A fall in the number of enforcement files can have several causes, and none of them on its own describes the situation of households. A file can be discontinued while the claim remains unpaid; a shorter debt-relief period does not say how many people actually enter the process. New proceedings can still be opened while the overall stock is falling, and a lower average per person can be produced by deleting old cases, not by paying them off.
1. Three curves that look like one
Imagine an inter-ministerial meeting on “improving the debt situation”. The finance minister reads a press release from the Chamber of Court Executors and sees that in 2024 more than 700 thousand ongoing enforcement cases disappeared. The justice minister notes that from 1 October 2024 debt relief lasts three years, not five. A mayor from the Ústí region says that in his municipality people still work off the books. A debt counsellor reports a queue that has not shortened. The president of the Chamber insists that discontinuing unsuccessful proceedings is legitimate. A sociologist shows that the number of unique persons in enforcement fell by tens of thousands, not by hundreds of thousands.
At the meeting, the word “improvement” hides several indicators. The finance minister watches the files in the Central Register of Enforcement (CEE), justice watches the length of the insolvency process, the mayor the return to a legal wage, the counsellor the capacity of services, the Chamber the state of the register, and the sociologist the number of people instead of the number of proceedings.
Each of these indicators can be correct, but their meanings must not be swapped. A year-on-year fall in files is not evidence that household incomes have risen. It can come from repayment, discontinuation for lack of assets, insolvency, deletion of low-value or unsuccessful proceedings, an extraordinary legal scheme, or a change in the new inflow.
Before we interpret the fall, we have to name the unit: a file, a person, a claim, or a completed debt relief. Each series answers a different question, and their year-on-year changes cannot be placed on one axis without conversion.
A file is not a debtor. A debtor is not discharged. Debt relief is not a return to a legal wage. And a fall in the register is not repayment.
— Jiný Kontext
What we know so far about three-year debt relief
We know: from October 2024 the standard three-year period applies, and in 2025 the number of authorised debt-relief cases rose.[5][8][9] We do not know: the first cohort under the new rules could not yet finish the full three-year cycle in 2026, so we have neither its final discharge rate nor a reliable figure for a return to insured employment. The number of authorised entries must therefore not be presented as the number of people successfully discharged.
2. The fall in files can happen in five different ways
The “number of enforcement cases” column is the stock of the register, not the sum of debts paid. It falls by several routes, and two equally large year-on-year changes can therefore mean different things for households.
Five layers of one number
- Claim: a legal duty to pay a debt. It exists whether or not enforcement is running.
- Enforcement title and application: a state-authorised attempt to recover a specific claim by force.
- Record in the CEE: an entry for a proceeding, including proceedings interrupted by insolvency or deferred.
- Discontinuation or fulfilment: the file ends because something was recovered, or because recovery stopped.
- Debt relief and discharge: a court process at the end of which the debtor may be exempted from paying claims that remain unsatisfied.
The claim determines what the debtor owes. It is not the only actor. The court executor can discontinue the file under Section 55 of the Enforcement Code without the creditor receiving the principal. The insolvency court can suspend recovery even though the CEE still lists the person. After the conditions of debt relief are met, the court may exempt the debtor from paying claims that remain unsatisfied; as a rule the claim does not legally disappear, but it cannot ordinarily be enforced against the discharged debtor. A new application can put the same person back into the statistics. A self-employed natural person can sit in a different column from an employee.
This is where the first error of public comparisons appears. The headline announces a fall in enforcement cases, but in reality it compares the number of files with the number of people, or the number of people with the number of principals recovered. An honest comparison must state not only the year, but also the unit, the population, whether it is the stock as of 31 December or the flow of newly opened proceedings, and whether the CEE includes latent files.
The reason for the fall has to be traced in the data that follow. A fall in files without a similar fall in persons may be linked to the discontinuation of old proceedings; a fall in persons without figures on debt relief does not show what happened to them next. Even a higher number of debt-relief cases does not, by itself, prove a return to insured work. The headline “enforcement cases are down” therefore needs a breakdown by reason and by unit.
Illustrative diagram: The same person can, in another column, with another title and another way of ending, produce a different fall.
3. Statistics must also show why the case ended
Readers usually look for the indicator with the largest year-on-year fall. That is reasonable, but incomplete. Equally important is what the error looks like, whether it can be recognised, how much it costs to correct, and whether the result is reversible. Discontinuing a fruitless file relieves the court. It is a cheap error if we read it as housekeeping. It is an expensive error if we read it as people getting richer. A recovered principal can be checked in the yields. A deletion for lack of assets will not appear in the yields.
In its press release of 17 July 2025 the Chamber of Court Executors itself says that the fall is linked mainly to the discontinuation of unsuccessful enforcement cases, not to an improving financial situation of the population. That is a professional body’s view of its own data, not an audit of principals recovered. It does show, however, that even an exact percentage can measure something other than what the headline reads.[1]
| Type of failure | What it looks like | How to test it | What limits the harm |
|---|---|---|---|
| Confusing the unit | A fall in files is read as a fall in people | Compare the stock of proceedings and the stock of unique persons in the same period | A mandatory pair of numbers in every release |
| Confusing flow and stock | Fewer running files are read as fewer new debts | Separate newly opened proceedings from the registered remainder | Publish both series side by side |
| Confusing an ending with fulfilment | Discontinuation is read as payment | Ask how many endings recovered the principal | A statement of yields, not only the number of closed files |
| Confusing the register with activity | An entry in the CEE is read as a wage deduction in progress | Separate latent files (insolvency, deferral) | A methodological note on every table |
In the assessment it is not the largest fall that decides, but its reason. Deleting an unsuccessful file can be legitimate management of the register; it must not be presented as satisfaction of the creditor. Likewise, a lower average of proceedings per person does not prove that people have returned to a legal payslip.
For each series it is therefore important to know why it ended, whether the person returned to a legal income, and the effect on contributions. A smaller year-on-year fall, accompanied by the share of claims actually recovered, can carry more information for public policy than a large deletion without that structure.
4. The number of files does not describe the state of households
The Chamber’s statistics are needed because they replace part of impression with a repeatable register. The problem begins when one number starts to stand for the whole health of households. The Chamber publishes the stock as of 31 December and as of mid-year, the number of newly opened proceedings, the average per person and several frequency bands. In the methodological notes on the open data it also recalls that the CEE includes latent enforcement cases that cannot be carried out because of insolvency.[2][3][22]
Every table contains hidden decisions. Does it cover natural persons including the self-employed, or legal persons as well? Is it a stock or a flow? Is a file counted when debt relief is already running? Is the district of permanent residence measured, or the seat of the executor? The answer to any of these questions can change the order on the map.
What must accompany a year-on-year fall
- The exact unit: a proceeding, or a unique natural person.
- The moment of measurement and whether it is a stock or a flow of newly opened proceedings.
- Whether the series includes latent files in insolvency.
- The reason it ended: recovery, discontinuation for lack of assets, a low-value case, debt relief.
- Regional and age distribution, not only the national average.
An overall ranking of “improvement” tells us less than a table with a precisely defined unit. A smaller audit sample of files with a reason for ending and the debtor’s later status can reveal what the aggregate hides; without a representative sample, however, it does not estimate the whole system.
5. Compare proceedings, not headlines
The market in interpretations can be described without a false podium of winners. The map below does not say which report is generally first. It shows the object around which each reading builds its argument, what kind of context it brings, and which failure the reader should look for while reading.
| Reading | Main object | What to check | Typical reason for the choice |
|---|---|---|---|
| Headline about a fall in files | Stock of proceedings in the CEE | Share of discontinuations vs. recoveries | Quick proof that “enforcement cases are down” |
| Headline about a fall in persons | Unique natural persons | Pace relative to files, band 1 vs. 10+ | Proof that the population of debtors is shrinking |
| Headline about debt relief | Authorised debt-relief cases in the period | Comparison with the target group, not only year-on-year | Proof that a second chance has opened |
| Headline about the grey economy | Contributions and declared wages | Whether it is an estimate or an administrative series | Proof that the deduction is pushing work off the books |
| Headline about the region | District share of persons in enforcement | Age, one vs. multiple, availability of advice | Proof that the Czech average says nothing about Most |
The table does not judge the quality of the institution or of the register as a whole. It recalls that the individual series have different definitions and that an amendment can change their meaning. The reader can then check whether the headline matches the actual unit and the reason for the change.
How to read the diagram: The size of the elements is not a ranking of success. The diagram separates the state of the register, the fate of the claim, the number of people and the return to the legal economy.
6. Discontinuation: the file disappears, the claim does not
In 2024, according to the Chamber, more than 910 thousand unsuccessful enforcement cases were discontinued. That is the flow of closed proceedings in a calendar year, not the number of people who stopped owing money. The Chamber links this to a systematic review of cases with no realistic chance of recovery. From the creditor’s point of view it often means the loss of a chance to obtain performance through enforcement, not certainty that the debt has vanished.[1]
The legal regime is narrower than the word “end” sounds. Discontinuation under Section 55 of the Enforcement Code ends the specific enforcement proceeding; by itself it does not automatically mean that the claim has been fulfilled. Whether the claim still exists and can be enforced depends on the reason for discontinuation, limitation, any insolvency proceeding and the other circumstances of the case. Section 55 alone is therefore not enough to assert that the claim always continues to bear interest or can be recovered again in the same way.[7]
In earlier years other tools also helped to clear the archive, and they must not be piled together with 2024. Low-value enforcement cases with a principal of up to CZK 1 500 could be discontinued from 2022; the Chamber then estimated hundreds of thousands of closed proceedings. Milostivé léto (Mercy Summer) allowed public-law debts to be settled by paying the principal without accessories; these were time-limited schemes with a different logic from fruitless recovery. The tools cleared part of the agenda. They lifted different burdens from the debtor: sometimes the accessories, sometimes only the enforcement order.[16][17]
What needs to be tested
The first test concerns the reason for the ending. Of the 910 thousand discontinuations in 2024, how many recovered the principal, how many were for lack of assets, how many for another reason? The public press release does not give that breakdown. The second concerns the person: how many unique people lost all their files, and how many only some of them? The third concerns return: how many discontinued claims reappear at a collection agency or in a new application? Without these three numbers, 910 thousand is an exact figure about office work, not about household debt relief.
For the courts this clean-up can be decisive. An overloaded apparatus would otherwise chase files from which nothing can be taken. For public debate, however, the sentence “enforcement cases are down” is not enough. It needs the type of ending, the owner of the remaining claim, and the line beyond which a person may return to a payslip without the deduction taking the entire remainder.
7. Multiple cases are not the same as fewer debtors
The core of the Czech debt agenda is not a person with one unpaid fine. The core is concurrence. That is a reasonable reading of the data, but only if we keep the definition.
According to the Chamber, as of 31 December 2024 some 614 721 natural persons, including the self-employed, faced enforcement. A year earlier the figure was 645 724. The fall is 31 003 persons, less than five percent. In the same period the number of proceedings fell from 4 059 035 to 3 350 486. The average per person fell from 6.28 to 5.43. Persons with one enforcement case numbered 143 587, almost the same as in 2023 (143 614). Persons with ten or more proceedings fell from 123 426 to 83 691; as of 30 June 2025 the Chamber listed 93 139 of them again. These are stocks in the CEE, not the number of people who repaid the debt.[1][3]
PAQ Research reports that at the beginning of August 2025 about 450 thousand people — roughly five percent of residents aged 15 and over — were in multiple enforcement, defined as at least two proceedings. Almost 400 thousand persons had at least three enforcement cases; according to the same text the average debt of that group exceeded one million crowns. This is an estimate tied to the Chamber’s open data and to a date other than the year-end stock of 2024. It is not the same number as 614 721 persons with all enforcement cases.[6]
Imagine Martin. This is not a report, but an illustration of the difference between a file and a person. He had eight enforcement cases, and in 2024 the executors discontinued his five oldest unsuccessful proceedings. In the file statistics that is a sharp fall, but three active enforcement cases remained. His deduction cannot be read from gross wages alone: the calculation depends on net income, dependants, priority claims and other statutory parameters. Five files disappeared; from the public total we cannot tell how his motivation for legal employment changed.[23]
It does not follow that the fall in persons is an illusion. It follows that 31 thousand unique people in 2024 is a different story from 708 549 proceedings. The first may be partly debt relief, partly death, partly a move into insolvency, partly deletion of the last file of a person who had ten. The second is mainly work on the archive.
8. The grey economy is a side account, not a footnote
A deduction from a legal wage can reduce part of the extra net income a debtor gains from higher official earnings. It is not true, however, that everything above the basic non-attachable amount is taken automatically: the remainder of the wage is split into thirds up to a statutory ceiling, and only the part above a further set limit is deducted without restriction; the result also depends on the type of claim and on dependants.[23] For some debtors this design can weaken the incentive to declare all income, but the extent of work off the books cannot be inferred from the deduction formula itself.
In a survey from March to May 2025 among 506 people in intractable debt — weighted to represent the population of persons in enforcement according to Chamber data from March 2025 — PAQ Research found that 55% of households in the sample live below the income-poverty line. In the CZSO Living Conditions survey (EU-SILC) for 2024 the at-risk-of-income-poverty rate in the whole population was 9.5%. That is a relative indicator, suitable for comparing groups inside one country, not for an international ranking of poverty. Half of the debtor sample showed screening signs of at least moderate depression or anxiety; in the general population of the same comparison the figure was 17%. Screening is not a medical diagnosis.[6][10]
The estimate of a fiscal shortfall is a weaker layer. In an interview for Czech Radio, Daniel Prokop gave a conservative estimate of a loss of around 12 billion crowns a year in unpaid contributions from people in multiple enforcement. The National Economic Council of the Government worked with an interval of 5 to 12 billion crowns and a population of 140 to 200 thousand people who limit declared income. These are model intervals, not a line in the state final account. The Czech Social Security Administration does not publish a series called “return from the grey economy after discontinuation of fruitless enforcement”. Without that series, a fall in files cannot be read as a rise in insurance contributions.[18][19]
For municipalities in the districts of Most, Ústí nad Labem and Chomutov, which the Chamber repeatedly lists as areas with the highest share of persons in enforcement, this layer is practical. The Czech average will not tell them how many local wages are visible. It will only tell them that their district sits at the top of the map.[1][12]
9. Three-year debt relief: a shorter period is not a shorter path
From 1 October 2024 the amendment to the Insolvency Act No. 252/2024 Coll. is in force. The standard debt-relief period for natural persons fell from five years to three. The amendment implements Directive (EU) 2019/1023 on preventive restructuring and a second chance. It also removed the requirement to repay at least 30% of the debt as an automatic gate to discharge. That is a change of rule. It is not automatically a change of behaviour.[8][9][11]
InsolCentrum calculated that from 1 October 2024 to 31 March 2025 some 12 041 natural persons entered insolvency proceedings, both self-employed and not. In 80% of cases the courts had authorised debt relief by the cut-off. Six months after the amendment of 1 June 2019, some 21 646 natural persons entered. The difference is 44%. This compares two six-month windows after two different amendments; it is not proof that three-year debt relief “does not work”. It shows that a shorter period by itself did not produce the rush that part of the debate expected.[4]
For the whole of 2025, according to the same source, the courts authorised debt relief for 22 770 persons, 13% more year-on-year. There were 25 336 insolvency applications, 9% more year-on-year. As of 31 December 2025 some 90 950 persons were in debt relief. In earlier years, according to PAQ, about 20 thousand persons a year entered debt relief. The figure 22 770 is an increase, not coverage of a group of hundreds of thousands with multiple enforcement cases.[5][6]
How to read the insolvency rise
The first test is the denominator. The 22 770 authorised debt-relief cases in 2025 should be divided by the population for which the tool is meant, not by the population of all residents. The second is a comparison of windows: 2019 had a different starting line, different information and a different 30% threshold. The third is the result at the end of three years, which we do not yet have in 2026. Authorised debt relief is not discharge. Discharge is not an insured job.
After the amendment there was therefore no jump comparable to the first half-year after the 2019 change; for the whole of 2025, however, the number of authorised debt-relief cases rose year-on-year. From these two comparisons one can say that the length of the process is not the only barrier to entry. One cannot tell from them how many people will in the end complete three-year debt relief successfully, nor whether the amendment will reduce the number of persons in enforcement in the long run.
10. Information about debt relief is not the same as entry
In the same spring 2025 survey PAQ found that 45% of those in intractable debt do not know the key reason to enter debt relief: that at the end of a successful process the remaining unpaid debt is forgiven. More than 40% do not even roughly know the basic rules, or have not heard of debt relief. Two thirds do not know that from October 2024 the 30% condition no longer applies. Two thirds fear that they would not manage the instalments in the long run. 57% do not know that non-profit organisations offer to prepare the application free of charge. This is a sample of 506 respondents, not a census of all 614 721 persons in the CEE.[6]
Willingness is higher than intention. More than 60% of the sample declared a hypothetical interest. Only 14% thought entry in the next year very likely. Fifteen percent of those interested said they would not manage even the minimum instalment. 76% said they would let an automatically prepared application be filed; some of them would still fail the conditions. This is not proof that people “do not want to work”. It is the result of one weighted survey of informational, income and psychological barriers, not a measurement of everyone in enforcement.[6]
Debt counselling was used by 22% of the sample, the internet as a source of information by 46%, and family or acquaintances by 37%.[6] A shortening of the period that does not pass through these channels may remain, for some people, only a change in the Collection of Laws. Insolvency courts also assess whether the debtor meets statutory duties; a shorter period does not cancel that check. A change of length alone therefore cannot be judged on the assumption that it will automatically bring everyone with several enforcement cases into debt relief.
11. The non-attachable amount is not a switch for work
Wage deduction is designed so that the debtor keeps a legally protected part of income. Its amount changes with legal parameters, net wages, dependants and the type of claim. After the non-attachable amount is subtracted, part of the remainder is split into thirds up to a set ceiling; only the part above the threshold of full attachability is deducted without restriction. A higher official wage therefore usually raises the deduction, but the simple rule “everything above the minimum goes to creditors” does not apply.[23]
Protection of the non-attachable amount prevents the deduction from taking the entire income. At the same time, in some income situations it can reduce the net gain from a higher official wage. Whether a particular person then moves into work off the books also depends on the size of the debt, the type of claims, the family, the job offer and the risk of a sanction; such behaviour cannot be predicted from the formula itself. The Ombudsman returns repeatedly to the practice of deductions and to the reading of the non-attachable minimum; that is oversight of application, not a substitute for insolvency.[14][23]
Before assessing the deduction
Find out how many concurrent proceedings per person remain after the archive has been cleared. Separate maintenance — in 2024 the Chamber recorded 4 400 newly opened enforcement cases for maintenance, of which 3 780 were for a minor child — from consumer debt. Maintenance has a different legitimacy and a different policy. Do not mix it with a microloan. And do not judge only the size of the deduction: record whether after it there is still a reason to sign a contract.[1]
CNB supervision of consumer credit targets providers and new credit contracts; by itself it does not delete old files in the CEE. The Responsible Lending Index of People in Need evaluates the present supply of loans, not the stock of historical enforcement cases. Both layers are useful, but they measure a different part of the problem.[15][20]
12. How to read numbers from press releases
Before the Chamber’s open data, the reader had to trust the summary. With the table the reader checks something harder: whether the column measures proceedings or persons, whether the period fits, whether latent files are sitting in the average, and whether the conclusion really follows from the cell to which it is attached. The number is a navigation to the evidence. It is not automatic evidence.
A fall of more than 700 thousand proceedings in 2024, a fall of 31 thousand registered persons and 12 041 entries into insolvency in the six months after the amendment describe three different movements. The summary conclusion “Czechia has been discharged” does not follow from them without figures on the reasons for ending. The half-year flow of 256 479 newly opened proceedings in the first half of 2025 is also not proof that the new inflow has stopped: the annual number was 460 573 in 2024 and 461 592 in 2023. The difference between those two years looks more like stagnation at a high level.[1]
A four-step check of the number
- Existence: the figure is in a primary table of the Chamber, InsolCentrum or the CZSO.
- Unit: a proceeding, a person, an application, an authorised debt-relief case.
- Support: a specific cell supports a specific sentence.
- Scope: the population, period, definition and the caveat about latent files all fit.
The most reliable procedure therefore separates searching from writing. First comes a table of claims, primary sources and limits. Only then the headline. If the reason why 910 thousand discontinuations ended cannot be opened, the claim is not upgraded to “paid off” merely because the number is large.
Three tests of the fall
The first is about the unit: did proceedings fall, or persons? The second is about the reason: did it fall by recovery, or by discontinuation? The third is about contradiction: in the same window, did newly opened proceedings or entries into debt relief rise in a way that would explain the fall differently? A product that announces the fall a day earlier need not be better if a month later no one can find out why it fell in that particular way.
13. The price of a file is the price of the raw material, not the price of the result
Justice uses several cost languages at once. The executor bills the costs of the proceeding. The creditor writes off the principal. The state counts contributions that did not arrive. The municipality counts benefits and lost purchasing power. The debtor counts what remains after the deduction. A direct comparison of “how many files disappeared” without knowledge of these accounts misleads.
Cheap deletion of a fruitless file can be economically useful for an overloaded office. The same deletion can be expensive if it is read as proof that counselling is no longer needed. Debt relief with an insolvency administrator can be disproportionate for one manageable claim. It can be useful for a person with many proceedings if the statutory conditions are met and a stable legal income becomes possible.
The real price of an accepted result
Clearing the archive + recovered principal + contributions from a declared wage + the cost of counselling + the price of a misread statistic + future new proceedings.
The result is cheap only when it passes a check of the unit and can be used to decide. Not when one column falls.
The costs also include fluctuation. If the same person, after five files are discontinued, returns with the third that remains, the average price of a “resolved debtor” is not the price of one deletion. If the system exhausts counselling capacity at the peak, the cost is waiting. If automatic conclusions are tied tightly to a press release, the cost is also a later correction.
It is therefore useful to measure the price per accepted output, not the price per file. Accepted means that a person’s status has changed — repayment, discharge, an insured job — and that this can be verified. That metric can reverse a ranking that looked clear from a year-on-year percentage. An IDEA study at CERGE-EI on the link between enforcement and voting behaviour in municipalities in 2001–2017 recalls that a file has costs outside the courts; it is an analysis up to 2017, not a verdict on the 2024 data.[21]
14. What to choose as an indicator
Before reading the headline it is necessary to choose the question and the matching control series. The table below does not rank “truths”; it shows pairs of indicators that must be compared in the same period, population and definition of an ending.
| Need | First series | Decisive test | Unacceptable shortcut |
|---|---|---|---|
| Whether the courts are being cleared | Stock and flow of proceedings in the CEE | Share of discontinuations vs. recoveries | Celebrate every fall in files |
| Whether there are fewer debtors | Unique natural persons | Pace relative to files, band 1 and 10+ | Mistake the average per person for a fall in people |
| Whether a second chance is opening | Authorised debt-relief cases per year | Denominator: the target group, not the whole CR | Read +13% as the end of the crisis |
| Whether wages are returning to the books | Contributions, insured jobs | Linking the person after discontinuation to a wage | Rely on an estimate of billions without a series |
| Whether the region is diverging | District map | Age, multiple cases, availability of advice | The national average as a local diagnosis |
| Whether the market is not feeding new debts | Newly opened proceedings, CNB supervision | The credit segment, not only the total | Declare stagnation the end of the inflow |
Maintenance, public-law debt and consumer credit need separate readings in the Czech context; one CEE number does not join their mechanisms. The check should include different districts and persons with one proceeding and with several, and it must allow the conclusion that the public source is not enough for the question asked.
One indicator can track the archive, another people, debt relief and contributions. That division is useful if the definitions are clear, the series are not added without conversion, and the closing sentence states which of them it rests on.
15. A short audit that says more than a year-on-year percentage
A short reader’s audit can show, before publication, whether one column can carry the intended headline. It looks for a swap of file for person, stock for flow and discontinuation for repayment, and it asks whether the newsroom or the office can correct it.
Fourteen days from impression to table
- Days 1–2: Write 20 to 40 real questions: file, person, new application, debt relief, contribution, district.
- Days 3–4: Prepare the same period, the same population and an anonymous scoring sheet.
- Days 5–7: Open each important series in the primary source and keep a register of swapped units.
- Days 8–10: Check latent files, the reason for ending, age and the frequency band.
- Days 11–12: Calculate how much of the fall you can assign to recovery, discontinuation and insolvency.
- Day 13: Run stress tests: contradictory series, an incomplete reason for ending, an old district average.
- Day 14: Decide which column may enter the headline, document the caveat and set a date for the next review.
If possible the assessor should not know which narrative the number is meant to support. For files, watch the reason for ending. For persons, the pace relative to proceedings. For debt relief, the denominator. For an estimate of the grey economy, whether an administrative series exists. Each failure is classified: a confabulation of the unit, an omission of latent files, an error of period, or a conclusion the cell did not carry.
The series should be compared to the same cut-off day and on the same population. If one includes latent files and the other does not, it is a comparison of different statistics, and that limit must be stated.
The pilot should also have exclusion conditions. For example: the single sentence “people got richer” without a share of recovery, the use of a half-year flow as an annual stock, or the impossibility of opening the CEE methodology. A high year-on-year percentage must not override an error whose impact the reader cannot accept.
The result of the audit should be a rule of use: which unit may enter the headline, what does not follow from it, who checks the methodology and when the conclusion is revised. Amendments, deduction tariffs and the definition of the CEE change, so every comparison must have a date.
16. A new inflow is not the same as the old stock
In 2023 some 461 592 enforcement cases were opened, in 2024 460 573. The difference is 1 019 proceedings. In the first half of 2025 the Chamber recorded 256 479 opened proceedings. That is a flow. It is not proof that the stock of 3.16 million files as of 30 June 2025 is “only dying off”. The stock can fall by deletion and at the same time fill with new applications at a pace that, for the whole of 2024, stayed above 450 thousand.[1]
Some of the new proceedings are maintenance, some public-law claims, some consumer credit. Without that cut one cannot say whether CNB regulation “worked”, or whether only the creditor changed. The Indebtedness Map of the Institute for the Prevention and Resolution of Over-indebtedness, which continues the older Map of Enforcement, shows the distribution down to municipalities. A nationwide flow will not replace it. A municipality with a high share of persons in enforcement may have a low flow of new applications, or the reverse. Those are two different policies: clearing the stock and braking the inflow.[12]
The average age of a debtor in the Chamber’s data rose from 46.2 years in 2023 to 46.8 in 2024 and to 47 years in the first half of 2025. That is the average of registered persons, not proof that poverty is ageing. Without an age pyramid of newly opened proceedings we cannot tell whether younger people enter the CEE less often, or older people remain in the stock.[1]
Stock and inflow should therefore be taken on deliberately: both series, a cut by creditor, and a test of whether the reason for the fall in the stock still holds when the inflow has not fallen.
17. A fall in the register does not yet change the debtor’s decision
The first generation of debt amendments sold the discontinuation of low-value cases. The next sold Mercy Summer. The next sold three-year debt relief. Today’s press releases sell something larger: the impression that the CEE curve is the state of society.
The contest is therefore not only about the best clean-up of files. It is about the debtor’s working context. About whether the person will sign a contract, whether the deduction leaves a reason for a declared wage, whether a counselling office can be found, and whether after five proceedings are deleted three remain — enough for the person to behave the same. The more comfortable it is to read one column, the higher the price of a mistake can be. Automatic narratives are hard to export back into the methodology.
Clearing the archive can have a substantial judicial benefit; it must not be mistaken for repayment of debts. Assessment needs a stable definition of the unit, a documented reason for ending, and a way to separate the person from the file.
A useful indicator is not the one with the largest fall or the highest number of discontinuations. Its unit must match the question asked, and any mistakes should be visible, correctable and acknowledged in the headline.
The moment the CEE only counts files, we are cleaning a register. The moment we assemble from the fall a report about people getting richer, about a return to work and about the end of debts, we are also assembling a new structure of responsibility for reading. And then the question is no longer: “By how much did enforcement cases fall?”
Before assessing the fall it is therefore necessary to find out: Which column fell, for what reason, who will see it in time, and who will bear the cost if we read a file as a resolved life?
— Jiný Kontext