A junior preparer runs Arelle, sees a clean validation report, and sends the file to the partner. The partner runs Arelle one more time with the calculation profile dialled up and gets back a single line: xbrl.5.2.5.2 calculation inconsistency on profit before tax. This is the moment most accountants meet the calculation linkbase, and it is rarely a friendly introduction.
Calculation arcs assert arithmetic. They wear the summation-item arcrole, and each arc carries a weight attribute. A weight of plus one means the child contributes to the parent with the parent's natural balance; minus one means the child reverses it. Cost of sales rolling into gross profit is a credit-balance child against a credit-balance parent: weight equals minus one because cost of sales reduces the parent. Asset rolling into a larger asset total: weight equals plus one. The sign convention sits inside the linkbase, not on the fact.
The validator's job is to take a parent fact and the same-context child facts, apply the weights, and check that the result agrees with the parent within the tolerance implied by their declared decimals. Same context here is strict: same entity identifier, same period, same units, same dimensional scenario. Cross-context summation is not permitted.
Two pitfalls account for most false positives. The first is decimals drift. Tag a parent at decimals equals minus three (rounded to thousands) and a child at decimals equals zero (whole units), and the rounding tolerance is computed from the looser side. Sub-thousand variations look like real inconsistencies even when the data is fine. The fix is to keep decimals consistent across the calc tree, or to migrate to Calculations 1.1, the 2024 Recommendation, which leans on Open Information Model rounding semantics rather than the older fact-level approach. ESEF accepts Calc 1.1; SEC EFM remains primarily on Calc 1.0 plus its own EFM rounding rules.
The second pitfall is duplicate facts. iXBRL practice routinely tags the same value in a summary table and again in a footnote. Calc 1.0 treats those as separate observations and gets noisy. Calc 1.1 expects this and merges them when their declared precision agrees. Knowing which validator profile is in play is half the diagnosis.
Calculation arcs carry summation-item arcrole and a weight attribute (typically +1 or -1).
A parent fact equals the weighted sum of its children, with same context (entity, period, dimensions) and matching unit. If they disagree beyond rounding implied by decimals, the validator reports calculation inconsistency (xbrl.5.2.5.2).
Weights are tied to balance: an asset-balance child rolling into an asset-balance parent uses weight=1; a credit-balance contra into a debit-balance parent uses weight=-1.
Calculations 1.1 (XBRL Recommendation 2024-02-14) is more permissive about duplicates, rounding, and dimensions; ESEF accepts it; SEC EFM is still primarily on 1.0 plus EFM-specific rounding rules.
Decimals drift across calc tree levels (parent at thousands, child at units) silently breaks rounding tolerance and produces false-positive xbrl.5.2.5.2 inconsistencies.
Children must share the parent's period type and dimensional context. Cross-context summation is not allowed.
<link:calculationArc xlink:type="arc"
xlink:from="profit-loss"
xlink:to="cost-of-sales"
xlink:arcrole="http://www.xbrl.org/2003/arcrole/summation-item"
weight="-1"
order="2.0"/>3 questions. Press 1 to 4 to select, Enter to confirm.
Completing this quiz marks the section done in your progress.
Tip: press 1 to 4 to select, Enter to confirm.