This chapter comments on tax administration performance in managing the collection of outstanding taxes, and describes the features of a modern tax debt collection function and the approaches applied by administrations to prevent debt being incurred.
Tax Administration 2025
Comparative Information on OECD and other Advanced and Emerging Economies
Abstract
The collection function involves engaging with, and potentially taking enforcement action, against those who do not file a return on time and/ or do not make a payment when it is due. Even with the growth in pre-filled or no-return approaches over past years (see Chapter 4), the filing of a tax return or declaration is still required in many jurisdictions participating in this publication. Although 2023 on-time filing rates (based on returns received) averaged between 85% and 93%, more than 130 million returns were not filed on time that year (see Tables 4.7. as well as A.47, A.51, A.55, and A.60). It is important therefore that administrations continue to focus efforts on improving the timely collection of late and outstanding returns.
Looking at the collection of late payments, all but one administration participating in the survey report that staff resources are being devoted to taking action to secure the payment of overdue tax payments (the Chilean tax administration reported not being responsible for debt collection; see Table A.19). Information provided by administrations attribute on average around 10% of total staff numbers to the collection function (see Chapter 9).
This chapter:
Takes a brief look at the features of a modern tax debt collection function, including the elements of a successful tax debt management strategy and the preventive approaches to debt that are being taken; and
Comments on tax administration performance in managing the collection of outstanding debt.
It should be noted that a detailed overview of debt collection powers and their usage by tax administrations can be found in the 2024 edition of this series (OECD, 2024[1]).
To maintain high levels of voluntary compliance and confidence in the tax system, administrations must ensure that their debt collection approaches are both “fit for purpose” and meet taxpayer’s expectations of how the system will be administered. This means not only taking firm action against taxpayers that knowingly do not comply, but also using more customer service style approaches where taxpayers want to meet their obligations, but for reasons such as short-term cash-flow issues, are not able to do so.
Increasingly, tax administrations are taking an end-to-end or systems view of their processes and researching the reasons why returns may not be filed or payments made. They are also using information about the taxpayer’s previous history, to identify patterns and/or anomalies. Box 7.1. highlights some developments in this area.
Belgium – Project ARGUS
Belgium’s General Administration of Taxes has embarked on a major digital transformation process. This has included creating a state-of-the-art debt database (FIRST), a taxpayer interaction platform – “MyFinaccount”, and a set of advanced, datamining-based, predictive analytics tools.
However, to manage the recovery process more efficiently, the ARGUS initiative was developed. It consists of 4 phases:
In Phase 1 a visual dashboard was built, presenting the solvency (overview of assets) of a debtor.
In Phase 2, the system was set-up to automatically complete all forms and documents related to asset seizures and garnishments.
Phase 3 will see the full integration of the existing predictive models into ARGUS, allowing it to propose recovery actions based on specific business rules. These models analyse large datasets at regular intervals in order to present the end-user with a multi-criteria and rule-based classification of debtors. For each class of debtors, a set of recommended recovery actions will be proposed. This is due to be completed by December 2025.
Finally, in a future Phase 4, Artificial Intelligence will be integrated in ARGUS to analyse all available data and identify payment evasion patterns, and to improve the prioritisation of the recovery action with the highest possibility of success.
Lithuania – New debt management strategy – targeted recovery
The State Tax Inspectorate of Lithuania (STI) has applied the OECD’s Debt Management Maturity Model (OECD, 2019[2]) to identify potential development for its debt recovery processes. This evaluation has influenced the STI’s long-term debt management goals.
The STI’s current approach focuses more on personalised, data-driven approaches, allowing the STI to adapt to constantly changing environments. An evaluation will be conducted to determine how debtors can be categorised based on their willingness/ capacity to pay. One potential approach is that categories could range from those who want to comply but lack the necessary knowledge or resources, to those who deliberately avoid paying tax despite having the means to do so. Different responses would then be considered for the different categories, ranging from educational outreach to strict enforcement measures.
Additionally, the STI will continue to leverage automation and advanced analytics to reduce the input from staff, ensuring efficient use of resources. A predictive model is in development to assess the likelihood of debt repayment. This will apply algorithms to analyse historical taxpayer data that identifies patterns which indicate which taxpayers are unlikely to repay their debts within the set timeframe. Beyond analysing an individual taxpayer’s past behaviours, the model will also compare similar taxpayers to identify the shared characteristics associated with a higher risk of non-compliance. The goal is to promptly identify taxpayers who should be subject to additional debt recovery measures at an earlier stage. Based on the predictions, the recovery actions will be prioritised accordingly.
Sources: Belgium (2025) and Lithuania (2025).
Essential features of a tax debt collection function
The 2014 report Working Smarter in Tax Debt Management (OECD, 2014[3]) provided an overview of the modern tax debt collection function, describing the essential features as:
Advanced analytics – that make it possible to use all the information tax administrations have about taxpayers to accurately target debtors with the right intervention at the right time.
Treatment strategies – the collection function needs a range of interventions, from those designed to minimise the risk of people becoming indebted, to support taxpayers to make payments and to take appropriate enforcement measures where appropriate.
Outbound call centres – which make it possible to efficiently pursue a large number of debts.
Organisation – debt collection is a specialist function and is usually organised as such. The right performance measures and a continuous improvement approach help drive desired outcomes.
Cross border debts – the proper and timely use of international assistance is crucial, particularly the “Assistance in Collection Articles” in agreements between jurisdictions.
The 2019 report Successful Tax Debt Management: Measuring Maturity and Supporting Change (OECD, 2019[4]) provides further insights into the elements of a successful tax debt management strategy, setting out four strategic principles that tax administrations may wish to consider when setting their strategy for tax debt management. These principles focus on the timing of interventions in the tax debt cycle, from consideration of measures to prevent tax debt arising in the first place, via early and continuous engagement with taxpayers before enforcement measures, to effective and proportionate enforcement and realistic write-off strategies. The underlying premise for these principles is that focusing on tackling debt early, and ideally before it has arisen, is the best means to minimise outstanding tax debt. The report also contains a compendium of successful tax debt management initiatives and an overview of a comprehensive Tax Debt Management Maturity Model which was subsequently published as a self-standing document (OECD, 2019[2]).
To make the maturity model easier to use and help to create a better global picture of maturity in tax debt management, the OECD published a new edition of the Tax Debt Management Maturity Model (OECD, 2025[5]). The 2025 edition is shorter and focuses more on the core elements of successful tax debt management. As Figure 7.1. shows the new edition has already been completed by 36 tax administrations, presenting a global picture that can be useful to administrations in examining their own future reforms and also in identifying where future collaborative work, such as capacity building, might be of the most value.
Source: OECD (2025), Tax Debt Management Maturity Model, OECD Tax Administration Maturity Model Series, 2025 Edition, https://www.oecd.org/content/dam/oecd/en/topics/policy-issues/tax-administration/tax-debt-management-maturity-model-2025-edition.pdf (accessed on 1 October 2025).
Preventive approaches
A key area of focus for tax administrations is to take actions that prevent debt from arising, as this offers efficiency opportunities over collecting outstanding arrears.
Box 7.2. illustrates the approaches taken by some administrations. Advances in predictive modelling and experimental techniques as reported in the OECD report Advanced Analytics for Better Tax Administration (OECD, 2016[6]), in the compendium of successful tax debt management practices contained in the OECD report Successful Tax Debt Management: Measuring Maturity and Supporting Change (OECD, 2019[4]), or in Annex B of the 2025 version of the Tax Debt Management Maturity Model (OECD, 2025[5]) are helping many administrations better match interventions with taxpayer specific risk. The approaches used fall into one of the following categories:
Predictive analytics, which tries to understand the likelihood of certain outcomes and, as regards debt collection, includes modelling the risk that an individual or company will fail to pay as well as models that attempt to assess the likelihood of insolvency or other payment problems; and
Prescriptive analytics, which is about predicting the likely impact of actions on taxpayer behaviour, so that tax administrations can select the right course of action for any chosen taxpayer or group of taxpayers. (OECD, 2016[6])
Many administrations are blending both practices and have trialled a variety of approaches aimed at changing “taxpayer behaviour”. As noted in Chapter 6, around half of the administrations report employing behavioural researchers, and the use of behavioural insight practices has the potential to transform the approach to tax debt as administrations move away from the ‘one-size-fits-all’ approaches (where it is cost-effective to do so) and instead try to identify:
Which cases should be subject to an intervention;
When to intervene (for example, even before a return or payment might be due); and
Which type of action would achieve the best cost-benefit outcome.
A new approach has been put in place that leverages data analysis techniques and IT tools alongside improved processes to improve efficiency in debt recovery. Under this approach, data analytic tools provide enhanced insights on the risk that a debt is not paid. Using this insight specialised teams can then apply specific strategies to triage risks, and conduct recovery actions on a large scale as well being able to take action to prevent debts occurring. An evaluation loop then analyses the outcomes of these actions to improve the service further.
Source: Spain (2025).
The total amount of outstanding arrears at fiscal year-end remains very large, in the region of EUR 2.7 trillion. For survey and comparative analysis purposes, “total arrears at year-end” is defined as the total amount of tax debt and debt on other revenue for which the tax administration is responsible that is overdue for payment at the end of the fiscal year. This includes any interest and penalties. The term also includes arrears whose collection has been deferred (for example, as a result of payment arrangements).
The total amount of “collectable arrears” at fiscal year-end was around EUR 990 billion. Collectable arrears is defined as the total arrears figure less (i) any disputed amounts for which collection action has been suspended pending the outcome, (ii) amounts that are not legally recoverable, and (iii) arrears which are unable to be collected, for example, where the debtor has no funds or other assets.
As a result, and despite efforts to make data comparable, care needs to be taken when comparing specific data points as the administration of taxation systems and administrative practices differ between jurisdictions.
In 2023, the average ratio for total year-end arrears to net revenue collected was 29.3% (see Table D.41). As in past years, it remains heavily influenced by the very large ratios of a small number of jurisdictions that show ratios above 50%. If these jurisdictions are removed, the average reduces to 11.5% of net revenue (see Figure 7.2. and Figure 7.3. as well as Table D.41). (It should be noted that the percentages mentioned in this paragraph are different from those in Table 7.1. as the table shows average arrears ratios only for those jurisdictions that were able to provide the information for the years 2018 to 2023.)
Note: The table shows average arrears ratios for those jurisdictions that were able to provide the information for the years 2018 to 2023. The number of jurisdictions for which data was available is shown in parentheses. Data for Bulgaria was excluded from the calculation as its data is not comparable across the period. Data for Kenya was excluded as the significant fluctuations would distort the averages.
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Table D.41 Arrears ratios: Closing stock and collectable arrears, http://isoradata.org (accessed on 1 October 2025).
When looking at the data over the period from 2018 to 2023, a small decrease of 1.5 percentage points in the average ratio of total year-end arrears to net revenue collected is visible, see Table 7.1. The noticeable increase of the ratio during 2020 was the result of the pandemic when many governments took action to support individuals and businesses by extending payment terms, or by suspending collection of outstanding debt (CIAT/IOTA/OECD, 2020[7]).
Table 7.2. shows the 10-year evolution of the ratio for total year-end arrears to net revenue collected. On average, the ratio has decreased by around 10% or 2.7 percentage points. The jurisdiction level data confirms this, with two-thirds of the jurisdictions showing decreasing ratios. (It should be noted that the table only takes into account information from jurisdictions that were able to provide data for both years 2014 and 2023, which explains the differences in 2023 averages shown in Table 7.1. and Table 7.2.)
Note: The table shows average arrears ratios for those jurisdictions that were able to provide the information for the years 2014 and 2023. The number of jurisdictions for which data was available is shown in parentheses. Data for Bulgaria and India was excluded from the calculation as their data is not comparable across the period.
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Table D.41 Arrears ratios: Closing stock and collectable arrears, http://isoradata.org (accessed on 1 October 2025), and OECD (2017), Tax Administration 2017: Comparative Information on OECD and Other Advanced and Emerging Economies, Table A.11, https://doi.org/10.1787/tax_admin-2017-en.
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Table D.41 Arrears ratios: Closing stock and collectable arrears, http://isoradata.org (accessed on 1 October 2025).
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Table D.41 Arrears ratios: Closing stock and collectable arrears, http://isoradata.org (accessed on 1 October 2025).
Looking at collectable tax arrears, the 2023 data shows that on average 55% of the total arrears are considered collectable. That is an increase of 10% or 5.2 percentage points to 2018. (See Table 7.1.) However, Figure 7.4. illustrates well the differences between jurisdictions: in some jurisdictions almost all arrears are considered collectable, while in others almost all arrears are considered not collectable.
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Table D.41 Arrears ratios: Closing stock and collectable arrears, http://isoradata.org (accessed on 1 October 2025).
Figure 7.5. shows the change of total year-end arrears between 2022 and 2023. In absolute numbers, the total year-end arrears increased in 35 out of 52 jurisdictions that were able to provide the information. (Note: This does not contradict the above observation that arrears ratios are decreasing. While in absolute numbers, arrears are going up in many jurisdictions, the ‘total arrears to net revenue collected’ ratio is decreasing as total revenue collections have increased even more.)
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Table D.42 Arrears ratios: Year-on-year change, http://isoradata.org (accessed on 1 October 2025).
Table 7.3. looks at the amount of arrears for the main tax types. In 2023, the average ratio of CIT arrears to CIT net revenue collected and the ratio for VAT are around 20% and 23%, respectively. Over the years, they have followed a similar pattern, including a significant increase in 2020 confirming the difficulties that businesses encountered during the pandemic. Since then, they are back to pre-pandemic levels.
At the same time, the ratio for PIT is noticeably lower. At 15% it remained steady over the 2018 to 2023 period. At around 7%, the ratio is the lowest for PAYE. However, this is expected, as employers are responsible for forwarding those taxes to the administration on behalf of their employees and have no right over the amounts.
Note: The table shows the average ratios for jurisdictions that were able to provide the information for the years 2018 to 2023. The number of jurisdictions for which data was available is shown in parentheses. Data for Bulgaria was excluded from the calculation of the average for the total year-end arrears as a percentage of net revenue collected as its data is not comparable across the period. Further, because they would distort the averages, data for Brazil and Greece was excluded in the calculation of the average for CIT and data for Malta was excluded in the calculation of the average for VAT.
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Tables D.44 Arrears in relation to collection by tax type: CIT and PIT, and D.45 Arrears in relation to collection by tax type: PAYE and VAT, http://isoradata.org (accessed on 1 October 2025).
Figure 7.6. takes a different look at year-end arrears to net revenue collected by illustrating the distribution of individual jurisdiction data for the main tax types. The boxes depict the 2nd and 3rd quartile (i.e. the central 50% of jurisdictions) with the middle line indicating the median value. This shows that:
For all tax types, the median values are significantly lower (around half) than the averages shown in Table 7.3; around, 7% for PIT and 2% for PAYE;
The ratios for CIT and VAT are similar with a median of around 10% and a greater distribution than the ratios for PIT and PAYE.
(It should be noted that the figure does not show the outliers that are above 100%.)
Note: The figure does not show the outliers above 100% even though those data points were used for the underlying calculations.
Source: ADB, CIAT, IOTA, IMF, OECD, International Survey on Revenue Administration, Tables D.44 Arrears in relation to collection by tax type: CIT and PIT, and D.45 Arrears in relation to collection by tax type: PAYE and VAT, http://isoradata.org (accessed on 1 October 2025).
References
[7] CIAT/IOTA/OECD (2020), “Tax administration responses to COVID-19: Measures taken to support taxpayers”, OECD Policy Responses to Coronavirus (COVID-19), OECD Publishing, Paris, https://doi.org/10.1787/adc84188-en.
[5] OECD (2025), Tax Debt Management Maturity Model, 2025 Edition, OECD, Paris, https://www.oecd.org/content/dam/oecd/en/topics/policy-issues/tax-administration/tax-debt-management-maturity-model-2025-edition.pdf (accessed on 1 October 2025).
[1] OECD (2024), Tax Administration 2024: Comparative Information on OECD and other Advanced and Emerging Economies, OECD Publishing, Paris, https://doi.org/10.1787/2d5fba9c-en.
[4] OECD (2019), Successful Tax Debt Management: Measuring Maturity and Supporting Change, OECD Publishing, Paris, https://doi.org/10.1787/e8fdb816-en.
[2] OECD (2019), Tax Debt Management Maturity Model, OECD, Paris, https://www.oecd.org/content/dam/oecd/en/topics/policy-issues/tax-administration/tax-debt-management-maturity-model.pdf (accessed on 1 October 2025).
[8] OECD (2017), Tax Administration 2017: Comparative Information on OECD and Other Advanced and Emerging Economies, OECD Publishing, Paris, https://doi.org/10.1787/tax_admin-2017-en.
[6] OECD (2016), Advanced Analytics for Better Tax Administration: Putting Data to Work, OECD Publishing, Paris, https://doi.org/10.1787/9789264256453-en.
[3] OECD (2014), Working Smarter in Tax Debt Management, OECD Publishing, Paris, https://doi.org/10.1787/9789264223257-en.