Methodology

The index ranks countries by one quantity: the number of paid days off a year the law makes an employer give a standard full-time employee. It adds two figures drawn directly from statute: the minimum paid annual leave an employer must give, counted in working days, and the number of national public holidays in the year. Rank 1 is the most days off.

Both figures are already counted in days off per year, so the index adds them rather than weighting them. That is a deliberate choice: any weighting would be an editorial judgment about whether a day of annual leave is worth more than a public holiday, and the index does not make one. Where we say “countries” for short, the set includes a number of territories and special administrative regions that set their own employment rules.

It measures statutory minimums, not what people get. Collective agreements, enhanced contractual holiday allowances, seniority-based extra days, and ordinary employer generosity all sit on top of the number shown here, and in many labour markets they are the norm rather than the exception. A reader taking a figure from this index as a description of working life in that country will be wrong. Taking it as the legal floor an employer has to clear will be right.

annual_leave_days is the statutory minimum paid annual leave for a standard full-time permanent employee at a normal length of service, counted in working days. Entitlements that scale with seniority are taken at the statutory base rate, not at the top of the scale.

public_holidays is the count of national public holidays in the year. Regional and optional observances are excluded where the source distinguishes them, which means the count can be lower than the number a resident of one particular region actually observes.

The composite is the percent-rank of the total within the ranked set, scaled to 0 to 100, where 100 is the highest total in the edition. It exists to make position legible at a glance. The days figure, not the composite, is the quantity to quote: the composite is relative to this edition’s set of countries, the days are not.

The two component percentiles shown in the table (annual leave and public holidays separately) are reported for context. They are not combined and they do not feed the composite.

Countries are ordered by total days, descending. Ties share a rank: countries with an identical total receive the same rank, and the next rank number is skipped (standard competition ranking), so a rank of 30 means 29 countries give more days off. The ranking is complete, not truncated.

Where the study page converts days into working weeks or into a career total, it divides by 5 working days a week and 260 working days a year. Those two constants are conventions of the presentation, not of the data.

The study page also states each country’s total as a share of the working year, because the reader it is written for is an employer deciding where to hire, and a statutory day off is a day of the year that is paid for and not worked. The denominator is 260 working days: a 5 day week across 52 weeks. It is a fixed convention of the presentation, exactly like the working-week and career figures above, and it is not adjusted per country.

Three limits of that arithmetic, stated rather than buried. It counts a public holiday as a working day lost, which is the index’s own definition of a day off but overstates the loss wherever a holiday falls on a weekend, and the number of those varies by country and by year. It takes no account of national working-time rules, so a country with a statutory week shorter or longer than 5 days is still divided by 260. And it is built on statutory minimums, so the real share in any particular contract is higher wherever the employer, a collective agreement or a seniority scale adds days, which is the norm in many labour markets.

The salary figures in the per-hire comparator are the same division applied to a salary the reader types: annual salary times total_days divided by 260, and that figure multiplied by a team size. We hold no salary data for any country and publish none; the salary is the reader’s own input and the default shown is labelled an example. The result is the share of a budgeted salary that buys statutory paid time off rather than worked days, not a cost added on top of salary, and it excludes social contributions, leave accrual rules and any payout of accrued leave on exit. The capacity figure alongside it is the same subtraction from the other direction: 260 working days less the statutory paid days off.

Public-holiday counts are available for far more countries than statutory annual-leave values are. Ranking a country on holidays alone against countries scored on both inputs would put it near the bottom of the table for a reason that has nothing to do with its law, so the index does not do it.

Countries holding an approved public-holiday count and no approved annual-leave value are published in a second table, clearly labelled, unranked, showing the one figure we have. This edition has 61 of them, including Åland Islands, Bangladesh, British Virgin Islands. Countries with neither input are listed as excluded, as are countries holding both figures where one of them has no government or official source yet. This edition ranks 150 countries and lists 39 unranked outside the holidays-only table.

Countries that set public holidays regionally are ranked with a footnote. Where public holidays are set regionally rather than nationally, the index counts the nationwide holidays plus the calendar of the most populous region, and names that region on the row. 5 countries are ranked this way in the 2026 edition: Spain (Andalusia), Australia (New South Wales), Switzerland (Zurich), United Kingdom (England and Wales), Canada (Ontario). Annual leave for all of them is the national statutory minimum, and the national range sits next to the figure in the note under the ranking table, so the spread the single calendar sits inside is visible rather than hidden.

A country is ranked only if it has an approved, sourced value for both statutory annual leave and public holidays. If either input is missing, it goes in the holidays-only table or the excluded list, named, with the field it is missing. We do not estimate, interpolate, or use AI to fill gaps.

A statutory zero is a value, not a gap. The United States sets no federal minimum paid annual leave, so it is scored at zero on that input and stays in the ranking. Removing it because the number is zero would quietly flatter the global picture and would hide the single most useful fact in the dataset.

A country is ranked only if it has approved, non-null values for both inputs (annual_leave_days and public_holidays). No gap-filling. A statutory zero is a value, not a gap: the United States sets no federal minimum paid annual leave and stays in the ranking at zero. Countries holding an approved public-holiday count but no annual-leave value are published in a separate, unranked public-holidays-only table. Where public holidays are set regionally rather than nationally, the index counts the holiday calendar of the most populous region, names that region on the row, and ranks the country with a footnote. A country whose public-holiday row is still flagged as regionally set without a named region is not ranked: there is nothing to count, and inventing a national figure would be a fabrication.

A country can hold more than one approved value for the same field, from different sources or from different dates. The selection rule is fixed, deterministic, and identical to the one the Employer Burden Index and the Termination Cost Index use: the most recent as-of date wins, with source identifier, source URL, and the value itself as pure tie-breaks. Nothing is averaged, and no human picks a favourite.

At an equal date the tie-break is source priority, in five tiers, applied identically across all three of our indexes:

1. The statute, or the government that administers it. Nothing outranks the primary legal text.
2. Official multilateral and government-agency compilations. ILO EPLEX, the OECD, the World Bank Employing Workers archive, ISSA and similar. Collected under a single documented methodology by bodies with a legal research function, which is what makes them comparable across countries.
3. Secondary compilers. The WageIndicator Foundation, PwC Worldwide Tax Summaries and similar. Real research processes, but commercial or NGO rather than official.
4. Machine-extracted rows. Pulled from a reachable page and editorially reviewed before approval. Shown when nothing better exists, but a country resting on one is not ranked.
5. Open community holiday calendars. Nager.Date. Shown when nothing better exists, never ranked.

Source identifier, source URL and value remain as last-resort tie-breaks, so selection is always total and reproducible from the dataset. Nothing is ever averaged.

Where the tier-1 values come from

For 57 countries we hold a separately audited set of government sources, collected for our country hiring guides: the labour code, the ministry page or the official holiday instrument, each captured with its publisher, its URL and the date it was read. The statute-tier figures for those countries are read from exactly those sources, and the quoted sentence and any unit conversion travel with the value in the dataset, so a reader can check the arithmetic against the same text we used.

The remaining countries have no audited government source in our set yet, so they use the next best tier available to them: an official multilateral compilation where one exists, then a secondary compiler. A country whose figure would come from a machine-extracted row or the holiday calendar is shown with its values but left out of the ranking until a government or official source is found. The source label on every row says which case a country is in.

Why the holiday calendar ranks last. That is the one placement in the list that is a judgment rather than an obvious ordering, so it is worth stating the reasoning. Nager.Date is an unreviewed community dataset of calendar observances, not a labour-law source, and its “nationwide holidays only” rule answers a different question from “how many paid public holidays does an employee get”. It gives 25 for El Salvador where the Código de Trabajo gives 9, and 5 for the United Kingdom and 6 for China where national practice gives 8 and 13. It is the right source for the many countries where we hold nothing else, and the wrong one wherever a statute or a national government list exists.

An earlier build of this edition fell back to comparing the source identifier alphabetically at an equal date, which is not a judgment about quality at all. It let the holiday calendar outrank a labour code on nothing but the letter n. Moving to the priority rule changed the selected value for 80 country-fields and the published total for 43 countries, and it left the two published sibling indexes untouched: not one notice, severance or employer-contribution value changes under it.

Calendar days versus working days

The annual-leave column is defined in working days. Several statutes express the entitlement in calendar days instead, and 30 calendar days is about 22 working days on a five-day week. Most rows in the fact store have already been converted, and the note on the row says so. Some have not.

The index does not convert anything at publication time. Converting a value the source did not convert would put a number in the table that appears in no statute and in no source, which is exactly the kind of quiet adjustment this index exists to avoid. Instead, a row whose stored value is still in calendar days is flagged deliberately, by whoever stored it: on the study page as a footnote under the table, and in the downloadable dataset as the unit_caveat column, alongside the note itself. No row carries that flag in this edition: every calendar-day statute now has its conversion recorded on the row. Brazil is the one to know about. The statute sets 30 dias corridos; the fact store records the conversion to 22 working days and the index publishes 22, with the Portuguese text and the arithmetic attached to the row.

Public holidays the law does not add on top

A public holiday adds to the total only where the law makes it a paid day off on top of annual leave. Two cases do not. In the United Kingdom the statutory 28 days, 5.6 weeks for a five-day week, can include the eight bank holidays: GOV.UK says bank holidays “do not have to be given as paid leave” and an employer “can choose to include bank holidays as part of a worker’s statutory annual leave”, so the UK total is 28. In the United States the Fair Labor Standards Act “does not require payment for time not worked, such as vacations, sick leave or federal or other holidays”, so for a private employer the federal holidays add nothing and the US total is 0.

A third case: in Norway only 1 May and 17 May are paid by statute, and in France only 1 May is a mandatory paid day off, so those totals add just the holidays the law requires as paid.

Each case is a stored fact, public_holidays_treatment, carrying the government text that decides it, and the dataset publishes it per country, so the total can be reproduced from the files. A country without that fact adds its public holidays to its leave. We are checking every ranked country against its own statute and update the edition as each one is confirmed. Both stored figures are always published as they stand; only the total follows the treatment.

Two studies cover nearby ground. Remote publishes a Global Life-Work Balance Index that blends around ten metrics across 60 countries into one score. Resume.io has published a paid-vacation study covering roughly 185 countries. Both answer a broader question than this index does, and both are worth reading.

This index is deliberately narrower. Five differences, stated plainly:

• Statutory entitlements only. Two inputs, both taken from law. No subjective inputs, no quality-of-life measures, no cost-of-living adjustment.
• No weights. The inputs share a unit, so they add. Nothing is blended and no weighting is applied or withheld.
• Every value sourced and dated. Source and as-of date travel with each figure, on the page and in the dataset.
• Open dataset. CSV and JSON under CC BY 4.0, a permanent archived record, and a changelog, so the ranking can be reproduced or contradicted by anyone who wants to.
• Gaps published, not filled. A country missing an input is named in a separate table rather than estimated into the main one.

One consequence of the coverage rule is worth flagging for anyone comparing headlines. A widely quoted finding elsewhere puts Iran at the top for vacation days. Iran holds an approved annual-leave value in our fact store but no approved public-holiday count, so it is excluded from this edition entirely and we can neither confirm nor contradict that headline from our own data.

Every value comes from the same sourced-and-dated statutory dataset that powers our country pages, our Global Employer Burden Index, and our Termination Cost Index. National statutes and labour codes supply the strongest values, with ILO EPLEX, the WageIndicator Foundation covering the long tail. The open Nager.Date public-holiday dataset fills gaps only in the unranked tables. This edition rests on 300 sourced statutory values across the two fields. Every row in the downloadable dataset carries its source and as-of date, so any number in the index can be traced back to where it came from.

On the public-holiday counts:almost every ranked country uses its government’s statutory list, the rest an official or secondary compilation, counted as days, with a holiday that falls on a weekend still counted. Where we hold only Nager.Date, which counts distinct nationwide holidays of type Public and excludes regional, optional and bank observances, the count is shown but the country is not ranked, because that rule can sit several days away from the statute.

The index is published in annual editions (2026 is the current one). The days figures are absolute and comparable across editions. The composite is not: it is a percentile within an edition’s set of ranked countries, so the same raw total can produce a different composite once the set changes. Compare days between editions, and positions within one.

Corrections policy: when a sourcing error is found within an edition, we correct the underlying value, rebuild the edition, and note the correction in the affected value’s provenance. We do not leave known-wrong numbers standing until the next edition. The dataset’s generated_at timestamp reflects the most recent rebuild.

The dataset is archived on Zenodo under Global Paid Time Off Index (Employ Borderless). The DOI for this edition is pending; until it resolves, cite the canonical URL of the study page.

The three indexes measure three different moments of the same employment relationship, from the same fact store. The Global Employer Burden Index scores what it costs to keep someone employed. The Termination Cost Index scores what it costs to end it. This one scores what the employee is entitled to take off in between. A country can sit high on one and low on another, and several do. For reference, the median ranked country here guarantees 29 paid days off a year.