Project Oakstead — 3-Hour Full LBO with an IPO Exit

A tranche issued below par raises less cash than its face value. Interest, amortization and repayment all run off the face amount, but only the net proceeds are available to fund the uses. Size the sources off face and the sources column overstates the cash raised, the equity plug is understated by exactly the discount, and the opening balance sheet does not close in any year. It is a small number that breaks a large tab, and because the equity check is the plug it hides inside the one figure the whole return is measured against.

Since the current presentation took effect, debt issuance costs and original issue discount are direct deductions from the carrying amount of the debt rather than deferred assets. The balance sheet therefore carries the paper at face less what has not yet accreted, and that number is smaller than par for the whole life of the deal. Two numbers, both correct, describe the same debt, and knowing which one belongs where is most of what this mechanic tests.

A fixed-rate bullet has a determinable set of cash flows, so it has a yield to maturity and the effective interest method applies: interest expense is the carrying amount times the yield, the cash coupon is the face times the coupon, and the difference is accretion that raises the carrying amount every year without a dollar being borrowed. A floating tranche that amortizes and is swept has no knowable balance path, so no yield exists to strike an effective rate on, and the credit agreement amortizes its discount straight-line over the remaining life instead.

Prepaying a term loan issued at a discount accelerates the pro-rata share of the remaining discount into that year's interest expense — and the same happens to the unamortized issuance costs. It is a real charge on the income statement in the year the principal goes out, it is non-cash, and it comes back on the cash flow statement. A model whose discount amortizes on a straight line and never reacts to a sweep has missed a charge that peaks in the year the offering's proceeds land on top of it.

Cash interest is what the business pays and what a credit committee sizes coverage on. Total interest expense is what the income statement carries, and on a structure like this one it is materially larger because four of its lines are non-cash: the payment in kind, the accretion of the discount, the amortization of the issuance costs and the write-off on prepayment. Print the two subtotals separately. A reader who cannot see the difference cannot tell whether the coverage ratio in front of them is the one a lender would quote.

Paper issued below par yields more than it pays. The gap between the quoted yield and the stated coupon is exactly what the discount is worth expressed as a rate, and it is provable in one row: discount the coupons and the principal at the quoted yield and you should recover the issue proceeds. Build that check. It takes ninety seconds and it certifies the whole accretion schedule underneath it.

Leverage tests, the excess cash flow calculation, the toggle election and the exit equity bridge are all struck on the par amount of the debt. The carrying value is a presentation matter. Netting the unamortized discount out of net debt when you bridge to equity adds it straight to the equity value, for money the company still owes — and the error survives every tie-out, because the balance sheet still balances. Using the carrying value where a lender or a buyer would use par is an expensive mistake.

Toggle notes pay cash or accrue depending on a test, and the test here is a leverage ratio at the PRIOR year end. Testing on the prior year end is what keeps the election out of a circular reference; testing on the current year end makes the file circular and forces iterative calculation on. Interest coverage can get worse in the year the balance sheet gets better, because deleveraging is what turns the cash coupon on. A model whose coverage ratios rise monotonically in every year has modeled a bond rather than a toggle.

Floating tranches price off a base rate subject to a floor, and a forward curve that declines through the projection will cross that floor. Build one effective base rate row as the greater of the curve and the floor and point everything at it. Wire the curve straight into the pricing formulas and the back years of the model are under-costed on every floating tranche at once — a mistake that is right for the first few years, which is why a spot check misses it.

When the capital program front-loads, depreciation lags it by years and the two move in opposite directions across the plan. Each vintage carries its own straight-line life with a half-year convention in the year it is placed in service, stacking on top of the existing asset base's own remaining charge. Running depreciation off revenue misses the ramp entirely: it understates the tax shield in the back half, overstates net property, plant and equipment, and will not tie to the balance sheet you are also being asked to build.

The add-backs at the front are the familiar half of the question: ask which figure the multiple is on, what the reconciliation contains, and whether each item describes something that stopped or something the business does. The other half is at the back. When a plan carries listed-company costs and stock-based compensation from the first year on the market, those costs sit inside the earnings the offering is priced on. Pricing the book on the year before them, or adding the stock compensation back because it is non-cash, values a company that does not exist.

A sale is one date and one price. A listing is a partial primary and secondary offering priced at a discount to a fully distributed value, a lock-up, and then a series of marketed secondaries at their own discounts and fees. The sponsor is paid on several dates at several prices, and the primary issuance dilutes it permanently at the offering's discounted price. Model each realization separately: the shares sold, the reference price, the discount, the rounded price, the fee and what is left.

Books price at clean increments. Take the indicative price — the reference value less the file-to-offer discount — and round it DOWN, never up, because rounding up prices the book above the discount the underwriters said they needed. The effective discount therefore comes out a shade wider than the stated one, and both belong on the page. The same rule applies at every secondary.

The geometric form — proceeds over invested capital raised to the power of one over the hold — assumes a single exit date. With one outflow and several inflows on different dates it is arithmetically wrong, and the return has to be solved. Lay a Newton iteration out as columns: a seed, the present value of the inflows at that rate, the function to be zeroed, its derivative and a corrected rate. Each column reads the one before it, so nothing is circular, no iterative calculation is needed and no banned function appears in the file.

At the offering a fund does not have a result. It has a partial realization and a retained stake, and it has to report both — cash received plus shares marked at the offer price, less what the management plan would take on the part it still holds. That mark is correct on the day it is struck. What happens afterward is more clock, more discounts, more fees and more settlements of the option ladder. Being able to say what stands between the mark and the money, and that all of it was knowable at signing, is what the question is asking for.

A management plan struck as one pool at one price takes a fixed share of the gain. A ladder of tranches at rising strikes does not: at a low realization price only part of it is in the money, and by a high one all of it is. The payoff is convex in the price, so a flat percentage over-charges the early realizations, under-charges the late ones and gets the total wrong. Settle it tranche by tranche at each realization price and take the share the sponsor actually monetized.

'What is the return?' and 'what would you pay?' are different questions. The second holds the return at the hurdle and solves back for the entry multiple. Work out first what the entry price changes: if the paper is fixed in dollars, the exit is priced off a trading multiple and the net debt the plan produces, and the share count and the option ladder are fixed by documents, then none of the sponsor's inflows moves with the price. The equity check that clears a required return is then the present value of those inflows at that return — one row of discount factors rather than a search.

A maximum price quoted at two decimals has to be truncated rather than rounded. Rounding a ceiling up produces a price at which your own model returns less than your mandate — a bound the analysis behind it does not support. A bound always rounds toward the side that keeps it true. It is a small piece of craft and it is the difference between a number you can defend in an investment committee and one you cannot.

Every realization in a staged sell-down is struck below where the stock is marked and pays a fee on top, and the whole program takes years rather than a day. Total those costs once, at the reference prices, and express them as a share of what the stake is worth. Then price the alternatives — one block, two blocks, everything into the offering, a trade sale — and compare them on BOTH measures, because selling faster raises the annualized return and lowers the money. Underwrite the route as well as the business.

The completed workbook and one page justifying the highest price you would pay. The page is not a summary of the model — it is the answer to question three, with the reasoning behind it. Budget twenty-five minutes for it and write it last, but decide what it is going to say while you are building, because the argument comes out of the returns tab rather than out of your head at the end.

Operating model, transaction, income statement down to EBIT, cash flow to the line before debt service, debt schedule in full, then back to fill the interest block, then the balance sheet, then returns. The balance sheet cannot close until the schedules that feed it exist, and the interest block cannot be filled until the debt tab does. Candidates who build strictly left to right stall at the interest lines with half the clock gone and no debt schedule.

Almost certainly because you funded the uses with the face amount of the debt rather than what it actually raises. Two tranches price below par, so the cash available to fund the transaction is smaller than the face outstanding, and the difference is exactly what your equity plug is short by. Read the issue-price column before you total anything, and remember that interest and repayment still run off face.

A deduction. Both the discount and the debt issuance costs are presented as direct reductions of the carrying amount of the borrowing, not as deferred assets, and that is also the presentation that makes the accretion visible. The balance sheet then carries the paper at face less whatever has not yet accreted, while every leverage test and the exit bridge use par.

Because only one of them has a knowable cash flow. A fixed-rate bullet has a determinable schedule, so a yield to maturity exists and the effective interest method applies. A floating term loan that amortizes and is swept has no knowable balance path, so there is no yield to strike an effective rate on and the credit agreement uses a straight-line convention over the remaining life instead — with a pro-rata write-off whenever principal goes out ahead of schedule.

Beginning, on every tranche, and the directions say so explicitly along with an instruction not to switch on iterative calculation. Average-balance interest is the market convention and is more precise, but it makes the model circular. Beginning balances remove the circularity, which is what lets every check row in the file be trusted. Say on the tab that you know why the convention was chosen, and say which way the difference runs.

Test it on the PRIOR year end. The election depends on a leverage ratio that the current year's interest would otherwise help determine, which is a circular reference; testing on the previous year's closing balance sheet breaks it cleanly and is how these agreements are actually drafted. Build the ratio as its own row, the election as its own row, and let both coupons multiply through it — then look at what happens to coverage in the year the election flips.

No. When the toggle flips from accrual to cash pay, cash interest jumps in the year the leverage test was passed. The balance sheet got better and the coverage ratio got worse, and the reason is that deleveraging is what turned the coupon on. If your coverage rises smoothly in every year, check whether your election row is doing anything at all.

Take the forward year's Adjusted EBITDA — the one the market will be looking at when the book prices — apply the trading multiple, deduct net debt at par, divide by the pre-offering share count for a fully distributed value per share, apply the file-to-offer discount, and round DOWN to the increment. Print the effective discount as well as the stated one; they are not the same number once you have rounded, and the difference is a real cost to the seller.

No. That form assumes one exit date, and here the sponsor is paid on four. It will give you a number, and the number will be wrong. Lay out the solve: a seed rate, the present value of the inflows at that rate, the function to be zeroed, its derivative, and a corrected rate — one iteration per column, each reading the one before it. Three or four columns is plenty, and it shows the reader you know why the shortcut does not apply.

The mark is what the fund reports at the offering date: cash received plus the retained stake valued at the offer price, less what the management plan would take on the part still held. The return is what the fund eventually receives, over four dates. Both are correct. Everything between them is time, discounts, fees and the option ladder, and all of it was disclosed in the terms the sponsor signed. Being able to state that distinction is worth more than either figure on its own.

Work out what the entry price actually changes. It moves the equity purchase price, the sponsor's check and goodwill — and goodwill is not amortized, so it never reaches the income statement. Every operating line, the whole debt schedule, the offering price and every realization are identical. If none of the inflows moves with the price, the equity check that clears a required return is just their present value at that return, and the enterprise value differs from it by a constant you can compute once. Write the algebra and invert it.

Truncated. It is a maximum, and rounding a maximum up quotes a price at which your own model returns less than your mandate. Round toward the side that keeps the bound true. It is a small thing that a reader with judgment will notice immediately, and it costs nothing to get right.

Yes, and naming it with its direction is free marks. At this level of leverage the interest limitation on business interest deductions would bite in the early years and a real structure would carry disallowed interest forward — modeling it would lower the return. No deferred tax movement is modeled, because no write-up is recognized; accelerated tax depreciation would raise it. No over-allotment option is exercised. And the option ladder is held fixed at the entry price rather than re-struck, which flatters the ceiling slightly. Also absent by design: no purchase price allocation, no management rollover, no covenant package and no Excel data tables anywhere.

The three-hour take-home is the version firms use when they want to see judgment as well as mechanics, and the staged-exit variant comes up whenever a sponsor's realistic route out is a listing rather than a sale. It is hard to fake: a discount that accretes correctly, an election that flips on the right test, and a return solved across four dates either work or they do not, and the written page shows whether the candidate understood what they built.