There are sixteen sheets in the Excel model template: LBO model, control, sensitivities, company, S&U, FCF, comp, DCF, beta, de-risking, football field, multiple progression, sell-side, DB credit, downside, and financials. Within each sheet, there are hundreds of cells, some of which are hidden. When I hit the small “+” sign next to the condensed hidden rows and columns, more numbers and data pour into the sheet.
Each cell can be identified by its row and column. Rows are sorted numerically, while columns are sorted alphabetically—thus, cell A 1 is the topmost left cell, cell B 1 is to its right, and cell A 2 is below it. There is data in cell QER 389 in the “multiple progression” tab. If the models I worked on in business school were at a high school JV level, and the model I “completed” in London training was Division III college level, what I’m looking at now is Olympic, like final-heat level—at least to me.
In the distance, I see the bright red EXIT sign next to the stairway. Part of me—the vast majority—wants to hightail it down the forty-four flights of stairs to freedom. Then I see Phuc, his BlackBerry in hand, looking like Ari Gold from Entourage.
As I contemplate my escape, Joel’s smiling face and behemoth body come into view. Passing by my cube on his way to the bathroom, he sticks out his right fist. I reciprocate, giving him a fist-bump. “My man,” he says, mid-stride, before disappearing into the bathroom.
With restored hope, I retrieve my headphones from my top drawer, queue up some country music, and methodically go through each sheet of the model, recalling Ted’s advice to focus primarily on the blue cells, the inputs, which will drive the outputs.
My first task is to separate the historical financials from the projected figures to avoid giving Leighton an aneurysm next time he looks at it. Then a new email pops up in my inbox from Arthur, the MD in lev fin, to the whole deal team: “Pls run 2 cases: 1) 6x—unfunded RCF, $1445 TLB, $360 bonds; 2) 6.5x—fund $150 of RCF (L+175); $1355 TLB, $450 bonds. LIBOR floor 1%, TLB OID 99.5, notes pricing 6.5%. 50% CF sweep. Tx.”
Before I can process the instructions, Arthur sends a follow up: “And try scenario 2 w/o RCF funded, 50 percent of TLB issued in Euro. Same floor, E+165. Tx.”
Slowly switching from sheet to sheet in the model, I stumble upon the sheet labeled “control,” which contains a series of headings containing the word “assumptions” along with columns of blue numbers. “Capital structure assumptions” reads cell D 43—this is the one. I celebrate the small victory by mouthing the words to the refrain of Kenny Chesney’s “Keg in the Closet,” which blares through my headphones.
Ten columns to the right are the interest rate assumptions and OID assumptions. Having no understanding of what OID actually means isn’t an issue. What matters is that the assumptions Arthur gives me are input into the model correctly.
After double-checking I input everything correctly, I scan up in the control sheet and toggle through the two cases. After setting the capital structure, operating assumptions, interest rate, OID, and cash-flow sweep all to “Case 1,” I hit the F12 key to refresh the spreadsheet. Unfortunately, F12 brings up the “Save As” box. I hit ESC and then hit F9, the key that refreshes the spreadsheet.
A few seconds pass. My screen dims slightly and my cursor disappears, a surefire sign Excel is freezing. “What in the shit?” I say, removing my headphones.
“Freeze?” I hear Ted say from his cube.
“All I did was toggle through the first case,” I plead. “I wasn’t—”
“Give it a second—should be fine. Go to options, formulas, then make sure ‘automatic except for data tables’ is selected.”
Seconds later, the screen brightens. I follow Ted’s instructions, deselecting “automatic” and selecting “automatic except for data tables.” I then toggle to case two, and the spreadsheet quickly updates without a hitch.
“That work?” he asks.
“How’d you know that was the issue?”
“When you toggle through cases and refresh the spreadsheet, everything gets refreshed. In models with a bunch of data tables, it takes forever to process, so best to just update data tables manually.”
Over the next three hours, I go through each of the sixteen sheets, carefully updating relevant cells in each sheet and tweaking formatting so output pages are consistent with the financials and data provided by the company. After updating the data for the comparable companies and precedent transactions, I check the football-field sheet.17 It looks reasonable, visually, and that’s what’s paramount. All ranges are relatively in line, with the management DCF showing the highest range and the precedent transactions next highest. The LBO, public comparables, and DB credit-case valuations are clustered around the same range, while the downside DCF shows the lowest valuation range.
John Bukowski: Any idea what LCAF is?
William Keenan: no, why?
John Bukowski: apparently I’m supposed to lead an LCAF call tmw
William Keenan: send me dial-in. will have popcorn ready
John Bukowski: you still lost in that model you’re working on
William Keenan: pretty much
John Bukowski: k good.
John Bukowski: also, pls take me off as “Jack DB” on your phone contacts
William Keenan: only if you offer to captain the banking kickball team
At 11:17 p.m., Arthur sends an email: “Where are we on model? How do numbers look???” It’s then that it occurs to me I’m not even certain what the ultimate goal is for this exercise. I spend the next hour compiling a set of pages, trying to summarize the data including valuation metrics and outputs, a comparison of the various assumptions used for each case, and a summary of the sources and uses and the pro-forma capitalization table.
I open up a new email response to Arthur just as the humming of the floor’s air conditioner shuts off. Attached to the email is a comprehensive six-page deck with every possible number I can think could be important. Within seconds, Arthur sends a response: “Don’t need all this. Just send paydown for both scenarios.”
My Ctrl-F search for “paydown” hits on the eighth sheet, the free cash-flow projections output page. After toggling through scenarios one and two, I copy the paydown figures (both 47 percent) and paste them into a new email to Arthur and the team. His email response comes within sixty seconds: “Must be bust in model. Not good enough. Do better.”
The bullpen smells like a mall food court when I walk in around 12:30 a.m. The first cube I pass emanates the distinct scent of low tide. On the desk, a brochure for Pad Thai pokes out of a greasy, ripped-open brown paper bag. The analyst who occupies the cube wears large headphones and periodically twists his plastic fork into the container of food, shoveling brown noodles into his mouth without taking his eyes off the monitor, which is teeming with colorful column-charts. On the counter in the middle of the bullpen is a pizza box with two slices of pepperoni pizza and an empty tray of bite-size Baked by Melissa cupcakes that threw the group into a frenzy when they were delivered earlier in the day.
“So how is the deferred tax-asset created?” asks Leighton in Joel’s direction.
“The difference in accounting methods,” says Joel before he spots me, and spins his chair in my direction. A paper napkin with blotches of vodka sauce splattered all over it is tucked into the collar of his shirt. “Wild Bill, I got an extra meatball if you want it. They’re delicious.” Joel picks up the plastic container and offers it up to me.
“Difference in accounting methods? Be more specific, Joel,” says Leighton. “Joel, are you listening to me?”
“No,” says Joel. “Seriously, Bill. These are like homemade meatballs, and this sauce is incredible, so creamy.”
“Then why did you ask me?” says Leighton.
“I didn’t. You overhead me talking on the phone to someone about it and then butted in,” says Joel.
Leighton shakes his head then turns back around in his chair. He’s the only person I’ve seen on the entire floor who only has one computer monitor, though it’s the size of two combined. I’m afraid to ask the rationale, but I don’t even need to. “People waste too much time twisting their head from monitor to monitor,” says Leighton as I approach his cube.
“Makes sense,” I say.
“Already eat?” I ask.
“I consume my daily caloric intake by 6:30 p.m. Only liquids thereafter.” With that, he pops open a tangerine-flavored LaCroix sparkling water, one of six on his desk. “Used to drink Poland Spring, but this has less carbonation and limits number of bathroom breaks.” Leighton then removes the black compression-sleeve from his right arm.
“What happened to your arm? From lifting?” I ask.
“He developed golfer’s elbow from modeling too hard,” says Joel from his cube.
“Something you know nothing about, since you can’t even tell me the basic components of a balance sheet,” says Leighton.
“You should get Tommy Modeling surgery,” adds Joel.
“Know you’re cranking on your sell-side, so not sure if you’ve been able to keep up with the email traffic,” I say. “Didn’t find any busts in the model.” Leighton scans through the most recent emails from Arthur, quietly mumbling to himself. His souped-up keyboard emits a distinct clicking sound with each tapped key.
When he arrives at my most recent email with the paydown figures, he shakes his head, “Yeah, that doesn’t work.”
“I’m not quite sure what Arthur means when he writes, ‘Do better,’” I say.
“What we care about is the percent of total debt paydown after year seven,” says Leighton. “This essentially determines whether it’s feasible for DB to underwrite the deal and market it to investors. We need DB credit case to be above fifty-three percent after year seven—that’s the key number. Thus, ‘do better’ means get to fifty-three percent.”
“Any idea how?” I ask.
Leighton looks at me as he unhooks the earbuds behind his ears and places them in his ears. “Any way you can.”
At my desk, I recheck every formula in the free cash-flow output sheet. Then I go back through the financials, both historical and projections, to ensure all the numbers are correct. Everything in the control tab, including debt assumptions, are exactly as Arthur instructed. Then Ted’s advice from earlier that evening hits me—work backward from the issue. He mentioned an Excel shortcut that quickly traces all precedents. I scan the list of shortcuts pinned in my cube by the last occupant—“Trace precedents: control, shift, left bracket,” reads one of the items near the bottom of the sheet. The paydown percent figure in year seven—Cell R 89—is my issue. I trace the precedent cells once, twice, three times, until finally I find the blue hardcoded figures that are the root numbers driving the cell’s output. Ultimately, the LTM EBITDA figure and the interest expense on the debt, which, combined, largely determine the free cash flow produced each year, can be used to pay down outstanding debt.
While Arthur’s email states the rate on the bonds, there is one variable that can be flexed—LIBOR. Interest rates on term loans are typically quoted at LIBOR plus an additional rate (L+175 bps). Hundreds of cells down in the control sheet of the model are the LIBOR rates. In the top right corner of the first LIBOR cell is a small red triangle, indicating a note. I hit Shift F2 and read the note that explains the LIBOR curve was last updated over six months ago—Aha! I refresh the FactSet code of the cell, which in turn updates the entire LIBOR curve. Except this doesn’t solve my problem. It worsens it. With interest rates rising over the past couple months, the LIBOR figure jumps twenty bps. This consequently lowers the paydown in both scenarios. I check the time in the bottom right corner of the monitor: 1:10 a.m.
“So answer this, Joel,” says Leighton as I pass by the bullpen to get water. “Is an NOL a tax credit or a tax deduction? And what are the key differences?”
“What is, ‘Who cares?’” I hear Joel say.
By 2:00 a.m., aside from the clattering of keyboards in the bullpen and some ambient light flooding out of its opening, the rest of the floor is dark and quiet.
In my cube, I stare at the LIBOR curve, for which there is a given rate for each duration, including overnight, one week, one month, two months, three months, six months, and one year. The LIBOR figure driving the rate in this model is the six-month rate. As the duration decreases, so too does the rate, which makes sense, since long-duration debt instruments typically command higher rates than shorter-dated ones due to greater interest-rate risk. What if the driver is the three-month LIBOR? I try it out—the paydown increases to 50 percent in both cases. The left side of my mouth curls into a half-grin. How about one-month LIBOR? Fifty-seven percent in both cases. The jump seems too big to justify. I settle on the three-month LIBOR18 then turn my attention to the LTM EBITDA figure.
The note (which I wrote) in the cell references the LTM EBITDA figure. It comes from the company’s latest confidential information memorandum. There’s no changing that figure, which is presented by the company. But in the company’s folder, I notice a recent management presentation I had only glossed through earlier. In the appendix on that presentation, there’s an EBITDA reconciliation with a series of adjustments, which are the same as the ones found in the CIM, excluding one. The omission of the one adjustment means the LTM EBITDA in the MP is $6 million higher than the figure shown in the CIM.
Replacing the LTM EBITDA from the CIM with the figure from the MP19 gets the paydown to 53 percent in both cases. At 2:48 a.m., I email Arthur and the team the updated paydown figures and then head home.
I arrive back at the office just over four hours later to dial into the previously scheduled 7:00 a.m. call, but it gets cancelled at 6:57 a.m. The corner offices are nearly all occupied at this time as MDs work the phones with their office doors closed, generating more work for those of us sitting vulnerable in the cubes. The floor itself is quiet, since junior bankers typically trickle in from 9:30 a.m. to 10:00 a.m.
Despite the call being cancelled, I sit at my desk from 7:00 a.m. till noon and field a maelstrom of emails from Arthur, instructing me to fund, defund, then refund the credit facility, increase the notes, decrease the term loan, and add a mezzanine piece. It’s virtually a one-on-one exchange between Arthur and me, with the rest of the team cc’d on the emails. My response time to update the model and resend the outputs decreases each time a new request comes from Arthur.
“Look at you, balls-deep in that model,” Ted says as he walks behind me at one point. I barely hear him as I send Arthur the most updated paydown figures.