Felix Salmon

If American Express is struggling with past-due loans, why shouldn’t it be struggling with performing loans too?

Last year our company spent almost $1 million with American Express, an average of over $81,000.00 per month. We paid the full balance every month – on time. Last week a representative from American Express called our controller to thank us for our spending last year. This week, with no warning, we have been cut-off after spending only $39,000.

When our VP of Operations was denied a charge after booking flights for many managers to attend a conference, he called the accounting department to find out why. We immediately called AMEX to resolve whatever problem so that our business could continue operating normally. What they told us was disturbing….

After living on the hold line for over an hour, they agreed to a compromise, we were to pay the current balance and they were to do nothing. When we picked ourselves up off the floor we asked what was going on. Why would American Express only want $300,000 of our business instead of $1,000,000?

It’s a good question; here’s a stab at an answer.

There are basically two different ways that a lender can judge creditworthiness. The first is credit history: when the borrower has borrowed money in the past, have they always paid it back on schedule? During normal times, it’s a reasonably safe assumption that someone who’s always paid their bills in full and on time will continue to do so going forwards.

These are not normal times, however, and lenders are revisiting a lot of the assumptions they made during the boom years (like "house prices don’t fall simultaneously across the USA"). And so now they’re asking themselves whether the credit-history assumption will hold.

For there’s another way of judging creditworthiness: simple ratios, like debt-to-income, or assets-to-liabilities. (Or, in the housing world, loan-to-value.) If those ratios start implying that the borrower won’t be able to repay the loan, then it might be a good idea to scale back the loan – even if the borrower has always repaid their loans in the past. After all, it’s a good idea to scale back before the borrower defaults, rather than waiting for the inevitable.

So what Amex is doing might actually make sense, although it does admittedly look peculiar at first blush. But it’s certainly another datapoint to add to the list of credit-crunch indicia.

Kate Haywood has some early speculation today on the subject of Microsoft’s bond spreads. Microsoft has never had any bonds – it doesn’t even have a credit rating – but its CFO has said, in the wake of the Yahoo bid, that "it’s likely we’re actually going to borrow for the first time".

Without a rating or any kind of benchmark out there, it’s very tough to work out just how attractive Microsoft really will be as a credit. Here’s Haywood:

One portfolio manager said he expects a 10-year Microsoft bond to sell with a risk premium, or spread, of less than 1.6 percentage point over U.S. Treasurys. A recently issued 10-year from triple-B-rated transportation company Union Pacific sold at a spread of 2.1 percentage points.

Union Pacific? I know it’s hard to find comparables in this market, but that’s straight out of left field. To me, 160bp over Treasuries seems quite wide. Microsoft had gross profit of more than $40 billion last year, and net income of $17 billion. The interest on $10 billion of notes paying a 5% coupon would be $500 million a year: pocket change, really. Microsoft bonds are a very safe bet in a time when there’s a large appetite for safety, and I wouldn’t be at all surprised if they came in much tighter than 160bp over.

That said, however, Treasuries are trading at extremely low nominal rates right now: the 5-year is at 2.64%, and the 10-year is at 3.54%. If Microsoft came at 136bp over the 5-year, that would mean a yield of just 4%, which is not exactly compelling from a buy-side perspective. In any case, Microsoft’s borrowed billions are unlikely to cost it much more than 5%, which is a pretty attractive rate to pay for a company as high-profile as Yahoo.

Yesterday, I applauded – on both sides – Merrill Lynch’s departure from the team raising money to fund Vale’s $90 billion takeover bid for Xstrata. This morning, I got my daily email from Latin Finance (highly recommended, sign up here, although I do wish they would publish it online) which has more details on the syndicated loan that Vale is trying to raise.

According to Latin Finance, the loan is pegged at $40 billion or more, and will consist of tranches ranging from 18 months to seven years. The benchmark will be the five-year loan.

Now Vale is no stranger to enormous syndicated loans, and when it bought Inco at the end of 2006 it paid 62.5bp over Libor on the five-year tranche. But things have changed since then. Now, the talk for the Xstrata acquisition finance is closer to 150bp over Libor – a startling indication of how much things have changed over the past year and a bit. (And remember, even at 150bp over, the pricing was still too tight for Merrill Lynch.) Latin Finance thinks that Vale will pay up:

Identifying a price that will clear the market and expedite the rest of the acquisition is of utmost importance, which suggests Vale may look to err on the generous side.

But I’d also note that in nominal terms, 150bp over Libor when this deal gets done might well be a lower rate of interest than 62.5bp over Libor at the end of 2006. Yes, this is floating-rate credit, which means the Inco loan is now costing Vale much less than it could ever have expected. But that might well just give Vale that much more room to pay a bit more this time around.

And I’d also note that Vale is still funding a huge part of the total acquisition price in the loan market, rather than bypassing the banks and going straight to the bond market instead. Banks might be feeling a bit of a capital crunch right now, but they’re still going to try their hardest to be there for their biggest and most valuable clients.

Update: the Latin Finance Daily Brief is online after all: it’s here.