What is folded in depends on the kind of credit, and this is where "just compare the APR" quietly stops being good advice. On a closed-end loan such as a mortgage or a car loan, the APR captures interest plus certain financing costs, which is why a mortgage APR is normally higher than the interest rate quoted beside it and why the two figures appearing together is not a contradiction. On a credit card the APR is essentially the periodic rate annualized, and the annual fee sits outside it. Two cards quoting an identical APR can therefore cost very different amounts to hold, and the comparison a cardholder needs is the fee schedule alongside the rate rather than the rate alone.
There is a third thing wearing the same name, and a reader who meets it on a deposit statement should not be surprised. Regulation DD, which governs deposit accounts, defines the interest rate as the annual rate of interest paid on an account which does not reflect compounding, and then permits that figure to be referred to as the annual percentage rate in addition to being called the interest rate. So on a loan the APR sits above the bare interest rate because it adds cost, while on a deposit it may lawfully be the bare nominal rate with compounding stripped out, sitting below the annual percentage yield. The label points in opposite directions on the two sides of a household balance sheet, and carrying it across produces a wrong comparison.
The most useful thing on this page is that Congress built a fourth, deliberately broader annual percentage rate for servicemembers, precisely because the Regulation Z figure leaks. The Military Lending Act's implementing rule caps the military annual percentage rate at 36 percent for covered borrowers, and 32 CFR 232.4(c)(1) sweeps into that calculation any credit insurance premium or fee, any charge for a debt cancellation contract or debt suspension agreement, any fee for a credit-related ancillary product, finance charges, most application fees, and participation fees. Then comes the clause that gives the game away: any such charge shall be included in the calculation of the MAPR even if that charge would be excluded from the finance charge under Regulation Z. A federal regulator wrote down, in a rule, that the ordinary APR omits costs worth capturing. The MAPR is not simply the APR with a ceiling attached; it is a wider measurement, though it keeps its own carve-out for genuine bona fide credit card fees under 232.4(d).
The accuracy tolerances are a disclosure standard, not a pricing allowance. Regulation Z treats a disclosed APR as accurate if it is within one eighth of one percentage point of the properly computed figure, widening to one quarter of one percentage point for an irregular transaction, which the rule defines as one involving multiple advances, irregular payment periods or irregular payment amounts. That is a rule about how precisely the paperwork must match the maths. It does not entitle anyone to charge an eighth of a point more than the deal says.
Because the calculation runs over the disclosed term, an APR describes a loan you keep. Paying discount points lowers the rate, and therefore the APR, by charging money at closing that is then spread across thirty years of disclosed payments. Sell or refinance in year four and the points were paid in full while only a fraction of the benefit arrived, so the APR that looked lower was describing a loan that did not happen. The same logic makes an APR on any short-held loan a poor guide, and it is the reason the break-even calculation below matters more than the comparison of two APRs.
One last boundary. On a mortgage the price-based tests that decide whether a loan is a qualified mortgage, and whether it is a higher-priced or high-cost loan, are keyed to how far its APR sits above the average prime offer rate for a comparable transaction. That machinery lives with the mortgage rules rather than with the definition of the figure, and the thresholds are indexed.