Defaults at the end of the general payment moratorium

Debt service obligations of the contracts subject to the moratorium were temporarily suspended, which also ruled out the possibility of becoming delinquent.

However, the accounting rules continued to require credit institutions to classify contracts into different categories (stages) for loan loss provisioning purposes, depending on the foreseeable future loss they may incur in relation to the contracts.

They could also assign a non-performing status if they had reasonable grounds to believe that, without the protection of the moratorium, the debtor would be unlikely to pay. The delinquency of clients that entered into moratorium with pre-existing delinquency remained unchanged for the duration of the moratorium and could only increase after exiting the programme.

In this paper, we consider the performing classification (performing vs. non-performing) of credit institutions as the main indicator of payment difficulties. In our view, this rating makes the most accurate use of the wide range of relevant circumstances, as credit institutions seek to use a variety of information in the rating process, including information that is not available to outsiders.

One of the possible alternatives is the extent of delinquency. This is not used because delinquency per se is insensitive to other relevant elements of payment difficulties, such as the size of delinquent amount. Another possibility could be some version of probability of default, but such a probability is difficult to define accurately, and the credit registry does not reliably contain such data for all institutions. Nevertheless, the non-performing classification has the disadvantage that a loan can be removed from the non-performing status even if the debtor’s solvency has not actually improved (for example, by selling the loan). We do not have good enough data to identify such outflows, but we try to mitigate their impact. Therefore, for any loan maturing after September 2022 that was missing a September 2022 non-performing classification, we impute the classification for each of the months missing until September that was contained in the last data observed in a previous month.¹² The change does not substantially alter the results of the regression analysis.

Among all contracts existing in October 2021, the ratio of non-performing loans jumped from 2.8 per cent at the end of the general payment moratorium in October to 4.0 per cent in November, and then rose slightly further (Figure 4, left panel).

The increase was mostly related not to contracts that left the general moratorium but to those that remained in the moratorium. In November, banks classified

performing, up from 9 per cent in October. This was presumably due to the fact that the rules had extended the programme only for vulnerable groups, and that they had to apply for it, which may have indicated poorer solvency. The non-performing ratio excluding those who remained in the moratorium barely increased after the general moratorium (2.4 per cent in October 2021 and 2.9 per cent in September 2022) and thus remained much lower than for those opting for the conditional moratorium.¹³ The non-performing stock in this group was around HUF 200 billion in the months after the end of the general moratorium, half of which was delinquent beyond 90 days. Behind this broadly unchanged stock over time, there was a larger inflow and outflow in 11 months than in the 19 months of the general moratorium (Figure 4, right panel). This suggests that a significant amount of meaningful additional information may have been used in the non-performing classifications after the general moratorium ended.

Figure 4

Non-performing household loan portfolio by delinquency and migration

HUF billions Per cent

Jun 2021 Jul Aug Sep Oct Nov Dec Jan 2022 Feb Mar Apr May Jun Jul Aug Sep

2 3 4

0 1 5

200 300 400

0 100 500

NPLs staying in moratorium NPLs leaving mor., 61–90 days due NPLs leaving mor., 1–30 days due NPL ratio, every loan (RHS)

NPLs leaving mor., 90+ days due NPLs leaving mor., 31–60 days due NPLs leaving mor., 0 day due NPL ratio, loans leaving mor. (RHS)

HUF billions

Mar 2020 – Oct 2021 Oct 2021 – Sep 2022

-50 0 50 100 150 200 250

Inflow Unchanged stock Outflow 55

144 124 100

-19 -46

Note: The left panel shows the volume and the share of non-performing loans within the outstanding debt at the given date of loans existing also in October 2021. The right panel shows the transitions between March 2020 and October 2021 and between October 2021 and September 2022 of the stock of non-performing loans within the loans that left the moratorium in October 2021 at the latest. It takes into account only the loans with observable non-performance classifications both at the beginning and at the end of the given period, and it calculates with the outstanding debt in October 2021.

Both the typical levels of non-performing ratios and their evolution around the end of the general payment moratorium differ significantly depending on whether and, in particular, how loans have previously participated in the moratorium.

Interestingly, the non-performing ratio among those that did not participate in the moratorium and those that exited the general moratorium before its end were similarly low, between 1.5 and 2.0 per cent around the end of the programme (Figure 5, left panel). The non-performing ratio was much higher among those that dropped out of the general moratorium in October 2021. This group is so overrepresented in the stock of non-performing loans that it accounts for more than half of it (Figure 5, right panel).¹⁴

Figure 5

Non-performing household loan portfolio by delinquency and participation in the general payment moratorium

Per cent Per cent

Jun 2021

Jul Aug Sep Oct Nov Dec

Jan 2022

Feb Mar Apr May Jun Jul Aug Sep 2 3 4

0 1 5 6 7 8 9

2 3 4

0 1 5 6 7 8 9

Dropped out at the end: 90+ days delinquency Dropped out at the end: 0–90 days delinquency Voluntarily left: 90+ days delinquency Voluntarily left: 0–90 days delinquency Never in moratorium: 90+ days delinquency Never in moratorium: 0–90 days delinquency

Every loan

Sep 2022 Nonperf.

loans

Dropped out at the end Voluntarily left Never in moratorium

19%

53% 16%

9%

38%

65%

Note: Only household contracts existing in October 2021 are shown. Dropped out at the end: They dropped out of the moratorium at the end of October 2021. Voluntarily left: They left the moratorium before October 2021. Never in moratorium: They have never been in moratorium. In the left panel, ratios of non-performing loans are shown within outstanding debts at the given date, the non-delinquent but non-performing stock is classified as “0–90 days delinquency”. The right panel calculates with

The positive correlation between the intensity of participation in the moratorium and subsequent non-performance is also observed at the district level. The correlation coefficient is relatively high: 47 per cent (Figure 6). In larger cities, moratorium intensity and non-performance rates in September 2022 are also typically among the lower ones. At the other extreme are the least urbanised districts of the south-western and eastern part of the country, where both indicators typically take high values. It is also noticeable that in almost all of the country’s north-western districts, the ratio of non-performing loans is typically relatively low.

Figure 6

Participation in the general payment moratorium and subsequent ratio of non-per-forming household loan stock by district

Moratorium intensity, %

NPL ratio, %

35.8 38.2 44.3

2.83.76.8

44 13 9

15 26 25

7 27 31

Note: The map does not take into account household loan contracts existing in October 2021 that remain in moratorium in November, nor does it take into account those contracts where the primary borrower could have applied the moratorium on a total instalment of more than HUF 20 million (around EUR 56,000 at the time) on all loans during the general moratorium. The horizontal axis of the legend shows moratorium intensity aggregated at district level. The vertical axis shows the non-performing share of the district-level outstanding debt in September 2022. The numbers on the axes are the tercile values separating each category and the maximum. The squares contain the number of districts in each category.