ebrd blog

a look at non-performing loans: the boomerang effect

by: ralph de haas, senior economist

posted on | july 16, 2009 |

authors: ralph de haas and stephan knobloch , 16 july 2009.

when the global financial crisis hit the transition region, worries among policy makers centred on the local banking systems and the potential for financial contagion from west to east. and when unemployment started to rise and output declined sharply as of q4 2008, the attention shifted towards the real-economic impact of the crisis.

now, notwithstanding more frequent discussions about green shoots and a bottoming-out of the crisis, the focus is moving back again to the financial sector. the main question is to what extent the problems of (credit-constrained) households and firms may backfire and lead to a second round of financial-system stress. the development of banks’ non-performing loans (npls) may provide a partial answer to the question of where we are going to end up in 2010.

to this end, this blog takes a look at recent detailed information on the development of npls in 21 of ebrd’s countries of operation. a methodological caveat upfront: widely differing definitions and limited data availability pose serious constraints to this kind of exercise¹.

given these challenges it is advisable to focus the analysis of npl ratios on relative rather than absolute changes. moreover, the npl numbers presented here only present the ‘official’ picture: rating agencies have repeatedly underlined that ‘true’ npls may be substantially higher in a number of countries. an important reason for the sometimes diverging official and unofficial statistics is that some banks pre-emptively restructure or roll-over bad loans. in such cases official npl figures are edging up only slowly. our focus on relative changes can alleviate but not solve this information problem. with that out of the way, a number of interesting patterns emerge:

first, figure 1 reflects the wide variation in npl dynamics across the transition region. roughly speaking there are – as yet – moderate increases in central europe, more pronounced dynamics in south-eastern and eastern europe, and stronger increases in the baltics, russia, central asia and mongolia. while it is too early for a final verdict, there seems to be a negative correlation between the increase in npls and the foreign ownership of local banking systems. latvia would be the main exception to this observation.

figure 1: relative changes of non-performing loan ratios since june 2008

the strongest dynamics are seen in russia, central asia, mongolia, georgia and latvia. in all of these countries npls increased more than two-fold between june 2008 and march 2009. for instance, mongolia’s npl indicator climbed to 3.6 times its june 2008 value, reflecting the liquidity and solvency crisis in the mongolian banking system in the wake of a protracted period of very high inflation, increasingly overindebted borrowers, and tugrug depreciation. also in latvia npls increased more than 3 times, followed by estonia (2.6x), russia (2.4x), kazakhstan (2.4x), and tajikistan (2.2x). georgia is an interesting outlier in the sense that npls doubled immediately after the armed conflict in august 2008 and then continued to increase at a slower pace.

second, and in sharp contrast to the above, the increases in the three central european countries in our sample have so far been much smaller. the hungarian npl ratio fell during the second half of 2008 and showed only a slight uptick in q1 2009. the slovak npl ratio increased to only 1.4 times its june 2008 level. the polish npl ratio has barely moved. the pessimist will note that this reflects a strategy of procrastination among banks as they consistently and persistently roll-over their dubious loans. the optimist, however, will point out that this pattern may also reflect that the upgrading of risk-management systems by foreign parent banks across central europe is bearing fruit in these difficult times. the jury is still out and more information will become available in the coming months.

third, the south-eastern and eastern european countries are sandwiched somewhere between central europe on the one side and the harder-hit countries further east on the other side. countries like fyr macedonia (1.1x), bulgaria (1.4x) and serbia (1.6x) appear to be closer to poland and the slovak republic, while albania, romania, turkey and ukraine saw their npl ratios double.

fourth, npl ratios increased across the board in april and may this year. on the low end, we find (again) poland and the slovak republic with only moderate increases. on the high end, we find a remarkable increase in kazakhstan from 15.2 per cent in april to 29.2 per cent in may, i.e. 5.7 times the june 2008 level. three factors are likely to have contributed to this sharp recent increase.

first, some of the large banks have been in debt restructuring talks and as such may have had to come clean about their portfolio quality. while kazakhstan was hit by the crisis as early as august 2007 npl ratios did not move much for about a year, a period during which some banks rolled over past due loans to both corporate and retail clients. second, the february 2009 tenge devaluation has gradually been feeding through the real economy as unhedged fx borrowers found it difficult to repay their bank loans. third, the increasing npls also reflect further price declines of kazakh real estate, a sector to which the main kazakh banks were overexposed.

fifth, the fate of the kazakh banks reveals a broader relationship between npl increases and real estate developments. we find that collapsing real house prices and relative increases in npls go hand in hand in other countries, too (figure 2).

figure 2: correlation between house price collapses and relative changes in non-performing loans

it is a well-known fact that loan quality lags the business cycle. during good times banks quickly expand their credit portfolios, the age of the average loan is low, and non-performing loans are few and far between. in contrast, during a business cycle downturn or a crisis, the inflow of ‘fresh’ loans is reduced, the average loan portfolio of banks matures and loan problems become increasingly apparent over time. npl ratios increase particularly fast as they combine the effect of weaker loan quality in the numerator with lower loan growth in the denominator. we expect therefore that during the next couple of months, when economic ‘green shoots’ will hopefully become increasingly visible, we may be confronted with the lagged legacy of the 2007-2009 crisis in the form of a further increase in non-performing loans.

1 only half of the countries use 90 days as the threshold after which an overdue loan is considered non-performing, for 6 countries there is no methodological information, and 9 countries offer no monthly or quarterly data. in this blog, we use march 2009 as the latest date since data availability is spotty afterwards.