The long effort to encourage mainstream investors in rich countries to lend more willingly to emerging markets took another step forward last week, with the publication of the latest Gems report on development banks’ lending — giving potential investors the fullest set of information yet.
Gems — the Global Emerging Markets Risk Database — is a collaboration between 29 multilateral and national development finance institutions (DFIs), in which they share information on how their loans perform.
The Gems Consortium’s latest annual statistical digest adds several dimensions of information it had never revealed before, including how loan performance varies between the internal rating categories used by DFIs themselves.
“The goal is that we want the general public to have access to exactly the same information that Consortium members have access to,” said Federico Galizia, vice-president for risk and finance of the International Finance Corp in Washington. “We are achieving that with [this] report.”
Galizia is co-chairman of the Gems steering committee — chief risk officers of seven major multilateral development banks.
Out of reach
Policymakers have wrestled for years with how to tap the vast pools of investable wealth in the Global North, to meet the acute development needs in many emerging markets.
The basic problem is most asset owners and asset managers do not want to take the risk of investing in emerging markets. Moreover, many of the critical investments needed in developing countries are small and cannot reach investment grade ratings — and they may require long term loans.
This is where DFIs specialise, but their capital is strictly limited, and a fraction of what private sector investors have. DFIs strive to ‘mobilise’ private capital to increase their firepower, but progress is limited by investors’ reluctance to accept the risks DFIs take, at the returns they receive.
One of the most effective ways to tackle that resistance is education — convincing investors that the risks are not nearly as bad as they think. This education is the public purpose of Gems.
In an IFC research note last year, Galizia and his colleague Susan Lund compared the Gems data on some 15,000 loans since 1994 with Moody’s and S&P default rates for global single-B bonds, including a large proportion from advanced economies.
Over the whole period, the average annual default rate of the Gems loans was 3.6%, against 3.3% for bonds rated B by S&P and 4% for those rated B3 by Moody’s.
Losses were comparable, but occurred at different times — the correlation was 0.46 with the S&P cohort and 0.33 with Moody’s. That means investors could reduce the risk of loss spikes by holding development finance loans in their portfolios.
When loans default, the DFIs recover 72% of the principal, compared with 70% on Moody’s global loans, 59% on Moody’s global bonds and 38% on bonds in the JP Morgan Emerging Market Bond Index.
Be specific
These strong characteristics have many causes — DFIs make loans carefully; often keep tabs on borrowers and support them during the loans’ lives through in-country staff; and borrowers try hard not to default on DFI loans.
But it is one thing to present this case to an investor — it’s another to convince them to invest. Offered the chance to plough $20m into a power station project in Cameroon, an investor is unlikely to be swayed by knowing that development finance loans in general are a decent bet. It will ask about the track record of power financing in Cameroon.
Gems has been trying, bit by bit, to fill that knowledge gap.
Started in 2009 as a channel for DFIs to share information among themselves, the Gems Consortium began in 2021 to make some of its data public — what Galizia calls “Gems 2.0” — starting with aggregate default rates.
DFIs have been cautious about opening up, but have been urged to do more by the G20, especially since its Capital Adequacy Framework report in 2022.
Investors have continually demanded more detailed information, and the Consortium has gradually revealed more in its annual digests.
The October 2024 report added recovery statistics for the first time. Investors could see default and recovery rates on a grid of sectors and regions — for example, utilities in sub-Saharan Africa. And not just simple averages, but 90% confidence ranges.
Still, investors wanted more. As part of its outreach efforts to draw attention to the data and get it used, the Consortium commissioned last year a survey of asset managers, banks and other investors.
Forty-seven percent of those asked refused to participate; 63% of the rest had never heard of Gems. But 81% of all respondents said DFI credit risk statistics were relevant to the private sector — and 80% of those who had heard of it said they wanted to make better use of it.
Getting granular
A year on, the Consortium’s new study has delivered much of what investors asked for. Readers can see how recoveries differ by seniority, from senior secured to subordinated — though only on a global basis.
And for the first time, default and recovery data are broken down by country and sector. You can also see how defaults in a given country have evolved over time.
From triple-A to… Gs10
Another new feature in the latest Gems report is the revelation of Gems’s own credit rating scale, similar to those used by major rating agencies.
The point is to harmonise the internal rating systems of the Gems members. The scale runs from Gi1 at the top down to Gi10 and then into 10 notches of the Gs band, analogous to speculative grade.
At Gs1 the annual probability of default is 0.8% and at Gs6 still only 4.6%, but it rises steeply after that to 25.5% for Gs10.
“Investors can easily take the Gems rating scale, which is calibrated on probability of default, and convert it to their own rating scales,” said Galizia. “This makes the scale usable.”
“We are satisfying the key requests from the investor survey, which was the most scientific attempt to date to collect a consensus on where Gems needs to go,” said Galizia.
So, what is the track record of power loans in Cameroon?
There’ s a catch. That cell in the grid is empty. In fact, Cameroon does not appear at all in the table of defaults by country and sector. Although the database contains 59 loans in Cameroon over 31 years, there are not enough in any sector for the Gems Consortium to publish the breakdown. Even at a regional level, there are some empty sectoral cells.
In general, there have to be 10 examples in a given category, over at least five years, for the Consortium to consider the information safe to publish. It wants to avoid two risks: accidentally revealing the identity of the borrowers; and giving out misleading statistics.
“If it were up to me, anything below 100 observations would come with a massive health warning,” said Galizia.
Asked whether investors should be trusted to evaluate this kind of data themselves, Galizia said there was too much risk of them drawing conclusions that were not statistically justified.
“Investors take signals very seriously, even if the signals are not statistically and econometrically well grounded,” Galizia said, “because investors are hungry for information.”
One improvement could be for Gems to declare — for example in the empty cell for energy in Brazil — whether there have been any such loans, and how many, even if their outcomes are not disclosed.
Show me the money
Many investors are still holding back. Do they need to be tempted further?
More information on cumulative default rates and the volume of defaulted debt, as opposed to numbers of borrowers, might be helpful. But next on some users’ wishlist is the other side of the risk/reward ratio that portfolio managers always consider.
“The more the MDBs can provide data on returns as well as risk, the more investors will get confident,” Manfred Schepers, CEO of ILX, told GlobalMarkets’ sister publication GlobalCapital in May.
Their wish may be answered — one day.
“I believe returns are key for Gems 3.0,” said Galizia. “We haven’t had a structured discussion about Gems 3.0, but you can imagine it’s been a lot of work to get to Gems 2.0. But to me, returns should be a fundamental feature.”
This would take a considerable time, he said. “Data quality will be an issue for a while. But to me that is the single most important dimension that can help investors — much more than filling in more empty cells in the tables.”
Good risks
What investors would see if the returns data were published, Galizia said, “is that not only are there creditworthy projects in all places — as demonstrated by the relatively moderate default rates and significant recoveries — but… You can actually price loans reasonably if it is a good loan with good security.”
Better information about how DFIs price their loans “will help set pricing expectations,” Galizia said, “in a way that’s going to maximise the impact we have on the ground. If you only invest at 15% [internal rates of return] you are not going to create many jobs. You should be able to invest in 5% or 3% IRRs — then you are going to create jobs. We should re-anchor price expectations toward single digit returns,” he said. “Emerging markets are able to provide a good risk/return profile with single digit returns.”