ODI hosted a terrific event about mobilising private investment recently, where the introductory talk first noted how much investment the world needs and then that the mobilisation ratios reported by DFIs were too low.
The conversation then turned to the mobilisation of institutional investors via securitisations and other structures in which the public sector injects some capital to shoulder some risk and create assets that are more palatable to them.
The mobilisation ratios in these structures look much higher – one pension fund manager spoke of being prepared to buy $1bn of DFI assets with only $50m of public first-loss capital being needed. That sounds much more like it. But alarm bells should be ringing. The mobilisation ratios in secondary transactions are not like those traditionally reported. Let me explain.
A primary transaction raises money for a business to use. In a secondary transaction, one investor buys an asset from another. [1] The original idea behind mobilisation was about raising for money for businesses: if the SDGs need $1 trillion of investment per year, and DFIs can deploy $200bn per year and mobilise $4 of private investment for each $1 of their own, we can get there. We might call this a 4:1 mobilisation ratio in primary transactions.
A pension fund buying $1bn of loans from DFIs in a secondary transaction, with $50m of public risk-bearing capital, looks like a “mobilisation ratio” of 20:1. [2] If DFIs can do that, would it mean that $1tn of investment on the ground could be achieved with only a $50bn outlay from DFIs? [3]
No. In secondary transactions, what looks like the mobilisation ratio is not describing a multiplier that turns $1 of investment by DFIs into a much larger investment on the ground. In a secondary transaction, the ratio of public to private money involved says something about the cost of helping DFIs raise money which they can then invest again impactfully.
Some forms of mobilisation are really about increasing the firepower of DFIs – development banks can issue bonds, they can insure risks in their portfolio to free up capital. DFIs can securitise or partially sell their assets, and so on. These methods – which could be called balance sheet management – are all ways of increasing DFIs’ pace of investment. They will accelerate progress towards the SDGs and climate goals if the problem is that DFIs are held back by a lack of investment headroom. [4]
Take for example a DFI that currently invests $5bn annually and has a portfolio of $25bn, from which it realises $5bn in receipts annually to finance those investments. But it wants to do more. If it was able to realise $20bn in receipts from its portfolio each year, by recycling its investments much more quickly, then it could maintain an annual investment pace of $20bn.
Suppose the DFI sells $20bn of investments each year to pension funds, in a structure that includes $1bn of public first-loss and $19bn of senior securities, for a 19:1 secondary mobilisation ratio. Unlike a primary mobilsation ratio, that 19:1 ratio does not tell us the $20bn of investments this DFI makes each year is being multiplied by 20. [5] There is not $400bn of investment happening.
Blurring the distinction between primary and secondary transactions
The argument so far is simple. Mobilisation in primary transactions is different to mobilisation in secondaries, and mobilisation ratios do not have the same interpretation. But to complicate matters, primary and secondary transactions can be tightly integrated in the mobilisation structures that some DFIs have established, or are in the process of developing. It can be hard to keep track of what counts as an investment by a DFI, and what mobilisation ratios should be applied to what. That gets a little complicated, so I will try to start simple and build up to it. There are two aspects to this – timing and pooling.
Let’s start with timing. A DFI finds a suitable investment on the ground that needs $20m in debt. It can offer that investment to a private partner, who may agree to put in $15m alongside $5m from the DFI’s own balance sheet. That primary co-investment would be reported as 3:1 mobilisation.
But the DFI could also make the initial $20m primary commitment itself, and then later partially sell the loan to a private investor, to raise $15m. Doesn’t that leave us in exactly the same place? If the DFI knew at the time of origination that this was going to happen, in principle a DFI might do something like record a $20m gross comittment but only a $5m net commitment, in expectation that the $15m is shortly to be taken by someone else. That would amount to a 3:1 mobilisation in a primary transaction, based on the DFI’s net comittment, even if the private money only arrives after something of a delay.
The fact that private capital can arrive sometime after origination means the distinction between primary and secondary can be blurry. But we have not yet arrived at anything like the 20:1 ratio in secondary transactions that this blog started with. For that, we must introduce pooling. Rather than partially sell its loans one at a time, DFIs can create pools of assets and then instruments such as a first-loss capital can be used to change the risk-return profile of the assets in that pool, so that private institutional investors are willing to buy them.
The SDG Loan Fund created by Allianz and the Dutch development bank FMO is a good example of this. Thus far we have only discussed a single instrument – a loan – but in these pooled structures the underlying loans are used to create a fund with two classes of asset. In the FMO/Allianz fund there will be $1bn of Class A shares, that will be owned by institutional investors, and $111m of first-loss Class B shares owned by FMO. [6] The timing in this case is that assets are added to the fund at origination by FMO. But pooled structures can also involve a delay when initial primary transactions are followed by secondary transactions at a later date. That can happen with securitisation structures that resemble collateralised loan obligations, where the originating DFIs could first build an inventory of primary transactions (loans) which are then pooled and securitised, later.
Returning to our example of a $20m loan, a DFI could raise $15m from partially securitising that loan in a pooled structure, which itself includes some public first-loss capital. Now what have we got? On a pro-rata basis (because this $20m loan is now part of a pool) that $15m could be split into a $14m senior tranche, bought by private investors, and $1m of first-loss capital provided by the public sector (perhaps the DFI, perhaps not). The $5m net commitment from the DFI, added to the $1m pro-rata first loss, gives us a total of $6m public money and $14m private. [7] Looking at the securitisation in isolation and comparing just the first-loss capital to the private capital it might look like a 14:1 mobilisation ratio, but that would not be the complete picture.
Because these structures can take different forms, keeping track of the complete picture could get quite complicated. There are two official methodologies for reporting mobilisation – OECD and MDB. The OECD does not currently cater for mobilisation in secondary transactions. The MDB methodology focuses on “mobilisation at origination”, which it defines as within 12 months of the MDB or DFI commitment. That allows for structures such as warehousing and securitisation that involve origination with the intent to quickly turn around and sell down. [8] Under this method a DFI that makes a temporary $20m gross commitment could report a $5m net commitment, with $15m recorded as mobilised. [9]
The 12-month cut-off is fairly arbitrary, and if mobilisation via secondary structures becomes more important over longer time horizons, reporting methodologies may evolve. However, while the distinction between balance sheet management and mobilisation at origination may get blurry, it is worth trying to preserve. The dividing line should be drawn somewhere between mobilisation that is tightly connected to a transaction that brings additional financing to an underlying company or project, as opposed to transactions that DFIs and MDBs undertake to generally increase their capacity to invest.
As these structures become more commonplace and confidence in them rises, the financial commitments required from DFIs may fall relative to the sums supplied by private investors, and mobilisation ratios may thus rise. But the long-term DFI commitment is unlikely to fall to zero. In other words, the model will be “originate to share”, not “originate to distribute”. Even in private structured finance transactions, the structuring bank often retains a share of the loans, because co-financiers want the originators to have an incentive to care about asset quality over its life, and because borrowers prefer to deal with one lender, so the structurer remains the lender of record. In the context of development finance, financiers may also want to benefit from DFIs relationships with governments, and for DFIs to monitor impact and ESG risk. DFIs also cover their overheads by returns generated by their portfolio and may not be able to survive on structuring fees alone.
Does it matter whether private investors give their money to DFIs upfront to invest on their behalf, or decide whether to participate in primary transactions at the time of origination, or come in later via secondary transactions? It depends on the objective. The big picture is that in some markets there is a need for capital that exceeds supply, and the impact of mobilisation lies in connecting institutional investors in OECD member countries with deep pockets to those investment opportunities, and hence increasing the quantity of urgently needed investment for climate and development. Whether private finance is supplied before, during, or after origination is not so important in that context.
But if DFIs are hoping to introduce hands-on private investors to new markets – not the big institutional investors whose job it is to allocate money into assets with certain characteristics, but the sort that gets involved in assessing business plans, has a dialogue with management and so forth – in the hope that they will get comfortable and then continue to originate new investments in those markets by themselves, then that requires participation in primary transactions at origination. Another thing DFIs might be trying to do is introduce local asset allocators, such as domestic pension funds, to local investment originators, and that too may call for different approaches.
There is of course much more to say about mobilsation, but the message of this blog is to avoid looking at secondary transactions in isolation and keep the complete picture in mind.
Acknowledgements
This article benefited greatly from conversations with Neil Gregory, Martine van Aalst, Hans Peter Lankes and Joel Bensoor.
Footnotes
[1] Secondary markets – such as the stock exchanges – are extremely important because they increase the volume of primary transactions. Investors are much more willing to get into something if they know they can get out again. Thinking about mobilisation more broadly, one of the most important things DFIs can do is help establish various secondary markets, to help investment originators realise their returns and recycle capital. Renewable energy infrastructure developers, for example, benefit from being able to sell operating projects to infrastructure investment trusts.
[2] Furthermore, by the pension fund’s calculations that $50m would have an expected return of 4%. That is below the equity-like returns a commercial investor taking the most junior tranche of a securitisation would demand, so it’s a subsidy, but in grant equivalent terms it’s tiny. Let’s say $5m. If the mobilisation ratio was calculated after first normalising the public contribution into grant equivalent terms, as it should be to avoid comparing apples to pears, that ratio would be enormous.
[3] The maths is a bit off here for the sake of using round numbers. $50bn from DFIs plus 20 x $50bn from private investors would actually be $1.05tn total.
[4] For a comprehensive survey of the various ways in which DFIs and MDBs can both manage their balance sheets to increase headroom and mobilise at origination, see the CCD policy paper: Taking Stock of MDB and DFI Innovations for Mobilizing Private Capital for Development by Neil Gregory.
[5] $1bn of public money is being added to by $19bn of private money for a total of $20bn. $1bn multiplied by 20 is $20bn.
[6] The A shares are lower risk and lower return than the B shares. FMO’s first-loss capital is credit-enhanced by an unfunded guarantee provided by the MacArthur Foundation.
[7] I am committing the sin of just adding up the face value of the public components and not adjusting for differences in their cost.
[8] Warehousing comes in various forms, but the general idea is a structure in which an inventory of primary investments can be accumulated for later sale or securitisation.
[9] Assuming the DFI is not also providing the $1m of first loss. Whether the DFI’s own annual reporting uses the $20m gross or $5m net figure would come down to that institution’s own rules, which are distinct from agreed mobilisation reporting methodologies.
