When Tradeteq, the official technology partner of the global bank and investor-led Trade Finance Distribution initiative, published its recent whitepaper, ‘Trade Finance in 2020: Asset Distribution – A Macro-economic Necessity’, its aim was to explain how the greater distribution of trade finance assets to non-bank investors would help SMEs access funding, both during and after the COVID-19 pandemic.
Smaller businesses have been badly hit by the enforced shutdown of trade. As these businesses are less likely to receive letters of credit (LCs) from banks during this period, the need for other options must be considered, argues Tradeteq CEO, Christoph Gugelmann.
His solution is to distribute trade finance assets to non-bank investors. He believes that doing so could unlock millions of dollars in liquidity, helping to plug the current US$1.5trn trade finance gap. Banks would also lower the risk on their balances, potentially creating new avenues for growth in the market.
The assets, in the form of trade receivables, could be bundled in a portfolio of possibly thousands of individual assets, which revolve every 30 days, explains Gugelmann. In risk terms, bundling would not be tranched, as CDOs are, but would instead aim for simplicity.
It requires a lot of automation to make this an efficient proposition for institutional investors. They typically would not have the operational infrastructure to access trade finance assets nor the experience of this market, hence it is not a common target asset for them. This, Gugelmann says, is where Tradeteq sits, acting as platform provider to facilitate such trades.
As with every asset, watertight contracts are also essential. The way to transfer risk from the bank to the investor could be directly through a contract, such as a standard Master Receivables Purchase Agreement (MRPA) or involving a Special Purpose Vehicle (SPV).
In practice, an instrument would be bundled as a matched funding transaction, creating a single note that shows exactly what is in the bundle, and, as a fully fungible asset, purchased or sold by the investor through the normal clearing mechanisms.
An investor investing for two years would see the underlying assets of the note revolving on a matched and ongoing basis. As such, that two-year investment requires no further consideration of exposures for that period, and no operational activity bar absorbing regular reporting. That said, a significant event reported within a portfolio (such as a major default) could materially adjust the quality of that portfolio, this being a signal for the investor to offload its investment.
Asset pricing would be done by the individual banks, taking a margin for their services, and offering a percentage to investors for taking the risk. As a wider spectrum of investors engage in trade finance as an asset, Gugelmann anticipates the ability eventually to precisely benchmark the risk assessment of the portfolio against price. A secondary market is also expected.
Rating the package is the most complex matter. Gugelmann notes that it’s not possible solely to use the obligor credit rating, not least since typically it’s only the larger organisations that have such ratings. And although large corporates produce a lot of data that credit agencies can consume (such as news and financial reports), smaller firms do not typically generate such useful, comparable data. “What you can observe though is their trades,” says Gugelmann. What Tradeteq offers is a transactional rating. “We rate the transaction, not the company per se.”
The transactional view is based on certain behavioural aspects which can be observed in trade networks, formed by every company in order to do business, which he says are “surprisingly interconnected”. It involves looking at not just the credit risk but also fraud risk, and dilution (uncollected AR). “It’s a holistic way for institutional investors to view risk.” For many investors, he says this view is “decisive”.
Indeed, many need to report on certain regulatory frameworks; without a risk assessment from a third party, they cannot access these portfolios. The main ratings agencies may not have a view on aspects such as dilution, he says, but Tradeteq is now meeting this need.
A trade portfolio may contain multiple assets (maybe as many as 10,000 receivables) from a number of obligors. Tradeteq has already examined a number of large corporate portfolios, connecting the aforementioned networks of trade counterparties (networks of networks, in fact) and looking at behavioural aspects.
It notes that when stressed, a corporate may start paying suppliers later or partially, or order fewer goods, for example. “Having the information on how behaviour changes from one company to another gives tremendous quantifiable information about the loss-absorption capability of a company,” explains Gugelmann.
It uses machine learning algorithms, developed in collaboration with the likes of Oxford University or Wroclaw University, which have been “several years in the making”. Tradeteq most recently received a grant from the Monetary Authority of Singapore in order to conduct further research in the field, together with the Singapore Management University and IBM.
The algorithm’s task is to undertake network analysis, analysing how credit events travel through them, and creating a risk view – the probabilities of default – of specific transactions. These probabilities can be mapped to every included transaction, every day, allowing formation of an average rating for the overall portfolio. This, Gugelmann believes, is the level of third-party rating that will open up investment for the institutional investor.
Of course, rating accuracy will be important to investors. The measure for accuracy of machine learning output – known as the ‘area under the curve’ – used by Tradeteq achieves over 90% prediction accuracy. Whilst this process cannot, for example, predict the rise or fall of oil prices, it can quantify the likelihood of the effect of price change impacting up or downstream participants in a supply chain.
The effects of commodities price shifts (or even something obvious like customers tailing off, as seen by airlines during the pandemic) will affect buyer behaviours (such as paying later or partially), which in turn will impact participants along the supply chain in certain ways. The machine learning algorithm judges millions of such data points daily, dynamically creating a risk-rating for each transaction in a portfolio.
Even if no trade has taken place between a buyer and supplier in some time, observation of the networks in which they participate can reveal the shocks likely to affect them, and their probable capacity to absorb those shocks. Somewhat counterintuitively, accuracy can “go beyond 90% with zero financial information,” says Gugelmann.
With bank-intermediated trade finance measured in trillions of dollars (the exact number is not known), even with a few hundred billion in outstanding investment flowing into trade finance from institutional investors, money market funds, wealth funds, family offices and so on (even retail, one day), it will not just be a game changer for global trade but also for many economies, believes Gugelmann.
With that volume revolving several times a year, it allows huge volumes of liquidity to become accessible, through banks, by their corporate customers. With bank trade risk lowered, and access to affordable trade finance increased for smaller firms, it is, says Gugelmann, a “win-win”.