Check a filing before submission

Who, when. You receive the package of a regulatory filing, COREP, FINREP, LCR, NSFR, and you have to say whether it can go. This guide chains the screens in the order of a review: upload, choose the controls, run, read, fix, compare. Each step links to the page that describes the screen in detail.

1. Prepare the perimeter

A perimeter groups the reportings of an entity or a consolidation level. Create it once, in Admin › Settings › Perimeters, and give it a preferred threshold: it will be proposed by default at every run (see Perimeters and reporting files).

Tip

One perimeter per reporting entity and level, for instance “Group — consolidated” and “Bank — individual”. Quarter-to-quarter comparisons, the analytical review and the Pillar 3 look for the previous filings in the same perimeter.

2. Upload the filing

In Analysis › Reportings › EBA reportings, click Upload files, choose the perimeter and drop the package (see How to upload reporting files).

  • Tick Official submission if this is the version that will be sent. Plan executions propose only official reports by default.
  • A ZIP archive of several files is split: each file is uploaded separately.
  • A file already uploaded in this perimeter is refused: “Already uploaded on this perimeter”.
Warning

From taxonomy 4.2 on, an XBRL-XML file is refused: upload the xBRL-CSV package (.zip) of the filing. An Excel workbook without a reference date waits until you complete it from the Processing tab.

Follow the processing in the Processing tab, then check the file in the Files tab: the Extraction incomplete mark flags data the taxonomy did not recognise, listed in the file’s Unmatched data tab.

3. Choose the controls

Three sources of controls combine in a plan (see Control libraries):

Source When to use it
VR-XL The EBA validation rules, as published. This is the base: a filing that breaks them is rejected by the supervisor.
EGDQ-XL The ECB data quality checks. They flag inconsistencies that do not block the filing but call for an explanation.
Custom controls Your own consistency rules: across templates, against the previous quarter, against your internal thresholds (see Controls).
Warning

The EGDQ and Validation rules libraries in the historical format are deprecated: build your plans with EGDQ-XL and VR-XL (see EGDQ controls and validation rules).

To write a custom control, two choices matter (see Control):

  • Match type: Exact when the formula must hold to the cent, for instance a total equal to the sum of its rows in the same template; Threshold when a difference is acceptable, for instance a reconciliation between COREP and FINREP, or a quarter-to-quarter variation.
  • Base reference: the member used as the denominator of the relative difference. Put on the left the reference value, the one you consider right.

To compare with the previous quarter, write {…}[T] and {…}[T-1] (see Time-dependent formulas).

4. Build the plan

In Analysis › Control plans, create one plan per review exercise, for instance “Quarterly COREP”, and tick the controls in its tabs (see Control plans and thresholds). Filter the published libraries by Templates and Modules to keep only the tables of the filing: a plan of every rule runs for a long time.

A plan only holds links: a corrected control is corrected in every plan that uses it, without changing past results.

5. Choose the threshold

The threshold applies only to Threshold controls and to == and != formulas (see Thresholds). It sets two bounds: up to the success value, the control is Pass; up to the warning value, it is Pass with reserve; beyond, it is Fail. A relative and absolute threshold also asks for a combination: Both must hold (AND) to require both deviations, Either is enough (OR) for one to suffice.

Tip
  • Relative 5 % / 10 %: the difference is measured as a percentage of the base member. Suited to reconciliations between reportings and to quarterly variations.
  • Absolute: the difference is measured as an amount, in the reporting’s unit. Suited to rounding differences between templates of the same reporting.
  • Relative & absolute: both bounds are set, so that a small amount does not raise a percentage alert.

A threshold is copied into each result: changing it later does not change past runs.

Exact controls, inequalities and the VR-XL and EGDQ-XL rules do not read the threshold: they pass or fail.

6. Run

On the plan page, Execute plan (see How to execute a control plan):

  1. the perimeter;
  2. the threshold, the perimeter’s one is preselected;
  3. the reporting date or dates. Untick Show only official reports if the filing under review is not yet marked official;
  4. the period type and the reference date: they give [T] and [T-1] their meaning;
  5. the files.
Warning

A [T-1] formula is only evaluated if the files of both dates are ticked. Otherwise the control gets No Data.

Check the selection flags, before the run, the templates read by your formulas that are missing from the ticked files. Tick Send me the result by e-mail when the execution ends to receive the Excel file at the end (see Receiving the result by e-mail).

7. Read the result

The result page (see Control plan results) starts with a summary card. Handle the statuses in this order:

Status What it says What you do
Fail The formula is false, or the difference exceeds the warning value. Expand the rule: the Comparison shows both members and the difference; the Data tab opens each datapoint in its template (see Open a datapoint in its template). Fix the filing or document the difference.
Pass with reserve The difference lies between the two bounds of the threshold, or the DPM-XL rule has severity “Warning”. To explain, not necessarily to fix.
Unable to evaluate Incorrect formula, division by zero, unsupported DPM-XL construct. The Errors tab gives the cause. Fix the control, not the filing (see Evaluation error messages).
No Data The cells read are not in the files. Check that the template is filed and that both dates are ticked.
Precondition is false The precondition excludes this case. Nothing: the control does not apply.
Not applicable The reporting’s release is outside the control’s window. Nothing, or widen the window if it is a mistake.
Pass Nothing to report.

The tiles filter the list, and Export … only (.xlsx) exports the filtered status, for instance the failures alone (see How to export a result).

8. Fix and run again

After the correction, upload the new version of the filing in the same perimeter, then run the plan again on that file. The ⋯ menu of a run offers Execute with the same parameters, but only as long as the original files exist: for a replaced file, use Execute plan and tick the new file.

The Vs previous run column of the Runs tab counts the statuses that changed compared with the previous run on the same dates.

9. Compare two versions

To see what changed between the corrected version and the previous one, tick both files in the reporting list and click Compare (see Perimeters and reporting files). The Pivot is the reference version; Hide Similar Data and the variation threshold keep only the differences. The Excel export applies the same filters.

When the filing goes out, mark the sent file as Official: it is the one that plans, dashboards and the Pillar 3 will propose by default.