Investment Banking puzzles, solved step by step

Rs 400 crore, and the coverage test binds. Leverage of 5.0x allows Rs 500 crore. Coverage of 2.5x caps interest at 100 divided by 2.5, which is Rs 40 crore, and at 10% that interest supports Rs 400 crore of debt. The borrower must pass both tests, so the lower number is the ceiling. Below an 8% rate, leverage would bind instead.

Why are there two answers, and why take the smaller?

A bank deciding your home loan asks two things: is the loan too large against your salary, and can you afford the monthly payment? You can pass one and fail the other. Lenders to a company ask the same pair. Leverage limits the size of the debt; coverage limits the cost of carrying it, and the borrower must clear both, so the binding test is whichever gives less. Compute both every time, then name the one that bites.

Leverage is one step: 5.0 times EBITDA of 100 is Rs 500 crore. Coverage works backwards from the payment. Interest coverEBITDA divided by interest expense: how many times over the operating profit pays the interest bill. of at least 2.5 means interest can be at most 100 divided by 2.5, which is Rs 40 crore. At a 10% rate, Rs 40 crore of interest corresponds to Rs 400 crore of debt.

When would the answer flip to the leverage test?

The coverage ceiling moves with the rate; the leverage ceiling does not. Set the two equal: interest of 40 on 500 of debt is an 8% rate. Below 8%, leverage binds; above it, coverage binds, which is why debt capacity shrinks when rates rise even if earnings do not move. At 7%, coverage would allow about Rs 571 crore, so the 500 leverage cap becomes the limit.

Say the limitation too. Real covenants often test cover on EBITDA less capex, or on a fixed charge basis that includes leases, and lenders size a deal with headroom below both ceilings rather than at them. The two-test logic stays the same.

About 71% for 30 people, and the odds pass 50% at 23. Work the chance that nobody matches: the second person avoids the first with 364/365, the third avoids both with 363/365, and so on. Multiply 30 of these and you get about 29.4%, so a shared birthday has a 70.6% chance. Repeating the product shows it first passes half at 23 people.

Why does the answer feel far too high?

Your instinct pictures yourself in the room: what are the odds someone shares my birthday? That is about 30 in 365, under 10%. But the question allows any two people to match. In a group of 30, every person can be compared with every other, and there are 30 x 29 / 2 = 435 such pairs. The number of pairs, not the number of people, drives the answer, and pairs grow roughly with the square of the group. Four hundred chances to hit a one-in-365 coincidence is a lot of chances.

How do you compute it without a list of 435 cases?

Count the opposite. Matching can happen in many overlapping ways, but nobody matching happens in exactly one: each new person must miss every birthday already taken. When at least one is messy, compute none and subtract from one.

For a quick estimate in your head, use pairs: the chance of no match is close to e raised to minus pairs over 365. With 253 pairs that is e to the minus 0.69, almost exactly a half, which is why 23 is the crossing point. The method assumes birthdays are spread evenly across the year; real birthdays bunch a little, which only nudges the odds higher. By 60 people the chance is 99.4%.

5,898 dollars. The fee is 1.5% of Rs 5,00,000, which is Rs 7,500, leaving Rs 4,92,500 to convert. Divided by 83.50 that is about 5,898.2. A quick route: 83.5 x 6,000 is 5,01,000, so Rs 5 lakh buys about 5,988 dollars before the fee, and taking 1.5% off 5,988 gives the same 5,898.

What is the question really testing?

These questions sit in the timed numerical section of an online assessment, and the arithmetic is easy. What they test is whether you apply each adjustment to the right base under time pressure. Think of a restaurant bill with a 10% service charge: it is 10% of the food, not of the food plus tax, and getting the base wrong gives a number that looks almost right. The test designers put the almost-right numbers in the options, so the base you apply the fee to decides whether you pick the correct one.

Does it matter whether you take the fee before or after converting?

Not for a percentage fee. Taking 1.5% off the rupees and then converting, or converting and then taking 1.5% off the dollars, multiplies by the same two numbers in a different order: 5,988 x 0.985 is also 5,898. Order matters only for a flat fee or a fee charged on a different base; a percentage fee on the same amount commutes with the conversion. Knowing this lets you pick whichever order makes the mental maths easier.

Where does 5,900 come from? Dividing by 1.015 answers a different question: how much could you convert if the 1.5% were charged on top of the converted amount. That base is smaller than Rs 5 lakh, so the fee comes out as about Rs 7,389 instead of Rs 7,500. On a test with options two dollars apart, that is a wrong answer.

About Rs 3.1 crore a year, on these assumptions. Take 5 shows a day and 200 seats. Weekday shows run about 25% full over 261 days and weekend shows about 55% over 104 days, giving roughly 122,450 tickets. At an average of Rs 250 a ticket that is about Rs 3.06 crore. Occupancy, not seat count, is the input that swings the answer.

How do you structure it so the interviewer can follow?

Think of a bus route: revenue is trips a day times seats times how full the bus is times the fare. A screen is the same machine with a different product. Revenue is capacity times utilisation times price, and the estimate is only as good as the utilisation guess. Lay the chain out first, then fill each slot with a number and a one-line reason.

Capacity: about 5 shows a day, since a film of two and a half hours plus cleaning takes roughly three hours a slot from late morning to past midnight, and about 200 seats in a mid-sized auditorium. Utilisation: a Tuesday afternoon show might be a fifth full while a Saturday night show is close to sold out, so split the year. With 104 weekend days at 55% and 261 weekdays at 25%, the blended occupancy is 33.5%. Price: Rs 250 as an average across morning discounts and premium evening seats.

Why is occupancy the swing input and not seats?

Each input moves revenue in proportion, so what matters is how wrong you could plausibly be. Seats are visible on a booking app's seat map and you will not be more than 10% off. Occupancy could reasonably be anywhere from 20% to 45% blended, and every 10 points is about Rs 0.91 crore, close to 30% of the answer. When you present an estimate, name the input you are least sure of and show what a fair range for it does to the total.

State the boundary of what you sized. This is gross ticket revenue before taxes, and the operator shares part of it with the film's distributor. Food, drinks and on-screen advertising sit outside it and can add a large share on top, so say which number you were asked for.

75 km. The trains close the 100 km gap at a combined 100 km/h, so they meet after exactly one hour. The bird flies without stopping for that hour at 75 km/h, so it covers 75 km. You never need to track the individual legs, though they do add up: 60 km, then 12, then 2.4, each a fifth of the one before, summing to 75.

What question should you ask first?

If a friend walks a dog in the park for an hour and the dog runs at 10 km/h the whole time, the dog covers 10 km however it zig-zags between trees. Nobody measures each dash. Distance is speed times time, so find how long the bird flies and the shape of its path stops mattering. The trains settle the time: they approach each other at 50 plus 50, which is 100 km/h, and 100 km apart means one hour.

If the interviewer insists on the legs, how do you sum them?

On the first leg the bird and the oncoming train close at 75 plus 50, which is 125 km/h, so they meet after 100 / 125 = 0.8 hours; the bird has flown 60 km. In that time the trains have closed 80 km, leaving 20 km between them. The next leg is the same problem a fifth the size: 12 km, then 2.4, and so on. Each leg is a fixed fraction of the last, which makes the total a geometric series that sums to 60 / (1 - 1/5) = 75 km.

Why ask this in a banking interview? It is a test of whether you look for the simpler quantity before diving into the mechanics, which is what a good analyst does with a messy model: find the number that settles the question and compute that. Answer the short way, then offer the series as a check if there is time.

At Rs 120, count it as 5 crore extra shares and not as debt; at Rs 80, count it as Rs 500 crore of debt and no new shares. Rs 500 crore at a Rs 100 conversion price is 5 crore shares. At Rs 120 those shares are worth Rs 600 crore, so the holder will convert. At Rs 80 they are worth Rs 400 crore, so the holder takes the Rs 500 crore back. Treat the bond as whichever it is most likely to become, never as both.

Why does the share price decide whether a bond is debt?

Suppose a friend lends you Rs 5 lakh and you agree they can take either the money back or a share of your shop. If the shop is booming, they will take the share; if it is struggling, they will take the cash. You plan around what they will choose. A convertible gives the holder a choice between repayment and shares, and the valuation should assume the choice a rational holder makes at today's price. The comparison is simple: conversion value, shares received times share price, against face value.

At Rs 120, converting gives 5 crore shares worth Rs 600 crore, more than the Rs 500 crore face value. The holder converts, so add 5 crore shares to the diluted count and remove the bond from debt. With, say, 50 crore existing shares, diluted equity value becomes 55 crore x 120 = Rs 6,600 crore. At Rs 80, converting gives shares worth only Rs 400 crore, so the bond stays as Rs 500 crore of debt and the share count stays at 50 crore.

What goes wrong if you count it the other way?

Counting it as debt at Rs 120 puts Rs 500 crore into the bridge when the holder will in fact take Rs 600 crore of value in shares, understating the claim by Rs 100 crore. The serious error is counting it twice, as debt and as diluted shares, which overstates enterprise value by the full Rs 500 crore. That happens when the share count comes from one source that already includes the conversion and the debt figure from a balance sheet that still shows the bond.

Say the limitation. Near the conversion price the choice is close and the bond carries option value either way, so some banks value it at fair value rather than flipping between debt and shares. Settlement terms also matter: some convertibles repay the face value in cash and deliver only the excess in shares, which needs a different share count.

About 84.58 rupees per dollar. Rs 83 invested at 7% grows to 88.81. Converted to 1 dollar and invested at 5% it grows to 1.05 dollars. Both routes are riskless once the forward is agreed, so they must end equal: 1.05 x F = 88.81, giving F of about 84.58. If the quoted forward sat far above that, you could borrow rupees, invest in dollars and sell them forward to lock in a gain.

Why is the forward set by interest rates and not by a forecast?

Imagine two banks next door, one paying 7% on rupee deposits and one paying 5% on dollar deposits, and a counter that promises today to swap your dollars back into rupees in a year at a fixed rate. If that promised rate were 83, everyone would ignore the dollar bank. The forward rate must exactly cancel the interest difference, otherwise one route earns more than the other with no risk, and the gap would be traded away. That rule is called covered interest parity, covered because the forward removes the currency risk.

What would you actually do if the forward were quoted at 86?

At 86, dollars are expensive in the forward market relative to parity, so sell dollars forward and build the dollars cheaply today. Borrow Rs 83 at 7%, owing 88.81 in a year. Convert to 1 dollar at spot and deposit it at 5%, which grows to 1.05 dollars. Sell those 1.05 dollars forward at 86 for Rs 90.30. After repaying the rupee loan you keep about Rs 1.49 per dollar, without taking any view on where the rupee goes. Traders doing exactly this pushes the quote back toward 84.58.

The forward rupee is about 1.9% weaker than spot, roughly the 2 point rate difference. A quick approximation is spot times one plus the difference, 83 x 1.02 = 84.66, close enough to sanity-check. The limitation: in practice, capital controls, transaction costs and limits on who can borrow in each currency let small gaps persist, so the arbitrage is cleaner on paper than on a dealing desk.

Yearly gives Rs 2.00, monthly about Rs 2.61, daily about Rs 2.71, and continuous compounding gives e, about Rs 2.718. Splitting 100% into n periods gives (1 + 1/n) to the power n. Each increase in frequency adds less than the last, and as n grows without limit the result levels off at the constant e rather than growing forever.

Why does compounding more often help at all?

Think of a savings jar where interest is added at the end of each month instead of once a year. From February onward, the interest from January is itself earning interest. More frequent compounding lets earlier interest start earning sooner, so the same headline rate produces more by year end. Monthly at 100% means adding 1/12 twelve times, (1 + 1/12) to the twelfth power, which is about 2.61 rather than 2.00.

Why does it stop at e instead of growing forever?

Because each extra slice adds interest on smaller and smaller amounts of fresh interest. Going from yearly to monthly adds 0.61; from monthly to daily adds only about 0.10; from daily to continuous about 0.0037. The gains shrink fast enough that their total converges, and the limit is the number e, about 2.71828. That is where e comes from: it is what 100% growth becomes when it compounds continuously.

Why it matters on a desk: the same quoted rate means different money depending on the compounding convention. A loan quoted at 12% compounded monthly costs about 12.68% a year in effective terms, so when comparing a monthly-pay loan with an annual-pay bond, convert both to an effective annual rate first. Continuous compounding is also why option and rates models use e: it is the clean limit of compounding very often.

Capitalising makes EBITDA 50 higher and net income 30 higher, but free cash flow is identical. Expensed, the 50 hits EBITDA and net income falls 37.5 after tax. Capitalised, only 10 of amortisation hits profit, so net income falls 7.5, and the 50 moves into investing cash flow as capex. Operating cash flow is 50 higher, capex 50 higher, and free cash flow is minus 37.5 under both.

If the same cash goes out, what actually changes?

Buying a laptop for work: you can think of it as this month's expense, or as a tool that lasts five years and costs a fifth each year. Your bank balance falls by the full price either way. Capitalising versus expensing changes which profit lines carry the cost and which cash flow section shows the outflow; it does not change the cash that left. That is why analysts compare companies on free cash flow when accounting choices differ.

Expensed: EBITDA falls 50, tax falls 12.5, net income falls 37.5. Capitalised: EBITDA is untouched, the year's amortisation is 50 over 5, which is 10, and net income falls 7.5. So capitalising shows EBITDA 50 higher and net income 30 higher in year one. In operating cash flow the capitalised company shows +12.5 against -37.5, a 50 swing, but it also shows 50 of capex in investing.

What assumption is doing the work on tax?

The question says the spend is deductible in year one either way, so the cash tax saving is 12.5 under both. In the capitalised books, tax expense follows the smaller 10 charge, and the 10.0 difference becomes a deferred tax liabilityTax that the accounts recognise as owed later because taxable profit was lower than book profit this year. that is added back in operating cash flow. If the tax authority instead required capitalisation too, the cash tax saving would fall to 2.5 and free cash flow under capitalisation would be -47.5, 10 lower: tax timing is the only route by which the choice reaches cash. State which rule you are assuming.

Why bankers care: two otherwise identical software companies, one capitalising development and one expensing it, will show different EBITDA margins. Valuing both on EV/EBITDA without adjusting flatters the one that capitalises. Look at capitalised development in the cash flow statement and either add it back to the expensing peer or deduct it from the capitalising one.

About 10%. The Gordon growth model says price equals next year's dividend over (cost of equity minus growth). Rearranged, cost of equity equals dividend yield plus growth. The dividend yield is payout over P/E, 0.6 / 20, which is 3%. Adding 7% growth gives 10%. If the 20x were on trailing earnings, the answer would be slightly higher, about 10.2%.

How does a price turn into a return?

If a flat costs Rs 1 crore and rents for Rs 3 lakh a year, rising 7% a year, a buyer is earning 3% in rent now and 7% a year from the rent growing, about 10% in total. A buyer of a growing income stream earns its current yield plus its growth, so a price plus a growth assumption tells you the return the buyer must be expecting. For a share, the income is the dividend.

Why is the dividend yield payout over P/E?

Dividend over price is (payout x earnings) over (P/E x earnings), and earnings cancel, leaving 0.6 / 20 = 3%. You never need the share price or the earnings per share, only the two ratios. That is the trick the interviewer is looking for: spotting that the units cancel.

How much should you trust the 10%?

Only as far as the growth input. The dividend yield is observable; the 7% perpetual growth is an opinion. Drop it to 6% and the same price implies 9%. Check consistency too: growth of 7% with 40% of earnings retained implies a return on retained equity of about 17.5%, since growth equals retention times return on equity. If the company cannot earn that, the 7% is too high.