference (P-value)
AskYLD (%) 9 1.44 0.85 2.13 –0.21 6.71 0.9443
AskYLD_M (%) 9 1.52 0.97 2.13 –0.36 6.72
BidYLD (%) 9 1.49 0.90 2.14 –0.15 6.81 0.9373
BidYLD_M (%) 9 1.57 1.02 2.14 –0.22 6.84
BidAskSP (%) 9 –0.05 –0.05 0.03 –0.10 0.00 0.6357
BidAskSP_M (%) 9 –0.06 –0.05 0.05 –0.14 0.00
TimeToMat (days) 9 1,622 1,369 1,029 21 3,084 0.9613
TimeToMat_M (days) 9 1,596 1,635 1,243 203 3,125
CPN (%) 9 2.20 1.85 2.05 0.38 7.38 0.8909
CPN_M (%) 9 2.33 1.50 2.01 0.38 7.25
AmtIssued ($) 9 972,772,222 874,875,000 465,366,847 306,000,000 1,633,100,000 0.9677
AmtIssued_M ($) 9 961,691,233 816,550,000 660,241,491 306,000,000 2,333,000,000
Source: Authors.
S. Hyun et al.
 7

Table A5: Descriptive Statistics for Others

Variables N Mean Median Std dev. Min. Max. Mean difference (P-value)
AskYLD (%) 5 2.53 2.92 1.21 0.94 4.09 0.9392
AskYLD_M (%) 5 2.47 2.97 1.41 0.42 4.08
BidYLD (%) 5 2.58 2.92 1.16 1.03 4.09 0.9339
BidYLD_M (%) 5 2.52 2.97 1.36 0.51 4.08
BidAskSP (%) 5 –0.05 0.00 0.07 –0.16 0.00 0.9511
BidAskSP_M (%) 5 –0.05 0.00 0.07 –0.14 0.00
TimeToMat (days) 5 1,838 1,650 917 1,222 3,432 0.9340
TimeToMat_M (days) 5 1,788 1,577 954 1,042 3,432
CPN (%) 5 2.93 2.88 0.94 1.75 4.38 0.9110
CPN_M (%) 5 2.85 2.88 1.11 1.25 4.38
AmtIssued ($) 5 452,560,000 500,000,000 156,816,770 179,550,000 583,250,000 0.9605
Differences Between Green Bonds Versus Conventional Bonds

AmtIssued_M ($) 5 446,226,000 500,000,000 227,422,977 47,880,000 600,000,000

Consumer discretionary = 3, consumer staples = 1, materials = 1.
Source: Authors.
149
 150

Table A6: Descriptive Statistics for US Dollar

Variables N Mean Median Std dev. Min. Max. Mean difference (P-value)
AskYLD (%) 45 2.21 2.08 0.50 1.58 4.09 0.8345
AskYLD_M (%) 45 2.23 2.14 0.51 1.51 4.08
BidYLD (%) 45 2.24 2.13 0.50 1.65 4.09 0.8506
BidYLD_M (%) 45 2.26 2.19 0.50 1.59 4.08
BidAskSP (%) 45 –0.03 –0.04 0.03 –0.12 0.00 0.7101
BidAskSP_M (%) 45 –0.03 –0.04 0.03 –0.09 0.00
TimeToMat (days) 45 1,376 1,189 959 21 3,560 0.8351
TimeToMat_M (days) 45 1,335 1,006 936 203 3,432
CPN (%) 45 2.03 1.95 0.60 1.00 4.38 0.9513
CPN_M (%) 45 2.03 2.00 0.65 0.88 4.38
AmtIssued ($) 45 624,444,444 500,000,000 276,673,056 300,000,000 1,500,000,000 <.0001
AmtIssued_M ($) 45 1,266,000,000 1,000,000,000 924,709,782 300,000,000 4,000,000,000
Source: Authors.
S. Hyun et al.
 7

Table A7: Descriptive Statistics for Euro

Variables N Mean Median Std dev. Min. Max. Mean difference (P-value)
AskYLD (%) 41 0.17 0.10 0.54 –1.22 1.76 0.5808
AskYLD_M (%) 41 0.24 0.22 0.54 –0.71 1.77
BidYLD (%) 41 0.25 0.15 0.50 –0.56 1.91 0.6785
BidYLD_M (%) 41 0.30 0.25 0.54 –0.50 1.91
BidAskSP (%) 41 –0.08 –0.05 0.17 –1.14 –0.02 0.5025
BidAskSP_M (%) 41 –0.06 –0.05 0.04 –0.24 –0.01
TimeToMat (days) 41 1,957 1,762 1,032 18 4,944 0.8960
TimeToMat_M (days) 41 1,987 1,852 1,090 87 5,176
CPN (%) 41 0.86 0.75 0.68 0.05 2.75 0.5682
CPN_M (%) 41 0.96 0.75 0.79 0.00 2.88
AmtIssued ($) 41 937,467,683 583,250,000 636,035,428 291,625,000 3,499,500,000 0.0144
Differences Between Green Bonds Versus Conventional Bonds

AmtIssued_M ($) 41 1,589,211,490 1,166,500,000 1,523,190,544 349,950,000 5,832,500,000

Source: Authors.
151
Table A8: Results of the Step 2 Regression (OLS): US Dollar Sample
152

Dependent variable: green bond premium (residual terms of the step 1 regression)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
AskYLD 0.046 0.005
(0.921) (0.030)
TimeToMaturity (years) 0.058 0.053
(1.629) (1.038)
TimeToMaturity_Square (years) –0.005 –0.005
(–1.524) (–1.220)
AmtIssued ($ billion) –0.285 –0.259 –0.176 –0.130 –0.077
(–0.605) (–0.534) (–0.370) (–0.263) (–0.142)
AmtIssued_Square 0.080 0.057 0.014 –0.003 –0.038
(0.302) (0.207) (0.051) (–0.010) (–0.125)
Rating_AA 0.011 0.038 –0.007 0.019 0.009
(0.136) (0.485) (–0.095) (0.231) (0.093)
Rating_A 0.002 0.127 –0.018 0.091 0.069
(0.035) (1.288) (–0.256) (0.893) (0.376)
Rating_BBB –0.043 0.046 –0.066 0.013 0.040
(–0.549) (0.447) (–0.827) (0.121) (0.207)
Group_Utilities –0.071 –0.101 –0.083 –0.086 –0.066
(–0.586) (–0.657) (–0.687) (–0.556) (–0.360)
Group_Financials –0.082 –0.165* –0.085 –0.142 –0.121
(–1.314) (–1.776) (–1.353) (–1.483) (–0.993)
Intercept –0.112 –0.114 0.131 –0.006 0.010 –0.006 0.139 0.115 0.084 –0.036
(–0.984) (–1.637) (0.742) (–0.170) (0.345) (–0.172) (0.759) (0.649) (0.454) (–0.101)
N 45 45 45 45 45 45 45 45 45 45
R2 0.0193 0.0600 0.0600 0.0091 0.0438 0.0836 0.0765 0.1070 0.1285 0.1678
Adjusted R2 –0.0035 0.0153 0.0152 –0.0634 –0.0017 –0.0339 –0.0419 0.0177 –0.0364 –0.0770
Source: Authors.
S. Hyun et al.
 7

Table A9: Results of the Step 2 Regression (OLS): Euro Sample

Dependent variable: green bond premium (residual terms of the step 1 regression)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
AskYLD –0.005 –0.079
(–0.105) (–0.624)
TimeToMaturity (years) –0.003 0.006
(–0.093) (0.112)
TimeToMaturity_Square (years) 0.000 0.000
(0.137) (0.121)
AmtIssued ($ billion) 0.013 –0.037 0.024 0.025 0.040
(0.094) (–0.229) (0.149) (0.127) (0.174)
AmtIssued_Square –0.006 0.003 –0.011 –0.011 –0.015
(–0.140) (0.059) (–0.236) (–0.212) (–0.244)
Rating_AA –0.012 –0.012 -0.042 –0.018 –0.008
(–0.164) (–0.143) (–0.426) (–0.163) (–0.066)
Rating_A –0.047 –0.033 –0.073 –0.040 –0.007
(–0.674) (–0.267) (–0.824) (–0.285) (–0.048)
Rating_BBB 0.013 0.054 –0.007 0.046 0.120
(0.154) (0.467) (–0.070) (0.334) (0.659)
Group_Utilities –0.046 –0.070 –0.052 –0.072 –0.097
Differences Between Green Bonds Versus Conventional Bonds

(–0.593) (–0.599) (–0.607) (–0.537) (–0.672)

Group_Financials –0.030 –0.005 –0.034 –0.005 –0.001
(–0.523) (–0.044) (–0.523) (–0.043) (–0.011)
Intercept –0.045 –0.042 –0.050 –0.028 –0.028 –0.028 0.022 –0.034 –0.032 –0.098
1.626) (–0.466) (–0.588) (–0.531) (–0.737) (–0.520) (0.155) (–0.315) (–0.186) (–0.439)
N 41 41 41 41 41 41 41 41 41 41
R2 0.0003 0.0009 0.0010 0.0205 0.0124 0.0348 0.0275 0.0158 0.0377 0.0518
Adjusted R2 –0.0254 –0.0517 –0.0516 –0.0590 –0.0396 –0.1031 –0.1114 –0.0935 –0.1664 –0.2643
Source: Authors.
153
154 S. Hyun et al.

References
ASEAN Capital Market Forum (2017) ASEAN green bond standards. ACMF Secretariat, Malaysia
Asian Development Bank (ADB) (2018) Promoting green local currency bonds for infrastructure
development in ASEAN+3. ADB, Mandaluyong
Barclays (2015) The cost of being green. Barclays, New York
CICERO (2015) Background report on long-term climate finance prepared for the German G7
presidency 2015 by CICERO and Climate Policy Initiative, Germany
Ehlers T, Packer F (2017) Green bond finance and certification. BIS Quarterly Review, September
2017. https://ssrn.com/abstract=3042378
European Union (2016) Study on the potential of green bond finance for resource-efficient
investments. Luxembourg
International Capital Market Association (ICMA) (2017) The green bond principles 2017. ICMA,
Paris
Östlund E (2015) Are investors rational profit maximizers or do they exhibit a green preference?
evidence from the green bond market. Stockholm School of Economics master’s thesis in
economics (21875), Stockholm
Petrova A (2016) Green bonds: lower returns on higher responsibility? Radboud University master
thesis (S4461746), Nijmegen
Standard & Poor’s Ratings Services (2016) The corporate green bond market fizzes as the global
economy decarbonizes. New York
VanEck (2017) What drives green bond returns? Market Realist. https://www.vaneck.com
Veys A (2010) The sterling bond markets and low carbon or green bonds. E3G
Wulandari F, Dorothea S, Andreas SC (2018) Liquidity risk and yield spreads of green bonds. DIW,
Berlin
Yoshino N, Abidhadjaev U (2016) Impact of infrastructure investment on tax: estimating spillover
effects of the Kyushu high-speed rail line in Japan on regional tax revenue. ADBI working paper
574. Asian Development Bank Institute, Tokyo
Zerbib OD (2017) The green bond premium. SSRN working paper. Zurich. https://papers.ssrn.com/
sol3/papers.cfm?abstract_id=2889690