Comments on the PIMCO report for Cyprus

Since the PIMCO report on the Banking System of Cyprus became public information, I've actually noticed much less comments regarding it than when it was still considered confidential. The reason might simply be that it is always much more difficult to speculate on something which is available to everyone than something which is supposed to be a secret. Thus, given the lack of comments on the subject, what follows is my take on the report.

The first thing you notice in the report is that PIMCO did an excellent job in identifying the major features of the Cyprus banking system. The most important of these features are:

1. The prevalence of asset-based lending practices

In essence, Cypriot banks lent out money only if you had some strong collateral to back your loan (in most cases real estate), with less attention given to whether the client had the ability to actually meet payments. If the borrowers got into trouble, they could always sell their property to repay their loans; as real estate prices increased for a very long time, this practice rarely yielded losses for the banks. Most importantly, borrowers who were not able to repay could always pledge more collateral and actually increase the amount of they borrowed.

2. Extended foreclosure and legal resolution timeline

Simply put, if a borrower could not repay his loan, then the whole procedure of obtaining the collateral and making a forced sale of the property ranged between 10 to 12 years. Add this to the previous feature and the reader may easily see that a borrower had no difficulty in pledging more collateral and obtaining more credit as there is almost zero downside on his part. This in its turn artificially increases loans and leads to the concentration of bad loans to few people (see land developers, hoteliers, etc).

3. Different provisioning methodology, impairment recognition and interest income practices

Notably, a fully-secured loan was not considered a non-performing one which makes the value of NPLs depend on the (subjective) valuation of the collateral. In addition, unpaid interest income was also considered to be income as a result of the historically appreciating property prices which made the probability of future losses very few (as also discussed in feature 1).

In addition to these, a high reliance on the the international banking operations for income and non-residents for funding was also reported. These features resulted in high cure rates for NPLs as well as high re-default rates (since extra collateral means a "cured" NPL but does not mean that the borrower's ability to repay has increased) and subsequently in high probabilities of borrower default but low loss-given-default rates due to over-collateralization.

The methodology of the exercise, will not be the subject of this article, yet, just simply comment that even though there has been much speculation, PIMCO does not appear to do anything different than standard procedure. The same holds for the base scenario which does not appear unrealistic given the European Commission (EC) forecast in Autumn 2012. Actually, compared to the -2.3%, -1.7% and -0.7% EC forecast for 2013, 2014 and 2015 respectively, the -3%, -0.6% and 0.8% is cumulatively much rosier. The same holds for the unemployment rate forecast.

The finding which most strikes out in the report is that Cyprus banks were unable to meet their capital needs even in the base scenario. In fact, not only would they need additional capital to meet regulatory needs but they would also end up with negative capital in 2015, making an additional case against the validity of the EBA stress test exercises in 2011. Again, I note that this is simply the base scenario. In the adverse scenario, the capital needs increase by more than 3 billion euros for the whole system.

The PIMCO numbers are supposedly the ones on which the decision on the percentage of the deposits haircut of the Bank of Cyprus  relied on. The 3.9 billion shortfall for the bank was very close to the 3.8bn obtained from the haircut. The problem, however, lies somewhere else: as stated in page 8 of the report "Greek loans represented approximately 40% of the defaulted balances. Moreover, in the adverse scenario [..] Greek loans represent 43% of total expected losses on Cyprus and Greek loans". In numbers, out of the total of 6.6 billion expected losses on loans and advances, around 2.8bn were from Greece. For those who have forgotten, MoU in March 2013 also included the forced sale of the banks' subsidiaries in Greece. This means that the banks were cut off from any potential losses in Greece thus lowering their capital needs. Even if we round the number of the PIMCO report to 2bn euros, the capital needs in the adverse scenario reduce to about 1.9 bn, without even taking into account the reduction in risk-weighted-assets which would further decrease needs.

Hence, the question which arises from the PIMCO report is: since the forecasts were made using the Greek branches as well, why was the amount employed in the haircut the same? In fact, since the Bank of Cyprus actually required a further re-capitalisation of close to 1 bn a few days ago, how low have the PIMCO estimates been? The only major difference which could make the actual outcome worse than the predicted is the fact that unemployment was 16% in 2013 rather than 13.8% in the forecast, while the house price forecast was much more pessimistic than the actual result. The change in the NPL definition does not matter at all, as PIMCO, in page 15, defines a non-performing loan as one which is 90 days past due, regardless of its collateral amount. The good Laiki couldn't have impaired the balance sheet by that much either since most of the losses were absorbed by the bad bank. Even in the worst case scenario of BoC obtaining all loans from Laiki, the exclusion of the Greek branches leaves capital needs for the domestic Laiki at less than 1bn. Strangely, the liquidation of Laiki was said to decrease Cyprus's needs by 4.2 billion which is in contrast to the PIMCO report in which the bank required 3.9 billion in total (including the Greek branches).

Concluding, the big question mark here is not if PIMCO over-estimated the capital needs of the banks, but why its numbers when it came to the Cyprus evolution of loans were so far off. If the BoC actually needed 4.8 bn (3.8bn of the haircut plus 1bn from the re-capitalisation) to pass the October stress tests why was the number less than 2 billion (excluding the Greek branches) in the PIMCO report? 

The PIMCO report, even though it did a great job in identifying and analysing the state of the Cyprus banking system did a very peculiar job in forecasting its capital needs. Truth be told, numbers and reports don't really add up.