Hurdle rate and cost of capital

Leo asks why Klasna discounts at 25% when the bank lends at 16%. Because the loan is only part of Klasna's money.

A quarter of Klasna's financing is debt (bank loans at 16% a year). Three quarters is equity: the founders' and investors' money. They took a much bigger risk and expect about 28% a year.

The cost of capital weights each source by its share. It is often called WACC, the weighted average cost of capital:

WACC = D ÷ (D + E) × r_D + E ÷ (D + E) × r_E

Klasna: 0.25 × 16% + 0.75 × 28% = 4% + 21% = 25%.

With a different mix, say 40% loans and 60% owners at the same rates: 0.4 × 16% + 0.6 × 28% = 6.4% + 16.8% = 23.2%.

A bet that earns less than the cost of capital doesn't pay the people whose money it uses. Interest usually lowers tax, which trims the debt part a little; we leave taxes out here. The slip is to use the cheapest source, the 16% loan, as if it paid for everything.

"So we borrow more and everything gets cheaper!" On paper, yes: with 40% debt instead of 25%, the weighted cost drops from 25% to 23.2%. But the formula assumed the owners keep asking for 28%, and they won't.

The bank is paid first, in a good month or a bad July. The more of the company's cash is promised to lenders, the more volatile what's left for the owners becomes. A riskier share means owners ask for a higher return, so part of the paper gain disappears.

There is a cash side too. A loan wants its repayment on schedule, including in August, when Klasna's cash is at its lowest. Equity has no such date.

How much debt is healthy is a separate question. The point for discounting is simpler: the cost of capital is what all of the money costs, not the cheapest piece of it. Using the loan rate for decisions treats owners' money as free.

The slip is to read 23.2% as a real saving without asking what happens to the owners' expected return.

A hurdle rate is the minimum return a bet must beat. It starts from the cost of capital and adds a premium for risk above the company's average.

What makes a bet riskier than average?

• Uncertain demand. A new feature only pays if customers use it more than they do today. Klasna's mock-exam mode relies on more maths families or families staying longer.
• New ground. A new market adds a new country, new sales and a new currency. Klasna for Teams in Poland has all three.
• Unknown costs. A build whose cost or timeline is guesswork.

What keeps a bet at average risk: known costs and known savings. Replacing a paid tool with a cheaper one at a known price, or paying a vendor in advance for a known discount, depends on almost nothing uncertain.

Klasna's ladder follows this: 25% for efficiency with known costs, 30% for product bets on current customers, 40% for new markets. The premium is about the risk of the bet, not about how confident the person presenting it sounds. The slip is to give a pet project the low rate because it "feels safe".

25% is the return for Klasna's average risk. Bets aren't average: cutting a known bill is safer than a new feature, and a new market is riskier still. So Klasna's policy sets hurdle rates, the minimum return a bet must beat, by kind of bet:

• 25% (the cost of capital): efficiency projects with known costs, such as replacing a paid tool;
• 30%: product bets on current customers, such as the maths mock-exam mode;
• 40%: new markets, such as Klasna for Teams in Poland.

The mock-exam mode had an NPV of +161,280 UAH at 25%. As a product bet it must clear 30%. Factors at 30%: 0.7692, 0.5917, 0.4552, 0.3501.

300,000 × 0.7692 + 600,000 × 0.5917 + 800,000 × 0.4552 + 800,000 × 0.3501 ≈ 1,230,034, minus 1,200,000: NPV +30,033.96 UAH.

Still positive, so it passes, but with a much thinner margin than at 25%. The slip is to keep judging every bet at the company-wide 25%.

Leo's next idea: Klasna for Teams in Poland. 2,000,000 UAH in, then 500,000, 900,000, 1,200,000 and 1,200,000. At 25% the NPV is +81,920. Approved?

Not at its own hurdle. A new market means a new country, new sales and a new currency: Klasna's policy asks 40%.

• NPV at 40%: −433,985.84 UAH;
• IRR: 26.97%, above the company average of 25%, far below the 40% a new market must beat.

So as it stands, Poland is rejected. The fix would be better cash flows (a cheaper launch, faster sales), not a friendlier rate.

Why not one rate for everything? Because with one rate, risky bets look better than they are and safe ones look worse. Year by year the company funds more risk without anyone deciding to. The hurdle depends on the bet, not on who pays for it.

The slip is to approve anything with an IRR above the cost of capital.

When a forecast looks too good, the tempting fix is to raise the discount rate "to be safe". It's the wrong tool.

The rate prices the risk of the kind of bet: how uncertain its outcome is compared with Klasna's average. The cash flows are the best honest estimate of what will happen. Mixing the two causes three problems:

• a padded rate punishes every year the same way, including the solid early years, while the doubtful year 3–4 numbers stay unexamined;
• nobody can see what was assumed: "we used 40%" hides "we don't believe 800,000";
• good bets fail quietly: at 40% the mock-exam mode drops to −179,800.08, though it clears its real 30% hurdle with +30,033.96.

So keep the rate tied to the bet's risk, and fix doubts where they live:

• ask where the number comes from (more maths families? families staying longer before the exams?);
• run a lower scenario, say 600,000 in years 3 and 4, at the same hurdle and see whether NPV stays positive.

The rate is for risk; the forecast is for the cash you expect.

Klasna's decisions use four rates, each with its own meaning. Mixing them up is the most common way to get a business case wrong.

• 12%: the deposit rate. What money earns almost without risk. It is the floor: no bet should earn less.
• 16%: the loan rate. What the bank charges. It is one ingredient of the cost of capital, not a hurdle on its own.
• 25%: the cost of capital. What lenders and owners expect, weighted: 0.25 × 16% + 0.75 × 28%. The hurdle for safe efficiency projects with known costs.
• 30%: the hurdle for product bets on current customers.
• 40%: the hurdle for new markets.

The extra points over 25% pay for the extra risk of the bet, not for doubts about the forecast.

And don't pad: a rate set too high is as wrong as one too low. At 40% for everything the mock-exam mode drops to −179,800.08 and a bet that clears its real 30% hurdle gets rejected. It just loses money more quietly. The slip is to pick whichever rate makes the bet you like pass.

Leo's last try: the bank lends at 16%, so why not judge Poland at 16%? Put the three answers side by side.

Teams in Poland, −2,000,000 then 500,000, 900,000, 1,200,000, 1,200,000:

• at 16% (the loan rate): +531,419.62 UAH;
• at 25% (the cost of capital): +81,920;
• at 40% (the hurdle for a new market): −433,985.84.

Same cash flows, a swing of almost a million hryvnias, from the best-looking bet on the list to a clear no. The longer and later a bet's cash, the more the rate matters: Poland's big years are years 3 and 4.

Why 16% is wrong twice:

• the loan pays for only a quarter of Klasna's money; owners, who put in the rest, expect about 28%, so all of the money costs 25%;
• a new country, new sales and a new currency carry more risk than Klasna's average, so Poland must beat 40%.

The slip is to pick the rate from the cheapest money available. The rate comes from what all the money costs and from the risk of the bet.