Katharina Allinger ¹
Elias Farnleitner

JEL classification: G28, G21, G51
Keywords: loan moratoria, household finance, COVID-19, policy evaluation, Central-, Eastern- and Southeastern Europe

During the COVID-19 pandemic, loan repayment moratoria for households (subsequently referred to as “moratoria”) were one of the relief measures implemented in many countries. These moratoria were largely complementary to other measures aimed at preventing household liquidity crunches and subsequent solvency issues. Studying the effectiveness of the relief measures taken to achieve this aim is central for policymakers and has therefore received most of the attention in the literature.

Our paper has a somewhat different aim, however, which has mostly been neglected in the existing literature: We study how the COVID-19 borrower relief measures implemented in nine Central-, Eastern- and Southeastern European (CESEE) countries ² were perceived by the public. In this context, we mostly focus on the following variables: awareness of the measures, usage of the measures and reasons for not using them.

All three aspects are important for different reasons and should concern policymakers: Being aware of a measure is clearly a prerequisite for being able to use it. Awareness can even matter for people who are not eligible for the measures, as this might affect their trust, expectations and, subsequently, decision-making. Regarding the usage of measures, we are most interested in the reasons people give for not using them, as this indicates whether people understood the measures correctly and thought they were eligible. In annex C, we also study how people attributed the measures to different institutions, which could be related to take-up. People might be reluctant to use measures offered by institutions they do not trust.

We analyze these topics along two dimensions: First, we study how different designs of moratoria are related to people’s perceptions and use of moratoria. We exploit the fact that while the objectives of the moratoria were largely the same in the countries covered, they were implemented very differently. Second, we analyze which observable characteristics of individuals help explain the variation in perceptions and take-up.

For our analyses, we use survey data collected in fall 2021. Based on these data, we can shed light on what individuals thought about the different aspects of the moratoria.

In our cross-country comparisons, we find marked differences in our variables of interest. Regarding people’s awareness of moratoria, we find that awareness is relatively low in some countries, even among the target group of borrowers. When put in the context of different moratorium designs, we conclude that, above all, awareness and use of moratoria are strongly related to design features. Moreover, the reasons given for not using moratoria vary across countries and show no clear pattern when analyzed against moratorium designs. In most countries, having “no need” for taking up moratoria was the answer given most frequently. In three countries, “not being eligible” was mentioned by even more respondents.

Finally, we study the correlation of our variables of interest with socioeconomic characteristics for the pooled sample and for each country separately. We find that despite some heterogeneity across countries, the patterns of correlations are relatively similar. Awareness increases with several characteristics, mostly related to respondents’ ownership of financial assets and liabilities as well as their level of education and financial knowledge. In some countries, there is a high degree of variation in awareness across regions, while in others, the shares are relatively similar across regions. For moratorium use, the most important factor was whether respondents were financially affected by the COVID-19 pandemic.

Our study belongs to the literature assessing borrower relief programs, which largely consists of studies on the effects of borrower relief on debt distress, debt taking, consumption and employment on a household or regional level (Agarwal et al., 2017, 2023; Cherry et al., 2021; Dobbie and Song, 2020; Dinerstein et al., 2023; Piskorski and Seru, 2021; Giné and Kanz, 2018; Kanz, 2016; Mukherjee et al., 2018; Fiorin et al., 2023). Recent studies related to the COVID-19 pandemic and CESEE include, e.g., the study by Aczél et al. (2023) who show that participation in moratoria in Hungary is correlated with subsequent defaults. Cesnak et al. (2023) use survey data for indebted households in Slovakia to study which households used the moratorium and how it impacted their finances. An earlier paper using data from the OeNB Euro Survey by Allinger and Beckmann (2021) finds that individuals who had exited moratorium programs by fall 2020 were not more likely to be in arrears with loan repayments than individuals who had not used these programs.

While these papers provide crucial evidence on the effectiveness of borrower relief measures, there are few papers on potential frictions in the transmission of such measures on the borrower side, such as low awareness, difficulties in understanding the measures and non-monetary costs. Johnson et al. (2019) combine administrative and survey data to study motives for not accepting refinancing offers of a US borrower relief program. They find that suspicion toward refinancing offers is significantly related to take-up, as is awareness of the offer and perceived eligibility. Allen et al. (2022) investigate two COVID-19 debt relief programs in Canada and find that take-up was low. They report that this is partially due to people’s low awareness of the programs. Jacob et al. (2023) complement their study on debt relief for US teachers with evidence from focus groups which suggests that administrative barriers and program complexity hindered take-up.

We are not aware of any papers that present evidence on public perceptions of COVID-19 loan moratoria in the CESEE region or borrower relief programs in a cross-country setting, linking public perceptions to the design of the policies. Thus, our study fills a gap in the literature. Moreover, it is very topical in the context of high inflation and interest rates, as some household finances are under pressure and a renewal of loan repayment moratoria has been discussed.

The study is structured as follows: In section 1, we provide a review of the designs of moratoria implemented in the CESEE countries and of the guidelines issued by the European Banking Authority (EBA) on moratoria. In section 2, we briefly discuss the data and methodology we use. In section 3, we present our data analysis, shedding light on people’s awareness and use of moratoria from a cross-country perspective. Section 4 focuses on a within-country perspective, using socioeconomic characteristics and geographic data. Section 5 summarizes and provides some policy conclusions.

1 Implementation of moratoria

This section first outlines what is meant by EBA-compliant moratoria and then proceeds to compare the designs of moratoria implemented in nine CESEE countries during the COVID-19 pandemic. We mostly use national sources, complemented with information collected by the EBA (2020d). Our task is complicated by the fact that the characteristics of moratoria changed in most countries over the course of the pandemic. Moreover, in some countries, several moratorium schemes existed in parallel, applying different conditions. On top of that, banks could always negotiate with clients bilaterally. Thus, the characteristics of moratoria could differ drastically even within a given country. ³

1.1 EBA guidelines on legislative and non-legislative moratoria

On April 2, 2020, the EBA published its guidelines on legislative and non-legislative moratoria on loan repayments applied in the light of the COVID-19 crisis (subsequently referred to as “EBA GL”; European Banking Authority, 2020a). The EBA GL set out the conditions for legislative and non-legislative general payment moratoria, which did not automatically trigger a reclassification of the exposure as forborne (in accordance with Article 47b of the Capital Requirements Regulation (CRR)) or defaulted (Article 178 of the CRR). These general payment moratoria stood in contrast to the usual regulatory forbearance approach, asking banks to carefully assess each borrower’s situation and tailor forbearance measures to the borrower. In fact, the COVID-19 moratoria had to be sufficiently broad in terms of both the participating creditors and the borrowers. The EBA GL thus excluded initiatives designed and implemented by a single bank, as well as solutions tailored to individual clients. The conditions offered by EBA-compliant moratoria needed to be the same for the same type of borrower or exposure. Thus, different conditions could only be specified for groups of borrowers or products, e.g. for mortgage loans. Only the payment schedule should be affected by the moratorium, while other terms (e.g. the contractually agreed interest rate) should remain unaffected. Contracts concluded after the start of the COVID-19 pandemic were not eligible.

The application deadline for moratoria under the EBA GL was extended twice. After the deadline had first been extended from June to September 30, 2020, the EBA decided in September 2020 to suspend its GL. However, due to the second COVID-19 wave, the GL were re-activated in early December 2020 and the application deadline was set to the end of March 2021. An additional condition was introduced, specifying that loan repayments could be deferred for a maximum of 9 months for the moratorium to remain compliant with the GL (European Banking Authority, 2020b, 2020c).

1.2 Moratoria in CESEE

Most CESEE EU countries modeled their moratoria at least partially on the EBA GL. However, compliance varied across countries and over time. Moreover, even while adhering to the EBA GL, there was substantial room for variation in the design of moratoria. We summarized some of the most important characteristics of, and differences between, moratoria in table B4 in annex B. In the subsequent paragraphs, we discuss some of the more distinctive features of the moratoria across countries.

Certainly, two of the more important distinctions were, first, whether respondents had to apply for, i.e., opt in to the moratorium or, second, whether the moratorium applied automatically unless clients actively opted out (or simply continued to make their loan repayments). Besides being more convenient for borrowers, opt-out moratoria were available to all borrowers. Opt-in moratoria in CESEE were mostly tied, directly or indirectly, to whether borrowers’ finances were affected by the COVID-19 pandemic.

Another distinction was whether moratoria were based on legal documents issued by governments, central banks or regulatory authorities, thus constituting public moratoria (see column 3 in table B4 in annex B), or whether they were based on private agreements, e.g., between members of banking associations. Public moratoria usually implied that participation was compulsory for banks and that any conditions of the moratorium outlined in legal texts or guidelines were followed closely, as they were legally binding. The latter is difficult to verify in retrospect and without insights into banks’ practices. However, the Polish central bank noted that “banks in Poland have not developed a uniform standard of loan moratoria. As a result, borrowers face various conditions on the suspension of loan repayment depending on the lending bank” (Narodowy Bank Polski, 2020, box 4.1.). This seems to support the theory that in the case of private moratoria, as in Poland, banks had more leeway when implementing the measures.

Along these two dimensions, the CESEE countries were split almost evenly. Three countries – Hungary, North Macedonia and Serbia – had public and, at least partially, opt-out moratoria. Another three countries – Czechia, Romania and Bosnia and Herzegovina – implemented public opt-in borrower relief programs. Finally, the policies in Bulgaria, Croatia and Poland can best be characterized as private and opt-in policies. ⁴

Of all the moratoria, the one in Hungary had the most generous terms, as it applied for a very long time and was changed from an opt-out to an opt-in moratorium relatively late. The Hungarian central bank was quite critical of the many blanket extensions of the moratorium granted by the government. Only from November 1, 2021, onward were the conditions of the moratorium tightened so that only specific groups (e.g. retirees, families with children) remained eligible. Overall, the moratorium applied until end-2022 (Ministry of Justice, 2020a, 2020b; Magyar Nemzeti Bank, 2021, 2022a, 2022b).

In Czechia, on the other hand, the government applied some of the tightest conditions among the CESEE EU countries by explicitly excluding revolving products and setting a comparatively early end-date for moratorium use, namely on October 31, 2020 (Act No. 177/2020, 2020).

Regarding private moratoria in Bulgaria, Croatia and Poland, these were largely established with strong involvement of the respective banking associations. Given their non-legislative nature, these moratoria were largely voluntary for banks, but information by the EBA suggests that in all three countries (almost) all banks participated. Bulgaria and Croatia definitely saw active involvement of their central banks. The Bulgarian central bank outlined the conditions of the moratorium on April 10, 2020 (Bulgarian National Bank, 2020), and these were then adopted by the Bulgarian banking association. In Croatia, the central bank sent several Circular Letters to the banks regarding the application of the EBA GL (Hrvatska narodna banka, 2020a, 2020b, 2020c, 2020d). According to the Polish banking association, the latter agreed on the moratorium with the Polish government (ZBP, 2020). ⁵

In the CESEE EU candidate countries, a special feature was that borrower relief was defined more broadly than just loan moratoria. In Serbia, for instance, the second part of the borrower relief program from mid-December 2020 onward required clients to opt in and was tied to eligibility criteria, i.e., to whether clients were negatively financially affected by the COVID-19 pandemic. Moreover, banks could choose from several options how to help borrowers in need (Narodna banka Srbije, 2020a, 2020b, 2020c). In North Macedonia, borrower relief generally included two offers made to clients (one in March and one in September 2020), providing for favorable changes in loan terms. The conditions of these changes were determined by the banks (National Bank of the Republic of North Macedonia, 2020a, 2020b). In Bosnia and Herzegovina, the banking agencies of the two entities adopted decisions in March 2020, establishing a temporary moratorium. The latter was intended to apply only until the end of the state of emergency (i.e. until May 2020) and mostly served to give banks and clients time to work out the right medium-term modalities for repayment. The decisions also detailed all modalities available, including the option to defer repayments for a maximum of six months. In August/September 2020, the banking agencies extended the application deadline for moratoria and other relief measures outlined in the decisions until end-2020, effectively allowing loan postponements until mid-2021 at the latest (ABRS, 2020a, 2020b; FBA, 2020a, 2020b; UBBIH, 2020).

2 Data and methodology

This section discusses the data and methodology used. It describes the construction of a design index for moratoria as well as key features of the data.

2.1 Constructing a design index for moratoria

The information contained in table B4 in annex B simplifies the complexity of COVID-19 moratoria. However, the information is still too detailed for further use in the paper, which is why we select three key characteristics from table B4 to construct a simple numeric index that captures certain design features of the moratoria discussed: i) the scope of eligible borrowers (opt-in/opt-out moratoria); ii) the binding nature of the moratoria (public/private); and iii) the duration of the moratoria. We chose these characteristics, as they seem to be good proxies for how generous the moratoria were for borrowers. The calculation of the index is shown in table 1. The results are displayed in the first panel of chart 1.

All other characteristics that we could have used to create more differentiation in the index across countries presented us with the following issues: The information available was incomplete across countries; the criteria were too unique and/or minor (e.g. only one country would get a score of 1 versus 0 for all other countries based on a minor aspect); or the criteria were collinear with characteristics already contained in the index. For instance, the latter would apply for eligibility criteria related to COVID-19, as these criteria existed in all opt-in countries, but not in the opt-out countries. In our opinion, information gathered through expert interviews with policymakers and bankers in the region would be needed to markedly improve the index.

Table 1

Design index for moratoria in CESEE  

	BG	HR	CZ	HU	PL	RO	BA	MK	RS
Opt-out (1)/opt-in (0)	0	0	0	1	0	0	0	1	0.5
Public (1)/private (0)	0	0	1	1	0.5	1	1	1	1
Maximum duration	0.5	0.5	0	1	0	0.5	0	0	0
Sum	0.5	0.5	1	3	0.5	1.5	1	2	1.5
Source: Authors’ compilation based on information provided by the EBA as well as various national competent authorities and banking associations. Note: Opt-in moratoria refer to moratoria for which borrowers needed to apply. Opt-out moratoria applied automatically unless borrowers opted out. Serbia has a score of 0.5, as borrowers had to opt out of the initial moratorium and opt in to its extension in 2020. Public moratoria refer to moratoria established by law, ordinances or decisions issued by governments, central banks or other financial authorities. Poland has a score of 0.5, as it had rather limited public and much broader private moratoria. The maximum duration refers to the date when the last moratoria expired and is judged relative to the EBA GL (maximum duration until December 31, 2021). Moratoria that were in place longer get 1 point, those in place shorter get 0 points. Moratoria in place for as long as indicated in the EBA GL receive 0.5 points.

Thus, the design index clearly contains many assumptions that have implications for our conclusions. However, instead of viewing the index as a perfect representation of how generous moratoria were in the countries, we consider it a necessary and helpful tool for subsequent analyses using publicly available information on moratorium designs. We provide some robustness checks in annex B.

2.2 The OeNB Euro Survey and module on borrower relief measures

The remainder of the paper uses data from the 2021 wave of the OeNB Euro Survey. ⁶ The OeNB Euro Survey is an annual survey among individuals in ten CESEE economies that has been conducted since 2007. The countries included in the survey are 6 EU member states, namely Bulgaria (BG), Czechia (CZ), Croatia (HR), Hungary (HU), Poland (PL) and Romania (RO) as well as four EU candidate countries, namely Albania (AL), Bosnia and Herzegovina (BA), North Macedonia (MK) and Serbia (RS). The sample for each OeNB Euro Survey wave consists of 1,000 randomly selected individuals per country and is designed to represent the adult population with respect to gender, age and regional distribution. Due to issues with data quality in Albania, the country is excluded from this study (Olbrich et al., 2024).

The OeNB Euro Survey wave conducted in October 2021 included a module on borrower relief during the COVID-19 pandemic. In this study, we present results for a couple of questions from the module (for more details, see table A1 in annex A):

• Awareness: “Are you aware of any measures your government or banks in [YOUR COUNTRY] adopted because of the pandemic to support borrowers (for example enabling borrowers to postpone repayments without penalties, offering borrowers favorable changes in loan terms)?”

• Usage: “Since the beginning of the COVID-19 pandemic, have you taken advantage of any measures that were adopted to support borrowers?”

• Reasons for non-usage: “Could you tell us why you didn’t make use of the measures? Please mention all reasons that apply.” Answer options: see table A1 in annex A.

Except for usage, the aspects listed above cannot be studied without survey data. However, survey data have some caveats: Given the international dimension of the survey, we needed to find a term suitable for all countries covered. As discussed in section 1, the EU candidate countries in our sample allowed for borrower support to take different forms. We therefore settled on the term “borrower relief” rather than “moratoria” for the survey module. Thus, using the terms “moratoria” and “borrower relief” interchangeably throughout the paper is not entirely precise in the case of the candidate countries. Moreover, while the OeNB Euro Survey is designed to represent the adult population in the surveyed countries, missing data and the fact that we occasionally work with quite small subsamples mean that we need to be careful when trying to interpret our findings for the entire population of a given country or subsamples of that population. For instance, given the lack of statistics on debtor characteristics for the respective countries, we cannot check the representativeness of our debtor sample or correct for imbalances ex post. ⁷ This is why we focus on the entire population of a given country rather than on subsamples of that population, wherever possible.

2.3 Methodology for cross-country and within-country analyses

In section 3, we present descriptive results for our questions on awareness , usage and non-usage of moratoria and discuss differences across countries. ⁸ While the cross-country heterogeneity is already interesting in itself, we hypothesize that policy design matters. We expect a positive correlation, meaning that the more borrower-friendly a measure, the higher people’s awareness and usage of the measure and the lower the share of people who did not use the measure because they were not eligible.

In section 4, we use a large set of available variables on respondents’ socioeconomic characteristics, preferences and beliefs to shed some light on within-country differences. We define binary dependent variables for each of our main questions of interest and estimate the following model(s) with probit regressions:

P(yi = 1) = Φ(βXi + εi)

where, depending on the model, P(y i = 1) stands for the probability that the respondent i is aware of borrower relief programs, or used the programs. X is a vector of explanatory variables and ε is an error term. Standard errors are clustered at the level of the primary sampling unit (PSU), which refers to a selected starting point for the random route of the interviewer. This level is chosen given the sampling design of the survey (Abadie et al., 2023; Cameron and Miller, 2015). Moreover, within-PSU correlation is likely, given potential interviewer and network effects. ⁹

Given the different dependent variables, we have different samples for each regression: for awareness, all respondents that answered “yes” or “no” to the corresponding question; for usage, all respondents with bank or nonbank loans or revolving debt, such as overdraft or credit card debt (subsequently referred to as “borrowers”).

With the exception of “having debt,” we use the same explanatory variables in the probit estimations for awareness, attribution and usage to facilitate comparisons. We use theoretical considerations and statistical methods and choose the following variables: having debt/loans, planning to take out a loan in the next 12 months, having no savings, owning investment products, having been negatively financially affected by the COVID-19 pandemic, living in the capital city, trust in banks and trust in the government, education, financial literacy, income ¹⁰ and employment status. Moreover, we also add further socioeconomic control variables that are not shown in the coefficient plots in section 4, namely age, gender, being married and household size. All pooled regressions contain country dummies. Correlations between the explanatory variables are rather low (see table A3 in annex A), as are variance inflation factors for the regressions shown in this study.

The main aim of the simple regressions is to provide some sense of the correlations between socioeconomic characteristics and awareness and usage of moratoria, respectively. The variables are selected to test different hypothesis for each dependent variable. We outline these hypotheses before presenting the results in section 4. We do not claim causality in the results we report, given the shortcomings of our design index as well as the fact that we cannot control for all relevant variables, e.g. different media landscapes/coverage or political factors.

3 Cross-country variation by moratorium design

In this section, we focus on cross-country variations in respondents’ average awareness and use of moratoria as well as their reasons for not using them, considering the different design features of moratoria.

The upper panel of chart 1 shows the results of our simple design index listed from highest to lowest value. Hungary stands out with the maximum value, followed by North Macedonia, which also had a public, opt-out moratorium. Serbia is on a par with Romania according to our index. So are Czechia and Bosnia and Herzegovina, both with a score of 1, while Bulgaria, Croatia and Poland come in last with a low average score of 0.5. This ranking of the countries is maintained in the middle and lower panels of chart 1, enabling us to see at first glance that while there is strong variation across countries, there seems to be at least some correlation between the design features of moratoria and respondents’ awareness and use of the latter. The Pearson correlation coefficients between the design index and the means of awareness and usage are 0.87 and 0.90, while the Spearman rank correlation coefficients are similar but slightly smaller.

3.1 Awareness of moratoria

At roughly 70%, the share of respondents aware of borrower relief measures is by far the highest in Hungary. Hungary is followed by the other two opt-out countries, namely North Macedonia and Serbia, both with shares of over 50%. In Czechia and Romania, the shares come to around 40%. In the remaining countries, the shares are close to or below 35%, with a low of 20% in Bosnia and Herzegovina. ¹¹ Since the relief measures are targeted at borrowers, we also plot the shares of borrowers aware of the measures in red. While these shares are higher in all countries and reach almost 100% in Hungary, they remain rather low in the last four countries displayed, i.e. in Bosnia and Herzegovina, Bulgaria, Croatia and Poland. If we consider an even smaller subgroup, namely borrowers affected by the COVID-19 pandemic, the results differ markedly across countries. Awareness is actually lower among borrowers affected by the pandemic than among those unaffected in six out of nine countries. Only in Czechia, Bosnia and Herzegovina and North Macedonia, affected debtors are more aware of the measures than unaffected ones (see table A4 in annex A).

In terms of design features, the ranking for awareness comes close to the results obtained for the design index. This suggests that more generous moratorium designs were related to higher awareness in the population. Intuitively, this makes sense, as more generous support measures were probably more present in the media. Also, opt-out moratoria certainly created more awareness among debtors, as most banks likely informed their debtors about the changes in their loan terms. Despite the intuitiveness of the correlation, it seems striking how large the variation between the countries is and how few people in some countries claimed to be aware of the measures taken.

Chart 1

Source: 2021 OeNB Euro Survey wave.
Note: Means are calculated with post-stratification weights. In the lower panel, the number of observations for the borrower subsample is indicated below the bars. The question on awareness was posed to all respondents.

The lack of awareness is potentially concerning from a policy perspective. While there may be good reasons to have tight eligibility criteria for a borrower support program, every borrower should at least be aware of the existence of the program to assess whether they are eligible and want to use it. If awareness is very low, potentially interested borrowers might not have been able to benefit from the measure, as they were simply not aware of it. A complementary explanation might be that since very low awareness mostly concerns private moratoria, the communication of such measures might have been different: They might have been communicated less through official channels and the media, or they might have simply not been communicated and noticed as broad-based policy measures related to COVID-19. After all, borrowers in difficulty can always discuss restructuring their loan with their banks.

3.2 Use of moratoria

The lower panel in chart 1 shows the use of borrower relief measures among borrowers in CESEE. Given the low absolute number of moratorium users in some countries, the means are subject to considerable uncertainty.

We see similar patterns as for awareness and an even higher correlation with the design index. Reported use of borrower relief measures was by far the highest in the opt-out countries, starting with around 55% of debtors in Hungary and around 30% in North Macedonia and Serbia. In Czechia, almost 20% of debtors reported using the relief programs, followed by around 10% in Poland. In the remaining countries, less than 10% of debtors in our sample used the moratoria. ¹²

We asked respondents who did not use the relief programs for the reasons behind not using them. Since we only asked debtors who were aware of the programs, we are left with few observations, ranging from 57 in Bosnia and Herzegovina to 188 in Croatia. Moreover, respondents could give more than one answer, even though the vast majority of respondents chose just one option. Keeping these caveats in mind, we nonetheless found some interesting cross-country similarities and dissimilarities evident from chart 2.

Chart 2

Source: 2021 OeNB Euro Survey wave.
Note: Countries are plotted in descending order based on the design index. The number of observations for each country are as follows: Hungary (N = 154), North Macedonia (N = 132), Serbia (N = 142), Romania (N = 118), Bosnia and Herzegovina (N = 57), Czechia (N = 137), Bulgaria (N = 71), Croatia (N = 188), Poland (N = 79).

In most countries, respondents not using the moratoria most often stated that they had no financial need to do so. In Bulgaria, Croatia, Czechia, Hungary and Romania, 50% or more of non-users gave this answer. In the remaining countries, around 35%–40% mentioned this reason. Moreover, non-users frequently stated that they were not eligible for the moratoria. In most countries, the shares of non-users mentioning eligibility ranged between 20% and 30%. However, in three countries, this answer was chosen most often, namely in Poland, Bosnia and Herzegovina and Serbia. It is interesting to note that Serbia is among these countries, given that Serbia initially had an opt-out moratorium. We can break down eligibility further to differentiate between debt type, criteria related to the COVID-19 pandemic and other eligibility criteria. In the case of Poland, Bosnia and Herzegovina and Serbia, the high shares of ineligible borrowers are largely due to respondents stating that the types of debt they held were not eligible for a moratorium. This is somewhat puzzling, as our reading of the design features of moratoria suggests that their debt types would have been eligible.

However, respondents might indeed not have been eligible if, e.g., they were in arrears on their loan in March 2020 or had taken out their loan after March 2020. We can also not exclude that respondents accidentally or deliberately gave false answers, not wanting to state the true reasons. The most worrying possible interpretation from a policy perspective is that respondents might have erroneously thought that they were not eligible. This could point to suboptimal communication by policymakers or banks. The data suggest that other borrowers potentially wanted to use the moratoria but were prevented from doing so due to the eligibility criteria defined or their interpretation of these criteria.

Having sorted the countries in chart 2 in descending order based on the design index, we find that there is no clear visual pattern based on moratorium design features for either “not eligible” or “not needed.” The computed correlation coefficients for eligibility are –0.44 (Pearson) and –0.54 (Spearman), indicating that a higher design index is associated with lower shares of respondents concerned about eligibility. However, the correlation is not significant. For “not needed,” the computed correlation coefficients are both around –0.1 and highly insignificant.

Finally, in most countries, around 5%–10% of respondents who knew about the moratoria but did not use them mentioned the complexity of the related application process. Particularly in North Macedonia, people also seemed to worry about their credit score, which deterred them from using the moratoria. In Croatia, Hungary, North Macedonia and Serbia, almost 20% of respondents also listed other (not specified) reasons.

4 Within-country variation by region and socioeconomic variables

In this section, we discuss the within-country variation in respondents’ awareness and use of borrower relief measures both with regressions using socioeconomic variables and, in the case of awareness, regional variation. Results for respondents’ attribution of the measures are reported in annex C.

4.1 Awareness of moratoria

With respect to awareness, we formulate several hypotheses about some of the variables we selected for our probit model, while other variables are primarily included as control variables and will therefore not be discussed in detail. We study the awareness of the entire population instead of just debtors for two main reasons: First, awareness of policy measures may have effects on debtors’ and non-debtors’ overall financial behavior. Those aware are potentially more likely to expect future bailouts by the government, which might alter their risk-taking behavior. Second, this allows us to exploit the full, representative population sample, which gives our statistical analyses more power. This is particularly relevant for the country regressions.

We assume that the following variables have a positive correlation with respondents’ awareness: i) having or planning to take out loans, as it is likely that debtors pay more attention to, and have a different stake in, borrower relief programs than non-debtors. Moreover, they may even have received personalized information from their banks, particularly in opt-out countries; ii) higher level of education and financial literacy, as both likely make it easier for respondents to understand financial policy measures and assess their usefulness and implications; iii) being negatively affected by the COVID-19 pandemic, as this may give respondents an incentive to be more aware of available support measures; and iv) living in the capital city, as this is usually where policies are decided in the CESEE countries and may therefore lead to increased awareness.

We believe that other variables of financial inclusion and sophistication, such has having no savings or owning investment products, are likely also important. However, we are uncertain about the expected direction of the effects. Both savings and investment products may, on the one hand, be an indicator of wealth and thus of the need for support measures. On the other hand, these variables may also be an indicator of financial inclusion and thus of being aware of developments in finance and banking in general.

Chart 3 shows the average marginal effects of several probit regressions. ¹³ The results of a pooled regression are shown in dark blue in addition to the results of country-specific regressions. We can clearly see that the magnitude and significance of the estimates varies across countries. Despite this heterogeneity, some common patterns can be identified.

For the variables that capture having loans, planning to take out loans and having higher levels of education and financial literacy, we find that they are strongly, positively and significantly related to awareness in almost all regressions. With respect to owning investment products or having no savings, the financial inclusion effect seems to dominate the wealth effect. Having no savings is associated with lower awareness of borrower relief measures, and owning investment products with higher awareness – again, this holds for most countries.

Interestingly, if respondents’ personal finances were negatively impacted by the COVID-19 pandemic, awareness levels were higher in the pooled regressions. However, the marginal effects appear relatively modest. Moreover, in country-specific regressions, these effects are mostly insignificant. Thus, the results for our initial hypothesis that COVID-19 affectedness correlates with awareness are mixed.

Chart 3

Source: 2021 OeNB Euro Survey wave.
Note: Dependent variable = 1 if respondent is aware of borrower relief measures. Average marginal effects from a probit model estimated by maximum likelihood. Standard errors are clustered at the PSU level. Full opacity means p-value of t-test < 0.1. Variables not shown include log(Age), Female (0/1), Married (0/1), Income: NA, Size of household.

Figure 1

Source: 2021 OeNB Euro Survey wave.
Note: NUTS 2, except for Bosnia and Herzegovina, where regions are defined according to Hijmans (2015). Please refer to table E1 for the numeric values and see also figure E1 in annex E.

Finally, we find evidence that respondents’ awareness is indeed significantly higher in a few country capitals and in the pooled sample, with the exception of Hungary, where awareness is lower in the capital. We cannot say, however, whether this is truly because of the proximity to policymakers, as we hypothesized, or some other, unobserved characteristic of respondents living in the capital city. ¹⁴

Related to this, we also show the geographic distribution of respondents’ awareness by country. In figure 1, we present the percentage points difference between the mean of a given NUTS 2 region and the mean of the corresponding country. The scale ranges from –40 to +40 percentage points, indicating considerable within-country fluctuations in awareness in some countries. Countries with an overall lighter, more transparent shade (e.g. Bosnia and Herzegovina, Hungary or Serbia) show less pronounced differentiation around their country mean than those with darker shades (e.g. Czechia, Romania or Poland). Figure E1 in annex E shows respondents’ awareness as predicted by our pooled probit model. Looking at both figures helps us better understand whether regional differences in awareness are due to observed or unobserved factors. In some countries (e.g. Bulgaria, Croatia), the figures point to similarities, suggesting that the observed socioeconomic characteristics can explain a large portion of the variation. In other countries (e.g. Poland, especially its eastern parts), the difference between the two figures is striking. Theoretically, there are many potential confounding factors, including media coverage or social networks, for which we cannot control and which might vary in importance across regions.

4.2 Use of moratoria

Regarding usage, our main hypothesis is that we expect to find broadly similar results to those found by Allinger and Beckmann (2021). In this study, the authors used a different question on moratoria included in the 2020 OeNB Euro Survey wave to assess socioeconomic determinants of moratorium use and the prevalence of arrears. Most socioeconomic control variables used by Allinger and Beckmann (2021) were insignificant in a pooled regression on moratorium use, pointing to the fact that usage was relatively broadly distributed among loan holders. However, several variables associated with the negative financial impact of the COVID-19 pandemic and with having no savings were significant. This makes sense given the larger need for support measures and the conditionality of moratorium programs in many countries.

Despite relying on a different survey wave and question, we find similar results in our current study compared to Allinger and Beckmann (2021). In chart 4 in the pooled regression, very few coefficients are significant. Being negatively financially affected by the COVID-19 pandemic increased the use of moratoria, which is not surprising. The coefficient on the capital city is also significant. Both variables have significant and positive coefficients in three country regressions.

Chart 4

Source: 2021 OeNB Euro Survey wave.
Note: Dependent variable = 1 if respondent took advantage of borrower relief measures. Average marginal effects from a probit model estimated by maximum likelihood. Standard errors are clustered at the PSU level. Full opacity means p-value of t-test < 0.1. Variables not shown include log(Age), Female (0/1), Married (0/1), Income: NA, Size of household.

5 Conclusions

This study compares moratorium designs across nine CESEE countries and uses survey data to analyze how certain aspects of borrower relief programs were perceived by the public. For this purpose, survey data are an excellent source, as they can shed light on individuals’ decision-making processes – something that loan-level data available to banks and financial authorities cannot do.

We find large heterogeneity across countries in respondents’ awareness of borrower relief measures, their attribution of the measures to different institutions, their use of the measures and their reasons for not using them. Regarding awareness, we find that in some countries, large shares of the overall population and almost all borrowers were aware of the relief measures put in place. In several other countries, however, less than 50% of respondents knew about the relief measures – even when considering the subsample of borrowers only. This could be a cause for concern, as awareness of a policy measure is a requirement for being able to decide whether or not to use it (e.g. Allen et al., 2022). Our findings suggest that awareness was higher in countries with a higher calculated design index for moratoria, which is our gross proxy for how borrower-friendly the implemented measures were. Particularly in countries with very low awareness, the public might not have perceived the implemented moratoria as different to the status quo (of bilaterally negotiating loan restructurings with banks), or banks and authorities may have provided (too) little information regarding the policy measures.

When looking at within-country variation in respondents’ awareness, we find relatively similar patterns across countries. Socioeconomic characteristics that proxy financial inclusion and sophistication (e.g. owning investment products) as well as general education and financial knowledge are strongly positively correlated with awareness. Thus, low financial inclusion or limited knowledge could also have contributed to lower aggregate awareness. Awareness also differed quite strongly across the NUTS 2 regions within some countries.

Regarding usage, we find a large dispersion across countries that is highly correlated with the moratorium design index. The more borrower-friendly the design of moratoria, the higher their usage. When looking at the correlations with socioeconomic characteristics, having been negatively financially affected by the COVID-19 pandemic seems to be the most important correlation. This makes sense given that this was one of the conditions tied to moratoria in many countries. When asking borrowers about why they did not use relief measures, they most often stated that they did not have a financial need to do so or that they were not eligible. There is some differentiation between countries regarding which of the two reasons was mentioned more often. However, these cross-country differences do not correlate with the moratorium design index. The fact that in several countries, the shares of non-users mentioning eligibility as an issue were quite high (above 50%), raises some concerns as to whether borrowers might have misunderstood the eligibility criteria defined by authorities and banks.

Overall, our study provides novel insights into differences in moratorium designs coupled with public perceptions of these moratoria. The findings should be evaluated jointly with studies on other aspects of moratoria, most importantly their effectiveness in preventing unnecessary defaults due to liquidity crunches. For the country sample covered in this study, evidence on loan arrears can be found in Allinger and Beckmann (2021). More work on the effectiveness and potential moral hazard implications of moratoria is envisaged based on the OeNB Euro Survey module used in this paper.

References

Abadie, A., Athey, S., Imbens, G. W. and Wooldridge, J. M. 2023. When should you adjust standard errors for clustering? The Quarterly Journal of Economics, 138(1):1–35.

ABRS. 2020a. Odluka o privremenim mjerama bankama za ublažavanje negativnih ekonomskih posljedica uzrokovanih virusnim oboljenjem “COVID-19.” Official Gazette of the Republika Srpska, 27.

ABRS. 2020b. Odluka o privremenim mjerama bankama za ublažavanje negativnih ekonomskih posljedica uzrokovanih virusnim oboljenjem “COVID-19.” Official Gazette of the Republika Srpska, 89.

Act No. 177/2020. 2020. Col. on some measures in the area of loan repayment in connection with the COVID-19 pandemic.

Aczél, Á., El-Meouch, N. M., Lakos, G. and Spéder, B. 2023. Household loan repayment difficulties after the payment moratorium: Hungarian experience from the COVID-19 pandemic. Financial and Economic Review, 22(1):21–56.

Agarwal, S., Amromin, G., Ben-David, I., Chomsisengphet, S., Piskorski, T. and Seru, A. 2017. Policy intervention in debt renegotiation: Evidence from the home affordable modification program. Journal of Political Economy, 125(3):654–712.

Agarwal, S., Amromin, G., Chomsisengphet, S., Landvoigt, T., Piskorski, T., Seru, A. and Yao, V. 2023. Mortgage refinancing, consumer spending, and competition: Evidence from the home affordable refinance program. The Review of Economic Studies, 90(2):499–537.

Allen, J., Clark, R., Li, S. and Vincent, N. 2022. Debt-relief programs and money left on the table: Evidence from Canada’s response to COVID-19. Canadian Journal of Economics, 55:9–53.

Allinger, K. and Beckmann, E. 2021. Use of loan moratoria by CESEE households: who are the users and how vulnerable are they? Focus on European Economic Integration, Q3/21:7–33.

Blinder, A. S., Ehrmann, M., De Haan, J. and Jansen, D.-J. 2022. Central bank communication with the general public: Promise or false hope?

Bulgarian National Bank. 2020. Procedure for deferral and settlement of liabilities payable to banks and their subsidiaries – financial institutions in relation to the state of emergency enforced by the national assembly on 13 March 2020 as a result of the COVID-19 pandemic. BNB Press Release.

Cameron, A. C. and Miller, D. L. 2015. A practitioner’s guide to cluster-robust inference. Journal of Human Resources, 50(2):317–372.

Cesnak, M., Cupak, A., Klacso, J. and Šuster, M. 2023. The impact of loan payment deferral on indebted households and financial stability during the COVID-19 pandemic: Survey results from Slovakia. Finance a Uver: Czech Journal of Economics & Finance, 73(3):220 –250.

Cherry, S. F., Jiang, E. X., Matvos, G., Piskorski, T. and Seru, A. 2021. Government and private household debt relief during COVID-19. Working Paper 28357, National Bureau of Economic Research.

Dinerstein, M., Yannelis, C. and Chen, C.-T. 2023. Debt moratoria: Evidence from student loan forbearance. American Economic Review: Insights (forthcoming).

Dobbie, W. and Song, J. 2020. Targeted debt relief and the origins of financial distress: Experimental evidence from distressed credit card borrowers. The American Economic Review, 110(4):984–1018.

European Banking Authority. 2020a. EBA/GL/2020/02 – Guidelines on legislative and non-legislative payment moratoria on loan repayments applied in the light of the COVID-19 crisis. Technical report, European Banking Authority.

European Banking Authority. 2020b. EBA/GL/2020/08 – Guidelines amending guidelines EBA/GL/2020/02. Technical report, European Banking Authority.

European Banking Authority. 2020c. EBA/GL/2020/15 – Guidelines amending guidelines EBA/GL/2020/02. Technical report, European Banking Authority.

European Banking Authority. 2020d. Notifications on general payment moratoria.

FBA. 2020a. Odluka o privremenim mjerama koje banka primjenjuje za oporavak od negativnih ekonomskih posljedica uzrokovanih virusnim oboljenijem “COVID-19.” Official Gazette of the Federation of Bosnia and Herzegovina, 22.

FBA. 2020b. Odluka o privremenim mjerama koje banka primjenjuje za oporavak od negativnih ekonomskih posljedica uzrokovanih virusnim oboljenijem “COVID-19.” Official Gazette of the Federation of Bosnia and Herzegovina, 60.

Fiorin, S., Hall, J. and Kanz, M. 2023. How do borrowers respond to a debt moratorium? Experimental evidence from consumer loans in India. Policy Research Working Paper 10358.

Giné, X. and Kanz, M. 2018. The economic effects of a borrower bailout: evidence from an emerging market. The Review of Financial Studies, 31(5):1752–1783.

Hijmans, R. J. 2015. First-level administrative divisions, Bosnia and Herzegovina, 2015. [shapefile]. Retrieved from: https://earthworks.stanford.edu/catalog/stanford-nj241sn6981 .

Hrvatska narodna banka. 2020a. Addendum to the circular of the Croatian national bank (reg. no. 180-020/19-03-20/bv) dated 19 March 2020. HNB Circular Letter. Reg. No: 348/020/20-7/BV.

Hrvatska narodna banka. 2020b. Adjustments of certain supervisory activities, extension of deadlines for the implementation of supervisory measures and supervisory expectations regarding the classification of exposures in response to coronavirus pandemic. HNB Circular Letter. Reg. No: 180-020/19-03-20/BV.

Hrvatska narodna banka. 2020c. Implementation of EBA guidelines on legislative and non-legislative moratoria on loan repayments applied in the light of the COVID-19 crisis. HNB Circular Letter. Reg. No: 348/020/20-8/MD.

Hrvatska narodna banka. 2020d. Implementation of EBA guidelines on legislative and non-legislative moratoria on loan repayments applied in the light of the COVID-19 crisis. HNB Circular Letter. Reg. No: 348/020/20-10/MD.

Jacob, B., Jones, D. and Keys, B. J. 2023. The value of student debt relief and the role of administrative barriers: Evidence from the teacher loan forgiveness program. Working Paper 31359, National Bureau of Economic Research.

Johnson, E. J., Meier, S. and Toubia, O. 2019. What’s the catch? Suspicion of bank motives and sluggish refinancing. The Review of Financial Studies, 32(2):467–495.

Kanz, M. 2016. What does debt relief do for development? Evidence from India’s bailout for rural households. American Economic Journal: Applied Economics, 8(4):66–99.

Magyar Nemzeti Bank. 2021. Financial stability report June 2021. Financial Stability Reports.

Magyar Nemzeti Bank. 2022a. Financial stability report May 2022. Financial Stability Reports.

Magyar Nemzeti Bank. 2022b. Financial stability report November 2022. Financial Stability

Reports.

Ministry of Justice. 2020a. Government decree 47/2020 (18 March) on immediate measures necessary for alleviating the effects of the coronavirus pandemic on the national economy.

Ministry of Justice. 2020b. Government decree 62/2020 (24 March) on the detailed rules for the payment moratorium.

Mukherjee, S., Subramanian, K. and Tantri, P. 2018. Borrowers distress and debt relief: Evidence from a natural experiment. The Journal of Law and Economics, 61(4):607–635.

Narodna banka Srbije. 2020a. Moratorium on debt payments. Press Release.

Narodna banka Srbije. 2020b. The national bank of Serbia enabled an additional delay in the repayment of debtors’ obligations – a new moratorium (Serbian only). Press Release.

Narodna banka Srbije. 2020c. NBS passes new measures to facilitate repayment to debtors hit by the COVID-19 pandemic. Press Release.

Narodowy Bank Polski. 2020. Financial stability report June 2020. Financial Stability Reports.

National Bank of the Republic of North Macedonia. 2020a. Banks can allow a second period of deferred payment of loan liabilities. National Bank of the Republic of North Macedonia Press Release.

National Bank of the Republic of North Macedonia. 2020b. Decree adopted: Faster and more straightforward implementation of changes to the loan agreement terms for citizens and businesses concerned. National Bank of the Republic of North Macedonia Press Release.

Olbrich, L., Beckman, E. and Sakshaug, J. W. 2024. Multivariate assessment of interviewer effects in a cross-national survey. OeNB Working Paper No. 253.

Piskorski, T. and Seru, A. 2021. Debt relief and slow recovery: A decade after Lehman. Journal of Financial Economics, 141(3):1036–1059.

UBBIH. 2020. Odgovori na najea pitanja klijenata vezana za moratorij na otplatu kredita COVID-19. Udruga Banaka Bosne i Hercegovine (UBBIH) Press Release.

Van der Cruijsen, C., de Haan, J. and van Rooij, M. 2023. The impact of high inflation on trust in national politics and central banks. Working Paper No. 762, De Nederlandsche Bank.

ZBP. 2020. Communique of the Polish Bank Association on aid measures taken by banks in connection with the coronavirus disease (COVID-19) pandemic. ZBP Announcement.

Annexes

The annexes can be found in the pdf version of this study.

---

1. Oesterreichische Nationalbank, Central, Eastern and Southeastern Europe Section, katharina.allinger@oenb.at . Opinions expressed by the authors of studies do not necessarily reflect the official viewpoint of the OeNB or the Eurosystem. The authors would like to thank Julia Wörz and Fabio Rumler (both OeNB) and an anonymous referee for helpful comments. ↩︎

2. Six CESEE EU countries (Bulgaria, Croatia, Czechia, Hungary, Poland, Romania) and three CESEE EU candidate countries (Bosnia and Herzegovina, North Macedonia and Serbia). ↩︎

3. Given these difficulties, information on moratoria had to be collected on a best-effort basis. ↩︎

4. However, there are some cases that are not entirely clear-cut, again speaking to the complexity of characterizing the moratoria. For instance, Poland briefly had a short legislative moratorium, and Serbia switched to an opt-in moratorium already in December 2020. ↩︎

5. From June 24, 2020, onward, there was also a brief legislative moratorium based on Articles 31fa-fc of the Act of 19 June 2020 on interest rate subsidies. The articles set out that borrowers could apply for moratoria of a maximum of 3 months. ↩︎

6. For details, see the OeNB Euro Survey website . ↩︎

7. See annex A for a description of all variables used in this study (including the corresponding questions) as well as summary statistics. ↩︎

8. For these analyses, we use the post-stratification weights of the OeNB Euro Survey calculated based on age, sex, education and region and additional variables in a few countries. ↩︎

9. For robustness, we cluster pooled regressions at a higher level, namely at the level of regions (74 clusters). While the standard errors are higher in this case (see table D2 in annex D), the change is not large enough to affect the graphic results in the main text. ↩︎

10. Income is included as dummies for income terciles and a dummy variable if the answer was “Don’t know” or “No answer,” given high income nonresponse. The results of the pooled regressions barely change when we exclude respondents with missing income information as a robustness check (results available upon request). ↩︎

11. Respondents who stated “Don’t know” or “No answer” are excluded from the total. Poland has by far the highest share of respondents stating “Don’t know” (20%). Higher shares were also reported for Bulgaria and Czechia (12%–14%). When including these respondents as not being aware of borrower relief measures, the shares of respondents aware of these measures would be lower in Poland (26%), Bulgaria (25%) and Czechia (36%). ↩︎

12. Regarding the question on how OeNB Euro Survey data on the use of borrower relief measures compares to data from other sources, we refer to Allinger and Beckmann (2021). In this paper, the authors discuss the difficulty of comparing OeNB Euro Survey usage data with the few other statistics available and provide a table comparing usage data from a variety of sources (see table A3 in annex A). ↩︎

13. For reasons of scope, not all coefficients are shown in the plot, but they are included in table D3 in annex D. ↩︎

14. As a small robustness check whether this is indeed a capital or large-city effect, we additionally add a dummy for large cities (we try cut-offs at 50,000, 75,000 and 100,000 inhabitants, respectively). Each dummy is insignificant in all regressions, while the capital city dummies remain significant (results available upon request). ↩︎