September 29, 1998

An Investment Opinion
by Dale Wettlaufer

Good Times on Borrowed Money

Long-Term Capital Management. Ah, what a delicious irony in the name of the fund involved in the biggest trading blowup in the U.S. this year. You couldn't come up with a trading strategy that is more short-term oriented than a black box driven $100 billion hedge fund. Let's see. What are some bad ideas I can come up with? A thousand-year Reich might be one. Or a long-term investing strategy that is based on leverage of 25-to-1 and the premise that all risks are knowable and controllable might be another. We know why the first was a really bad idea. Why the second?

First, there's the leverage issue. We've dealt many times before with the mistaken premise that a high return on equity (ROE) is an absolute good. And maybe we shouldn't go so lightly as before on those that promulgate that idea, because the Long-Term Capital episode is a good example of why that premise is mistaken. Using a model that is a great financial jackknife and with which readers of this column are familiar, the idea that all high ROE enterprises are not good can be explained this way:

ROE = Asset turnover * margin * leverage.
Asset turnover = revenues / average assets
Margin = net income / revenues
Leverage = Average assets / average shareholders' equity

Expressing the idea above in another way, Return on Equity =

Revenues       Net Income     Avr. Assets 
 ----------- * ------------ * ------------- 
 Avr. Assets     Revenues      Avr. Equity
Mathematically inclined readers will notice that the numerators and denominators cancel out each other to produce the regular ratio by which we know ROE -- Net Income divided by Average Shareholders' Equity. Now, what does this have to do with Long-Term Capital? As hedge fund manager James Cramer wrote on Monday morning:

"Let's stop the pretense that these guys were any good at all. Ever. Now that we know what kind of leverage they were using their returns sucked. The whole time. They should have been shooting the lights out with the amount of money they were borrowing. If borrowing means taking on risk, which it does by the way, regardless of what these clowns tell you up in Greenwich, these guys should have been up 100% to 150% without a problem. So stop praising their returns ex-this year. They were disappointing given the leverage."

Numerous reports have put the company's balance sheet when healthy at equity of $4 billion and assets of $100 billion. That's a leverage ratio of 25-to-1. Allowing for a little slippage in facts from playing telephone with a story that is still not fully detailed, let's assume it had $80 billion in assets to its $4 billion in equity. That's still a leverage ratio of 20-to-1. What that means is this: For every 100 dollars of assets, by definition, the company had $4 in equity and $96 in liabilities. Cramer is saying that a return of 60% sucks because that works out to an unleveraged return of 3%.

Think of it this way. Since ROE = ROA times leverage, then we just need to divide ROE by the leverage ratio to arrive at that number. Here, it's ROE of 60% divided by a leverage ratio of 20. By expecting a return of 100% or 150%, then we're talking about an unleveraged return of 5% to 7.5%. Since this fund can roam the world and invest in any class of assets, that sort of return is not unreasonable. And, for a moment, let's just take note that people citing modern portfolio theory by looking backward at returns of various indexes should get a life. Modern portfolio theory uses expected return to adjust return versus risk. Would fund managers please stop using it to excuse the fact that they might have lost a little less money than the index against which their returns are measured?

In other words, Long-Term Capital's absolute return on assets and expected returns on assets and equity all blew chunks. If you're going to engage in seemingly risk-free global arbitrage, then your return on assets really should beat the risk-free return on Treasury bills. Come on. There is an analogy to be drawn between this and the S&L debacle. When times are going well, a focus on return on equity can make a bad operator look like a damned genius. Even a focus on return on assets without paying attention to the risks involved can make a marginal operation look good.

Say I'm an S&L with lots of junk bonds and commercial construction loans on the balance sheet. If I have $10 billion in assets and $750 million in shareholders' equity, then I have to fund the difference with deposits and borrowings. Let's say the yield is 13% on the junk bonds and 11% on the commercial loans (the junk bonds and loans are weighted equally at 20% apiece, the rest is in run-of-the-mill mortgages and things yielding 7 1/2%), and my overhead is 50% of revenues. And let's say the cost of funding these assets works out to 4% of the liabilities. Further, loan loss provisions are being taken to the tune of 1 1/2% of assets. My income statement would look like this:

Interest income�$930 million
Interest expense�$370 million
Net interest income before
loan loss provision�$560
Loan loss provision�$100 million
Noninterest income�$140 million
Overhead�$350 million
Pre-Tax income�$250 million
Net income�$162.5 million

On average equity of $750 million, I'm a genius. I've taken my backwater S&L and turned it into a premier company generating a great 1.65% return on assets and a 21.7% return on equity. Roll out the awards.

What happens, though, when the commercial real estate market turns into quicksand and the companies issuing the junk bonds can't service the debt? I don't need much havoc in these markets to turn me insolvent. If the junk bond market crashes by 30% and my portfolio mirrors that, then all of a sudden I take a big hit to shareholders' equity. After tax, that's a $390 million hit to shareholders' equity. I better hope that the commercial real estate market doesn't succumb.

Oh, but too bad! The market has gone illiquid! Congress just changed the rules of the tax game and now everyone's running like a bunch of damned idiots. Triple-A office space is going for 40% of the best projections and the occupancy on the rest of the portfolio is ridiculously low -- and that assumes that all your builders didn't leave you with just skeletons of buildings. All your projections are now running into human nature and market forces. You have to mark down your real estate portfolio by 20%. After-tax, you take another $260 million hit to shareholders' equity. You're now down to $100 million in shareholders' equity.

You better hope that job market holds up. Oh, too bad -- corporate downsizing. Good thing your bank serves a good, conservative clientele. Yes, true, but 4% of your customers can't escape the carnage and declare personal bankruptcy. After-tax, that's a $156 million hit to shareholders' equity. What do we have for him, Bob? That's right! A free trip to receivership and a wipeout of your shareholders! Yes, that's right, you can win lifelong enmity in your community. But don't worry, some of the guys at the country club will still think you're a bigshot.

Now take that scenario and make the unleveraged risk-adjusted return on assets worse. Then make it so that the liabilities are variable. You're not just paying passbook accounts 4%, your liabilities are made up of Treasury notes and bonds, interest rate swaps, futures, and options. And make your assets the sovereign obligations of a bumbling new democracy that did a reverse split of its currency last year. All of a sudden you've got a higher proportion of your assets blowing up at a worse loss than our S&L example, and you've got liabilities that are, in a manner of speaking, sending your funding costs through the roof.

What would a good risk manager do here? Maybe cool off a little bit? Step back and not press it so much? Not at Long-Term Capital. Not with black boxes programmed by Nobel laureates. You press it. Shake the dice baby, 'cause we can't be wrong! Not when the premises behind the black box are based on hundreds of years data from financial interaction. There's a value for everything, they think, and that value can be calculated to five decimal places, enough so that huge arbitrage trades can make up in volume what they lack in profit margin. The problem is, this isn't the grocery business. It's not Wal-Mart. Every fifth box on the shelves of Safeways around the country isn't going to explode, and the interest rate on their commercial paper isn't going to all of a sudden triple. "Making it up on volume" in the supermarket business is a heck of a lot different than it is in the global arbitrage business.

The problem with black boxes and modeling financial interactions is that the human world is a complex, messy thing. You can model until you're blue in the face and even come up with lots of viable-looking things, but projecting that forward with a degree of certainty can really get messy. You can turn out statistically significant stuff and still lose. Try to model in the flux of markets and real world slippage and liquidity potholes. Capture all of that stuff. Sooner or later, the real world rips to shreds even the most well-tested algorithms. And that's especially true when geniuses lose the flexibility to consider that maybe they are wrong.

For individuals going to 150% investment in their margin accounts or putting their nest eggs into model portfolios based on shabby statistical work, watch out.