What is a green bond?

Green bonds refer to debt securities issued to raise capital earmarked for green projects. The exact definition, however, varies depending on what constitutes green projects. For example, the Green Bond Principles (GBPs), which were established in 2014 and are maintained by the International Capital Market Association (ICMA), provide guidelines and green project categories (ICMA 2021). The Climate Bonds Standards (CBSs), built on top of the GBPs by the Climate Bonds Initiative (CBI), provide a sector-specific definition of “green” and are used for the certification of green bonds by CBI (Climate Bonds Initiative 2019).

For example, the CBSs categorize eligible projects into eight groups: energy, transport, water, buildings, land use and marine resources, industry, waste and pollution control, and information communications and technology.¹ Thus green bonds cover a wide range of environmental activities, some of which could be broader than climate objectives. As reporting requirements, the CBSs ask the issuers to document the use of proceeds, the process for evaluation and selection of projects and assets, and the management of proceeds, both before the issuance and annually after the issuance. A green bond issuer can obtain certification if the issuer pays fees to one of the verifier organizations and it confirms that the CBSs are met. This is, however, a private initiative, so compliance by bond issuers to the CBSs is voluntary.

Green bond data can differ across databases. ICMA (2017) explores four databases (Bloomberg, Environmental Finance, Dealogic, CBI) and discusses the difference in the definitions. For example, Bloomberg tags the “Green Bond” label when an issuer self-labels its bond as green or declares its compliance with the GBPs on the use of proceeds.² The Green Bond Database by Environmental Finance lists all bonds that are self-labeled as “Green.” Eikon is another database that provides green bond data, whose definition is aligned with the CBSs; the data are reviewed by CBI. Thus the analysis of green bonds, in general, should be understood with caveats on the data. The analysis in this paper relies on Eikon as it is consistent with the CBSs and has been used extensively in the literature of the sovereign green bond (for example, Doronzo, Siracusa, and Antonelli 2021). For sovereign green bonds, Eikon and Bloomberg are comparable.

The public sector has accelerated its development of definition and regulatory framework following private initiatives. For example, the People’s Bank of China issued guidelines in 2015 and a catalog in 2021, defining the projects that are eligible for green bond issuance (People’s Bank of China 2021). The European Union adopted the regulation of EU taxonomy in 2020 that defines environmentally sustainable economic activities. In 2021, the European Commission proposed the legislation of the European green bond standard (European Commission 2021). Many other countries have issued green bond guidelines and frameworks as summarized by CBI (2022).³

Evolution of sovereign green bonds

Sovereign green bonds are a recent phenomenon, starting in 2016. The literature often cites the bond issued by the European Investment Bank in 2007 as the first green bond (Cortellini and Panetta 2021; OECD 2021).⁴ Since 2007, international organizations, municipalities, and private sectors have increased the issuance. Until 2015, although the annual issuance of green bonds had reached $40 billion, no issuance by central governments was recorded. In 2016, building on the momentum of the Paris Agreement adopted in 2015, Poland became the first issuer of sovereign green bonds.

A wide range of sovereigns has issued green bonds since 2016. Figure 1, panel 2, shows the green bond in Eikon issued by the central government.⁵ The list of green bonds used for the figure is provided in Annex 1. Poland was the first country to issue a sovereign green bond in 2016, followed by France in 2017, and the issuance recorded nearly $80 billion in 2021. Most issuance was by advanced economies until February 2022 (Figure 1, panel 1).⁶ Geographically, the cumulative issuance from 2016 to 2021 is mostly concentrated in European countries ($161 billion), followed by Asian Pacific countries ($9 billion), Western hemisphere countries ($8 billion), the Middle East and Central Asian countries (less than $1 billion), and African countries (less than $1 billion).

View Full Size

Green bonds issued by central government in billions of US dollars

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Source: Eikon and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

France has issued nearly $48 billion for green projects and is the largest issuer as of February 2022. In terms of cumulative green bond issuance relative to GDP at the country level, Chile has the highest ratio of 2.37 percent (relative to its GDP in 2021). Others are all below 2 percent as of February 2022. The average maturity as of issuance is 12.6 years with a standard deviation of 8.4 years. Although the holder’s information is not available from Eikon, Doronzo, Siracusa, and Antonelli (2021) suggest that real money investors, such as pension funds, sovereign funds, and insurance companies, invest their money with a long-term perspective and a buy-and-hold strategy in Europe.

The main purpose of issuing green bonds in the sample is clean transport (Figure 2). The classification of purpose is not necessarily mutually exclusive, but the available data in Eikon shows that clean transport is the main purpose of green bonds. The share of climate change adaptation and aquatic biodiversity conservation is also significant. One caveat is that the purpose of green bonds is classified by Eikon and may not reflect all purposes of each bond issued.

View Full Size

Green bond issued amount by purpose

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Source: Eikon and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Costs and benefits of sovereign green bond issuance

Several costs associated with sovereign green bond issuance have been discussed in the literature. For example, Doronzo, Siracusa, and Antonelli (2021) discusses three types of costs relative to the costs of the conventional bond:

• 1. Green bond requires more disclosure and tracking for the use of proceeds. For example, if a green bond issuer wants certification from CBI, documentation to show that the CBS is met and engagement with verifiers is needed. But more information provision could lead to less uncertainty for buyers, so the net impact is not clear.

• 2. The reputation of the issuer could be damaged if the green project that the green bonds finance fails or is perceived as greenwashing (falsely claiming that the financed investment is green). The net impact is again ambiguous since the green bond plays the role of a commitment device and thus can lower the probability of failure by motivating better planning and governance of the projects.

• 3. The issuance of green bonds can crowd out that of conventional bonds, resulting in lower liquidity and higher funding costs for both segments. Doronzo, Siracusa, and Antonelli (2021) summarize Danish and German techniques to mitigate the liquidity problems. For example, the Germany Finance Agency mitigates the impact on the liquidity of conventional bonds by increasing its stock of conventional bonds at the time it issues green bonds by the same amount. The additional own holdings in conventional bonds can be used on the secondary market for repo transactions or for lending activities.

OECD (2021) also points to gaps in supply constraints. A pipeline of green projects needs to be established to sustain the supply and liquidity of the green bonds. OECD (2021) argues that the supply constraints can be mitigated by utilizing technical assistance from experts and aggregation of small-scale projects with securitization.⁷ Another obstacle to sovereign green bonds is that most sovereign debt legal frameworks do not allow the earmarking of proceeds.

The literature discusses a wide range of benefits. OECD (2017) points out their reputational benefits and their role as a commitment device, among other benefits. Unlike conventional bonds whose proceeds can be used for general purposes, the proceeds from green bonds need to finance green projects, tying the hands of the issuer. This commitment to finance green projects can send signals and improve the reputation of the issuer, leading to a higher price of the issuer’s nongreen bonds (halo effect). For sovereign issuers, Doronzo, Siracusa, and Antonelli (2021) mention that the issuance of sovereign green bonds can encourage other issuers to enter the green bond market as it provides a market benchmark. Doronzo, Siracusa, and Antonelli (2021) also argue that green bonds tend to be issued with a long maturity, so the refinancing risk is lower, and the benefit could be larger for emerging or less-developed countries that have less stable demand for extra-long maturities.

A central benefit associated with green bonds has been labeled as the green premium (greenium). When a green bond exhibits a lower yield compared to a similar conventional bond without the green label, the green bond is said to exhibit positive greenium.

Greenium = yield of similar conventional bond – yield of green bond

A positive greenium implies that the price of the green bond is higher than that of a similar conventional bond.

Theoretically, the greenium can take either positive or negative signs. On one hand, the issuance amount and liquidity are smaller than the conventional bond, which could lead to a negative greenium. On the other hand, environmental, social, and governance investors’ demand for green bonds and more information on the use of proceeds can justify a positive greenium.⁸ Thus whether a green bond is traded at greenium is an empirical question.

The literature on sovereign green bond greenium is limited. Doronzo, Siracusa, and Antonelli (2021) discuss that the evidence of sovereign greenium reported by private financial institutions is mixed and estimate greenium in both the primary and secondary market using Eikon’s data. They show that the greenium is negative in the primary market but is slightly positive (0.5 bps) in the secondary market. IMF (2021a, 2021b) shows that the greenium of 5- and 10-year green bonds are around 3 to 5 bps and that the greenium implied by swap spreads from 1 to 7 bps for six EU countries. In the context of the US municipal bonds, Karpf and Mandel (2018) find that the greenium was negative but has turned positive recently, suggesting that green bonds have become more attractive to investors in recent years. Baker and others (2021) also find that the greenium is positive except when it is issued simultaneously with ordinary bonds from the same issuer; in that case, a premium emerges over time on the secondary market.

How large is the sovereign greenium?

Since the literature on the sovereign green bond is scarce, this section provides greenium estimates. We first present the result of Germany since it issues twin bonds for the purpose of measuring greenium. For other countries, we impose additional assumptions and estimate the greenium.⁹

Germany

Germany has issued twin bonds since 2020 to provide a benchmark of greenium. Twin bonds consist of a conventional bond and a green bond that share the same maturity date and coupon. The main difference is that the use of proceeds from the green bond is limited to green projects. They are, however, also different in that the green bond’s issuance volume is smaller and the issuance date is later. For example, in the twin bonds with maturity in 2030, the conventional bond was issued in August 2019 with a size of €30 billion, while the green bond was issued in May 2021 with a size of €6 billion. Through the issuance of twin bonds, Germany aims to establish the yields of green federal securities as the reference for the Euro green finance market (German Finance Agency 2022).

Germany’s greenium oscillated between 2 to 5 basis points. As of February 2020, four twin bonds are on the market with maturity dates in 2025, 2030, 2031, and 2050. Coupons are zero for all bonds. Figure 3 shows that although the behavior of greenium differs across maturity dates and yields of the twin bonds can be positive or negative, the greenium is consistently positive.¹⁰ The greenium does not seem to react much to large uncertainty shocks such as the Russian invasion of Ukraine in February 2022.

View Full Size

Germany’s Twin Bonds by Tenor

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Source: Eikon.

• Download Figure
• Download figure as PowerPoint slide

Other countries

Since twin bonds are not available in other countries, greenium needs to be estimated with a different approach. The rest of the section presents summary statistics of green versus conventional bonds, and greenium estimates using regression analysis.

Summary statistics of green versus conventional bonds. Annex Table 3.1 in Annex 3 shows the summary statistics of issuance size, yield-to-maturity, spread, and maturity of the green and conventional bonds in the sample, separately for advanced and emerging market and developing economies. Annex Tables 3.2 and 3.3 show the statistics for euro- and USD-denominated bonds, respectively.

• Issuance size. Green bond issuance is still relatively small, about 2.0 percent of the total issuance (2016–2022 average), but growing over time from 2.6 percent in 2018 to 3.2 percent in 2021. The share of green bond issuance and its growth are larger for emerging market and developing economies than advanced economies.

• Maturity. In the whole sample, the average maturity is 12.9 years for green bonds and 12.3 years for conventional bonds. This pattern is consistent with the idea that green bonds help countries extend their debt maturity profiles. The longer maturity of green bond debt is more pronounced in emerging market and developing countries; the difference is almost three years for EMDEs.

• Yield. The summary statistics already indicate a degree of greenium: the average yield of green bonds is lower than conventional bonds. The regression analysis will better estimate the size of the greenium, by controlling for relevant variables such as maturity and liquidity.

Greenium estimate. The average greenium is 3.7 and 30.4 basis points for euro- and USD-denominated bonds, respectively, as shown in Annex Table 3.4.¹¹ The difference is partly explained by the fact that a larger portion of USD-denominated green bonds is issued by emerging markets. Also note that the sample sizes of the two groups differ by around five folds.

Greenium is larger for emerging market and developing economies than advanced economies, for all currencies of debt denomination. Specifically, the greenium estimates for emerging markets are 49.3 and 12.5 basis points for the USD- and euro-denominated bonds, respectively, compared to 5 to 6 basis points for advanced economies as shown in Annex Table 3.5. There can be various reasons behind the difference, and formal analysis of the determinants with richer data is warranted and left for future research.

Time-series variation. Figure 4 plots the estimated greenium separately for euro-denominated and USD-denominated bonds in each year, from 2018 to 2021. The greenium of both USD- and euro-denominated bonds— though they started small—has been increasing over time. In the case of euro-denominated bonds, the greenium increased from on average -2.0 basis points in 2018 to 6.8 basis points on average in 2021.

View Full Size

Greenium Estimate

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Sources: Eikon; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Catastrophe Bonds

Extreme weather is expected to be one of the consequences of climate change, and will result in both physical and fiscal damage. There are many ways to mitigate fiscal risks that originate from extreme weather events. For example, OECD and World Bank (2019) list both ex ante and ex post financing tools to mitigate the fiscal risks (Table 1).¹² Debt with maturity extension provisions such as hurricane clauses is another alternative.¹³ A debt instrument with a unique structure is catastrophe bonds.

^{Table 1.}

Examples of Mitigation Tools for Residual Fiscal Risk

Source: Organisation for Economic Co-operation and Development and World Bank (2019).

^{Table 1.}

Examples of Mitigation Tools for Residual Fiscal Risk

Ex Ante Financing	Ex Post Financing
Dedicated reserve fund Contingency budget Contingent financing (credit/grant) Sovereign risk transfer Insurance of public assets Catastrophe bonds	Budget reallocation Debt financing Taxation Multilateral/international borrowing International aid

Source: Organisation for Economic Co-operation and Development and World Bank (2019).

View Table

^{Table 1.}

Examples of Mitigation Tools for Residual Fiscal Risk

Ex Ante Financing	Ex Post Financing
Dedicated reserve fund Contingency budget Contingent financing (credit/grant) Sovereign risk transfer Insurance of public assets Catastrophe bonds	Budget reallocation Debt financing Taxation Multilateral/international borrowing International aid

Source: Organisation for Economic Co-operation and Development and World Bank (2019).

What is a catastrophe bond?

A catastrophe bond is a debt instrument that allows the cedent (the insured) to get funding from the capital market, if and only if catastrophic conditions, such as an earthquake or hurricane, occur. From an economic point of view, the instrument insures the cedent against the loss from catastrophic events (called peril) by shifting risks to the holders who bet on the nonoccurrence of catastrophic events. The insurance against natural disasters can be considered an adaptation policy for countries with exposure to climate change risks.¹⁴

The catastrophic conditions can be defined by various types of triggers. For example, a trigger based on actual monetary losses experienced by the cedent is called an indemnity trigger, the one based on industrywide losses is called an industry loss trigger, and the one based on noneconomic catastrophic conditions such as the magnitude of an earthquake or wind speed of a hurricane is referred to as parametric index trigger. The advantage of indemnity type is that it insures cedents against the actual loss, while a disadvantage is time-consuming loss verification since the damages need to be assessed. In contrast, the parametric type may not insure cedents against the amount of actual loss, but it has the advantage of speedy settlement since parameters such as wind speed and magnitude of an earthquake are easier to measure. The idea can be extended to noncatastrophic conditions, such as mortality rates, in which case the concept of catastrophe bonds is generalized to insurance-linked securities (ILS).

The legal structure of a typical catastrophe bond is designed to minimize counterparty risk. Specifically, a special purpose vehicle (SPV) is set up, and the cedent (often called the sponsor) enters an insurance agreement with the SPV. Cedents pay premiums upfront in exchange for future reimbursement conditional on catastrophic events. The SPV issues catastrophe bonds to the holders in exchange for cash, promising future principal and interest payments conditional on the nonoccurrence of catastrophic events. Thus, what an SPV does is to collect cash from cedent and investors, keep the cash typically in safe assets, and disburse it to either cedent or investors depending on the occurrence of catastrophic events. In this way, an SPV can secure the cash for later distribution, and who owns the bond does not affect the capacity to pay cedents, so the bond can be traded in the secondary market (Figure 5).

View Full Size

Legal structure of catastrophe bonds

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Source: IMF staff.Note: SPV = special purpose vehicle.

• Download Figure
• Download figure as PowerPoint slide

The legal structure has financial, statistical, and economic implications. Financially, the catastrophe bond is insulated from the cedent’s financial condition, so the credit rating is different from that of the cedent. Statistically, the catastrophe bond is issued by the SPV and not by the cedent, so the cedent’s debt does not increase. Economically, the cash proceeds are kept by the SPV, so they cannot be used by the cedent to spend on items including consumption, investment, etc., until triggered.

A notable difference from the traditional bond is the modeling of catastrophe risks (Figure 6). In addition to credit ratings, the risk modeling is prepared by a third-party risk modeler, such as AIR Worldwide (or Verisk), and the results including the expected loss are disclosed in the bond’s offering documentation. Investors can ask questions to the modeler in the marketing process of the bond, and they often have their own modeling team to assess the risks. The modeler also calculates the actual loss after catastrophe events. The modeling often involves the assessment of extreme but nontail events since many catastrophe bonds are structured in the way that investors incur loss only if the loss from catastrophic events exceeds a certain threshold (attachment point), and the investors’ loss is bounded by the principal (exhaustion point) (White 2020). The typical maturity is three to five years, so the long-term impact of climate change risks may not be fully reflected.¹⁵

View Full Size

CAT bond and ILS risk modelling

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Source: Artemis.

• Download Figure
• Download figure as PowerPoint slide

Evolution of sovereign catastrophe bonds

The catastrophe bond issuance by the public sector is increasing over time. Figure 7, panel 1, shows that the nominal amount of issuance is in an upward trend. This is also the case for the number of cedent countries. The largest player is the United States, accounting cumulatively for nearly $5.6 billion, followed by Mexico, Chile, Turkey, etc. Some of them, including the California Earthquake Authority, are local state agencies, but the central governments themselves can be the cedents, including the recent examples of Jamaica (2021), Mexico (2020), and the Philippines (2019). The list of the catastrophe bonds used in the analysis is provided in Annex 2.

View Full Size

CAT bond issued amount in millions of US dollars

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Source: Artemis.

• Download Figure
• Download figure as PowerPoint slide

The perils covered by public sector catastrophe bonds have been mainly US earthquakes (Figure 7, panel 2). However, climate-related catastrophic events such as storms and hurricanes also constitute a significant share. Importantly, pandemics can be the perils: in 2017, the World Bank issued a five-year coverage, and then received payment for the COVID-19 pandemic.

Pricing of sovereign catastrophe bonds

Empirically, the cedents pay more than they receive in expectation. According to Artemis, from which only the aggregated data of private and public catastrophe bonds is available, the investors’ average return to catastrophe bond in coupon is around 2 to 4 times the expected loss (Figure 8). Difiore, Drui, and Ware (2021) note that risk spreads have widened materially following major catastrophes in the past, such as 2006 following a US hurricane and 2012 following earthquakes in Japan and New Zealand.

View Full Size

CAT bond and ILS issuance average expected loss and coupon

Citation: Staff Climate Notes 2022, 004; 10.5089/9798400210006.066.A001

Source: Artemis.

• Download Figure
• Download figure as PowerPoint slide

Whether the positive spread is expensive depends on the benchmark to compare. For example, self-insurance can save the coupon payment to investors, and thus can be an inexpensive alternative. The cash, however, needs to be stored in a dedicated fund and cannot be used for other illiquid purposes. Therefore, for countries with large opportunity costs, self-insurance can be more expensive than catastrophe bonds. A comparison can also be made with traditional reinsurance. Michael-Kerjan and others (2011) argue that the premiums that traditional reinsurance charges range from 3 to 5 times the expected loss, which is not very different from catastrophe bonds.¹⁶

Some literature argues that the catastrophe bond market is inefficient. In theory, the premium that cedents pay should be independent of the cedents’ credit risks since the SPV stores and disburses cash. Chatoro, Pantelous, and Shao (2021) and Gotze and Gurtler (2020), however, argue that, in both the primary and secondary markets, the risk premium depends on the cedent’s characteristics, such as the cedent’s length of the time in the market, credit ratings, etc. Thus, new cedents can face challenges not only due to positive spread but also due to higher premiums than other experienced cedents.

Intermediation by the World Bank has mitigated the challenges that countries face in utilizing catastrophe bonds. Since 2016, all the sovereign catastrophe bonds in the data set compiled by Artemis have been intermediated by the World Bank. By providing the service of an SPV, the World Bank simplifies the procurement process as setting up an offshore SPV could be a legal barrier for countries. Anecdotally, the reputation and experience of the World Bank also contribute to narrowing the spread. Thus, catastrophe bond issuance through the World Bank offers an attractive venue for the countries that seek insurance against natural disasters. For example, a case study by the World Bank on its collaboration with Mexico can be found in World Bank (2020a).

Holder of sovereign catastrophe bonds

Although the holders’ information is scarce, it is available for some catastrophe bonds intermediated by the World Bank. Table 2 summarizes the investor information from three recent issuances (World Bank 2019, 2020b, 2021). Most investors are in Europe, and the type of investors are concentrated on institutional investors with expertise such as ILS funds and insurance companies.

^{Table 2.}

Holders of Recent Sovereign Catastrophe Bonds

Source: World Bank. Note: ILS = insurance-linked securities.

^{Table 2.}

Holders of Recent Sovereign Catastrophe Bonds

Cedent (issuance year)	By Geography	By Investor Type
Jamaica (2021)	Europe 60% North America 24% Bermuda 15% Asia 1%	ILS fund 66% Insurer/reinsurer 17% Asset management 14% Pension fund 3%
Mexico (2020)	Europe 52% North America 42% Bermuda 5% Asia 1%	ILS specialist fund 61% Asset management 16% Pension fund 15% Insurer/reinsurer 8%
Philippines (2019)	Europe 58% North America 25% Asia 13% Bermuda 4%	Asset management 50% ILS fund 29% Insurer/reinsurer 13% Pension fund 8%

Source: World Bank. Note: ILS = insurance-linked securities.

View Table

^{Table 2.}

Holders of Recent Sovereign Catastrophe Bonds

Cedent (issuance year)	By Geography	By Investor Type
Jamaica (2021)	Europe 60% North America 24% Bermuda 15% Asia 1%	ILS fund 66% Insurer/reinsurer 17% Asset management 14% Pension fund 3%
Mexico (2020)	Europe 52% North America 42% Bermuda 5% Asia 1%	ILS specialist fund 61% Asset management 16% Pension fund 15% Insurer/reinsurer 8%
Philippines (2019)	Europe 58% North America 25% Asia 13% Bermuda 4%	Asset management 50% ILS fund 29% Insurer/reinsurer 13% Pension fund 8%

Source: World Bank. Note: ILS = insurance-linked securities.

Broadening the investor base to the public sector can help correct the market failure of climate change and improve crisis management. Since climate change has a heterogeneous impact on different regions on the earth, the cost of climate change in one region may not be internalized by other regions. Investing in catastrophe bonds can be an effective mechanism to internalize climate risks that are physically far away from the investors. In other words, by investing in catastrophe bonds, governments can show commitment by putting their “skin in the game” while earning a positive return on average. From a crisis management perspective, countries often pledge financial support after natural disasters in other regions, but aid pledges made while media attention is at its peak may not always be disbursed, could take a long time to arrive, or may replace previously pledged aid (Becerra, Cavallo, and Noy 2012). For example, the US Government Accountability Office (2011) reports that, 20 months after the 2010 Haiti earthquake, only 0.8 percent of the $412 million that the US government pledged for infrastructure construction activities was expensed. Catastrophe bonds offer an ex ante mechanism for countries to pledge financial support for natural disasters and timely disburse funds in catastrophic events.¹⁷