Leverage is a calendar, not a ratio.

The denominator problem, again

Net debt to EBITDA is the standard measure and it has the same defect as a price to earnings ratio. The numerator is a fact. The debt is the debt. The denominator is an output of wherever the business happens to sit in its cycle.

So it moves, and it moves in the wrong direction. Take the company from the last note. Same business, same fixed cost base, same ordinary cycle. Give it $400.0M of net debt at 6%, which costs $24.0M a year, and a covenant at 4.5 times.

Not one dollar of debt is repaid or borrowed anywhere in that table. The ratio moves from 1.80x to 5.97x, a factor of 3.3, entirely because the denominator went home.

Read the top row properly, because it is the whole note. At the trough this company earns $17.0M and owes $24.0M of interest. It cannot pay its lenders out of operations. Nothing has gone wrong at the business. This is simply what an ordinary cycle looks like from inside a levered balance sheet.

The cushion that is not a cushion

The usual defense is headroom. At mid-cycle this company needs $88.9M of EBITDA to stay inside its covenant and produces $130.0M, a cushion of $41.1M, or 32% of EBITDA. That reads as comfortable. Board packs are full of that number.

It is the wrong comparison, and the right one takes one line. A normal cycle, not a crisis, not a pandemic, just the ordinary breathing of this industry, takes EBITDA down 48% from mid to trough. The cushion is 32%.

The cushion covers about 65% of an ordinary downturn. It is not protection against a bad year. It is protection against most of an average one.

Run the same arithmetic forwards instead and it produces the prescription. Debt this business could carry all the way through a cycle, sized off trough EBITDA rather than mid-cycle EBITDA, is $301.5M. It borrowed $400.0M. It is over-borrowed by $98.5M, which is 33% more debt than the cycle supports, and at mid-cycle every ratio on the page says that is fine.

Two identical companies, one calendar apart

Now the actual argument. Take two companies with the same business, the same $400.0M of debt, the same 3.08x mid-cycle leverage, the same everything. The only difference between them is the year printed on the maturity.

Company A's debt comes due in a normal year. It refinances at roughly the old cost and nothing happens. There is no story. Nobody writes it up.

Company B's debt comes due in the trough year. Its EBITDA has halved, it is through its covenant, and it is refinancing into a market that can see all of that. Say it gets done at 5 points wider.

• A keeps paying $24.0M a year
• B now pays $44.0M a year
• The difference is $20.0M a year, permanently, which is 25% of mid-cycle EBIT

A quarter of normal-year operating profit, gone forever, transferred to lenders. Not because anybody misjudged the business. Because of the year on a piece of paper, decided years earlier by somebody optimizing a coupon.

And when refinancing is not available at all

The worse version. B cannot refinance on any terms it will accept, so it does what companies in that position do and raises equity to get back inside the covenant.

At mid-cycle these shares are worth $8.00. In the trough, with an enterprise value down 30% and $400.0M of debt sitting in front of the equity, they are worth $4.40. Debt does not just add risk to the equity. It concentrates the entire swing of enterprise value into a smaller slice, which is why the equity of a levered cyclical falls so much further than its business does.

That 30% haircut is an assumption rather than a derived figure, and it is the load-bearing one in this section. It is on the mild side of what actually happens to a levered cyclical at the bottom, which makes the numbers below conservative rather than dramatic.

So it raises $150.0M at $4.40.

• Shares issued: 34.1M
• Share count goes from 100M to 134.1M
• Existing holders diluted by 25.4%
• When the cycle recovers, the shares are worth $7.08 instead of $8.00
• $91.5M of value, permanently moved out of the people who owned it and into the people who turned up at the bottom

That is the mechanism people mean when they say a stock never recovered. The business recovered completely. Every operating number went back to where it started. The claim on it was permanently reduced, at the worst available price, on a date that was set before anybody knew what the cycle would do.

What this changes about reading a balance sheet

Stop asking how much and start asking when. The maturity schedule is in the debt note of every annual report, it takes two minutes to find, and it tells you something the ratio structurally cannot. Two companies at identical leverage, one with nothing due for six years and one with a wall next year, are not comparable and the ratio cannot tell them apart.

Test the covenant against a real downturn, not a mild one. The question is never "is there headroom." It is "is there more headroom than a normal cycle removes." Here there was not, and the number that told you was 32% against 48%.

Size the balance sheet off trough earnings. A cyclical that borrows against mid-cycle EBITDA has borrowed against a number it does not earn most of the time. Trough EBITDA is the honest denominator, the same way mid-cycle earnings are the honest denominator for a multiple.

Watch for leverage and peak margins in the same company. That is the specific combination that kills. The last note showed that peak earnings make a stock look cheapest exactly when it is most dangerous. Peak EBITDA does the same thing to a leverage ratio, at the same moment, in the same company. A screen would show a low multiple and conservative leverage, both wrong, and wrong for the identical reason.

A balance sheet does not have a condition. It has a schedule, and solvency is just the question of whether a bad year and a due date arrive together.