Case 080Fixed income and creditCore
Lending case: Sonvarsha Castings wants a Rs 120 crore loan. Probability of default is 2.5% a year, loss given default 45%, EBITDA Rs 60 crore, total debt Rs 180 crore and interest cover 2.4. What is the expected loss, what spread covers it plus a 1.5% cost of capital, and would you lend?
1The situation
Sonvarsha Castings, a mid-market maker of engine and pump castings, asks for a Rs 120 crore five-year term loan to add a foundry line. After the loan its total debt will be Rs 180 crore against EBITDA of Rs 60 crore, and EBITDA covers interest 2.4 times.
The bank's rating model gives Sonvarsha a 2.5% annual probability of default and, for an unsecured loan, a loss given default of 45%. The bank wants every loan to earn 1.5% a year over its funding cost for the capital it must hold against unexpected losses. Sonvarsha's treasurer has asked for a spread of 2.25% over the bank's funding cost and offers a first charge on the new plant if that helps.
2Your task
Compute the expected loss and the spread the bank needs, test whether the credit metrics make the default estimate believable, and give a lending view.
Quick check
What spread over funding does the bank need, before any profit on top?
Worked solution
Try it on paper, then open one step at a time.
30-second answerThe answer to give first
Expected loss is 1.125% a year, Rs 1.35 crore, so the bank needs at least 2.625% over funding; the 2.25% asked works only with security. A first charge that cuts loss given default to 30% brings the required spread to exactly 2.25%. Leverage of 3.0x and cover of 2.4x are sound today, but a 25% EBITDA fall takes them to 4.0x and 1.8x. Lend secured, with covenants.
Step 1What does the bank expect to lose each year?
Think of lending Rs 1,000 to each of 40 friends, knowing from experience that one of them will fail to repay each year and that you will get about half back from that one. Expected loss is the chance of default times the share lost when it happens, times the amount lent. For Sonvarsha that is 2.5% x 45% x Rs 120 crore, Rs 1.35 crore a year, or 1.125% of the loan. A single default would cost Rs 54 crore; the expected figure is what a portfolio of such loans loses on average, not what this one loan will do.
| PD | probability of default in a year, 2.5% |
| LGD | share of the loan lost if default happens, 45% unsecured |
| EAD | exposure at default, Rs 120 crore |
| s | spread over the bank's funding cost |
Step 2How do you set the spread, and can the ask work?
The spread has two jobs. First it must replace the expected loss; second it must pay a return on the economic capitalMoney a lender sets aside to absorb losses worse than the average, sized from how bad a bad year could be. held against losses worse than average. That is 1.125% plus 1.5%, so 2.625% over funding, before any operating costs. The treasurer's 2.25% falls short by 37.5 basis points. But the shortfall can be closed on the loss side: a first charge on the new plant raises recoveries, and if it cuts loss given default to 30%, expected loss falls to 0.75% and the required spread to exactly 2.25%.
Step 3Do the credit metrics make a 2.5% default rate believable?
The price is only as good as the default estimate, so test it against the business. Debt of Rs 180 crore on EBITDA of Rs 60 crore is 3.0x leverage, and interest of about Rs 25 crore is covered 2.4 times: a leveraged borrower, not a distressed one. Now stress it. Castings demand follows the auto and pump cycles; if EBITDA falls 25% to Rs 45 crore, leverage rises to 4.0x and cover falls to 1.8x. EBITDA would have to fall 58% before it failed to cover interest at all. That profile fits a moderate default probability, so 2.5% is plausible, but the loan's safety rests on EBITDA holding up through one downturn.
Step 4So what is the lending view?
Lend, on terms. Accept the 2.25% spread only with a first charge on the plant, and set covenants at a leverage ceiling of 3.5x and an interest cover floor of 2.0x, so the bank can act while Sonvarsha is still solvent. An amortising schedule reduces exposure each year as the new line starts to earn. Without security, hold at 2.625%. The limit to say plainly: the 45% and 30% loss figures come from the bank's model and the value of a foundry line in a forced sale, both of which are uncertain, so a security package is worth checking with a valuation rather than taking on trust.
Where candidates lose it
The usual miss is pricing the loan at the expected loss alone, 1.125%, and calling it covered. That leaves nothing for the capital held against a bad year, and a bank that prices this way loses money whenever losses run above average.
The second is computing the spread and never testing the default probability against leverage and cover. The interviewer wants the number and the judgement on whether the number can be trusted.
What the interviewer asks next
- What spread would you need if the default probability were 4% but the loan were secured?
- How would a covenant breach at 3.5x change the bank's expected loss?
- Why might two banks quote different spreads for the same loan?
- How would you price a five-year loan if the default probability rises each year?
Asked at Deutsche Bank, Sales and Trading, New York, 2024 (Wall Street Oasis): The case study was relatively simple, focused on lending to a middle-market corporate
Company names and figures are illustrative.