Two retailers can run the same stores, sell the same goods at the same gross margin, and show completely different valuation multiples. Not because one is a better business. Because one bought its property and the other rents it, and most of the screens you use treat only one of those decisions as an obligation.
This is the single largest comparability problem in retail and airline names, and it is entirely mechanical. You can fix it in about twenty minutes per company with a pocket calculator, and once you have done it a few times you stop trusting an unadjusted EV/EBITDA on any asset-light-by-choice business again.
A lease is a debt with better manners
Sign a fifteen year lease on a store and you have committed to pay a fixed amount every year whether or not the store makes money. Borrow money to buy the same store and you have committed to pay a fixed amount every year whether or not the store makes money. The two commitments differ in tax treatment, in who carries the residual value, and in how easy they are to walk away from in a bad year. Economically they are close cousins.
Current lease accounting standards bring most of this onto the balance sheet as a right-of-use asset with a matching liability, which helps. It does not finish the job. Older filings still push rent straight through operating expense. Short leases and variable, turnover-linked rent are frequently outside the capitalised figure. Non-US filers you meet on a global screen may be reporting on a different basis from the US name you are comparing them against. So the reported numbers are not automatically consistent, and if you are running a comparison across a mixed universe you should assume they are not.
The consequence is specific. A company that rents shows lower EBITDA, because rent sits inside operating expense, and shows less debt, because part or all of the obligation never made it to the balance sheet. Lower EBITDA and less debt push the EV/EBITDA multiple in opposite directions, and which one wins depends on how the multiple compares to the capitalisation rate you apply to rent. Most people assume the effect is small. It is not small, and more importantly it is not neutral across a peer group.
The two identical retailers
Made-up round numbers, chosen so the arithmetic is visible. Two chains with identical store economics.
The owner bought its property for 3,200 and borrowed 3,200 to do it. It pays no rent, so EBITDA is 800. Market capitalisation is 3,000. Enterprise value is 3,000 plus 3,200 of debt, so 6,200, and EV/EBITDA is 7.75 times.
The renter leases identical property for 400 a year. Rent sits in operating expense, so EBITDA is 400. Market capitalisation is also 3,000. It carries no debt, so enterprise value is 3,000 and EV/EBITDA is 7.5 times.
On the unadjusted screen the renter is the cheaper stock. It is also apparently debt free, which is the sentence that does the real damage, because it is the sentence a private investor repeats to themselves when the position goes against them.
Now capitalise. Take the annual rent and multiply it by eight, a common convention and as good a starting point as any. That gives a lease liability of 3,200. Add it to net debt. Take the rent back out of operating expense, so EBITDA rises to 800. The renter's enterprise value becomes 6,200 and its EV/EBITDA becomes 7.75 times. Identical to the owner, which is what you would expect from two identical businesses, and which is the entire point of doing the adjustment.
The ranking on the screen never moves
Worth being direct about what this board does and does not carry, because the adjustment above needs columns that are not there. The visible columns in this capture are Price, Mkt Cap, P/E, a composite score, a verdict, three sub-scores and a mispricing figure. There is no enterprise value column and no EBITDA column, so there is no EV/EBITDA to re-rank. You cannot make the lease adjustment on the screen and watch the order shuffle.
What you can do is use the screen as a shortlist generator and do the adjustment yourself. Filter to consumer discretionary or to whichever bucket holds the lease-heavy names you care about, take the shortlist, and for each name pull two numbers from the annual report: the rent or lease expense, and the lease liability if the balance sheet carries one. Then build enterprise value by hand as market capitalisation plus debt plus the lease liability minus cash. It is a spreadsheet with four columns and it takes an evening for a ten name list.
Where the adjustment earns its keep, and where it lies
The identical-twins example above is deliberately rigged so the adjustment lands exactly. In practice the eight times multiple is a convention, not a measurement. If the remaining lease term is three years, eight times rent overstates the obligation badly. If it is twenty five years on prime sites, eight times understates it. Where the filing gives you an undiscounted schedule of future lease payments, use that instead and discount it at something near the company's borrowing rate. The convention is a fallback for when the disclosure is thin.
Two more places it misleads. Turnover-linked rent, where the landlord takes a percentage of sales, is genuinely not a fixed obligation and capitalising it as though it were overstates the risk. And a company with a short average lease term has real optionality, since it can shrink the estate at renewal without paying to break anything. That flexibility is worth something and the capitalised number does not show it.
The rule I use for whether the unadjusted multiple is usable at all is crude and it works. Take capitalised leases, divide by EBITDA before rent. Under about one times, the adjustment is a rounding difference and you can skip it. Between one and three times, do the adjustment before you compare the name to anything. Above three times, the unadjusted EV/EBITDA on that company is not a number, it is a typographical accident, and any screen that ranks on it is telling you about accounting rather than about value.
What the adjustment does to a 4,000 dollar position
Percentages hide the part that matters, so put it in money. You buy 4,000 dollars of the renter because it screens at 7.5 times against a peer group averaging 9 times, and you expect the gap to close. After the adjustment the renter is at 7.75 times and the peer group, once you have done the same work on them, is at 8 times. The discount you were buying went from 17 percent to about 3 percent. Nothing about the company changed. Your reason for owning it did.
The second half is the one that actually hurts. Before the adjustment the renter had zero net debt against 400 of EBITDA. After it, the renter has 3,200 of obligations against 800 of EBITDA, which is four times. That is not a mild balance sheet. In a year when comparable sales fall ten percent, the owner can refinance, sell property, or negotiate with a bank that would rather be repaid slowly than foreclose. The renter owes the same rent on the same day, and the landlord has a much simpler set of options. The equity of a four times levered retailer in a bad year behaves like an option, not like a share of a business.
The two lines to pull this week
Pick the most asset-light name you own, meaning the one that rents the most and owns the least. Open its annual report and find two things. First, the total lease or rent expense for the year. Second, the maturity schedule of lease payments if it is disclosed. Write down capitalised leases divided by EBITDA before rent.
If that ratio is above three, go back to whatever comparison convinced you to buy the position and redo it with the liability included. You may well conclude the position is still fine. What you will not do again is describe that company as debt free, and on a business where the fixed obligations are four times cash profit, dropping that phrase from your thinking is worth more than the multiple ever was.