API Available

UPA File Generator

Build UPA (Uniforme Pensioenaangifte) test files for Dutch pension administrations. Set the employer and the schemes, choose the life events you want in the population, decide which optional elements to fill, leave empty or omit, and download one XML file per declaration period plus a participant overview for your CRM. Up to 10,000 employments per period. Everything runs in your browser and all data is fictional.

Every setting on this page is also a field in the API. POST /api/v1/tools/upa-file-generator, or generate_upa_files from an AI assistant. Useful when a regression suite has to rebuild last month's declaration, or when you want a file with a deliberate defect to test the rejection path. Full parameter reference.

A matching BRP/GBA register file is included by default. It holds the exact same people as the UPA files, same BSN, name, date of birth, address and partner, so every person in a declaration can be looked up in the register while you test. It downloads as brp-gba-test-data.csv inside the ZIP, in the wide GBA Totaalfile persoonslijst layout that BRP import tools expect (a readable compact format is also available). See also the BRP Test Data Generator.

UPA schema version

CdRdnEindArbov code 31 was removed and codes 32, 33 and 34 were added.

Employer name (NmLcr / NmIP)
Payroll tax number (LhNr)
Contact person (ContPers)
Phone (TelNr)
SWO relation number (RelNr)
Payroll package (GebrSwPakket)
Supplier id (IdLcr)
Administration id
Message id prefix (IdBer)
Affiliation number (RelNrAansl)
Collective agreement (CAO)
Sector (Sect)
Wage tax table (LbTab)
Job description (FuncOms)
Norm hours per week (AantUNormWk)
Franchise per year
Maximum pensionable salary per year

The franchise and the maximum pensionable salary are round placeholders, not legal figures. Replace them with the values from your own scheme before you use the premium amounts.

Scheme type

UPA has no element for the scheme type. This choice only sets the defaults below and is recorded in the participant overview.

RegKnmrk
RegVrnt list
Premium %
RegKnmrk
RegVrnt list
Premium %
RegKnmrk
RegVrnt list
Premium %
Start period
Start year
Period type (TvkCd)
Declaration type
Consecutive periods
Split into files
Employments per period
Seed
Youngest age18
Oldest age67
Holiday allowance in period5
Lowest annual salary24000
Highest annual salary90000
Address abroad12%
Unusual names15%
Unusual street names15%
Participant overview format

The one to feed into another tool, a script, pandas or R. No byte order mark and quotes only where a value truly needs them, because an importer that does not strip quotes would otherwise look for a column called BSN and find one called "BSN" instead. A Dutch or German Excel shows it as one column unless you import it through Data, From Text.

BRP/GBA file format

Downloaded as brp-gba-test-data.csv, inside the ZIP with the UPA files and also available on its own with the button above. The Totaalfile format is the one most GBA import tools accept.

Optional GBA metadata (dates, act numbers, source documents)

Off by default, so the file stays identity-only. Switch a group on to fill every related field consistently (each date lands in the right category next to its act number and source document, never half-filled).

Press Generate to build the files.

UPA test files for Dutch pension administrations

UPA (Uniforme Pensioenaangifte) is the message Dutch employers send to a pension provider every payroll period. It carries the employment, the wage components and the pension scheme data for every employee, and it also carries civil status, partner name and date of death, so most life events reach the pension administration through this file.

This generator produces a whole run at once: one XML per declaration period, with the same population carried across the periods so joiners, leavers, deaths and civil status changes land in the period where they belong. Element names and code lists follow the Pensioenfederatie and SIVI product description. Every BSN passes the elfproef and every IBAN passes the mod-97 check, so a clean file gets past the first line of validation.

Edge cases are the point of a test file. Long surnames, accented characters, name prefixes, house number additions and addresses abroad break letter templates and fixed width columns. The deliberate defects do the opposite: they break one employment on purpose so you can see what the receiving system says about it.

Everything runs in your browser and nothing is uploaded, so the tool works on a locked down corporate laptop. All generated data is fictional.

Dutch rules toolkit

This is one of 21 tools built on actual Dutch regulations rather than generic maths: salary and severance, notice periods and WW duration, BTW and mortgages, public holidays, AOW age, and GDPR-safe test data for Dutch systems.

See all Dutch tools
Dutch test data toolkit

This generator is part of a set of GDPR-safe Dutch test data tools: test BSNs, IBANs, document numbers, BRP records, and UPA files. All of them work in the browser, and several are available as a REST API and MCP server for CI pipelines and AI agents.

View the Dutch test data toolkit

How the UPA File Generator Works

  1. 01Fill in the employer header: name, payroll tax number (LhNr), affiliation number and the payroll package
  2. 02Pick the scheme type and set RegKnmrk and RegVrnt for the base scheme, plus a supplementary scheme or social fund if you need them
  3. 03Choose the start period, how many consecutive periods you want, and whether to split the result into smaller files
  4. 04Set the population size and the life event mix, or pin specific people by hand or from a CSV
  5. 05Set each optional element to fill, empty or omit, and switch on any deliberate defects
  6. 06Pick the participant overview format, press Generate and download the ZIP with one XML per period plus that overview

Building a UPA Test Set That Actually Finds Defects

Start small and clean: two or three periods, about 25 people, no defects. Once the message is accepted, raise the share of unusual names and foreign addresses, add the life events you care about, and extend the run to twelve periods so the year boundary and the holiday allowance peak are covered. Then make a second set with deliberate defects switched on and check that every one of them is rejected with a message you can act on. Write down the seed value: it makes both sets reproducible, so a defect found today can be recreated exactly next week.

When to Use the UPA File Generator

Use it when you need test data for a pension administration, a payroll to pension provider integration, a Wtp or FPR migration rehearsal, life event testing across a chain of systems, correspondence and letter testing, or a volume test that needs thousands of employments across many periods.

Common Use Cases

  • Life event testing: joiners, leavers, re-employment, death in service, marriage, divorce, unpaid leave, sickness and early IVA in one population
  • Chain testing: feed a run of consecutive periods through payroll, pension core and CRM in one go
  • Migration rehearsal: build a population that spans a full year to check the year boundary and the annual recalculation
  • Correspondence testing: long names and foreign addresses show how letter templates and address labels hold up
  • Error handling: inject a deliberate defect and check the rejection message the provider returns
  • Volume testing: up to 10,000 employments per period across as many as 24 periods, split into files of any size
  • Version testing: generate the same population in UPA 2024, 2025 and 2026 to see how the receiving system handles the element and code list differences

Expert Tips

  • Note the seed value in your test report, so the exact same file set can be regenerated later
  • Keep the first run clean, then make a second run with defects switched on rather than mixing the two
  • Set the holiday allowance period to the month your payroll pays it, so the cumulative maximum is actually reached
  • Use one file per employment when a test needs a single letter or notification to be traceable to one person
  • Feed the participant CSV back in through Import to pin the same population for a repeat run. The importer detects comma, semicolon and tab on its own
  • Switch the schema version once your settings are set: the field list greys out anything the selected version does not have

Frequently Asked Questions

What is a UPA file?→
UPA stands for Uniforme Pensioenaangifte, the Dutch standard for the periodic message an employer sends to a pension provider. It contains one block per employment with the person's details, the income period, the wage components and the pension scheme data. It is owned by the Pensioenfederatie and managed by SIVI, and nearly every Dutch pension administrator uses it.
Does one file contain one person or many?→
Many. A UPA message covers one administrative unit for one declaration period and normally holds every employee of that unit, so the default is one file per period with the whole population in it. You can override that: split into files of a fixed size, or produce one file per employment when you want a single letter or notification to be unambiguous.
How many employments can one file hold?→
Up to 10,000 per period here. The default is 100, which is the right size while you are still checking that the message is accepted. Above a few thousand the live preview switches off and you press Generate when your settings are ready, because building that much XML takes a moment.
Can UPA carry life events like marriage or death?→
Partly. The message holds CdBurgSt and DtIngBurgSt for civil status, DatOvl for a date of death, and DatEind with CdRdnEindArbov for the end of an employment, so those events do reach the pension administration through UPA. The partner name fields and the bank account number were removed from the standard in UPA 2022, so partner details have to come from another source. The generator writes what the standard still holds and keeps the rest in the participant overview.
What changed between UPA 2024, 2025 and 2026?→
Less than you might expect, because the big clean-up happened in 2022 when the partner group, RekNr, ExtrSal and OpgRchtExtrSal were removed and TaalVoorkeur and IndEindDln were added. Across the three versions offered here there are two differences that matter: VerlofUReg was added in UPA 2025, so it is not written when you pick 2024, and the CdRdnEindArbov code list changed in UPA 2026, which dropped code 31 and added 32 for retirement, 33 for death and 34. The tool applies both automatically and greys out any element that does not exist in the version you selected.
Which participant overview format should I pick?→
Plain comma CSV is the default and is the right choice when another tool, a script, pandas or R has to read the file. If you want to open it in Excel, pick the comma variant with the sep= line: it is still comma separated but Excel splits the columns correctly whatever language version you run. Semicolon and tab are there for tools that ask for them, and the Excel workbook removes the question entirely because a .xlsx has no separator or encoding to get wrong. Every text format is written as UTF-8 with a byte order mark, so accented names survive.
Where do I choose between a flexible and a solidary scheme?→
The scheme type selector sets the defaults, but it is not written into the XML. UPA identifies a scheme by RegKnmrk and RegVrnt only, and the pension provider knows which scheme type belongs to those codes. The choice you make is recorded in the participant overview so your test data stays self-explanatory.
Are the generated BSN numbers real?→
No. They are randomly generated numbers that satisfy the elfproef, which is what validation software checks. They are not issued to any person. The deliberate defect option can also produce a BSN that fails the elfproef on purpose.
What is the difference between empty and omit?→
Omit leaves the element out of the XML entirely. Empty writes the element with no content, for example a Voorl element with nothing between the tags. Validation code often handles those two cases differently, and the second one is the one that tends to be forgotten, so both are worth testing.
Can I correct an earlier period?→
Send a new full declaration for the same period with a later creation date. UPA has no separate mutation message: each period carries the current position, and the provider treats the most recently created declaration as the valid one. A supplementary declaration (AanvullendeAangifte) is also available for adding only the employees you want to top up.
What is in the participant CSV?→
The same people as the XML, plus everything that is useful in a CRM but not always in the message: civil status, partner name and BSN, whether the partner is in the same fund, another fund or has no pension at all, ex-partner, number of children, IBAN, document number, the life event assigned to each person and any deliberate defect. You can also feed that CSV back in to pin the same people again.
Does my data leave the browser?→
No. Generation runs entirely in JavaScript on your own machine and nothing is sent to a server. The tool also works offline once the page has loaded.

Related tools

12 suggested

From the blog

Further reading