Crazy Safe Arbitrage

In a past life, I used to run this structured trade finance program for some of the world’s largest agricultural companies. It was essentially a carry trade on cross-border interest rates.

These guys traded vast volumes of physical commodities. At any given point they had billions of dollars worth of inventory being ferried across the high seas. This is called trade flow — and is rather handy when borrowing from banks.

For this setup, the financing instrument utilized is a Letter of Credit (LC). It serves to remember the following about LCs:

• The de-facto financing method for international trade. A buyer will get his bank (Issuing Bank) to issue a LC in favor of the seller’s bank (Advising Bank).
• The banks act as intermediaries, de-risking physical trade — ensuring the seller gets paid and the buyer only pays once he gets the goods. LCs are issued and settled against documents provided to the banks pertaining to the goods, transport, and sale/purchase between parties.
• Depending on the type of LC, it also enables credit periods for buyers and prepayments to sellers.
• For the Issuing Bank, the risk is on the buyer — their customer. They are on the hook to pay the Advising Bank come what may. For the Advising Bank, this is FI risk — they are taking a call on the Issuing Bank’s creditworthiness.
• LCs are “non-funded” instruments — i.e. the actual movement of funds is contingent on an event (in this case delivery of goods). Hence minimal fees and 0 interest is charged.

When the macro environment is conducive, tens of millions can be added to the bottom line of these large commodities traders from arbitraging interest rates using trade flows. This is effectively free money.

So lets call the company in this example GLENGILL INC. It has a bunch of subsidiaries around the world called GG1, GG2 etc.

The transaction went something like this:

1. Find a country thirsty for US dollars. Lets say one where banks are paying at least 5% p.a. for deposits.
2. GG1 places a deposit in said bank. Lets say $100 million for 1 year. So at the end of the year the bank returns $105 million to GG1.
3. This bank is then requested to issue a (1-year) LC against the deposit in the amount of $105 million — LC settlement lines up with deposit maturity.
4. To enable this, GLENGILL INC needs to allocate one of their shipments that GG2 will sell to GG1 — a synthetic trade between its own subsidiaries.
5. GG1’s bank issues the LC in the value of $104.5 million (net fees) in favor of GG2’s bank — which will usually be in a country flush with dollars and healthy monetary environment.
6. GG2’s bank is willing to finance the LC (assuming it has lines on GG1’s bank) at 3% all in. Meaning it will release the value of the LC net of interest cost to GG2 today, and the exposure will be settled 1 year later when GG1’s bank pays out against the LC.
7. So, $104.5 million less 3% = $101.365 million.

End-to-end the entire deal takes about a day or two. So for locking up $100 million for a couple of days, GLENGILL INC earns $1.365 million, or an annualized return of around 248%.

Not bad, eh?

Technically, this is risk free. Arbitrage is supposed to be, but in financial market trades it entails all sorts of operational risks. In this case settlement lines up quite nicely because of the involvement of banks.

The Issuing Bank relies on the deposit its already holding. The Advising Bank relies on the Issuing Bank’s LC. GLENGILL INC walks away with its cash and a kicker with no further obligation.

Not to say that things can’t go wrong. Banks always bake in some sort of recourse on their customers. If the shipment sinks and insurance doesn’t pay out etc. it can create a cascade of problems. Plus these transactions hardly exist in a vacuum. The banks willing to accommodate these deals usually have a more entrenched relationship with these companies. Their fates are often tied to some extent.

Also, one of the dollar-thirsty countries at the time that was usually tapped up for issuing banks was none other than — Sri Lanka! Not entirely sure what the fate of foreign depositors was as a result of the big D. But had it happened in the two days it takes to execute such a trade, that hundred mill might be toast.

All things considered, if you quantify the risks I would imagine it most certainly entails a return far far below ~2.5x p.a.

A couple of years ago this arbitrage structure was adopted toward a retail bond issued by the Government of Pakistan. Here’s the product paper for reference: