PreprintPDF Available

Facts and Fantasies about the Green Bond Premium

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

The purpose of this study is to determine if investors are rewarded with lower yields when they invest in Green Bonds in the secondary market. To make the assessment of whether this premium exists we consider green bonds belonging to an established green bond index, we make use of two methodologies: an intra-curve method at the security level and a matching method at the asset class level. We show that on average there is a statistically significant overall negative premium on green bonds compared to their associated regular bonds, and that the premium is more pronounced in certain fixed income sub-segments. We also run a panel regression with control variates to investigate the determinants of the premium. After neutralizing the liquidity factor in the search of the premium, we find an even lower premium.
Content may be subject to copyright.
Facts and Fantasies about
the Green Bond Premium
Mohamed Ben Slimane
Quantitative Research
Amundi Asset Management, Paris
mohamed.benslimane@amundi.com
Dany Da Fonseca
Credit Portfolio Management
Amundi Asset Management, Paris
dany.dafonseca@amundi.com
Vivek Mahtani
Alpha Fixed-income Solution
Amundi, London
vivek.mahtani@amundi.com
November 2020
Abstract
The purpose of this study is to determine if investors are rewarded with lower yields when
they invest in Green Bonds in the secondary market. To make the assessment of whether this
premium exists we consider green bonds belonging to an established green bond index, we
make use of two methodologies: an intra-curve method at the security level and a matching
method at the asset class level. We show that on average there is a statistically significant
overall negative premium on green bonds compared to their associated regular bonds, and
that the premium is more pronounced in certain fixed income sub-segments. We also run a
panel regression with control variates to investigate the determinants of the premium. After
neutralizing the liquidity factor in the search of the premium, we find an even lower premium.
Keywords: Green bond, risk premium, credit rating, liquidity, environmental, ESG.
JEL classification: C23, G12, Q56
The authors are grateful to Thierry Roncalli, Takaya Sekine, Jean-Marie Dumas and Herve Boiral for their
helpful comments.
1
Facts and Fantasies About the Green Bond Premium
1 Introduction
Green Bonds (GB) are fixed income securities whose proceeds are earmarked exclusively for new
and existing projects with environmental benefits focused on renewables, energy efficiency, water,
clean transport, and climate change mitigation and adaptation. They are part of the broader uni-
verse of socially responsible investments, which include bonds and equities from issuers identified
by environmental, social and governance (ESG) standards.
According to OECD (2016), a green bond is differentiated from a regular bond by commitment
to use the funds raised to finance or refinance green projects, assets, or business activities. As it
is an emerging financial instrument, there is not yet a commonly accepted definition of a green
bond. Green bond investors face the issue of distinguishing a genuine commitment on the part
of the issuer to use the proceeds in a greenly way from a simple greenwashing1. That said, green
bond issuance, having started in June 2007 with a EUR 600 million Climate Awareness Bond
issued by the European Investment Bank, has seen explosive growth in the recent years to more
than USD 950 billion in outstanding issues as of September 20202, thanks to the progress made
on standards and especially to the introduction of the Green Bond Principles (GBP) in 2014.
The GBP are the most widely accepted standards3and represent voluntary procedural guide-
lines developed by key market participants and put forward by the Zurich-based International
Capital Market Association (ICMA). They consist of four core components that recommend dis-
closure and transparency and promote integrity in the development of the green bond market.
These components are Use of proceeds, Management of proceeds, Process for project evaluation
and selection and Reporting. They define the subjects of the projects financed by the proceeds and
the management of these proceeds, which should be tracked in an appropriate manner through a
transparent process. For instance, Green bond proceeds should be credited to a sub-account that
is financially separate from other business accounts, so all transactions can be easily identified.
The GBP also provide guidance on the provision of information by green bond issuers regarding
the process for project evaluation and selection. Reporting, the fourth component of the GBP,
specifies the ongoing information that the issuer should provide after the issuance of a green bond
regarding the status and the use of the proceeds. Issuers of green bonds are also encouraged to
obtain an external review such as a second opinion, a rating or a certification rating4(Ehlers
and Packer,2017) to provide an objective assessment of the project’s compliance with the GBP,
thereby reducing the asymmetry of information between borrowers and investors. For instance,
certifications under the Climate Bond Standard (Almeida et al.,2019) administered by the non-
profit organization Climate Bonds Initiative (CBI), certify the full alignment with the GBP and
the goals of the Paris Climate Agreement to limit global warming to under 2 degrees.
1Flammer et al. (2018) shows how green bonds may allow a firm to portray an environmentally responsible
public image without actually doing so. In a similar way, Schmuck et al. (2018) noticed how misleading advertising
about the environmental features of product affects how consumers perceive ads and brands.
2Environmental Finance, https://www.bonddata.org/.
3Along with the Climate Bond Standard of the Climate Bond Initiative (CBI). However, China and the Eu-
ropean Union have taken an interest in developing their own standards as well. China, has drawn up its Green
Bond Endorsed Project Catalogue, and the European Union is in the process of developing the EU Green Bond
Standards(EU Technical Expert Group on Sustainable Finance,2019)
4The external reviewers guarantee the sustainable use of proceeds at issuance, however only certification
providers monitor practices post-issuance.
2
Facts and Fantasies About the Green Bond Premium
From the issuer’s point of view, it seems clear that issuing a green bond is more expensive than
a regular bond, given the costs of possible external review, regular reporting and holding separate
accounts for the proceeds. Hachenberg and Schiereck (2018) report that the costs entailed are
estimated at between 0.3 and 0.6 basis points for a USD 500 million issue. Even through CBI, a
certification costs 0.1 bps5. From the investor’s perspective, the question that arises is whether
the green label influences the price that investors are willing to pay for a bond, that is, whether
investors are willing to accept a lower yield spread for a green bond relative to a conventional
bond with the same characteristics.
Throughout the paper, we will use the term premium, to refer to the excess yield on the bond
due to its green characteristic6. At first sight, there is no fundamental rationale for the green label
to influence the yield of a Green Bond. Green bonds rank pari-passu with bonds with the same
rank and issuer. The Green Bonds holder does not own any additional right on the underlying
projects and is subject to the same market dynamics. A green premium for the issuer is therefore
somewhat of a market anomaly7.
In this paper, we try to answer the question on the secondary market for bonds belonging to
an established green bond index using two different methodologies: a top-down approach using
the index and a bottom-up method focusing on the individual green bonds in the index. After a
review of related literature in Section 2, Section 3 describes the data we use in the pricing analysis.
Section 4 and 5 introduce the methodologies cited above while Section 6 discusses the results and
offers some conclusions and implications.
2 Literature review
Academics have studied the prices of GBs from different angles and periods, either at issuance or
in the secondary market, relying on a set of bonds or on special types of bonds (for instance EUR
denominated or US municipals8). Regressions of the yields (or the yield spreads) or analysis of
the yield curves are the main methods used and are often performed with the support of matching
processes. The matching, i.e. pairing the green bond with its regular equivalent that has similar
bond price determinants, allows the green label effect to be isolated as the differences in the pricing
of the two bonds can stem only (presumably) from this different determinant. The findings show
contrasting evidence and do not offer a definitive answer. Some studies report GBs trading at
a negative premium (i.e. at lower yields) than regular bonds. Others document no significant
difference in yields or even higher yields for GBs. Table 1displays 18 recent studies. For each of
them, we report the universe of green bonds, the type of market, the period of observations, the
final number of green bonds after a potential data cleaning or matching, the method used and the
findings. Mixed results are presumably attributable to differences in time periods, samples, and
methodologies.
5https://www.climatebonds.net/certification/get-certified.
6In the case of a negative premium, this implies giving up yield.
7For ease of reading, we will use interchangeably the terms green bond premium and green premium.
8US Municipal bonds are issued in series at the same time by the same issuer, with the same official statement
and use of proceeds covering each series of bonds, this aspect of the muni market give the unique opportunity to
compare the green bonds with their direct vanilla equivalents.
3
Facts and Fantasies About the Green Bond Premium
Table 1: Overview of GB pricing in the literature
Study Market #GBs Universe Period Method Premium estimate
Bachelet et al. (2019) Secondary 89 Global 2013 - 2017 OLS model 2.1 to 5.9 bps
Baker et al. (2018) Secondary 2 083 US Municipals 2010 - 2016 OLS model -7.6 to -5.5 bps
19 US Corporates 2014 - 2016
Bour (2019) Secondary 95 Global 2014 - 2018 Fixed effects model -23.2 bps
Ehlers and Packer (2017) Primary 21 EUR & USD 2014 - 2017 Yield comparaison -18 bps
Fatica et al. (2019) Primary 1 397 Global 2007 - 2018 OLS model
Gianfrate and Peri (2019)Primary 121 EUR 2013 - 2017 Propensity score matching -18 bps
Secondary 70 – 118 3 dates in 2017 -11 to -5 bps
Hachenberg and Schiereck (2018) Secondary 63 Global August 2016 Panel data regression Not significant
Hyun et al. (2020) Secondary 60 Global 2010 - 2017 Fixed effects GLS model Not significant
Kapraun and Scheins (2019)Primary 1 513 Global 2009 - 2018 Fixed effects model -18 bps
Secondary 769 +10 bps
Karpf and Mandel (2018) Secondary 1 880 US Municipals 2010 - 2016 Oaxaca-Blinder decomposition +7.8 bps
Larcker and Watts (2019) Secondary 640 US Municipals 2013 - 2018 Matching & Yield comparaison Not significant
Lau et al. (2020) Secondary 267 Global 2013 - 2017 Two-way Fixed effects model -1.2 bps
Nanayakkara and Colombage (2019) Secondary 43 Global 2016 - 2017 Panel data with hybrid model -62.7 bps
Ostlund (2015) Secondary 28 Global 2011 - 2015 Yield comparaison Not significant
Partridge and Medda (2018)Primary 521 US Municipals 2013 - 2018 Yield curve analysis -4 bps
Secondary Small but below 0
Preclaw and Bakshi (2015) Secondary Index Global 2014 - 2015 OLS model -16.7 bps
Schmitt (2017) Secondary 160 Global 2015 - 2017 Fixed effects model -3.2 bps
Zerbib (2019) Secondary 110 Global 2013 - 2017 Fixed effects model -1.8 bps
Source: this information is retrieved from the discussed articles.
4
Facts and Fantasies About the Green Bond Premium
Only two of the above studies (Ehlers and Packer,2017;Fatica et al.,2019) are fully dedicated
to the primary market. Mainly, due to the fact that primary market yields express a market
price at time t, which can be influenced by the imbalance of demand and supply, most studies
focus on the secondary markets that signal the stability of premia and also indicate windows of
opportunity to issue new bonds. Kapraun and Scheins (2019) examine both primary and secondary
market effects and find that green bonds listed on the London and Luxembourg secondary markets
with a dedicated green bond segment are traded on average 7 bps lower9. This highlights that
issuers benefit from the reduction of information asymmetry on the secondary market, which will
undeniably influence primary market yields.
As can be noted, most cited studies find negative premia whose estimates are close10 to zero.
Lau et al. (2020) find that a relatively large premium tends to suffer from a sample being too small
or biased, a yield comparison without a sound matching process or a lack of controls for bond
features. For instance, as mentioned by Baker et al. (2018), the positive yield spread of 7.8 bps
by Karpf and Mandel (2018) is a result of neglecting the effect of taxation in the US municipal
securities market. According to Zerbib (2019), the low premia “emphasize the low impact of
investors’ pro-environmental preferences on bond prices, which does not represent, at this stage,
a disincentive for investors to support the expansion of the green bond market.
The studies, using matching procedures, control for bond risk factors such as default and curve
risks but not all of them control for the liquidity risk. This risk arises because the green bonds
market is smaller and less liquid than the conventional bond market. For instance, Zerbib (2019),
Bour (2019) and Hyun et al. (2020) define the premium as the residual part of the difference in
spreads after controlling for the difference in liquidity proxy.
Five studies point out the importance of external reviews in the pricing of green bonds.
Kapraun and Scheins (2019) (resp. Fatica et al. (2019)) find investors are more likely to pay
a premium (i.e. accept lower yields) for corporate (resp. non-financial) green bonds when they
are certified as such by a third party. Kapraun and Scheins (2019) even mention the term of cred-
ibility in the green label. Fatica et al. (2019) show that among the financials, only institutions
that have declared a clear commitment to environmental principles (i.e. those subscribing to the
United Nations Environment Programme Financial Initiative) issued green bonds at a premium.
Although Larcker and Watts (2019) do not document a significant overall premium, they find
that green bonds carrying CBI Certifications exhibit lower premia. Baker et al. (2018) assess that
the negative premium doubles or even triples for GBs that are externally certified and publicly
registered with the CBI. Hyun et al. (2020) estimate that GBs enjoy 7 bps discount if they have
an external reviewer and 9 bps if they obtain a CBI certification.
Along with credibility, reputation is relevant. Kapraun and Scheins (2019) find that premia are
observed in both primary and secondary markets only when bonds are issued by governments or
supranational entities, denominated in EUR or USD, or corporate bonds with very large issue sizes.
These bonds and their issuers can be considered as more credible in terms of a better potential
implementation or a greater impact of the financed green project. Fatica et al. (2019) show that
green bonds issued by repeat issuers benefit from an additional negative premium compared to
9In the full sample, they report mixed results: a negative premium in the primary market and a positive premium
in the secondary market.
10If we except Nanayakkara and Colombage (2019).
5
Facts and Fantasies About the Green Bond Premium
those issued by one-time issuers in the green market.
3 Data
We define as green bonds all bonds that are both self-labelled as green bonds or sustainability
bonds by their issuer and are part of the Bloomberg Barclays MSCI Global Green Bond Index,
hereafter denoted as the “Green Index”. The independent screening conducted by MSCI guar-
antees that all bonds in the studied universe comply with the basic requirements of most ESG
investors on the definition of green bonds such as
the proceeds being exclusively and formally applied to projects or activities that promote
climate or other environmental sustainability purposes11 ;
the bonds complying with the four dimensions set by the Green Bond Principles;
at least 90% of the use of proceeds falling within at least one of seven eligible environmental
categories defined by MSCI ESG Research (alternative energy, energy efficiency, pollution
prevention and control, sustainable water, green building, climate adaption, and other)12 ,
with operational or research & development expenses excluded.
In addition to the robustness of the green screening, the mentioned index also ensures that we
are focusing on Green Bonds of a minimum size and other characteristics that guarantee that they
are tradable and as such quoted with consistent market prices. The bonds enter the index after
issuance and not before month-end, which ensures that when included in our studied universe, the
bonds have been trading for some time in the secondary market. All the post issuance effects on
prices (like issuance premium/concession, allocation adjustments) have then been mostly removed.
Figure 1: Breakdowns per sector & currency
Agencies
CMBS
Covered
Financials
Industrials
Local Authorities
Sovereign
Supranational
Treasury
Utilities
0
10
20
30
40
50 Benchmark Green portfolio
(a) Sectors
AUD CAD CHF CNY EUR GBP JPY KRW SEK USD
0
10
20
30
40
50
60
70 Benchmark Green portfolio
(b) Currencies
The “Green Index” consists of 532 green bonds as of 25 September 2020. These green bonds
are or have been in the Bloomberg Barclays Global Aggregate Bond Index13 (hereafter referred to
11Bloomberg Barclays MSCI Global Green Bond Index factsheet.
12ibid.
13The Bloomberg Barclays Global Aggregate Bond Index is a flagship measure of global investment grade debt
from twenty-four local currency markets.
6
Facts and Fantasies About the Green Bond Premium
as the “benchmark”) before falling below their minimum maturity requirement of one year14. In
our study, we discard bonds with less than 1-year to maturity from the green index to form our
rebased portfolio of green bonds (referred to as the “green portfolio”).
Figure 1illustrates the breakdowns per sector and currency of both the “benchmark” and “green
portfolio” as of 25 September 2020. We note in Figure 1a that the “green portfolio” tends to
have disproportionately more supranationals, agencies, financials, and utilities. We have reported
in Figure 1b the weight of currencies with at least 0.50% presence in the “benchmark” or “green
portfolio”. We note that the “green portfolio” is mainly EUR-denominated (more than 65%
compared to 20% in “benchmark”). Around 20% is USD-denominated (compared to 40% in the
“benchmark”) and the remainder is denominated in 11 other currencies.
Figure 2: Breakdowns per maturity & rating
1 - 3 yrs 3 - 5 yrs 5 - 7 yrs 7 - 10 yrs 10 - 20 yrs >20 yrs
0
5
10
15
20
25 Benchmark Green portfolio
(a) Maturity
Aaa Aa A Baa
0
5
10
15
20
25
30
35
Benchmark Green portfolio
(b) Rating
In Figure 2, we report the breakdowns per maturity and rating. We note that the green issuance
is relatively higher between 5 years and 20 years compared to “benchmark” and especially on the
10-20 years bucket. The ratings are balanced between the first three categories. Compared to the
“benchmark”, the “green portfolio” is overweighted in Aa and Baa rated bonds.
4 First method: Top-down approach
We first consider a method which should be of particular interest to a macro focused investor
who is concerned with the cost or benefit of green bonds in the context of a pure top down fixed
income allocation. We have noted that there are significant compositional differences between
conventional indices and green indices hence we cannot simply make “off-the-shelf” comparisons.
We therefore adopt the same approach as Fender et al. (2019) to compare two matched indices.
We dissect the portfolio of GBs into each currency c, sector s, quality q, and maturity m. We build
then a synthetic conventional portfolio from the “benchmark” that matches the same dissection
by applying the weights of the green index.
14The “Green Index” does not have a 1-year minimum time to maturity and will hold bonds until final maturity.
The inclusion of green bonds to maturity is designed to accommodate this market practice by not forcing unwanted
turnover.
7
Facts and Fantasies About the Green Bond Premium
Here, we define the premium as the excess weighted average OAS15 on our universe of green bonds
versus the “benchmark” which has been re-weighted to match the characteristics of currency,
sector, quality and maturity. We make use of the Barclays Level 2 sector field where 12 categories
are available16. For quality, we classify bonds according to their category of ratings AAA, AA, A
and BBB. Regarding the maturity, bonds are split into 6 buckets17.
The premium is the weighted excess spread. Indeed, we write
P remium =OASGOASB
=
N
X
i=1
ωGiOASGi
N
X
i=1
ωGiOASBi
=
N
X
i=1
ωGi(OASGiOASBi)
where Nis the number of quadruplets Qiof (Currency, Sector, Quality, Maturity), ωGiis the
weight of Qiin the “green portfolio”, OASGi(respectively OASBi) is the spread over govies of the
Qiin the “green portfolio” (respectively the “benchmark”).
4.1 Evolution of the green premia & returns
In what follows, we use the “global bond index” to refer to the synthetic index built from the
“benchmark” where the weights are matched with those of the “green portfolio” according to the
available quadruplets of (currency, sector, quality, maturity).
Table 2: OAS & Premium
OAS Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
Green portfolio 73.41 14.77 55.13 71.13 132.88 1.75 4.74 4.97 ***
Global bond index 78.07 15.57 58.10 75.96 143.36 1.97 5.92 5.01 ***
Premium -4.66 1.37 -10.47 -4.40 -2.54 -2.03 6.49 -3.40 ***
Note: * p 0.1; ** p 0.05 ; *** p 0.01
In Table 2, we report different statistics of the OAS of the “global bond index” and “green
portfolio” and the premium over the last 4 years18. Premium is negative, has a mean of 4.66
and has evolved between 10.47 and 2.54. The metrics of skewness and kurtosis indicate the
presence of outliers to the left of the distribution.
Figure 3a displays the change in both OAS over the studied period. Both OAS widened in 2018
and 2020 and peaked during the recent turmoil. On the other hand, Figure 3b shows that the
premium reached its minimum during the recent Covid-19 crisis before reverting to its mean.
15OAS, the acronym for Option Adjusted Spread, is the constant spread above the treasury curve that compen-
sates for credit and liquidity risks but excludes the premium for the option risk.
16They are Financials, Corporates, Sovereigns, Utilities, Local Authorities, Agencies, Government, ABS, CMBS,
Supranationals, MBS and Covered.
17They are 1 3 yrs, 3 5 yrs, 5 7 yrs, 7 10 yrs, 10 20 yrs and >20yrs.
18From September 2016 to September 2020
8
Facts and Fantasies About the Green Bond Premium
Figure 3: Global OAS & Premium
2017-01
2017-07
2018-01
2018-07
2019-01
2019-07
2020-01
2020-07
60
80
100
120
140
OAS (bps)
Global bond Index
Green portfolio
(a) OAS
2017-01
2017-07
2018-01
2018-07
2019-01
2019-07
2020-01
2020-07
10
9
8
7
6
5
4
3
Premium (bps)
Premium
mean
(b) Premium
Table 3: Durations
Duration Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
Green portfolio 7.41 0.70 5.80 7.35 8.56 -0.62 0.35 10.55 ***
Global bond index 7.21 0.57 5.57 7.36 8.09 -1.72 2.64 12.60 ***
∆ OAD 0.19 0.32 -0.27 0.07 0.76 0.61 -1.21 0.60
We make use of the OAD19 to calculate the duration of the bonds in the “global bond index”
and the “green portfolio” and we measure the duration of both portfolios as the weighted average
OAD. Table 3shows that on average the duration of the “green portfolio” is +0.19 years longer.
Figure 4: Global OAD & ∆ OAD
2017-01
2017-07
2018-01
2018-07
2019-01
2019-07
2020-01
2020-07
5.5
6.0
6.5
7.0
7.5
8.0
8.5
OAD (in years)
Global bond Index
Green portfolio
(a) OAD
2017-01
2017-07
2018-01
2018-07
2019-01
2019-07
2020-01
2020-07
0.2
0.0
0.2
0.4
0.6
0.8
OAD (in years)
OAD
mean
(b) ∆ OAD
Table 4reports the metrics of excess returns20 of both portfolios. Although, lagging in term of
returns, the “green portfolio” exhibits the same Sharpe ratio and thus a relative lower volatility.
19Option Adjusted Duration or effective duration refines the duration of bonds that contain call features by
incorporating the probability of issuers exercising their call options.
20Excess return is the return in excess of the total return of a risk-matched basket of governments or interest rate
9
Facts and Fantasies About the Green Bond Premium
Negative skewness and high kurtosis indicate the presence of outliers on the leftmost side of the
distributions of returns. We note that these metrics are relatively low for the “green portfolio”.
Table 4: Excess returns - Statistics
Portoflio Excess return (%) Std dev.(%) Sharpe Skewness Kurtosis
Green portfolio 1.08 2.59 0.42 -3.01 15.53
Global bond index 1.15 2.70 0.42 -3.13 16.39
Figure 5a shows the yearly excess returns of the “global bond index” and the “green portfolio”
and Figure 5b reports the differences in yearly returns between these two portfolios. The “green
portfolio” outperformed in 2018 and 2020, two years of elevated risk aversion and extreme uncer-
tainty. While in 2020, the Covid-19 outbreak caused global economic disruption and the largest
global recession since the Great Depression, investors saw a significant market sell-off in 2018 when
concerns about the slowdown in the global economy, the ramping up of trade tensions between the
US and China, and the unexpected tightening of the FED monetary policy caused major market
jitters. The first crisis manifested most notably during the last thirteen weeks of 2018 whereas the
second crisis manifested strongly between February and April 2020. We showed above the OAS
spikes during these two periods.
Figure 5: Outperformances
2016 2017 2018 2019 2020
2
1
0
1
2
3
4
Performance (%)
Global bond Index
Green portfolio
(a) Yearly returns
2016 2017 2018 2019 2020
50
40
30
20
10
0
10
20
30
Outperformance (bps)
(b) Outperformance
We report in Figure 6the weekly outperformances of the “green portfolio” and emphasize the crisis
periods using dashed boxes. We note (Figure 6a) that in 8 weeks out of 13, the green portfolio
has outperformed by 6 bps on average peaking at 14 bps in the midst of the 2018 crisis. On the
other hand, Figure 6b shows that during the first six weeks of the Covid-19 crisis that began in
mid-February, the green portfolio has outperformed 6 weeks out of 6 by 4.6 bps on average peaking
again at the height of the crisis.
swaps, thus neutralizing the interest rate and yield curve risk and isolating the portion of performance attributed
solely to credit and optionality risks.
10
Facts and Fantasies About the Green Bond Premium
Figure 6: Weekly Outperformances in bps
(a) 2018 returns (b) 2020 returns
The findings on the behavior of the “green portfolio” during the 2018 and 2020 crises are strength-
ened by the payoff charts displayed in Figure 7. Here, the outperformance21 of the “green portfolio”
is plotted against the “global bond index” performance where one point corresponds to one month
of data (Figure 7a) or one week of data (Figure 7b). To illustrate the payoff, a local regression line
is added. The put profile of these payoffs, reinforces the belief that the green portfolio has a very
interesting contrarian feature in market downturns. This result is consistent with the findings of
Silva and Cortez (2016) and Nofsinger and Varma (2014). Silva and Cortez (2016), who evaluate
the performance of green mutual funds invested globally, find that the green funds performed
worse than the benchmark. However, their performance increased in crisis periods compared to
non-crisis periods. Nofsinger and Varma (2014) report that the outperformance of socially re-
sponsible funds during periods of market crises compared to matched conventional mutual funds
comes at the cost of underperforming during non-crisis periods. The nature of green bonds issuers
and investors may be put forward to explain the resilience during crises. Issuers of green bonds
have historically been large, such as development banks, with established governance structures
and thus better armed to cope with crises. Secondly, green bond investors include a large share
of buy-and-hold investors, such as pension funds and insurance companies (Cochu et al.,2016),
which are unlikely to move from green investments in a crisis.
21We show in Appendix A.5 that the outperformance is driven by the difference in spread durations.
11
Facts and Fantasies About the Green Bond Premium
Figure 7: Payoffs
4321 0 1
Global index performance (%)
15
10
5
0
5
10
15
20
Green index outperformance (bps)
(a) Monthly returns
1.51.00.5 0.0 0.5 1.0
Global index performance (%)
15
10
5
0
5
10
15
Green index outperformance (bps)
(b) Weekly returns
4.2 Breakdown of the Green premium
In what follows, we report the breakdown of green premia per sector, per currency, per rating and
per maturity bucket. First, we define the broken-down premium as
P remium (c, s, q, m) = X
iQB
ωGi(OASGiOASBi)
X
iQB
ωGi
where QB={(currencyj, sectorj, qualityj, maturityj)Q:currencyj=csectorj=s
qualityj=qmaturityj=m}22
4.2.1 Currency
Table 5: Currencies - 2016 – 2020
Currency Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
All currencies -7.30 2.39 -13.16 -7.82 -3.41 -0.45 -0.28 -3.06 ***
EUR -7.30 2.39 -13.16 -7.82 -3.41 -0.45 -0.28 -3.06 ***
USD -0.98 5.26 -21.33 -0.93 10.82 -0.95 4.09 -0.19
GBP -9.28 9.94 -40.52 -6.49 2.55 -1.09 0.81 -0.93
CAD 0.05 6.34 -12.00 0.00 13.41 0.16 -0.63 0.01
AUD -1.07 2.14 -5.99 -0.69 3.06 -0.43 -0.50 -0.50
The Euro, which is the main currency of the “green portfolio”, has a negative premium with
22Obviously, if one characteristic is omitted, its condition is omitted too.
12
Facts and Fantasies About the Green Bond Premium
an average of 7.30 and standard deviation of 2.39. Table 5shows that this premium is only
significant compared to the premia of other currencies including the US Dollar.
Figure 8: EUR & USD premia (in bps) – 2016 - 2020
2017-01 2017-07 2018-01 2018-07 2019-01 2019-07 2020-01 2020-07
20
15
10
5
0
5
10 All currencies
EUR
USD
Figure 8shows the trend of EUR and USD premia since September 2016. The EUR premium
fluctuated in a negative range between 13 bps and 3 bps. The impact of the 2018 crisis is more
striking than the impact of the recent crisis. The USD premium, which is twice as volatile as the
EUR premium, had been oscillating around 0 bps until September 2018, it then rose to 10 bps
during the 2018 crisis and has decreased since, slipping down to 21 bps during the recent crisis
before bouncing back.
4.2.2 Maturity
If we look at the average monthly premium broken down per time to maturity displayed in Table
6, we see that it is only the maturities between 5 and 10 years that have a significant negative
premium. We note also that when the time to maturity is below 10 years, the lower the maturity
bucket, the higher the premium. Figure 9confirms this last finding where we see that most of the
time, premia of lower maturity buckets trend above those of higher maturity buckets.
Table 6: Maturities - 2016 – 2020
Maturity Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
All maturities -4.66 1.37 -10.47 -4.40 -2.54 -2.03 6.49 -3.40 ***
1 - 3 yrs 0.82 4.45 -15.83 -0.38 10.90 -0.31 3.31 0.18
3 - 5 yrs -1.26 3.93 -8.41 -1.40 10.96 0.94 1.53 -0.32
5 - 7 yrs -6.84 3.32 -13.26 -6.91 2.53 0.72 0.77 -2.06 **
7 - 10 yrs -9.25 4.08 -24.72 -9.22 -3.67 -1.49 3.88 -2.27 **
10 - 20 yrs -5.45 7.78 -21.19 -3.42 18.27 0.44 0.92 -0.70
Beyond 20 yrs -5.41 6.28 -20.57 -5.40 16.63 0.91 2.76 -0.86
13
Facts and Fantasies About the Green Bond Premium
Figure 9: Maturities’ premia in bps - 2016 – 2020
2017-01 2017-07 2018-01 2018-07 2019-01 2019-07 2020-01 2020-07
25
20
15
10
5
0
5
10
All maturities
1 - 3 yrs
3 - 5 yrs
5 - 7 yrs
7 - 10 yrs
4.2.3 Sector
Table 7reports the average monthly premium per sector on the entire period of observation. All
sectors except industrials and local authorities have negative premia on average, however only
financials and agencies exhibit significant negative premia with more marked negative premia for
the former.
Table 7: Sectors - 2016 – 2020
Sector Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
All sectors -4.66 1.37 -10.47 -4.40 -2.54 -2.03 6.49 -3.40 ***
ABS -4.78 3.44 -10.45 -4.67 5.00 1.10 1.90 -1.39
Agencies -9.08 3.11 -19.31 -8.94 -4.16 -1.19 2.40 -2.92 ***
CMBS -19.83 13.65 -61.91 -13.89 -7.44 -1.95 3.39 -1.45
Covered -5.27 4.95 -29.99 -4.29 -0.56 -2.87 12.23 -1.07
Financial-Institutions -14.30 5.76 -30.16 -12.95 -6.50 -0.98 0.34 -2.48 **
Industrials 28.09 23.52 -30.81 21.93 86.06 0.66 0.78 1.19
Local-Authorities 2.68 3.98 -2.73 1.31 12.97 1.05 0.54 0.67
Sovereign -4.52 9.58 -26.02 -4.62 13.27 -0.12 -0.52 -0.47
Supranational -1.05 1.27 -3.25 -0.99 1.74 0.45 -0.37 -0.82
Treasury -4.26 5.69 -13.26 -4.18 6.25 0.15 -1.17 -0.75
Utilities -2.25 4.91 -14.35 -0.73 7.89 -0.31 -0.17 -0.46
Figure 10 shows the trend of the four main sectors that make up the “green portfolio”. Three kinds
of trends are shown. Supranationals and agencies see their premia fluctuate in a range around 0
bps for the former and around 10 bps for the latter, albeit with a different thickness. Regarding
Financials, their premium rose from 30 bps to 7 bps by the end of 2017, before entering a
range between 7 bps to 19 bps, with the lowest values being reached during the recent crisis.
The observed trend is that of utilities. It is a decreasing trend that begins with positive values
but worsens amid the Covid-19 turmoil to reach 15 bps.
14
Facts and Fantasies About the Green Bond Premium
Figure 10: Sectors’ premia in bps - 2016 – 2020
2017-01 2017-07 2018-01 2018-07 2019-01 2019-07 2020-01 2020-07
30
25
20
15
10
5
0
5
All sectors
Agencies
Financials
Supranationals
Utilities
4.2.4 Rating
In Table 8, we report the average monthly premium per rating category. Two observations can
be drawn: The lower categories of ratings (A & Baa), exhibit a significant negative premium.
We note also that the lower the category of rating, the lower the premium and the higher the
volatility.
Table 8: Ratings - 2016 – 2020
Rating Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
All ratings -4.66 1.37 -10.47 -4.40 -2.54 -2.03 6.49 -3.40 ***
Aaa -0.50 0.88 -2.21 -0.40 1.43 0.15 -0.49 -0.57
Aa 0.57 3.28 -3.60 -1.22 9.46 0.56 -0.84 0.17
A -6.54 2.80 -12.73 -5.78 -1.36 -0.61 -0.33 -2.34 **
Baa -15.76 7.29 -37.15 -14.55 -3.84 -1.13 1.54 -2.16 **
These observations are illustrated in Figure 11. It depicts a huge range for the Baa premium and
a smaller range for the A premium. The evolution of Aaa premium is the exact replica of the
evolution of the supranationals’ premium. The Aa rating premium, mainly composed of French
and Belgian sovereigns and agencies, has an atypical evolution with regard to the rating categories
having generally seen a upward trend in recent periods.
It is important to note that in this method we assume that on either side of the comparison,
we are comparing well-diversified portfolios, as we do not explicitly control for issuer effects. This
will not be true for all sub-categories hence we attempt to address this in the next section.
15
Facts and Fantasies About the Green Bond Premium
Figure 11: Ratings’ premia in bps - 2016 – 2020
2017-01 2017-07 2018-01 2018-07 2019-01 2019-07 2020-01 2020-07
30
20
10
0
10
All ratings
Aaa
Aa
A
Baa
5 Second method: Bottom-up approach
In this method, we assess the level of premium of green bonds compared to non-green bonds by
performing an intra-curve estimate of the premium for every green bond in the universe and using
the aggregate data by the relevant fixed income asset sub-categories. In comparison with the first
method, we are now solely focusing on the intra-curve green premium, i.e. the premium for buying
the green bond format. Any potential broader effect in the overall risk premium of an issuer that
issues green bonds (often referred as “Green Halo”) will thus be excluded.
The premium will not be estimated for green bonds that are not in the “green portfolio”.
Even though ESG investors may see the proceeds as being green, green bonds that are not rated
Investment Grade, not of a benchmark size or for which the use of proceeds does not match the
taxonomy or transparency requirements of the index will be excluded.
We define the green premium as the difference in the reference spread (S) between the green
bond and a comparable conventional bond with the same issuer23, the same currency, the same
seniority24 and the same modified duration (MD). The reference spread is either the Z-spread25 or
the G-spread26 according to the reference curve communicated when the issuance was announced
27.
P remium =SGB SCB
23The same name and the same Bloomberg ticker.
24They are Covered, Contingent Convertible, Hybrid Corporate, Lower Tier Two, Senior and Senior Non Pre-
ferred.
25The Z-spread or zero volatility spread is the constant yield spread over the entirety of the swaps spot curve
such that the present value of the cash flows matches the clean price of the bond.
26The G-spread or nominal spread is defined as the difference between the yield of the corporate bond and the
interpolated yield of the treasury bond of the same time to maturity.
27See Table 23 on page 36
16
Facts and Fantasies About the Green Bond Premium
For a given green bond, the spread of the comparable conventional bond is determined using a
linear interpolation of the spreads of the nearest two conventional bonds CB1and CB2, picked
from the “benchmark”. The proximity between two bonds B1and B2sharing the first three
characteristics is defined as |MDB1M DB2|.
CB1GB CB2
CB1CB2GB
GB CB1CB2
Figure 12: Cases of interpolation
Three cases of interpolation, as depicted by Figure 12, are possible: the green bond may be
surrounded by CB1and C B2, or may have two close bonds with a lower or higher modified
duration.
To keep homogenous curve segments and to minimize the differences in the slope of the credit
spread, the interpolation is performed under the two following conditions of proximity:
|MDGB M DNG| ≤ |M DCB2M DCB1|
MDGB M Dmax M DGB
2,3 years
where NG is the nearest neighbor among {CB1,CB2}to the green bond and M D is the average
of MD of CB1and CB2.
The first condition particularly targets the second and third cases of interpolation as it is obviously
satisfied for the first case. If we denote Dthe distance between CB1and CB2, the green bond
should not be more than Daway from CB1and CB2. The second condition constrains the MD
of the green bond to be half its value at most away from MD. For the lowest values of MD (i.e.
6 years), we impose at most 3 years of distance.
The interpolation supposes the spread as an increasing function of the modified duration. If SCB1
is higher than SCB2, we pick if possible two new non-green bonds, as we cannot determine which
bond is “mispriced”. The linear interpolation does not take into consideration the local concavity
of the curves that we considered as negligible given the condition put on the length of the interval
[MDC B1, M DCB2]28.
CB1nCB1GB CB2CB2n
For the study, We use Bloomberg BVAL quote as price source. Some Green Bonds may not
quote as readily. To avoid misleading figures or an impact of stale pricing, the bond is excluded
from the set of studied green bonds. The same approach applies to the comparable non-green
bonds. The market for Green Bonds has its own liquidity features that are key to consider when
studying the hypothesis of a premium. On the one hand, Green Bonds tend to be smaller in size
than conventional bonds and tend to be held more by buy-and-hold investors when compared to
28The concavity (of the spreads to duration) slightly lowers the premium.
17
Facts and Fantasies About the Green Bond Premium
conventional bonds, which slows down their circulation. On the other hand, they are under the
spotlight of the bond markets and facing strong demand: active investors and traders can rely on
their attractive selling traits. Finally, they are on average newer and more likely to quote close to
par, which is another factor favouring better liquidity.
We adopt the same approach as Zerbib (2019) in order to reflect the difference in liquidity
Liquidity between the green bond and the comparable conventional bond:
Liquidity =Liquidity(GB)Liquidity(CB) (1)
The liquidity of bonds is traditionally assessed using the bid-ask spread (Chen et al.,2007). Fong
et al. (2017) show that the percent quoted bid-ask spread29 is the best low-frequency measure
for liquidity. Like Zerbib (2019) and Bour (2019), we adopt this proxy for the green bond and
its nearest conventional neighbours whereas the liquidity proxy for the comparable conventional
bond is defined as:
Liquidity (CB) = d2
d1+d2
Liquidity (CB1) + d1
d1+d2
Liquidity (CB2) (2)
where d1(respectively d2) is the distance between the green bond and CB1(respectively CB2).
Along with the definition of equation 1where the green bond is more liquid than its comparable
if ∆Liquidity is lower than 0, we impose that two bonds B1and B2have the same liquidity if the
liquidity of one bond does not exceed by 150% the liquidity of the other bond. In other words,
the liquidity ratio satisfies the following inequality:
1
1.5Liquidity(B1)
Liquidity(B2)1.5 (3)
5.1 Results
We report in red in Figure 13 the change from April 2019 to September 2020 in the number of
green bonds in the “green portfolio”. Their number rose steadily from 318 to 509. In the same
figure, we report the ratio of the calculated premia to the total number of green bonds. We note
that this ratio is around 51% on average reaching its lowest values between March and April 2020.
Table 9: Statistics
Metric Mean Std dev. Min 25 (%) Median 75 (%) Max
Total green bonds 420.29 56.46 318 383 425 469 509
Calculated premiums 213.51 27.72 171 193 208 228 271
No possible interpolation 106.15 13.58 76 99 108 121 125
Not so near related bonds 56.95 10.60 36 48 59 64 80
Incoherence in spreads 49.78 13.06 18 44 53 58 72
29BidAsk spread = 2 ·Ask price Bid price
Ask price +Bid price
18
Facts and Fantasies About the Green Bond Premium
Figure 13: Coverage
In Table 9, we report the reasons that did not allow the premia to be calculated. In roughly half
of cases, no interpolation is possible since no two related non-green bonds can be found. When
available, these two related bonds do not comply with the conditions of proximity in 25% of cases.
The last 25% of cases is attributed to the spread inconsistencies since no new comparable bonds
can be retrieved.
Table 10: Results
Metric Mean Std dev. Median Skewness Kurtosis T-statistic
CB Spread 59.59 55.34 45.36 2.18 7.80 15.73 ***
GB Spread 57.42 55.16 43.79 2.25 8.24 15.21 ***
Premium -2.17 10.74 -1.04 2.35 111.79 -2.95 ***
Duration 5.83 3.63 5.08 1.93 5.16 23.44 ***
We report in Table 10 the metrics of the spreads of the related non-green bond, the associated
premium and the modified duration. On average, the premium is negative (around 2.17 bps)
and is significant at 99%. The skewness metric that measures the asymmetry of the distribution
of premia is positive indicating that the mass of the distribution is concentrated to the left of the
mean and that most of the outliers are present on the right side of the distribution. The high
value of kurtosis confirms the presence of heavy tails in the data set of premia. Figure 16 displays
per date the number of outliers that are above 4 times the standard deviation or below -4 times
the standard deviation. The Covid-19 crisis exacerbated the number of outliers and in particular
the number of those in the right of the distribution. There may be a reason for this: in times
of sell-off, liquidity is poorer, and prices take more time to adjust. Under the assumptions that
Green Bonds are more in demand ceteris paribus, in such an environment they are the first ones
19
Facts and Fantasies About the Green Bond Premium
to be sold (and to have their price adjusted)30.
Figure 14 shows that the premium has evolved in a narrow range around 2 bps over the last
year. The recent crisis knocked it out of its range peaking at 3 bps in mid-March 2020 before
retracing.
Figure 14: Evolution of the premium (in bps)
If we look closer and drill down per currency (Table 11), we note that premia are negative and
that among the main currencies, only EUR and USD have a significant premium, albeit at 95%
and 90% confidence levels. These results are in line with those of Kapraun and Scheins (2019).
Compared to the EUR premium, the USD premium is tighter and twice as volatile. Its distribution
is right-skewed (i.e. most of outliers are on the right) whereas the distribution of EUR premia is
moderately left-skewed. The high level of the kurtosis is reflective of the presence of outliers in
both currencies.
Table 11: Breakdown per currency
Premium Av Spread Av MD
Currency Mean Std dev. Skew Kurtosis N. Obs T-stat
EUR -1.62 8.02 -1.41 19.85 9 194 -2.19 ** 53.70 6.39
USD -3.74 16.27 3.04 74.26 4 609 -1.77 * 73.95 5.16
CAD -2.19 6.98 -0.01 0.05 1 015 -1.13 67.47 6.12
AUD -1.01 2.79 -0.93 1.31 834 -1.19 40.07 4.80
Other Currencies -0.95 6.07 -2.12 9.46 1 002 -0.56 19.74 4.34
Table 12 shows that premia are negative if we perform a breakdown per category of rating.
However, only Aaa bonds have a statistically significant premium at a 95% confidence level. We
note that the volatility of the premium is an increasing function of the rating. According to the
30Second-method shows different results during this period.
20
Facts and Fantasies About the Green Bond Premium
skewness, all ratings except the A rating, exhibit asymmetric distributions of premia: left-skewed
for Aaa and right-skewed for Aa and Baa. The higher level of kurtosis of Aa compared to the
other ratings (of order of x9 to x10) indicates the presence of its outliers on the rightmost side.
Table 12: Breakdown per credit rating
Premium Av Spread Av MD
Rating Mean Std dev. Skew Kurtosis N. Obs T-stat
Aaa -1.94 6.97 -5.44 56.11 4 991 -2.22 ** 18.66 5.42
Aa -1.38 6.70 13.75 562.17 3 986 -1.47 44.51 6.18
A -3.13 13.30 0.56 77.38 4 264 -1.74 * 67.97 5.92
Baa -2.25 14.74 3.43 56.05 3 399 -1.01 116.33 5.93
Table 13: Breakdown per sector
Premium Av Spread Av MD
Sector Mean Std dev. Skew Kurtosis N. Obs T-stat
Covered -0.21 1.23 0.15 1.43 967 -0.61 3.73 5.18
Financials Corporates -1.19 13.16 5.97 146.38 4 258 -0.67 73.64 4.19
Non-financial Corporates -3.64 14.42 0.20 29.78 3 803 -1.76 * 85.37 6.84
SSA -2.23 7.01 -3.90 37.51 7 626 -3.14 *** 41.23 6.33
In Table 13, we report a sectoral breakdown. We distinguish four main groups: SSA (Suprana-
tional, Sovereigns and Agencies), Financial corporates, Non-financial corporates and Covered31.
We note the negative sign of the premium for each sector group. However, only the premium of
the SSA is significant at 99% and Non-financial corporates at 90%. It is interesting to note that
within the credit universe, green bond investors give a lower premium to Non-financial corporates
in comparison to Financials (3.6 bps vs 1.2 bps, when the average spread of the asset classes
are close at 85.4 bps vs 73.6 bps). These findings hold in the EUR universe32 as detailed in Table
14 with lower levels of skewness and kurtosis.
Table 14: EUR universe: Breakdown per sector
Premium Av Spread Av MD
Sector Mean Std dev. Skew Kurtosis N. Obs T-stat
Covered -0.28 1.11 0.32 1.26 880 -0.85 2.81 5.28
Financials Corporates -1.15 7.32 -0.25 3.82 3 042 -0.98 66.02 4.08
Non-financial Corporates -3.36 13.05 -0.92 9.20 2 151 -1.35 71.46 6.31
SSA -1.24 4.08 -1.50 11.89 3 121 -1.93 * 43.81 9.00
In Table 15, we report the breakdown per sector in a more granular way. We split off the Spec
finance sector from the financial sector. This sector encompasses bonds from Real-Investment
31Covered bonds are secured bonds issued by banks. Most issuances use mortgages as collateral.
32In Appendix A.3, we detail all the breakdowns in the EUR universe.
21
Facts and Fantasies About the Green Bond Premium
Trusts or infrastructure owners. We notice that only supranationals, agencies and utilities exhibit
significant negative premia at 95% for the former and 90% for the latter. As we can expect,
the volatilities of premia of covered bonds, sovereigns, agencies and to a lesser extent those of
supranationals are lower than the respective volatilities of corporates. Financials and spec finance
appear to be the only sectors with positive skewness and higher kurtosis indicating right-skewed
distributions with outliers on the rightmost side.
Table 15: Breakdown per sector
Premium Av Spread Av MD
Sector Mean Std dev. Skew Kurtosis N. Obs T-stat
Agencies -1.72 5.54 -1.94 11.91 4 443 -2.34 ** 48.85 6.23
Covered -0.21 1.23 0.15 1.43 967 -0.61 3.73 5.18
Financials -1.01 11.33 3.97 201.45 3 161 -0.57 60.36 3.65
Other Corporates -2.34 12.09 -0.54 4.30 674 -0.57 85.71 4.94
Sovereigns -0.47 5.87 -0.24 3.88 699 -0.24 70.12 9.72
Spec finance -1.73 17.39 6.95 82.33 1 097 -0.37 111.88 5.75
Supras -3.63 9.12 -4.37 33.94 2 484 -2.25 ** 19.48 5.55
Utilities -3.92 14.86 0.30 31.68 3 129 -1.67 * 85.30 7.24
If we focus per region (Table 16), we note that all regions33 exhibit a negative premium but only
Europe has a statistically significant premium. If we compare American bonds to European bonds,
we draw the same conclusions as for the currency breakdown, in terms of averages, volatilities,
skewness and outliers.
Table 16: Breakdown per region
Premium Av Spread Av MD
Region Mean Std dev. Skew Kurtosis N. Obs T-stat
Europe -1.88 7.70 -1.43 20.21 10 735 -2.86 *** 46.64 6.05
North America -3.07 16.94 2.51 66.11 3 513 -1.21 78.28 6.06
Asia & Pacific -3.16 10.61 -1.09 4.60 1 659 -1.37 84.21 4.37
Others 0.09 10.32 20.74 519.84 747 0.03 54.81 4.89
An interesting result is found in the breakdown by the Amundi ESG rating34 (Table 17). For
a large share of ESG investors, Green Bonds are a unique opportunity to incorporate into their
portfolios issuers that are lagging in terms of ESG scoring35 . The rationale is that the efforts made
by the green bond issuers to add green projects, bring transparency, and update the green strategy
with the green issuance are actually a way to spot ESG improvers. Besides the fact that premia,
if we exclude unrated bonds by Amundi, are negative, we find a decreasing relation between the
33To supranationals, we assign the region of their country of domicile.
34See Appendix A.1 for the definition of the Amundi ESG rating.
35Aside from general poor practices, lack of transparency in extra-financial communication maybe one factor for
a low ESG rating
22
Facts and Fantasies About the Green Bond Premium
spread and the ESG rating: the lower the ESG rating, the higher the spread36, the higher the
premium and the higher the volatility of the premium. Compared to the best in-class, all ratings,
and in particular E-F ratings37 exhibit a higher excess yield. We obtain the same results (Table
24) if we perform the breakdown by the Environmental pillar of the Amundi ESG rating38.
Table 17: Breakdown per ESG-rating
Premium Av Spread Av MD
Rating Mean Std dev. Skew Kurtosis N. Obs T-stat
A -3.04 8.29 -4.26 38.19 3 226 -2.36 ** 30.69 5.79
B -2.35 7.30 -5.17 52.99 2 620 -1.86 * 48.53 6.04
C -2.15 10.30 3.22 86.20 5 750 -1.79 * 58.70 5.89
D -2.20 13.42 3.31 123.27 2 624 -0.95 82.78 6.05
E-F -1.73 16.90 2.83 59.60 1 527 -0.45 96.01 4.87
NR 0.65 5.76 -3.74 40.65 907 0.38 31.68 5.98
In terms of time to maturity39, all premia are negative on average (Table 18). Only the 5-7 years,
7-10 years and 10-20 years buckets have significant premia at 95% or 90%. If we exclude the last
bucket, the premium seems to be a decreasing function of the maturity: the higher the bucket,
the lower the premium. We note that only the 5-7 years bucket is not skewed. All other buckets
are skewed: those below 10 years are right-skewed however, long-term buckets are left-skewed.
Table 18: Breakdown per time to maturity
Premium Av Spread Av MD
Maturity Mean Std dev. Skew Kurtosis N. Obs T-stat
1 - 3 yrs -1.03 12.54 3.11 166.03 2 910 -0.50 32.18 2.07
3 - 5 yrs -1.38 7.60 8.69 323.87 4 434 -1.36 48.98 3.89
5 - 7 yrs -3.01 9.87 0.11 103.84 3 404 -2.02 ** 52.62 5.60
7 - 10 yrs -2.79 11.33 3.04 58.78 4 073 -1.78 * 72.21 7.72
10 - 20 yrs -5.23 9.85 -2.55 7.19 972 -1.87 * 66.77 12.99
Beyond 20 yrs -0.39 16.57 -0.75 7.20 861 -0.08 124.67 12.45
Table 19 reports the implication in terms of premia if the green bond is more or less liquid than its
comparable conventional bond. We note that liquid green bonds tend to exhibit a lower premium
(0.88 bps). This result suggests the existence of a bond liquidity premium since the bond
liquidity is positively correlated to the green premium. Wulandari et al. (2018) argues that this
liquidity premium is due to the insufficient supply and excess demand in the green bonds market,
which implies a thin market.
36This result is consistent with Ben Slimane et al. (2019) who find that ESG has a positive impact on the cost
of debt and this relationship has become stronger since 2014.
37We regroup E and F into one cluster, as the number of F ratings is too small.
38ibid
39We apply the next-call date convention regarding callable bonds and in particular perpetual bonds.
23
Facts and Fantasies About the Green Bond Premium
Table 19: Breakdown per liquidity
Premium Av Spread Av MD
Liquidity Mean Std dev. Skew Kurtosis N. Obs T-stat
GB less Liquid -1.71 9.72 5.31 142.44 7 936 -1.77 * 56.13 5.56
GB more Liquid -2.59 11.58 0.76 93.29 8 703 -2.36 ** 58.65 6.08
In the meantime, Table 20 shows that if we adjust the sample with greens that are comparable
with non-greens of same liquidity, we lower the premium from 2.17 bps in the full sample to
2.88 bps. Duration and spread metrics of the samples are close to each other (5.93 years vs 5.83
years and 56.13 bps vs 57.42 bps).
Table 20: Breakdown per liquidity
Premium Av Spread Av MD
Liquidity Mean Std dev. Skew Kurtosis N. Obs T-stat
Different Liquidity -1.64 8.46 0.02 45.20 9 496 -2.13 ** 58.44 5.76
Same Liquidity -2.88 13.14 3.08 105.94 7 143 -2.10 ** 56.13 5.93
Table 21: Breakdown per certification
Premium Av Spread Av MD
Certification Mean Std dev. Skew Kurtosis N. Obs T-stat
Not Certified -1.80 15.19 5.17 94.54 2 531 -0.67 79.53 6.01
Certified -2.23 9.73 0.17 93.74 14123 -3.09 *** 53.46 5.80
Certified by CBI -2.92 11.23 5.72 148.86 2 290 -1.41 72.15 6.41
Certified by others -2.10 9.40 -1.64 70.17 11 833 -2.75 *** 49.84 5.68
Table 21 shows the breakdown per certification. We split the certified bonds into two groups:
Those certified by CBI40 and those certified by other reviewers. Compared to uncertified bonds,
we note that certified bonds exhibit a lower and significant premium and that among these certified
bonds, CBI certified bonds even have the lowest premia on average (1.12 bps vs 0.40 bps).
5.2 Determinants of the green premium
To investigate the determinants of the green premium, we run a panel data regression model with
fixed time effects using the bonds for which we calculate a premium and discarding those whose
premium does not belong to the [4σp,4σp] range, σpbeing the premium distribution standard
deviation. The percentage of discarded premia does not exceed 0.72% of the total number of
premia.
40CBI is one of the external reviewers.
24
Facts and Fantasies About the Green Bond Premium
Let Premiumi,t be the green premium of Bond iat time t. We assume that the premium depends
linearly on the bond’s intrinsic characteristics such as age, duration, natural logarithm of the size,
currency, sector and country or on external features such as certifications or Amundi E-ratings41:
Premiumi,t =αt+βmd ·MDi,t +
NSector
X
j=1
βSector (j)· Sectori,t (j) +
βdm ·Domestici,t +
NRating
X
k=1
βRating (k)· Ratingi,t (k) +
βag ·Agei,t +
NRegion
X
l=1
βRegion (l)· Regioni,t (l) +
βliq ·∆Liquidityi,t +
NSeniority
X
m=1
βSeniority (m)· Seniorityi,t (m) +
βsz ·ln Sizei,t +
NCurrency
X
n=1
βCurrency (n)· Currencyi,t (n) +
βcf ·Certifiedi,t +βcbi ·CBIcertifiedi,t +εi,t (4)
where MDi,t, Sizei,t , ∆Liquidityi,t, Agei,t are respectively the modified duration, the size of the
bond, the liquidity proxy and the age of Bond iat time t. All the other variables are categor-
ical variables. When Bond iis denominated in local currency, Domestici,t equals 1 otherwise 0.
Sectori,t (j), Ratingi,t (k), Regioni,t (l), Seniorityi,t (m) and Currencyi,t (n) are dummy variables
for the jth sector42, the kth E-rating43 , the lth region44 , the mth seniority45 and the nth currency46.
Finally, we introduce two additional variables regarding the external certifications. Certifiedi,t
(resp. CBIcertifiedi,t) equals 1 when the Bond iis certified by an external reviewer (resp. when
Bond iis certified by CBI) otherwise 0.
For the purpose of this regression, we group together the two lowest Environmental ratings E and
F into E-F. The proposed model does not include categorical variables related to credit ratings as
sectors and ratings are highly correlated: supranationals are AAA-rated, utilities are BBB-rated.
We omit the financial sector, the European region, the senior seniority, the EUR currency, and the
A E-rating dummy variables to avoid the multi-collinearity problem as the model already includes
a constant. The betas associated with one specific sector (resp. region or seniority or currency or
E-rating) represent the excess premium with respect to the financial sector (resp. the European
region or the senior seniority or the EUR currency or the A E-rating).
From regression (4), we derive four models whether or not we include the external certification
or the CBI certification and report in Table 22 several statistics of the models and the different
41The results using Amundi ESG ratings are similar and reported in Table 25 on page 37.
42We have Sectori,t (j) = 1 if the ith corporation belongs to the jth sector at time t, otherwise zero.
43We have Ratingi,t (k) = 1 if the ith corporation has the kth E-rating at time t, otherwise zero.
44We have Regioni,t (l) = 1 if the ith corporation has its main country risk in lth region at time t, otherwise
zero.
45We have Seniorityi,t (m) = 1 if the ith bond has the lth seniority, otherwise zero.
46We have Currencyi,t (n) = 1 if the ith bond is denominated in nth currency, otherwise zero.
25
Facts and Fantasies About the Green Bond Premium
Table 22: Panel regression statistics using E-ratings
No certif. Certified CBI Certified All
Intercept 1.95** 3.92*** 3.28*** 4.62***
Modified Duration -0.10*** -0.11*** -0.05*** -0.07***
Age -0.02 -0.05 -0.02 -0.04
Size 0.03 0.06 0.15 0.16
Domestic -2.18*** -2.26*** -2.31*** -2.36***
∆ Liquidity 4.27*** 4.20*** 4.34*** 4.29***
Certification
CBI Certified -3.08*** -2.85***
Certified -1.56*** -1.14***
Seniorities
Subordinated 6.17*** 6.35*** 6.21*** 6.34***
Secured -0.56** -0.55** -0.31 -0.32
Non-preferred Senior -2.83*** -2.80*** -3.42*** -3.36***
Countries
North America 1.17*** 0.96*** 1.05*** 0.91***
Asia & Pacific -0.97*** -1.07*** -0.39 -0.51*
Other countries 0.25 0.28 2.69*** 2.54***
Sectors
Agencies -0.97*** -0.78*** -1.00*** -0.86***
Sovereigns 1.12*** 1.38*** 0.23 0.49
Spec finance -1.73*** -1.68*** -1.89*** -1.84***
Supras -2.23*** -2.21*** -3.23*** -3.14***
Utilities -1.96*** -2.03*** -2.42*** -2.44***
Other Corporates 0.14 0.45 -0.28 -0.03
Currencies
AUD -1.20*** -1.40*** -1.16*** -1.30***
CAD -1.51*** -1.70*** -1.61*** -1.74***
USD -3.30*** -3.58*** -3.35*** -3.55***
Other currencies -1.09*** -1.19*** -1.26*** -1.32***
E Rating
B 0.56** 0.42 0.72*** 0.60**
C -0.33 -0.52* -0.70** -0.81***
D -0.28 -0.58* -0.53* -0.73**
E-F 2.21*** 1.53*** 2.24*** 1.75***
NR 3.40*** 3.36*** 3.12*** 3.11***
Stats
R2(%) 8.59 8.96 9.69 9.88
F-stat 59.4 59.9 65.32 64.35
VIF (max) 5.58 5.62 5.63 5.65
VIF (mean) 2.13 2.11 2.14 2.13
N. Obs. 16 534 16 534 16 534 16 534
betas. The outcome of the F-test confirms the rejection of the null hypothesis H0:βmd =... =
26
Facts and Fantasies About the Green Bond Premium
βcf =βcbi = 0. The VIF statistic is the acronym of the variance inflation factor, a measure of
multi-collinearity of two exogenous variables. As a rule of thumb (O’Brien,2007), a VIF lower
than or equal to 5 indicates a low dependence between the independent variables. We verify that
VIF47 is low and remains almost unchanged even after the inclusion of the certifications’ variables.
The coefficient of determination R2, that calculates the explanatory power of the model, is in the
[8.5%, 10%] range for 16 534 observations. The relatively low R2represents the scatter around
the regression line48 and means only that the model cannot be used to perform precise predictions
regardless of significance of the coefficients of the independent variables. Obviously, the “All”
model has the highest R2as it includes all variables. We note that the inclusion of the CBI
certification variable to the first model with certification has a relatively higher impact (+110
bps) on R2than including the certification by an external reviewer (+37 bps).
The outcome of the analysis shows a statistically significant negative relationship between modified
duration and premium. GBs issued with short durations tend then to have higher excess yields.
However, this is to be tempered as the level of beta is relatively small. The ∆Liquidity, in
the contrary, is an increasing function of the premium. GBs with different liquidity than their
comparable CBs exhibit higher premia. The above relationships are statistically significant at the
1% significance level. These two results are in line with the findings of Tables 18 and 20. The
analysis shows, albeit being non-significant, a positive sign for size and a negative sign for age.
Unlike Kapraun and Scheins (2019), we observe lower premia in the secondary market when bonds
have smaller issue amounts. Aged issues seem also to be prized too. The shortage of supply and
the large demand may be a reason.
If we focus on the Domestic variable, we come to the same conclusion as Nanayakkara and Colom-
bage (2019) on bonds denominated in local currencies. Their credit spreads are tighter than for
bonds issued in foreign currencies. This can be explained by the fact that investments in bonds
denominated in local currencies are low-risk investments even when investing overseas.
As documented in other studies, we find that certifications lower the premium. All other things
being equal, being externally certified lowers the premium by 1.14 to 1.56 bps. The CBI certifi-
cation lowers the premium by an additional 3 bps. These findings are in line with those of Baker
et al. (2018), Hyun et al. (2020) and Kapraun and Scheins (2019) mentioned above. The CBI
certification, one of the most stringent form of certification, dispels worries of investors of investing
in green bonds that do not bring any sustainable benefit, generating then a buying pressure for
these bonds and thus lower yields and premia.
If we look at currencies, non-EUR denominated bonds exhibit lower premia compared to EUR-
denominated bonds. This result is significant at 99% whatever model we take. Finally, we have
confirmation of our finding on the worst-rated bonds in terms of E-ratings: Their premium is
significant and higher than those of the best-rated bonds.
Regarding seniorities, subordinated bonds (resp. non-preferred seniors and secured bonds) exhibit
higher premia (resp. lower premia) compared to senior bonds. In terms of geography, compared
to bonds whose country of risk is in Europe, bonds whose risk is located in Asia and Australia
show lower premia, whereas North-American bonds have a higher premium (+1.17). This last
result seems to be in contradiction with the findings of Table 16.The positive beta associated with
47We report the highest and the mean VIF of each pair of exogenous variables.
48The closer to the line, the higher coefficient of determination.
27
Facts and Fantasies About the Green Bond Premium
American issued bonds is in fact a EUR-denominated beta, so for instance, to assess the level of
beta associated with American USD-denominated bonds, one should add the beta associated to
USD to find a negative beta of 2.13 = 1.17 3.30.
In terms of sectors, all sectors except sovereigns, have lower premia compared to financials. If we
rank them from bottom to top, we find supranationals, utilities then spec finance.
6 Conclusion
6.1 Findings
The purpose of this study is to determine if investors are rewarded with lower yields when they
invest in Green Bonds. We present two methods to assess whether a green bond premium exists.
We consider the green bond constituents of the Bloomberg Barclays MSCI Global Index. In the
first method, we build a synthetic conventional portfolio from a global aggregate bond index by
dissecting it using four criteria and then re-weighting, applying the weights of the portfolio of
green bonds. The premium is defined here as the difference in OAS of the green bond portfolio
and re-weighted aggregate bond index. In the second method, we match, when it is possible, each
green bond with two conventional bonds that have the same issuer, the same currency and the
same seniority under some constraints of proximity in terms of duration. We either interpolate or
extrapolate the spreads to find the spread of a theoretical bond having the same duration as the
green bond. The difference in spreads is then the premium we seek.
Both methods show small negative and significant premia of respectively 4.7 bps and 2.2
bps. The second method by asset sub-class breakdowns are the following: 2.2 bps for Suprana-
tionals, Sovereigns and Agencies significant at 99%, 3.6 bps for Non-Financial Corporate issuers
significant at 99%, 1.2 bps for Financials and 0.2 bps for Covered Bonds, with the last two
categories being not statistically significant. The same order of magnitude of premia is observed
in the EUR universe, which is the main currency of the considered index.
According to both methods, this premium is significant in several market segments: EUR-
denominated bonds, A rated bonds, bonds issued by agencies and bonds whose time to maturity
is between 5 and 10 years. In our first method, we confirm the findings of Zerbib (2019) regarding
bonds issued by financial institutions and low-rated bonds whose negative premia are more pro-
nounced. We take advantage of the presence of the synthetic conventional portfolio to compare
its performances with that of the green portfolio and observe the existence of a put payoff for
the outperformance49 of the green portfolio. If we focus on the first method, the premium is
also significant for USD-denominated bonds, bonds specifically of supranationals, and utilities,
bonds rated Aaa, bonds whose country of risk is located in Europe and bonds certified by external
reviewers. We also found that the premium and its significance increases with the ESG quality
of the issuer, i.e. beyond the use of proceeds, green bond investors reward a more negative pre-
mium to issuers with better extra-financial standards at the company level. Finally, within the
Credit segment, the premium is three times lower for Non-financial corporates than for Financial
issuers, with comparable sample size and average spread of both segments. Moreover, using a
panel regression with control variates, we find that the premium is a decreasing function of the
49Or a call payoff for the performance.
28
Facts and Fantasies About the Green Bond Premium
duration and that domestic bonds and bonds certified by the CBI tend to show low premia. We
show further that all sectors except sovereigns (resp. that the AUD, CAD, and USD currencies)
exhibit lower premia compared to the financial sector (resp. to the EUR). We also find that the
green premium lowers with smaller-sized bonds, as well as with age, although not significantly.
Finally, as described broadly in this paper, the green bond market has different liquidity
features than the overall market. As such, it would be arguable that the existence of a liquidity
bias can explain part of the green bond premium found. To circumvent this argument, we filtered
our universe to retain bonds with similar liquidity characteristics. After neutralizing the liquidity
factor in the search of the premium, we find a green bond premium, which is even lower at 2.9
bps compared to 2.2 bps. At least, we deduce from this specific result that the overall green
premium found is not hiding a liquidity premium.
6.2 Discussion
There are several arguments in favor of a negative premium. Green bonds provide two major
features to investors: a commitment to dedicate a least an equivalent amount raised to a pool
of specific, detailed green projects and transparency on the use of the proceeds, with reporting
on the impact of those projects. The issuance of a green bond also has a financial cost for the
issuer50. Against a backdrop of strong demand, one may argue that green bond issuers may seek
financial compensation to at least offset the additional cost of issuance. The issuance of green
bonds compared to total issuance is still limited today.
In the meantime, the demand for “Green” or “Impact” investments more generally is increasing
at its a pace independent of supply. This potential mismatch of supply and demand can trigger
scarcities and thus larger negative premia. However, most ESG investors are not ready to give
up returns to hold green bonds, which may counter-balance the pace of demand for green bonds.
Furthermore, in the long-run, whilst one can expect that the Green Bond market is far from having
reached its maximum size; considering that best practices require green bonds to finance tangible
green projects, there must always be a significant size gap between the green bond universe and
the broader bond universe.
As we have discussed in our study, green bonds show strong liquidity features in favor of the
seller. It is easier to find a buyer of a green bond than an equivalent non-green with the same
characteristics.
Finally, and perhaps the strongest argument for a negative premium, climate risk is increasingly
a concern (Hong et al.,2020) for institutional investors (Krueger et al.,2020), governments and
public policymakers. As such, it is foreseeable that at some point in the future regulators or public
investors, will actively distort the markets to better price climate risk in asset prices51, including
50To assess the full impact on cost of debt for green issuers, rather than the secondary market, we would need
to consider the primary issue prices. Indeed, on average, yields at issuance incorporates an additional New Issue
Premium (NIP) in favour of investors. Not covered by this study, this NIP may be actually lower for Green Bonds
than for conventional bonds.
51Le Guenedal et al. (2020) identify that outside the power generation sector, corporates’ emission intensity
trajectories are not in line with a 2 degrees scenario. This confirms the necessity for asset-owners and investment
managers to keep their focus on the climat transition and a long-term assessment of corporates’ emission intensity
track-record.
29
Facts and Fantasies About the Green Bond Premium
in the fixed-income space. In terms of regulatory forces, one can imagine tax discounts52 or green
adjusted capital requirements to financial institutions. In terms of investment forces, like any
ESG investor, large public investors can play a role. The European Central Bank which has a
massive Asset Purchasing Program to channel its monetary policy is often referred to as the next
game changer53. Furthermore, in terms of regulatory forces, one can imagine tax discounts or
green adjusted capital requirements to financial institutions. For now, no such groundbreaking
public measures have been implemented and there appears little anticipation by financial markets
given the small size of the negative premium that we have found. In this context perhaps an
active investor should not be dissuaded by this negative premium, as this could be compensated
by future excess returns.
Yet, despite these different potential arguments justifying a green bond negative premium,
there are still other more convincing ones against it. When buying a green bond, the green
investor does not own any rights to the projects to be financed. On the contrary, the investor
bears the exact same Credit and ESG risks as the owner of a non-green bond with the exact same
financial characteristics. In addition, green bonds create high and potentially onerous expectations
among ESG investors. Namely in terms of alignment of the use of proceeds and reporting with
the commitment at the issuance. As such, and this is important to consider, unlike other ESG
bonds54 , a green bond bears the additional risk of controversy on the use of proceeds (and thus
the risk of suffering a bond-specific sell-off ). This controversy risk approach reduces the rationale
for a negative green premium.
In this study, we focused on the prices in the secondary market. The levels we have seen
indicate that although the negative premium on Green Bonds is significant, it is still marginal.
If the idea of a green premium has been so popular among bond investors, this is probably due
to the lower average new issue premium offered by green bonds over non-green bonds at issuance
(Cuilliere et al.,2020). There are even few recent memorable cases of issuers coming to market
with green bond at a huge discount to the regular secondary curve. According to Bloomberg, the
e1 billion 10-year green bond issued by the automaker Daimler AG priced more than 13 basis
points tighter than its conventional spread curve. Likewise, Volkswagen AG sold eight-year and
12-year green benchmarks with a volume of e2 billion, 15.4 and 13.6 basis points lower in yield
versus the rest of its bonds55
However, considering that we have found a relatively small negative premium in the secondary
market, could what investors observe as a negative green bond premium bond negative premium
is in fact more generally represent a negative premium on green issuers, whether the particular
issue is green or not? The results of method one vs method two is somewhat consistent with this
as issuers of green bonds can be compared to non-green issuers. With the wave of ESG integration
into bond markets, research has found that ESG is increasingly part of the premium prices into
bonds. Through an integrated ESG-credit pricing model, Ben Slimane et al. (2019) find some
52https://www.climatebonds.net/policy/policy-areas/tax-incentivess
53However, even if climate risk is priced into purchasing programs, this is very unlikely that a climate filter would
be applied at the format level rather at the issuer level.
54Namely Sustainability-Linked Bonds that offer an insurance premium or step-up coupon in case the issuer does
not meet its ESG commitment.
55In the case of Daimler, Amundi Portfolio Management models assess the premium 13 days after the issuance
at only 4 bps (from 13 bps at issuance).
30
Facts and Fantasies About the Green Bond Premium
evidence that ESG affects the cost of capital in a positive way: issuers with higher ESG scores
have lower costs of capital than issuers with lower ESG scores for the same credit rating. Among
ESG risks, the Environmental or Climate risk is viewed at the same time as more material, more
likely to materialize and more damaging for debt issuers. An issuer coming to market with a green
bond is offering the whole market (and not only green bondholders) both transparency and an
update on its green strategy and commitment towards green projects. Whenever the green risk
is material, either because it is important in a specific sector, or because a controversial issuer is
on a path to green redemption by the market: a green tightening becomes financially rational.
The correct question to ask then is, does the green benefit accrue to the entire issuer curve or is
it restricted to the green bonds?
31
Facts and Fantasies About the Green Bond Premium
References
Almeida, M., Filkova, M., Harrison, C., and Sette, P. (2019), Green Bond European In-
vestor Survey, Climate Bonds Initiative, November 2019.
Bachelet, M.J., Becchetti, L., and Manfredonia, S. (2019), The Green Bonds Premium
Puzzle: The Role of Issuer Characteristics and Third-Party Verification, Sustainability, 11(4),
pp. 1-22, February.
Baker, M., Bergstresser, D., Serafeim, G., and Wurgler, J. (2018), Financing the Re-
sponse to Climate Change: The Pricing and Ownership of US Green Bonds, National Bureau
of Economic Research, 25194.
Ben Slimane, M., Le Guenedal, T., Roncalli, T. and Sekine, T. (2019), ESG Investing
in Corporate Bonds: Mind The Gap, Amundi Working Paper, 94, www.research-center.
amundi.com.
Bour, T. (2019), The Green Bond Premium and non-Financial Disclosure: Financing the Future,
or Merely Greenwashing?, Master’s thesis, Maastrict University, Maastricht, The Netherlands.
Cochu, A., Glenting, C., Hogg, D., Georgiev, I., Skolina, J., Eisinger, F., Jespersen,
M., Agster, R., Fawkes, S., and Chowdhury, T. (2016), Study on the Potential of Green
Bond Finance for Resource-efficient Investments, Report to the European Commission, Novem-
ber 2016.
Cuilliere, T., Jan, R., Levy, J. and Machado, J. (2020), What Future for Green & Social
Bonds After Covid-19?, Green Bonds Review, July 2020.
Chen, L., Lesmond, D.A., and Wei, J. (2007), Corporate Yield Spreads and Bond Liquidity,
Journal of Finance, 62 (1), 119-149.
Ehlers, T., and Packer, F. (2017), Green Bond Finance and Certification, BIS Quarterly
Review, September.
Fatica, S., Panzica, R., and Rancan, M. (2019), The Pricing of Green Bonds: Are Financial
Institutions Special?, Technical report to the European Commission.
Fender, I., McMorrow, M., Sahakyan, V., and Zulaica, O. (2019), Green Bonds: The
Reserve Management Perspective, BIS Quarterly Review, September.
Fong, K. Y. L., Holden, C.W., Trzcinka, C.A. (2017), What Are the Best Liquidity Proxies
for Global Research?, Review of Finance, 21(4), 1355-1401.
Flammer, C. (2018), Corporate Green Bonds, SSRN,www.ssrn.com/abstract=3125518.
Gianfrate, G., and Peri, M. (2019), The Green Advantage: Exploring the Convenience of
Issuing Green Bonds, Journal of Cleaner Production, 219, pp. 127-135.
Hachenberg, B., and Schiereck, D. (2018), Are Green Bonds Priced Differently from Conven-
tional Bonds?, Journal of Asset Management, 19(6), pp. 371-383.
32
Facts and Fantasies About the Green Bond Premium
Hong, H., Karolyi, G.A., and Scheinkman, J.A. (2020), Climate Finance, Review of Financial
Studies, 33(3), pp. 1011-1023.
Hyun, S., Park, D., and Tian, S. (2020), The Price of Going Green: The Role of Greenness in
Green Bond Markets, Accounting & Finance , 60(1), pp. 73-95.
Kapraun, J., and Scheins, C. (2019), (In)-Credibly Green: Which Bonds Trade at a Green
Bond Premium?, SSRN,www.ssrn.com/abstract=3347337.
Karpf, A., and Mandel, A. (2018), The Changing Value of the ‘Green’ Label on the US Mu-
nicipal Bond Market, Nature Climate Change, 8(2), pp. 161-165.
Krueger, P., Sautner, Z., and Starks, L.T. (2020), The Importance of Climate Risks for
Institutional Investors, Review of Financial Studies, 33(3), pp. 1067-1111.
Larcker, D., and Watts, M. (2019), Where’s the Greenium?, Stanford: Rock Center for Cor-
porate Governance at Stanford University.
Lau, P., Sze, A., Wan, W., and Wong, A. (2020), The Economics of the Greenium: How
Much Is the World to Pay to Save the Earth?, Hong Kong Institue For Monetary and Financial
Research.
Le Guenedal, T., Girault, J., Jouanneau, M., Lepetit, F., and Sekine, T. (2020), Trajec-
tory Monitoring in Portfolio Management and Issuer intentionality Scoring, Amundi Working
Paper, 97, www.research-center.amundi.com.
Nanayakkara, M., and Colombage, S. (2019), Do Investors in Green Bond Market Pay a
Premium? Global Evidence, Applied Economics, 51(40), pp. 4425-4437.
Nofsinger, J., and Varma, A. (2014), Socially responsible funds and market crises, Journal of
Banking & Finance, 48(C), pp. 180-193.
O’Brien, R.M. (2007), A Caution Regarding Rules of Thumb for Variance Inflation Factors,
Quality & Quantity, 41(5), pp. 673-690.
OECD (2016), Green bonds: Country Experiences, Barriers and Options.
Ostlund, E. (2015), Are Investors Rational Profit Maximisers Or Do They Exhibit a Green
Preference?, Master’s Thesis in Economics, Stockholm School of Economics.
Partridge, C., and Medda, F. (2018), Green Premium in the Primary and Secondary US
Municipal Bond Markets, QASER Laboratory, University College London.
Preclaw, R., and Bakshi, A. (2015), The Cost of Being Green, In Credit Research: US Credit
Focus. New York: Barclays, September 18, pp. 1-12.
Schmitt, S. (2017), A Parametric Approach To Estimate the Green Bond Premium, NOVA -
School of Business and Economics London.
33
Facts and Fantasies About the Green Bond Premium
Schmuck, D., Matthes, J., and Naderer, B. (2018), Misleading Consumers with Green Ad-
vertising? An Affect-reason-involvement Account of Greenwashing effects in Environmental
Advertising., Journal of Advertising, 47(2), pp. 127-154.
Silva, F., and Cortez, M.C. (2016), The Performance of US and European Green Funds in
Different Market Conditions., Journal of Cleaner Production, 135, pp. 558-566.
Technical Expert Group (TEG) on Sustainable Finance (2019), Report on EU Green Bond Stan-
dard, Report to the European Commission, June 2019.
Wulandari, F., Dorothea, S., Andreas, S., Chen, S. (2018), The Impact of Liquidity Risk
On the Yield Spread of Green Bonds. Finance Research Letters, 27, pp. 53-59.
Zerbib, O.D. (2019), The Effect of Pro-environmental Preferences on Bond Prices: Evidence from
Green Bonds, Journal of Banking & Finance, 98, pp. 39-60.
34
Facts and Fantasies About the Green Bond Premium
A Appendix
A.1 Amundi ESG scores and ratings
We consider the scoring system provided by the Amundi ESG Research department. For each
company and each month, we assess the ESG score and its three components: E (environmental),
S (social) and G (governance). These scores are based on the data of four external providers and
are reviewed and validated by internal ESG analysts. The scores are normalized sector by sector
to obtain a z-score shape, implying that they generally have a range between 3 and +3. This also
means that the scores are sector-neutral, and they are approximately distributed as a standard
Gaussian probability distribution. An example is given in Figure 15, which shows the empirical
distribution of the global ESG score at the end of December 2018. The Gaussian approximation is
very good even though we observe that the empirical distribution exhibits a low positive skewness.
On average, the z-score is then equal to zero if we consider all the corporations together or if we
consider a specific sector. The sector-neutrality of z-scores is an important property of many ESG
scoring systems.
Figure 15: Empirical distribution of the ESG score (December 2018)
-3 -2 -1 0 1 2 3
0
1
2
3
4Gaussian approximation
We define the ESG rating as a letter grade by mapping the z-score as shown below. This procedure
is performed also on each pillar of the ESG score.
Rating z-score
A +2.5z-score
B +1.5z-score <+2.5
C +0.5z-score <+1.5
D0.5z-score <+0.5
Rating z-score
E1.5z-score <0.5
F2.5z-score <1.5
Gz-score <2.5
35
Facts and Fantasies About the Green Bond Premium
A.2 Full sample : Additional tables
Table 23 indicates the type of spread used, depending on the currency and the sector.
Table 23: Z-spread or G-Spread
Currency ABS Agencies & Supras Banking Industrials Covered & Real-Estate Sovereigns
EMU Non-EMU
AUD G G Z Z Z G Z
CAD G G G G G G G
CHF Z Z Z Z Z Z Z
CNY G G G G G G G
DKK Z Z Z Z Z Z Z
EUR G G Z Z Z G Z
GBP G G G G G G G
HKD G G G G G G G
JPY G G G G G G G
NOK Z Z Z Z Z Z Z
SEK Z Z Z Z Z Z Z
SGD G G G G G G G
USD G G G G G G G
Table 24 reports the breakdown per E rating. We note that the premium is negative and the
worst-rated bonds exhibit higher excess yields than the best-in class rated bonds.
Table 24: Breakdown per E-rating
Premium Av Spread Av MD
Rating Mean Std dev. Skew Kurtosis N. Obs T-stat
A -3.52 8.91 -2.13 8.55 1 326 -1.63 33.78 4.68
B -2.05 8.28 -2.48 73.87 4 666 -1.92 * 41.77 5.78
C -2.17 9.35 3.19 100.05 6 379 -2.10 ** 58.50 6.19
D -2.66 17.61 3.62 69.79 2 461 -0.85 93.16 5.79
E-F -2.29 12.13 -3.92 71.61 915 -0.65 93.33 5.21
NR 0.65 5.76 -3.74 40.65 907 0.38 31.68 5.98
36
Facts and Fantasies About the Green Bond Premium
Table 25 reports the results of the panel data regression when we consider ESG ratings instead of
E ratings. We notice that all significant results obtained in Table 22 hold.
Table 25: Panel regression statistics using ESG ratings
No certif. Certified CBI Certified All
Intercept 2.83*** 4.78*** 4.03*** 5.42***
Modified Duration -0.08*** -0.10*** -0.04* -0.05***
Age -0.06 -0.08* -0.06 -0.08
Size -0.05 -0.02 0.06 0.07
Domestic -2.35*** -2.43*** -2.46*** -2.50***
∆ Liquidity 4.24*** 4.16*** 4.31*** 4.24***
Certification
CBI Certified -2.81*** -2.59***
Certified -1.58*** -1.20***
Seniorities
Subordinated 6.34*** 6.54*** 6.41*** 6.56***
Secured -0.63*** -0.62*** -0.36 -0.37
Non-preferred Senior -2.54*** -2.51*** -2.95*** -2.89***
Countries
North America 0.89*** 0.76*** 0.69*** 0.61**
Asia & Pacific -1.70*** -1.69*** -1.28*** -1.30***
Other countries 0.13 0.18 2.28*** 2.15***
Sectors
Agencies -1.28*** -1.14*** -1.42*** -1.30***
Sovereigns 0.62 0.96** -0.38 -0.04
Spec finance -1.56*** -1.59*** -1.68*** -1.69***
Supras -2.92*** -2.95*** -3.76*** -3.72***
Utilities -2.37*** -2.49*** -2.95*** -2.99***
Other Corporates 0.30 0.55 -0.13 0.09
Currencies
AUD -1.30*** -1.51*** -1.25*** -1.42***
CAD -1.31*** -1.56*** -1.35*** -1.54***
USD -3.37*** -3.67*** -3.39*** -3.61***
Other currencies -1.27*** -1.38*** -1.44*** -1.51***
ESG Rating
B -0.85*** -0.89*** -0.92*** -0.95***
C -1.25*** -1.35*** -1.46*** -1.51***
D -1.55*** -1.77*** -1.75*** -1.90***
E-F 1.20*** 0.67 1.23*** 0.82**
NR 2.53*** 2.59*** 2.32*** 2.38***
Stats
R2(%) 8.71 9.09 9.66 9.88
F-stat 60.26 60.82 65.1 64.31
VIF (max) 6.37 6.38 6.38 6.39
VIF (mean) 2.36 2.34 2.36 2.34
N. Obs. 16 534 16 534 16 534 16 534
37
Facts and Fantasies About the Green Bond Premium
A.3 EUR Universe
1. Bottom-up approach
(a) Table 26: This Table is the EUR-equivalent of the overall results presented in Table 10
on page 19. We note that the premium is significant at 95% and that the distribution
of premia is left-skewed.
(b) Table 27: EUR-equivalent of the breakdown per credit rating presented in Table 12 on
page 21. The premium is negative but only Aaa is significant at 95%
(c) Table 28: EUR-equivalent of the breakdown per sector presented in Table 15 on page
22. The premium is negative but no sector is significant.
(d) Table 29: EUR-equivalent of the breakdown per region presented in Table 16 on page
22. The premium is negative and significant for bonds labelled in EUR and whose
country of risk is located in Europe.
(e) Table 30: EUR-equivalent of the breakdown per ESG rating presented in Table 17 on
page 23. The premium is higher for worst-rated bonds.
(f) Table 31: EUR-equivalent of the E rating presented in Table 24 on page 36. Like the
ESG rating, the premium is higher for worst-rated bonds.
(g) Table 32: EUR-equivalent of the breakdown per time to maturity presented in Table
18 on page 23. The premium is negative but only significant for bonds whose time to
maturity is between 10 and 20 years.
(h) Table 33: EUR-equivalent of the breakdown per liquidity proxy presented in Table 19
on page 24. The result differs from the one obtained for the full sample. Here, less
liquid GB exhibit significant lower premia due to a negative skewness.
(i) Table 34: EUR-equivalent of the breakdown per liquidity proxy presented in Table 20
on page 24. Thes results are in line with the result obtained for the whole universe.
The premium is 24 bps lower than in full sample.
(j) Table 35: EUR-equivalent of the breakdown per certification presented in Table 21 on
page 24. The results are in line with the result obtained for the whole universe.
2. Top-down approach
(a) Table 36: EUR-equivalent of the breakdown per time to maturity presented in Table 6
on 13.
(b) Table 37: EUR-equivalent of the breakdown per sector presented in Table 7on 14.
(c) Table 38: EUR-equivalent of the breakdown per rating presented in Table 8on 15.
38
Facts and Fantasies About the Green Bond Premium
Table 26: EUR Universe: Results
Metric Mean Std dev. Median Skewness Kurtosis T-statistic
CB Spread 55.32 47.14 44.42 2.19 7.80 12.74 ***
GB Spread 53.70 46.45 43.67 2.20 7.57 12.55 ***
Premium -1.62 8.02 -0.88 -1.41 19.85 -2.19 **
Duration 6.39 4.00 5.40 1.79 4.05 17.32 ***
Table 27: EUR Universe: Breakdown per credit rating
Premium Av Spread Av MD
Rating Mean Std dev. Skew Kurtosis N. Obs T-stat
Aaa -1.18 2.88 -0.77 1.80 1 843 -1.99 ** 20.90 7.34
Aa -0.80 3.61 -0.28 2.06 2 593 -1.29 41.03 7.02
A -2.08 8.91 -1.85 6.69 2 704 -1.37 56.36 5.87
Baa -2.43 12.56 -0.51 10.47 2 054 -0.99 95.64 5.43
Table 28: EUR Universe: Breakdown per sector
Premium Av Spread Av MD
Sector Mean Std dev. Skew Kurtosis N. Obs T-stat
Agencies -1.21 4.04 -2.34 17.02 2 152 -1.57 47.47 8.17
Covered -0.28 1.11 0.32 1.26 880 -0.85 2.81 5.28
Financials -1.11 6.94 -0.80 2.96 2 447 -0.90 57.04 3.73
Other Corporates -2.93 10.13 -2.41 11.42 231 -0.50 68.80 4.05
Sovereigns -0.30 4.39 0.36 2.59 454 -0.17 44.64 12.25
Spec finance -1.31 8.73 0.91 4.63 595 -0.41 102.96 5.53
Supras -2.22 3.72 -0.28 0.36 515 -1.53 27.78 9.63
Utilities -3.41 13.36 -0.84 8.90 1 920 -1.27 71.78 6.58
Table 29: EUR Universe: Breakdown per region
Premium Av Spread Av MD
Region Mean Std dev. Skew Kurtosis N. Obs T-stat
Europe -1.91 8.17 -1.35 20.11 8 250 -2.41 ** 53.32 6.58
North America 2.27 4.37 0.26 0.47 359 1.11 62.65 5.23
Asia & Pacific -0.54 8.26 -1.62 5.24 346 -0.14 63.44 4.62
Others 1.19 3.09 1.07 2.80 239 0.67 39.55 4.08
39
Facts and Fantasies About the Green Bond Premium
Table 30: EUR Universe: Breakdown per ESG-rating
Premium Av Spread Av MD
Rating Mean Std dev. Skew Kurtosis N. Obs T-stat
A -1.63 4.97 0.32 5.58 1 212 -1.30 51.48 7.87
B -2.87 8.04 -5.19 45.66 1 944 -1.78 * 54.29 6.67
C -1.74 9.03 -0.40 10.98 4 396 -1.45 53.21 6.04
D 0.56 6.22 1.01 6.83 1 310 0.37 58.26 5.94
E-F 0.21 10.10 -2.05 4.82 161 0.03 65.82 3.52
NR -2.36 2.36 -0.54 -0.21 171 -1.48 29.20 7.87
Table 31: EUR Universe: Breakdown per E-rating
Premium Av Spread Av MD
Rating Mean Std dev. Skew Kurtosis N. Obs T-stat
A -1.66 4.15 -0.34 2.92 626 -1.13 50.87 4.90
B -1.90 7.51 -4.59 43.36 2 950 -1.56 48.45 6.40
C -1.56 8.71 -0.29 11.75 4 563 -1.37 55.62 6.80
D -0.59 8.95 -0.13 4.70 723 -0.20 67.97 5.43
E-F -1.61 7.16 -1.77 8.92 161 -0.32 68.62 3.16
NR -2.36 2.36 -0.54 -0.21 171 -1.48 29.20 7.87
Table 32: EUR Universe: Breakdown per time to Maturity
Premium Av Spread Av MD
Maturity Mean Std dev. Skew Kurtosis N. Obs T-stat
1 - 3 yrs -1.55 7.22 -2.42 13.80 1 302 -0.88 36.08 2.07
3 - 5 yrs -1.12 5.70 -0.73 4.52 2 316 -1.07 50.02 3.99
5 - 7 yrs -1.55 6.33 -0.35 3.50 2 043 -1.25 50.09 5.70
7 - 10 yrs -1.36 5.46 -0.61 6.60 2 083 -1.29 51.76 8.05
10 - 20 yrs -5.44 10.13 -2.51 6.80 873 -1.80 * 60.76 13.11
Beyond 20 yrs 0.87 18.53 -0.85 6.58 577 0.13 117.41 12.04
Table 33: EUR Universe: Breakdown per Liquidity
Premium Av Spread Av MD
Liquidity Mean Std dev. Skew Kurtosis N. Obs T-stat
GB less Liquid -2.07 7.68 -2.02 26.81 4 208 -1.98 ** 54.72 6.23
GB more Liquid -1.23 8.28 -1.02 15.37 4 986 -1.19 52.85 6.52
40
Facts and Fantasies About the Green Bond Premium
Table 34: EUR Universe: Breakdown per Liquidity
Premium Av Spread Av MD
Liquidity Mean Std dev. Skew Kurtosis N. Obs T-stat
Different Liquidity -1.49 8.03 -1.61 21.74 6 021 -1.63 53.84 6.08
Same Liquidity -1.86 8.01 -1.03 16.39 3 173 -1.48 53.44 6.97
Table 35: EUR Universe: Breakdown per Certification
Premium Av Spread Av MD
Certification Mean Std dev. Skew Kurtosis N. Obs T-stat
Not Certified -1.26 6.03 -0.43 0.61 559 -0.56 54.26 4.43
Certified -1.64 8.13 -1.42 19.94 8 635 -2.12 ** 53.67 6.52
Certified by CBI -3.58 9.61 -2.87 9.62 1 422 -1.59 48.89 7.10
Certified by others -1.26 7.75 -0.81 23.36 7 213 -1.56 54.61 6.40
Table 36: EUR universe: Maturities - 2016 – 2020
Maturity Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
All maturities -7.30 2.39 -13.16 -7.82 -3.41 -0.45 -0.28 -3.06 ***
1 - 3 yrs -1.25 3.21 -7.43 -1.86 6.17 0.60 0.09 -0.39
3 - 5 yrs -0.73 6.04 -9.38 -1.47 21.23 1.70 4.06 -0.12
5 - 7 yrs -9.52 5.22 -23.24 -10.40 3.79 0.18 1.11 -1.82 *
7 - 10 yrs -14.64 4.51 -24.83 -14.31 -7.19 -0.24 -0.61 -3.25 ***
10 - 20 yrs -6.20 5.41 -18.85 -5.68 2.28 -0.30 -0.96 -1.15
Beyond 20 yrs -5.20 4.71 -11.86 -6.24 5.38 0.89 0.15 -1.10
Table 37: EUR universe: Sectors - 2016 – 2020
Sector Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
All sectors -7.30 2.39 -13.16 -7.82 -3.41 -0.45 -0.28 -3.06 ***
Agencies -12.25 4.38 -24.82 -11.05 -5.71 -0.85 0.37 -2.79 ***
Covered -6.83 5.33 -29.99 -5.61 -2.19 -3.46 13.39 -1.28
Financial-Institutions -17.99 8.05 -43.98 -15.05 -8.33 -1.55 2.04 -2.23 **
Industrials 4.01 7.81 -16.26 4.52 18.00 -1.03 1.41 0.51
Local-Authorities 4.42 6.86 -3.99 2.22 24.62 1.66 2.58 0.64
Sovereign -4.16 9.46 -28.04 -6.19 13.27 -0.24 0.02 -0.44
Supranational -0.72 2.35 -4.04 -1.62 5.92 1.50 1.54 -0.31
Treasury -4.26 5.69 -13.26 -4.18 6.25 0.15 -1.17 -0.75
Utilities -2.61 8.73 -22.66 -1.73 14.40 -0.30 -0.01 -0.30
41
Facts and Fantasies About the Green Bond Premium
Table 38: EUR universe: Ratings - 2016 – 2020
Rating Mean Std dev. Min Median Max Skewness Kurtosis T-statistic
All ratings -7.30 2.39 -13.16 -7.82 -3.41 -0.45 -0.28 -3.06 ***
Aaa -1.23 1.97 -4.03 -1.50 2.87 0.49 -0.88 -0.62
Aa 0.64 4.52 -4.80 -1.18 15.86 0.89 0.78 0.14
A -9.33 4.38 -23.94 -8.18 -2.38 -1.15 1.79 -2.13 **
Baa -27.37 13.68 -66.24 -24.49 -7.24 -1.02 0.75 -2.00 **
A.4 Figures
Figure 16 shows per date the number of the outliers whose premium is below 4σor above 4σ,
with σis the premium distribution standard deviation.
Figure 16: Overall Premium: Number of outliers
A.5 Mathematical results
For a given bond, the relationship between the excess return R and the changes in spread yield
δy follows in first approxmiation the relation:
R=MD ·δy (5)
where MD is the spread duration.
We apply Equation(5) to a portfolio of bonds and we calculate a weighted spread change δyp.
42
Facts and Fantasies About the Green Bond Premium
δyp=PiQωi·MDp
i·δyp
i
PiQωi·MDp
i
(6)
where ωi,MDp
i,δyp
iare the market weight, spread duration and spread change for sector i, and
the sums are over all portfolio sectors.
At time t, ∆ return, the difference in excess returns between the green portfolio and the global
bond index, can be written then as follows:
return =RGRB=M DG·δyGM DB·δyB(7)
where RG,MDG,δyG(resp. RB,M DB,δyB) are the excess return, the spread duration, and the
overal change in spreads of the green portfolio (resp. the global bond index).
Equation(7) can be rearranged to show an allocation return and a selection return:
return =(M DGM DB)·δyB
| {z }
Allocation return
MDG·δyGδyB
| {z }
Selection return
(8)
Table 39 displays the correlations between the three returns defined above. We make use of the
weekly excess returns of both green portfolio and the global bond index. We understand that
the ∆ return is driven by the selection return, as the correlation with the selection return is high
(+0.82) while the correlation with the allocation return is null (0.08).
Table 39: Correlations between returns
return Allocation return Selection return
return 1
Allocation return -0.08 1
Selection return +0.82 -0.64 1
43
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Central banks’ frameworks for managing foreign exchange reserves have traditionally balanced a triad of objectives: liquidity, safety and return. Pursuing these objectives involves explicit tradeoffs. Recently central banks have shown interest in incorporating environmental sustainability objectives into their reserve management frameworks. Rather than a triad, central banks may analyse (and weigh) a tetrad of reserve management objectives in allocating part of their foreign exchange reserves to green bonds.
Article
Full-text available
We examine the pricing difference of Green Bonds (GB) and conventional bonds (CBs) in capital markets worldwide. Credit spread is used to observe whether investors would like to pay a premium for GBs over CBs. This study uses panel data regression with hybrid model to analyse daily observations over the period 2016 to 2017. We employ Option-Adjusted spread (OAS) to measure the credit spreads of bonds while controlling for bond specific, macroeconomic and global factors that influence the spread. With the hybrid model used in the panel data analysis, we were able to capture the fixed-effects of variables in a random effect model. We find that GBs are traded at a premium of 63 basis points (BPS), compared with a comparable corporate bond issue. We find that the green label provides issuers an incentive to raise funds through issuing GBs while providing investors an opportunity to diversify their investments returns. Our findings provide several implications to the major stakeholders driving the GB market to scale up the market to finance the required level of global green investment needs. We stress an urgent need to support the growth of the GB market to achieve sustainable development through mitigating climate change challenges. Abbreviation GB: Green Bond; CB: Conventional Bond; YS: Yield Spread; BPS: Basis Points; OAS: Option-Adjusted Spread; PCSE: Panels Corrected Standard Errors; CPI: Consumer Price Index; GBPs: Green Bond Principles; CBS: Climate Bond Standard
Article
Full-text available
If we examine the characteristics of a sample of green bondsmatched with their closest brown bond neighbors, we encounter a challenge. Green bonds have higher yields, lower variance, and are more liquid. The institutional/private issuer and the green third-party verification/non-verification breakdowns help explain this puzzle. Green bonds from institutional issuers have higher liquidity with respect to their brown bond correspondents and negative premia before correcting for their lower volatility. Green bonds fromprivate issuers havemuch less favorable characteristics in terms of liquidity and volatility but have positive premia with respect to their brown correspondents, unless the private issuer commits to certify the "greenness" of the bond. An implication of our findings is that the issuer's reputation or green third-party verifications are essential to reduce informational asymmetries, avoid suspicion of green (bond)-washing, and produce relatively more convenient financing conditions.
Article
Climate finance is the study of local and global financing of public and private investment that seeks to support mitigation of and adaptation to climate change. In 2017, the Review of Financial Studies launched a competition among scholars to develop research proposals on the topic with the goal of publishing this special volume. We describe the competition, how the nine projects featured in this volume came to be published, and frame their findings within what we view as a broader climate finance research program.
Article
In this paper, we empirically investigate how greenness information is priced in the green bond market. Our comparison of liquidity‐adjusted yield premiums of green bonds versus synthetic conventional bonds indicates that, on average, there is no robust and significant yield premium or discount on green bonds. However, green bonds certified by an external reviewer enjoy a discount of about 6 bps. Furthermore, green bonds that obtain a Climate Bonds Initiative certificate show a discount of around 15 bps. The findings suggest that a universally accepted greenness measure can benefit the development of the green bond market.