The exposures a quarterly pack tends to hide

A system integrator's earnings move in steps. One troubled fixed-price contract can remove a large part of a year's EBITDA in a single reporting period, because an expected loss is recognised as soon as it becomes evident rather than spread across the remaining work. I therefore treat covenant terms as part of project risk, not as a treasury formality. Before a refinancing I model our order book with one plausible bad contract and compare the headroom each proposed definition would leave. I prefer half-yearly testing, a liquidity measure beside the debt ratio, and distribution restrictions I can explain to shareholders without embarrassment. I also share the fixed-price project register with lenders each half-year, so that a provision is never their first news of a contract. Cheaper money with brittle tests rarely stays cheaper once a waiver fee is counted.

Agencies in our markets generally pay reliably once an acceptance certificate is signed; the uncertainty lies in reaching that signature. I manage public-sector receivables from the documentation backwards. Our acceptance desk tracks each deliverable's test evidence, sign-off owner and committee date, and I review the ageing of completed but uncertified work beside invoiced balances, because the uncertified amount is usually larger. At bid stage the price must carry the financing cost of the stated payment terms and of realistic certification delays. Retention sums and performance bonds are treated as committed cash until their release. Selling certified receivables can be sensible, but only when the arrangement truly transfers the credit risk and costs less than the problem it removes. Neither condition held when a bank last offered us one, so we kept our invoices.

Our Ho Chi Minh City centre earns US dollars and pays salaries in Vietnamese dong. Exchange movements matter, but pay for experienced developers has risen faster than the currency has moved, and that cost does not wait for a contract renewal. I keep the two exposures apart. Currency risk on contracted receipts is hedged within a written policy, up to 60 per cent of the next twelve months, and uncontracted forecasts are never hedged. Wage risk is handled in the contract through an annual rate review linked to an independent salary survey, with a cap and a floor the client can budget around. Each quarter I review attrition, pay positioning and centre margin together, because a margin protected by underpaying engineers tends to reappear later as resignations, rework and missed delivery dates.

On fixed-price work we recognise revenue as progress is made, measured by cost incurred against total expected cost, so every change in the estimate of remaining effort alters reported profit immediately. Much of my career has gone into making that estimate dependable. Each month, delivery leads present remaining effort by workstream, the movement since the previous month and the reason for it; finance challenges the assumptions rather than the engineers. Variations enter the forecast only after the client has signed them. When a contract becomes loss-making, we provide for the whole expected loss in that period and tell the board at once. I also review every bid above S$10 million for its effect on the fixed-price share of the order book, so that no single contract can decide a covenant test on its own.