What ten years of reading credit proposals taught me about financial red flags

My grandmother — God rest her soul, a woman who raised eight children on a small piece of land in Nyeri and never once took a loan from anyone — had a saying she deployed whenever a stranger came to the compound with a proposal that seemed, in her words, "too smooth for the road it came on." She would squint at the man, tilt her head slightly to the left the way a hen eyes a suspicious piece of grain, and say: "The goat that arrives fat in the dry season has been eating from someone else's shamba."

She was, in the language of modern credit analysis, describing a red flag.

I spent close to a decade reading credit proposals at Standard Chartered Bank and I&M Bank — corporate facilities, SME loans, personal lending, mortgages, auto loans — across six African markets, and I want to tell you that the instinct my grandmother was describing is not mystical. It is pattern recognition. The proposals that went bad had tells. Not always, not every time, not so obvious that you could point to one line and say "here, this is where the goat came from." But fairly consistent tells, the kind that accumulate across hundreds of files until you develop what I can only describe as a smell for the thing. You open the folder, you start reading, and something — some combination of ratios and dates and narrative that does not quite cohere — makes the hairs on the back of your neck do what my grandmother's hen did when it spotted that suspicious grain.

Which brings me to what years of work actually taught me. Not the textbook version — the textbooks will tell you about debt-service coverage ratios (DSCR) and gearing and current ratios and all the metrics that are perfectly correct and also perfectly gameable by anyone who has spent twenty minutes with a cooperative accountant and a copy of Excel. I am talking about the things underneath the numbers. The tells the numbers are trying to hide.

Let me begin from the beginning, which is to say: the narrative section.

Every credit proposal has one. It is supposed to tell you who the borrower is, what the business does, why they need the money, and how they intend to repay it. In Wanjiku-speak: the story. And the first thing i learnt is that a story that is too clean is not a story. It is a performance. The borrower who can account for every shilling of revenue growth, who can explain every dip and recovery with a tidy causal chain — "we experienced a temporary downturn in Q2 2019 due to the election cycle, recovered strongly in Q3, and have maintained 18 per cent annual growth since" — that borrower has rehearsed. Real businesses are messier. Real businesses have a quarter where the owner's brother-in-law ran off with the petty cash, or a supplier who delivered substandard goods and refused to refund, or a customer who owed Sh4 million and went quiet for eight months. Real businesses have scar tissue. A proposal without scar tissue is a proposal that has been buffed to a shine, and I want to know what is underneath the polish.

The second tell is projected revenue that grows in a straight line.

I cannot tell you how many proposals I have seen where the financial projections look like someone held a ruler against the last two years of actuals and simply extended the line. Revenue grows at 20 per cent per year for five years. Costs grow at 12 per cent. EBITDA expands neatly. The Sh50 million facility is comfortably repaid by year three. Everything is proportional, everything is reasonable, and nothing — nothing — in the projection accounts for the possibility that Kenya will have an election, a drought, a fuel price shock, a global pandemic, or a Managing Director who decides in year two that he would prefer to run a boutique hotel in Diani.

In Kenyan-speak: the projection assumes the road is tarmacked all the way to the destination, with no potholes, no diversions, no bodaboda cutting across the lane, and perfect weather throughout. This is not a plan. This is a wish dressed in a spreadsheet. And the tell is not that the projection is optimistic — most projections are optimistic, that is the nature of the exercise — but that it is optimistic in a perfectly uniform, mathematically-suspiciously-tidy-and-round-number kind of way that suggests nobody has seriously stress-tested what happens if revenue in year two comes in at 60 per cent of forecast.

The third tell — and this one I learned the hard way, meaning I approved a facility once before I had fully internalised it — is the concentration problem.

A business that derives 70 per cent or more of its revenue from one customer is not a business. It is a department of that customer, with the added disadvantage that it bears its own overheads. The customer sneezes, the borrower catches pneumonia. The customer decides to bring the function in-house, or switches to a cheaper supplier in China, or simply pays late for three months running — and suddenly the Sh30 million working capital facility that looked adequately covered is being serviced from reserves that were never meant for that purpose. I have seen this play out. It does not end with the borrower calling the loan officer to say "I have a problem." It ends with the loan officer receiving a memo from the recoveries team.

Related to concentration, and equally dangerous, is the connected-party transaction that nobody wants to talk about.

Good people, I am going to say something that will make a few company directors adjust their cufflinks uncomfortably. A significant portion of the credit losses I observed in my banking career were connected, in ways that were sometimes direct and sometimes required a fairly determined squint at the corporate structure, to transactions between the borrowing entity and companies owned by the same directors or their families. The borrower buys raw materials from a company owned by the chairman's wife at prices that are — how shall I put it — "market-related" in the loosest possible interpretation of the phrase. The borrower leases its premises from a company in which the CEO holds a silent stake, at a monthly rental of Sh850,000 for space that the market would price at Sh400,000. The profits that should be retained in the business to service the loan are, by this mechanism, quietly siphoned into adjacent pockets before the auditors arrive for their annual tea and biscuits.

The tell is not always the connected-party transaction itself, which a borrower is technically required to disclose. The tell is when the disclosure note is buried on page 38 of the audited accounts in type that required me — I am not joking — to borrow a colleague's reading glasses, and the explanation is a single sentence reading "the transaction was concluded on arm's-length terms." Arm's-length. I once traced an arm's-length transaction between a Sh200 million borrower and a supplier owned by the borrower's brother that generated gross margins for the supplier of 340 per cent. At arm's length, presumably, the arm in question belonged to a basketball player.

Let me pause here and say something about audited accounts generally, because experience taught me to read them rather differently from how they are typically presented. The audit opinion at the front — the one that says the accounts present a "true and fair view" — is an opinion about whether the accounts comply with the applicable accounting standards and are free from material misstatement. It is not a certificate of business health. It is not a warranty that the numbers will hold. An auditor's job is to confirm that what is on the page is consistent with the underlying records and with the relevant accounting framework. A company can have audited accounts that present a perfectly true and fair view of a business that is lurching toward insolvency. The accounts do not lie. The story they tell may simply not be the story you think you are reading.

This is the part of credit analysis that no formula captures and no checklist fully replicates. It is the part where you put down the proposal — the clean narrative, the straight-line projections, the arm's-length disclosures on page 38 — and you ask yourself: does this business, as described, make biological sense? Not financial sense. Biological sense. Can it breathe? Does it have enough oxygen — working capital, customer diversification, management depth, market position — to survive a bad season? Or is it a plant that looks healthy in the nursery because it has been watered twice a day and kept out of the wind, and will keel over the moment you put it in actual soil?

The proposals that worried me most were the ones where the borrower could not answer simple operational questions without referring back to documents. How many employees does the business have? "Let me check the proposal." What is your largest single customer's current outstanding balance? "I'll have my accountant send you that." When did you last draw a management account? "We do quarterly." It is possible for a business owner to not have these figures at his fingertips on any given Tuesday. It is also possible that the business, as presented in the proposal, is a more polished version of the business that actually exists, and the owner is navigating the gap between the two in real time.

In these dying column minutes, let me tell you what I did not learn from textbooks and did not learn from training programmes and did not learn from any of the bank's internal credit policy manuals, which are excellent documents that I read carefully and commend to you. What I learned from experience has is that the single most reliable red flag in a credit proposal is the one that makes you feel slightly embarrassed to raise. The projected revenue that is technically possible but seems aggressive. The connected-party transaction that is disclosed but not explained. The auditor's note that qualifies something on receivables in language so measured you almost read past it. The borrower's narrative that accounts for every failure as temporary and every success as structural. The proposal that has clearly been prepared by a very good consultant who has done this before and knows exactly which questions the committee will ask and has answered all of them just clearly enough to pass.

My grandmother would squint at that proposal, tilt her head to the left, and say: this goat arrived fat.

The question the loan officer has to answer — the question I asked for ten years, and ask still when I review financial data for publication — is not whether the goat is fat. It is: whose shamba has it been eating from?