Next Article in Journal
Water allocation analysis of the Zhanghe River basin using the Graph Model for Conflict Resolution with incomplete fuzzy preferences
Next Article in Special Issue
Fiscal Decentralization, Local Competitions and Sustainability of Medical Insurance Funds: Evidence from China
Previous Article in Journal
An Assessment Framework for Grassland Ecosystem Health with Consideration of Natural Succession: A Case Study in Bayinxile, China
Article
Peer-Review Record

The Green Bonds Premium Puzzle: The Role of Issuer Characteristics and Third-Party Verification

Sustainability 2019, 11(4), 1098; https://doi.org/10.3390/su11041098
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Sustainability 2019, 11(4), 1098; https://doi.org/10.3390/su11041098
Received: 23 December 2018 / Revised: 3 February 2019 / Accepted: 14 February 2019 / Published: 19 February 2019
(This article belongs to the Special Issue Social Impact Investments for a Sustainable Welfare State)

Round  1

Reviewer 1 Report

- PARAGRAPH 2 - LINE 131: please, see also the recently published Report of the TEG available at the following link: https://ec.europa.eu/info/sites/info/files/business_economy_euro/banking_and_finance/documents/190110-sustainable-finance-teg-report-climate-related-disclosures_en.pdf

- PARAGRAPH 3 - LINES 175, 177, 179, 204: more than using the expression "investors willing to pay a premium/price premium" better to put it the other way round as for example "investors willing to receive lower yeld"

- PARAGRAPH 3 - LINE 235-237: it could bring added value to the analysis to compare the green bonds of the sample with "climate aligned" green bonds and not only with brown matched ones;

-  PARAGRAPH 3 - LINES 264-271, 283 AND TABLE 1: in order to define the bond couples do you consider the date of the issue of the two matched bonds?  

- PARAGRAPH 3 - LINES 267-271: please give explanation for choosing 2 years lag and +/- 0.25%

- PARAGRAPH 3.4: no consideration has been paid to the financial structure of the bonds in terms of amount, tranching of the notes and guarantees offered to investors: it may be the case that institutional issues can be featured as the ones of larger amounts, different categories of investors protected by public collaterals.

- IN GENERAL it would be worth restructuring paragraf 3 titled "Research Hypothesis" but including also findings of the work.

Author Response

Dear reviewer,

Thanks a lot for the useful comments and suggestions that helped us to improve the quality of our paper. In what follows we explain how we took into account all the points raised.

We submit a final version with all changes in revision mode (included those of the English editor’s revision suggested by the Editor and one cleaner without revision mode

Best regards

 

- PARAGRAPH 2 - LINE 131: please, see also the recently published Report of the TEG available at the following link: https://ec.europa.eu/info/sites/info/files/business_economy_euro/banking_and_finance/documents/190110-sustainable-finance-teg-report-climate-related-disclosures_en.pdf

 

RESPONSE

We follow the advice and update the section reviewing the institutional action and suggestions on green bond standards by including insights from the TEG report.

More specifically we mention that in the recently published European Commission TEG report on Sustainable Finance in 2019, it has been suggested to companies to consider disclosing their green bond ratio (the proportion of the total green bond outstanding amount over the total bond outstanding amount) or their green debt ratio (total amount of  green bond or green debt instruments over the total debt amount) in order to transparently communicate the intensity of their  low carbon transition plan supported by debt financing activity.

This consideration stresses the relevant point of the importance of having information to assess the overall corporate effort that may not be clear when focusing on the specific bond initiative.

The Report is much wider concerning investigation on different financial activities (lending, mortgages, asset management, insurance) on different types of climate risks (underwriting risk, market risk, strategic risk, investment risk, operational risk, reputational risk). We consider from it what is crucial in the specific point of the report related to green bonds.

 

- PARAGRAPH 3 - LINES 175, 177, 179, 204: more than using the expression "investors willing to pay a premium/price premium" better to put it the other way round as for example "investors willing to receive lower yield"

RESPONSE

Amended in all of the four lines

 

- PARAGRAPH 3 - LINE 235-237: it could bring added value to the analysis to compare the green bonds of the sample with "climate aligned" green bonds and not only with brown matched ones;

RESPONSE

The advantage of working on the green versus brown bond comparison stands in the clarity of definition of the groups we compare.

We actually have three groups:

i)                green bonds that have been externally certificated,

ii)              non certified bonds that have been labelled as green by the issuer and that respect the CBI principles and

iii)             the brown bond counterparts of each green.

By definition climate aligned bonds are bonds that are financing green/climate assets that help enable a low-carbon economy but have not been labelled as green by the issuing entity.

(as from the definition from the PRI (Principle for Responsible Investment) website https://www.unpri.org/climate-change/low-carbon-investing-and-green-and-climate-aligned-bonds/3284.article)

The absence of a label and of clear thresholds that distinguish climate aligned bonds from brown bonds make it difficult to make a comparison of green bonds with climate aligned bonds (or of climate aligned bonds with brown bonds).

The difficulty is increased since it is more likely that money raised by climate-aligned bonds will be used for other destinations or investment given that the issuer does not officially commit to use money raised for green investment.

In view of these problems related to a rigorous taxonomy when dealing with climate-aligned bonds we decided to work on the three groups as defined above

Note that the reviewer’s concerns lead us to clarify the distinction between group i) and group ii) using the taxonomy (third-party) verified green bonds/non-verified green bonds instead of labelled-unlabelled green bonds.

 

-  PARAGRAPH 3 - LINES 264-271, 283 AND TABLE 1: in order to define the bond couples do you consider the date of the issue of the two matched bonds?  

RESPONSE

Our choice is to follow Zerbib (2019) approach and limiting issue date differences within six years. Note however that green bonds are compared with brown bond counterparts having closer maturity (2-year maximum difference) and in the same trading day.

There is obviously a trade-off between the available number of matched pairs and closeness between green and brown twins. The more we restrict criteria, the lower the pair of available “twins”. Our choice in this trade-off follows the approach used in the mentioned paper on this point.

Note as well that the difference in issue date between a green bond and its brown “twin” is time invariant. Hence it is controlled for in our fixed effect estimates.

We as well check the average (green vs brown “twin”) issue date difference in our sample and find that the median is zero and the mean 0.3 with standard deviation 2.6.

It would be almost impossible to find a match also on this third characteristic (date of issue) when matching the other two. This is why more degrees of freedom in terms of matching the date of issue are used by both us and Zerbib (2019). In addition to it, we expect that issue date should not matter significantly as regressor in the estimate (the variable is omitted by Zerbib, 2019).


 

Reference

 

Zerbib, Olivier David. "The effect of pro-environmental preferences on bond prices: Evidence from green bonds." Journal of Banking & Finance 98 (2019): 39-60.

 

- PARAGRAPH 3 - LINES 267-271: please give explanation for choosing 2 years lag and +/- 0.25%.

 

As explained in the previous response we must choose a point in the trade-off.

The two thresholds are a reasonable compromise between the need of closeness between green and brown bond and the need of having a reasonable number of observations.

Despite of the above intervals, summary statistics show that the matching is well done because there are not big differences in the matched bonds amount and coupon. In addition to it, the remaining difference in amount, coupon, liquidity and maturity date for each couple is controlled for with the ΔLiquidity, ΔAmount, ΔCoupon and ΔMaturity variables.

Furthermore, the 2-year lag threshold is the same threshold Zerbib (2019) used. For what concerns differences in the coupon rate, Zerbib (2019) and many other relevant papers in the field that used bond matched pairs to control for issuer credit risk (i.e. Helwege and Turner, 1999, Dick-Nielsen et al. 2012) do not take into account this characteristic in the matching. When choosing our matching criteria, we obviously face a trade-off between number of matched bonds and accuracy of the matching. In particular, we choose a threshold that is slightly larger with respect to Helwage et al. (2014), that, in order to study the effect on liquidity on the yield spread, fixed as maximum difference in coupon rate 1.5%.

References

Zerbib, Olivier David. "The effect of pro-environmental preferences on bond prices: Evidence from green bonds." Journal of Banking & Finance 98 (2019): 39-60.

 

Helwege, Jean, and Christopher M. Turner. "The slope of the credit yield curve for speculative‐grade issuers." The Journal of Finance 54.5 (1999): 1869-1884.

 

Dick-Nielsen, Jens, Peter Feldhütter, and David Lando. "Corporate bond liquidity before and after the onset of the subprime crisis." Journal of Financial Economics 103.3 (2012): 471-492.

 

Helwege, J., Huang, J. Z., & Wang, Y. (2014). Liquidity effects in corporate bond spreads. Journal of Banking & Finance, 45, 105-116.

 

- PARAGRAPH 3.4: no consideration has been paid to the financial structure of the bonds in terms of amount, tranching of the notes and guarantees offered to investors: it may be the case that institutional issues can be featured as the ones of larger amounts, different categories of investors protected by public collaterals.

RESPONSE

We consider the amount as requested by the reviewer in the matching procedure as it is customary in this literature (see table 1). In addition to it, given the interval of discretion allowed, we control for ΔAmount in our estimates. The variable is therefore accounted for.

We as well double check and specify in this new version of the paper that there are not any differences in investor guarantees in the green and brown bonds. In particular, by construction of the matching pairs, beyond being issued by the same issuer, they also have the same level of seniority (no differences of protection). The bonds have also been matched looking at their structure.

- IN GENERAL it would be worth restructuring paragraf 3 titled "Research Hypothesis" but including also findings of the work.

RESPONSE

We are afraid we did not fully understand this point. We generally formulate our research hypotheses, describe data and empirical methodology followed and then present and comment empirical findings. Findings are as well anticipated in abstract and introduction and mentioned again in conclusions.  

We think it is preferable to follow this order. But if the reviewer believes it is an important issue we ready to reconsider


Author Response File: Author Response.docx

Reviewer 2 Report

Overall, I found the manuscript interesting, the topic of high importance and relevant as well as the presented research results. However, I think that the presentation & the use of English is relatively weak and require improvement.

Minor Comments:

(a) Mentioning twice the word 'green' in the title is redundant.

(b) There is no comment in the paper regarding the zero values of R-squared shown in several tables. Why the authors do not discuss any diagnostics tests of the estimated models?

(c) The authors need to discuss and argue further on their rationale for choosing OLS and FE models, and not other panel model specifications.

(d) In my view, an important issue regarding the green bond market has to do with the fact that, currently supply fails to keep with demand; green bonds are not yet considered a mainstream investment product. Obviously, lack of supply according to market reports has a role to play in the authors' research and deserves discussion in the paper.   

(e) The role of the ratings of green bonds needs to be discussed, if not researched further in the paper.

(f)The distribution of green bond issuances by country/sector/rating is not very clear as shown in Fig. 2. Could it be that pie-charts improve visibility?

(g) Terms like 'greenwashing' need to be properly explained in the text.

(h) Several references seem to be incomplete.


Author Response

Dear reviewer,

Thanks a lot for the useful comments and suggestions that helped us to improve the quality of our paper. In what follows we explain how we took into account all the points raised.

We submit a final version with all changes in revision mode (included those of the English editor’s revision suggested by the Editor) and one cleaner without revision mode

Best regards

 

 

Overall, I found the manuscript interesting, the topic of high importance and relevant as well as the presented research results. However, I think that the presentation & the use of English is relatively weak and require improvement.

Minor Comments:

(a)   Mentioning twice the word 'green' in the title is redundant.

 

We changed our title to:

The Green Bonds Premium Puzzle: The Role of Issuer Characteristics and Third-Party Verification”. We replace also Certification with Verification for better consistency with our approach explained in Figure 1

(b)  There is no comment in the paper regarding the zero values of R-squared shown in several tables. Why the authors do not discuss any diagnostics tests of the estimated models?

We agree with the reviewer that more has to be said on this point.

First of all, note that for a specification with fixed effects and just the intercept without other controls the variance of the residuals is equal to the total variance and for this reason the R-square is equal to 0. We use this specification (in which we compute the mean and the variance) only as a first benchmark for comparison with the augmented specifications that follow. We now comment in the paper that the introduction of the two liquidity variables in Table 4 is responsible for a substantial increase in goodness of fit (.2 in the R-Square).

The close to 0 R-square in Table 5, column one (even when adding controls) in the liquidity estimate implies that the daily difference in liquidity is purely random with the exception of the significance of the intercept.

Note as well that poor goodness of fit in our regressions is expected since regressions where the dependent variable is in first differences have poor goodness of fit and regressions where the first difference is regressed on presumed differences between “twins” as in our case is expected to have an even poorer goodness of fit. In other terms it is highly likely that daily first differences between twins are random walk

 We introduced part of these considerations in the paper.

(c)   The authors need to discuss and argue further on their rationale for choosing OLS and FE models, and not other panel model specifications.

We prefer FE model to RE effect model as it is customary when time invariant unobservable omitted variables could be correlated with both our depended and independent variables thereby producing biases RE estimates. We as well show in the paper that results are still statistically significant when we depart from the OLS normality assumption on the dependent variable by using bootstrapped standard errors (Table 9). Finally, we also show that findings are robust when we take into account the presence of bounds in our dependent variable and use the Tobit estimator that it is estimated through maximizing the likelihood function (Table 10).

 

(d)  In my view, an important issue regarding the green bond market has to do with the fact that, currently supply fails to keep with demand; green bonds are not yet considered a mainstream investment product. Obviously, lack of supply according to market reports has a role to play in the authors' research and deserves discussion in the paper.   

Excess demand (or lack of supply) should produce lower yields for green versus brown bonds. We however observe that the green vs brown bond yield differential is in on average positive and significant in our overall sample. It becomes negative in the sample of institutional issuers. We comment that this finding should be given by the higher reputation of the institutional issuers thereby attracting higher demand. Hence, more than the general excess demand in the green versus brown bond market what seems to emerge from our results is a demand differential between institutional versus private green bond issuers.

 

(e)   The role of the ratings of green bonds needs to be discussed, if not researched further in the paper.

The role of ratings has already been investigated in the literature. Zerbib (2018) shows that green bond premium is greater for low-rated bonds. On the other side, Hackenberg and Schiereck (2018) show that rating classes AA-BBB of green bonds trade marginally tighter with respect to brown bonds. These results are consistent with ours. In particular, borrowers with high credit scores are more trusted by the market. On the other side, the literature shows that low quality borrowers are more likely to be less environmentally responsible and for this reason the risk of greenwashing is higher.

 

We follow the reviewer’s advice and introduce further discussion in the literature. More specifically we quote and comment papers that worked on this point previously

 

References

Zerbib, Olivier David. "The effect of pro-environmental preferences on bond prices: Evidence from green bonds." Journal of Banking & Finance 98 (2019): 39-60.

 

Hachenberg, Britta, and Dirk Schiereck. "Are green bonds priced differently from conventional bonds?." Journal of Asset Management 19.6 (2018): 371-383.

 

(f)The distribution of green bond issuances by country/sector/rating is not very clear as shown in Fig. 2. Could it be that pie-charts improve visibility?

We follow the advice and create the new figure using pie-charts

(f)    Terms like 'greenwashing' need to be properly explained in the text.

With the term “greenwashing” we mean companies that declare a commitment to environmental responsibility higher than the reality. The benefit of greenwashing (temptation function) is the advantage gained in terms of reputation and extra demand from consumers willing to pay of environmental responsibility. The cost of greenwashing (punishment function) is the “sanction” arising when the public opinion becomes aware of the gap between declarations and facts. As an example of these costs consider the case of Volkswagen whose stock recorded a 20 percent loss on 21 September 2015 after the EPA’s notice of violation became public. The price remained 30 percent lower at one-year distance.

We introduce the point in a footnote.

(g)   Several references seem to be incomplete.

References have been completed (ie. Zerbib) and updated using the style indicated by the editor of the journal (ACS format instead of Chicago format)

 


Author Response File: Author Response.docx

Back to TopTop