Using an SDR Bond Creatively to Boost Concessional and Climate Finance
from Follow the Money and Greenberg Center for Geoeconomic Studies

Using an SDR Bond Creatively to Boost Concessional and Climate Finance

Combining an SDR with budget funding is the most efficient way to stretch budgeted dollars (and euros) and increased the global supply of concessional financing. 
A worker walks between solar panels at Centragrid power plant in Nyabira, Zimbabwe.
A worker walks between solar panels at Centragrid power plant in Nyabira, Zimbabwe. Philimon Bulawayo/Reuters

This is a joint post by Brad Setser and Stephen Paduano, the executive director of the LSE Economic Diplomacy Commission and a PhD candidate at the London School of Economics. 

In a recent speech on “renewing American economic leadership,” U.S. National Security Advisor Jake Sullivan signaled that the United States intends to “close the global infrastructure gap in low- and middle-income countries” and create an “international financial system that enables partners around the world to reduce poverty and enhance shared prosperity.” His comments pick up on Treasury Secretary Janet Yellen’s calls to “meet the enormous infrastructure needs of low- and middle-income countries without trapping them in cycles of debt.” French President Emmanuel Macron, who is hosting a summit at the end of June that aims to build momentum for a new “global financial pact,” is no less ambitious.

More on:


International Monetary Fund (IMF)


Greenberg Center for Geoeconomic Studies

These agenda items—closing the global green infrastructure gap, reducing global poverty, and reimagining the global financial architecture—won’t happen without real funding from the governments of advanced economies.

Relying entirely on private markets to turn billions into trillions won’t work. As Barbados’ Prime Minister Mia Mottley has noted, private sector financing is simply too expensive to make most green investments viable in low-income countries.      

Indeed, for much of Africa, the option of funding infrastructure with long-term private financing is now entirely hypothetical. No African country has issued a Eurobond in a year, bonds issued 5 or 10 years ago are now coming due, and China’s vaunted financing for African infrastructure has been in continuous decline since 2016—drying up almost entirely after 2020. To make matters worse, loan payments on most Chinese projects ratchet up after 5 years, leaving a real risk that capital is now flowing back to China from low income countries.   

These global challenges can only be addressed through a substantial expansion of concessional financing. But concessional financing requires scarce budget resources, and advanced economies are themselves facing pressure to bring down deficits that were allowed to widen during the pandemic and Russia’s invasion of Ukraine. The deal on the debt ceiling reintroduces caps on the growth of non-discretionary spending in the United States. Just last week, the International Monetary Fund (IMF) called for long-term fiscal consolidation in the United States of 5 percent of GDP “to put public debt on a decisively downward path by the end of this decade.” The IMF almost certainly is over-doing it a bit (not for the first time). But the reality now is that development budgets of advanced economies around the world are under close scrutiny. 

There is thus a clear need to find the most cost-effective way to expand the availability of concessional financing. Otherwise, talk of a new financial pact will ring hollow—and the Biden administration’s efforts to show that democracies can deliver on global infrastructure investment and global poverty reduction will seem empty. 

More on:


International Monetary Fund (IMF)


Greenberg Center for Geoeconomic Studies

Fortunately, there is a real financial solution here that would only take a modest amount of budgeted dollars and euros:

  • Multilateral development banks (MDBs) need to be willing to stretch their balance sheets in a financially prudent way by issuing a bond that is denominated in the IMF’s reserve asset, the Special Drawing Right (SDR), but settled in hard currency such as dollars.*
  • Holders of the world’s surplus SDRs need to commit to invest their idle SDR reserves into the MDBs’ SDR-linked bonds. There is no shortage of potential funding: the United States alone has $164 billion in SDRs, the euro area countries have $200 billion, Japan another $58 billion, and the UK another $46 billion. There is no budget cost from using SDRs in this way.** 
  • The world’s advanced economies should also commit to provide budget resources sufficient to pay the interest on the SDR bond. The SDR interest rate is currently 3.8 percent, but it is expected to fall over time as inflation recedes in the United States and Europe, and interest rates come down too. With standard interest rate swaps, the budget cost of this could be locked in at roughly 3 percent of the bond’s face value. That would allow all of the SDRs raised to be lent out at fully concessional (zero) interest rates to fund a wave of clean energy investments in low- and even middle-income countries.

With this structure, a $50 billion fully concessional facility would have a budget cost of between $1.5 and $2 billion a year. That is an effective way to scaling up concessional funding in an era of increasingly tight budgets.   

The only financial innovation required is a large new issue of SDR bonds, as we have proposed and explained previously. Fortunately, this innovation has robust precedent, as the World Bank has already issued SDR bonds on a smaller scale. 

The other component of the proposal would mirror the mechanism already put in place by the IMF. Aid dollars would be used to cover the interest rate on what would otherwise be a normal loan, turning standard lending into concessional financing.

Why use a new financing model?

Part I: Lessons from the IMF

Pairing an SDR bond issuance with a grant contribution that covers the SDR interest rate would build upon the financing model already used by the IMF’s Poverty Reduction and Growth Trust (PRGT). 

The PRGT works as follows: donor countries lend SDRs to the PRGT’s loan account and also provide grants to the PRGT’s subsidy account. The subsidy account allows the PRGT to be fully concessional, with subsidy resources offsetting the SDR interest rate, which currently stands at 3.77 percent.

There is a bit more technically, as the PRGT has a third account, a reserve account, which provides a further layer of security for donors by helping with on-time repayment in the event the recipient country makes a late payment. The reserve account is also intended to generate income, which can help contribute to the subsidy account.

But the basic idea is simple: put currently under-used reserves to work, and make the funding concessional by tapping into budget dollars from donors.

Using this structure to expand the concessional lending capacity of the World Bank or another MDB only requires one change to this financing model: replacing SDR loans with SDR bonds.

The difference between SDR loans and SDR bonds may seem small, but there are important legal and technical distinctions between loans and bonds (securities) for the World Bank’s key shareholders. Notably, the U.S. Congress, the UK Parliament, and the EU’s Court of Justice have all permitted the Exchange Stabilization Fund, the Exchange Equalisation Account, and eurosystem central banks to purchase securities, i.e. bonds. They have not clearly extended the same authorization for the provision of loans, meaning that the largest SDR holders would find it more difficult to make use of their SDRs if they could only do so through SDR loans.***    

Structuring the financing as the sale of a security is also the easiest way to ensure that it remains a reserve asset, which is a stipulation of the IMF’s Articles of Agreement and a key challenge in rechanneling discussions. Securities are meant to be traded and can be sold into the private market—particularly if the securities are only denominated in SDRs and actually settle in hard currency (dollars, euros, pounds, yen, and yuan). Loans aren’t meant to trade, and assuring the loans extended to the IMF’s trust funds remain reserve assets has taken a bit of work and creative financial engineering. This will become even more important as the main vehicle of mobilizing SDRs needs to move away from the IMF to the MDBs.

Part II: Lessons from IDA

The World Bank’s International Development Association (IDA) also has moved beyond its old financing model where all of its concessional loans were financed purely out of its substantial equity base ($179 billion, as of 2022) to a hybrid financing model where conventional market borrowing extends the IDA’s lending capacity.

However, IDA’s new hybrid model hasn’t been used extensively—IDA has only issued $26 billion in bonds. IDA’s reluctance to boost its market borrowing reflects its need to keep its borrowing costs as low as possible so that its lending can remain fully concessional. It is worth noting that the increase in G-10 interest rates created complications for the hybrid model (see Clemence Landers), as the “grant” contribution earmarked for interest on these loans was often the function of annual appropriations and the interest cost on new fixed rate bonds has gone up.***

Combining SDR bonds (which would provide at cost financing directly from IDA’s donors) and budget resources would offer IDA a better model for using debt financing than pure market borrowing. 

Of course, the subsidy contribution would need to be locked in upon the issuance of an SDR bond to avoid creating balance sheet risks for the MDB (when subsidies and market borrowing are not fully synchronized, there is a risk that rising interest rates outstrip subsidy contributions—creating limits on the MDBs’ operations). Directly pairing the two would ensure that the World Bank could offset the SDR interest rate upon every SDR bond issue. An SDR bond would need to pay a floating rate, but it is technically easy for the World Bank to use the swaps market to convert floating into fixed or vice versa—so a fixed annual contribution could easily match floating rate borrowing. With an inverted curve, this even saves a bit of money.

Why bother with SDRs and the extra complexity?

Some may be left wondering why this should be done in SDRs rather than normal currencies.

The easy answer is because there are far more SDRs than there are current uses of SDRs.  

The IMF’s rechanneling mechanisms can only absorb $63 billion in SDRs—less than the $100 billion worth of SDRs the United States and G-20 countries have committed to rechanneling, and far less than the $935 billion in SDRs in existence. An SDR bond issue that effectively mobilizes a portion of these funds is thus not only a good way to use currently under-used resources, it is an obviously scalable model for the MDBs to raise future funds.

The other reason is that the U.S. Treasury has more SDR reserves than hard currency reserves, and more authority to buy bonds than make loans. Other countries could in theory use their existing hard currency reserves to satisfy their SDR rechanneling pledges (as Germany intends to do, given the Bundesbank’s opposition to SDRs). But the United States really can not. The U.S.’ Exchange Stabilization Fund (ESF) currently holds only $17 billion in dollars. By contrast, the ESF holds $164 billion in SDRs. SDR rechanneling options to date have been designed more to meet the institutional needs of the IMF and to meet the capital needs of the MDBs than to meet the rules governing the use of SDRs in the major advanced economies. This is part of the reason why the United States is struggling to contribute to the existing rechanneling mechanisms, the IMF’s PRGT and RST. By contrast, Congress has already authorized the ESF to purchase securities (such as SDR-denominated, dollar-settled bonds).

Just as an SDR bond is beneficial for those that will purchase it, it is also beneficial for those that will issue it—the MDBs. An SDR bond would tap a largely unused, captive market of up to $935 billion. Credit rating agencies should view an SDR bond sold to the MDBs’ existing shareholders, with the expectation that it would roll over in perpetuity as it perfectly matches the shareholdings existing SDR liability to the IMF, as a substantially safer and more scalable form of financing than conventional market borrowing. A commitment to buy SDR bonds at par by the major holders of SDR (perfectly matching the current return on the SDR balances at the IMF) would assure the Bank of the lowest possible borrowing cost. In effect, an SDR bond would function as a permanent, low cost contribution to the MDBs’ resources. As a result, it could help get the credit rating agencies on board with proposals to stretch the MDBs’ balance sheets further, by substantially reducing the risk of prudent reductions in the MDBs’ leverage ratios (e.g., beyond the uninspiring one percentage point notch the World Bank is pursuing).

Two suggested new facilities that would mobilize the world's SDRs

To make these ideas concrete, it helps to work through the numbers for a concrete proposal

Facility #1: The IDA Post-Pandemic Facility (IDAxSDRs)

IDA would be authorized to raise its leverage rate by issuing up to [50 billion SDRs/$60 billion USD] in SDR linked bonds at current exchange rates.

Donors would increase their contributions to IDA replenishments by a sum equal to the fixed interest rate equivalent of the SDRs current floating rate (3.8 percent), which we estimate would be around 3 percent for a 10 year bond (and should be a blend of 10y GBP/ USD/ German bund and JPY rates). That is an annual commitment of roughly $3 billion (or in the worst case scenario $4 billion).****

IDA would increase its fully concessional (zero interest rate) lending by SDR 50 billion ($60 billion). This would expand IDA’s lending from its current $184 billion to $245 billion a real increase. It would also be an ideal mechanism for funding facilities designed to rapidly scale up IDA lending in the event of natural disasters or other unanticipated shocks.

The scale of the needed financial commitment (something less than $6 billon over IDA’s 3 year replenishment cycle) should be seen as both highly manageable on the part of shareholders and—as such contributions would unlock $60 billion in new borrowing—a highly efficient way to stretch shareholder’ contributions further than ever before.

Facility #2: A Green Infrastructure Facility for Middle-Income Countries (The Partnership for Global Infrastructure x SDR facility)

The World Bank would commit to issue [SDR 50 billion/ USD $60 billion in new bonds] for a new lending window for green infrastructure investments by middle-income countries. This would move the World Bank’s leverage ratio from 4:1 (20 percent) to close to 6:1 (15 percent), still a safe number.

The World Bank’s major shareholders would commit to buy these bonds at cost.

The World Bank would retain $10 billion of the $60 billion raised as part of a special Green Infrastructure Facility investment portfolio designed to generate investment income (investing in U.S. Agencies, for example, offers a government-backed return well above the current SDR interest rate) and help defray costs as well as to provide the facility with a financial buffer to assure repayment of any bond issues. Over time, the investment portfolio could be invested in various safe green bonds around the world. This of course is analogous to the reserve tranche in the PRGT.

The World Bank would establish a Green Infrastructure Lending Facility in the IBRD, which would be available to finance green investments in middle income countries. Such loans would carry the SDR interest rate, which is considerably below the cost of borrowing in foreign currencies for most middle income countries.

This provides concessional funding, though not fully concessional funding. But it would effectively put green infrastructure projects funded through this facility on par (financially speaking) with green projects in advanced economies that have a government guarantee, which would be a major improvement for middle-income countries. 

Last year, the World Bank provided $14.7 billion in gross infrastructure finance—a welcome 50 percent increase from 2021, but still far below the world’s infrastructure needs. This facility would increase gross lending by roughly $10 billion a year over the next five years—a material sum.

This facility is designed to have no budget cost to the participating creditor countries, who would only need to commit to buying an SDR-linked bond at par. The SDR bond, which would settle in dollars, euro, yen, pounds, or yuan, would be pari passu with all other IBRD bonds, and clearly qualify as a reserve asset.

Technically, this proposal requires that the World Bank accept a larger change in its leverage ratio than is currently on the table. The commitment of SDRs from the Bank’s existing shareholders and the additional buffer built into the facility would make this a safe way to stretch the World Bank’s existing capital. Of course, a modest increase in the World Bank’s capital would provide an alternative mechanism for balance sheet expansion—but in our view it is not necessary.

It also should go without saying that SDR bonds could also be deployed directly to support existing plans to raise IDA's leverage or increase the Bank's overall lending capacity.

Moving forward

Pairing these two facilities would directly address concerns that opening a concessional “green” window for middle-income countries would cut into funding for the development of low-income countries.

Compared to some proposals, our suggestions lack ambition as they do not allow SDRs to be leveraged—which is what the African Development Bank and Inter-American Development Bank hope to do with their “hybrid capital” proposal. But an SDR bond and the two facilities described here should score highly on three other critical dimensions:       

One, they don’t depend on mobilizing private capital—they rely entirely on available funds from official sources plus a modest commitment of new budget resources. Low- and middle-income countries would see that these funds are real, not aspirational.

Two, they are doable with a realistic commitment of resources from the advanced economies and other countries with surplus SDRs or a current budget windfall.    

Three, they offer a straight forward way to safely stretch the Bank's existing capital.

Our proposals have been sized in a way that would generate a material increase in concessional flows while increasing multilateral lending on terms that debt servicing constraints seriously.   

If they succeed—given the nearly trillion dollars in SDRs outstanding that are mostly held on strong balance sheets—these kinds of facilities clearly could be scaled up over time.


* Stretching refers to lowering the leverage ratio and allowing the MDBs existing capital to support lending. Right now, the World Bank’s $20 billion of in paid in capital and $30 billion reserve (part of its equity base) supports about $230 billion of lending (a leverage ratio of just over 20 percent). IDA, the World Bank’s concessional arm, is almost entirely equity funded, with only $179 billion of borrowing supporting $184 billion of lending. The extent the Bank needs to stretch its balance sheet would of course be reduced through the addition of additional paid-in or hybrid capital. 

** The SDR bond would have no fiscal impact on the countries that purchase it. IMF members pay the SDR interest rate on their SDR “allocation” and receive the SDR interest rate on their SDR “holdings.” This means that when countries don’t use their SDRs at all, what they pay is equal to what they receive, and they face no budget cost. Once countries actually start to use their SDRs, their holdings drop below their allocation and they pay the SDR interest rate on the difference. The budgetary appeal of the SDR bond is that the bond would pay the SDR interest rate, meaning it would perfectly offset what the countries would owe the IMF for using their SDRs.

*** The United States’ Exchange Stabilization Fund is explicitly permitted in its governing statute to purchase securities while the Bretton Woods Agreement Act creates conditions around lending to the World Bank (and the IMF). The UK’s Exchange Equalisation Account, is similarly authorized to purchase any security, while no such authority exists for His Majesty’s Treasury to act freely in writing loans out of the Exchange Equalisation Account. The same distinction between loans and bonds exists for Eurozone countries. The ECB and the eurosystem’s national central banks have clear authority under the Public Sector Purchase Programme (one of the ECB’s quantitative easing program) to buy the bonds of supranational organizations such as MDBs (e.g., the European Investment Bank. Although the ECB can and should adopt a more accurate understanding of “monetary financing” that doesn’t arbitrarily define the IMF as the only acceptable channel for using SDRs, the current ECB interpretation may make it impossible in the near term for eurozone countries to rechannel SDRs through the MDBs in any way other than purchasing SDR securities. 

**** In IDA’s April 2021 pound issuance, a seven-year bond that raised GBP 1.5 billion, IDA explained its denomination decision by noting that IDA has a “natural need” to raise funding in SDR currencies as the “SDR is IDA’s functional currency.” This could have spurred more creative thinking about SDR denominated issuance. Issuing bonds in the smaller SDR currencies is helpful, but IDA can generate SDRS more directly by issuing directly in SDRs.

***** IDA has a net investment portfolio of $42 billion, with 65 percent of its holdings in securities that are scored as AA and above. The purpose of the investment portfolio is "to supply liquidity for operations," which can double as securing repayment of SDR denominated bonds in the event a need arises—which will solidify the reserve asset status of SDR bond purchases—and could also augment subsidies when need be.

Creative Commons
Creative Commons: Some rights reserved.
This work is licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) License.
View License Detail