Credit risk

Imagine a company that must repay you 100 USD, or US dollars, on an agreed date. We now have a debtor, an amount and a deadline; only then can we assess whether the obligation was met. The contract and applicable rules also determine when a delay or another event counts as default. There is no universal waiting-day count for every debt. If a bond's price meanwhile falls from 100 to 95, that alone does not prove non-payment: the market may have reacted to interest-rate changes. Equally, a company's ability to pay may deteriorate long before its first missed payment. We therefore examine its finances as well as the particular debt's terms. [1]

An event can be rare and still be costly. Bank for International Settlements (BIS) uses PD for probability of default; LGD for the fraction lost if default occurs; and EAD for exposure at default, the amount of the claim at that point. A claim is money the other party owes you. Our simplified annual model uses PD 2%, LGD 60% and EAD 100 USD. The calculation 0.02×0.60×100 = 1.2 USD gives expected loss under those assumptions. It does not mean exactly that amount will be written off every year. It is an average, not a loss ceiling: an individual loan can turn out much worse. This is a simple model, not a complete bank-capital calculation or a substitute for assessing the borrower. [2]

Knowing that a troubled company still owns assets is not enough. We need to know who is entitled to them. Our liquidation example has assets worth 70 USD, a secured claim of 50 backed by those assets and an unsecured claim of 40. Under the assumed priority and without costs, the first creditor receives 50 and the second the remaining 20. The latter recovers 20/40 = 50% of its claim. Costs, delays, other priority claims or invalid security can change the real outcome. Collateral is not a magic word: its legal effect and others' rights over the same asset matter. [3]

A credit rating is an agency's opinion of a borrower's or a particular debt's credit quality. Investor.gov warns that even AAA does not guarantee payment. A company's rating may also differ from the rating on its particular loan or bond. Ratings do not address every price risk or whether you can sell an investment quickly. Check who issued the rating, when and what exactly it covers; the grade alone is not enough. Registration with the SEC, the US securities regulator, does not mean it endorses the agency’s ratings. Even highly rated debt may be offered at a price you do not want to pay. [4]

If a company offers interest on bitcoin you hand over, ask where the interest comes from and who must return your deposit. Our contract promises 1 BTC plus interest of 0.05 BTC. After default, however, you recover only 0.4 BTC. Compared with the original deposit alone, 0.6 BTC, or 60%, is missing; compared with the entire promise, the shortfall is 0.65 BTC. These are amount comparisons, not a contractual allocation of repayments between principal and interest. This is not native Bitcoin protocol yield: it depends on the company's activities and ability to pay. Identify the legal borrower, any onward lending, insolvency rights and the currency of the claim. A positive number in an app does not prove the company can release the bitcoin. [5]

Spreading money across companies can help if the same problem does not hit them all. Our two loans of 50 USD each, however, go to firms dependent on the same market. If both default and each returns only half, the loss is 25+25 = 50. Different names did not remove the shared shock. Companies may also share ownership links, collateral or a guarantor. A guarantor is another party promising, under specified conditions, to pay instead of the borrower. A guarantee therefore transfers some risk to that party; its ability to pay and the enforceability of its promise matter. Count not just the contracts, but also the risks they have in common. [1]

In financial statements you may encounter IFRS 9, the international financial-instruments standard, and ECL, expected credit losses. Twelve-month ECL concerns defaults possible in the next 12 months after the reporting date, but includes their lifetime cash consequences, weighted by probability. If our hypothetical borrower defaulted in month 10 and some debt was recovered only in month 24, that does not mean ignoring later recovery. The timing of money also matters. Lifetime ECL considers possible defaults over an instrument's entire expected life; the approach used depends on the instrument’s rules and credit status, including any significant rise in risk since it was first recorded in the accounts. These accounting requirements cannot automatically be applied to US GAAP, the US accounting rules, or to every crypto product. [6]

A borrower's receipts can shrink just as collateral prices fall and collection takes longer. The problems then reinforce one another in practice. Our stress model raises PD to 10% and LGD to 80% with EAD 100 USD: 0.10×0.80×100 = 8 USD instead of the original 1.2. This is not a forecast; it tests changed assumptions. That is why we need current receipts and payments, maturities, legal protection and reliable information. A simple average is a useful starting point, not a complete picture of extreme losses. Every promise of yield eventually brings us back to the same question: who will pay, and from what resources? [1] [2]