Data dumps and pig farms; Inside Europe's stress tests

By Laura Noonan

LONDON (Reuters) - In his youth, Barrie Wilkinson studied how to stress test bridges. Now he is stress testing the euro zone's biggest banks.

The European Central Bank's year-long review of the currency bloc's largest banks will be a game-changer for the euro zone, either drawing a line under its financial crisis or causing a new one if the scrutiny is deemed insufficient or the holes in banks' books place too great a burden on indebted states.

This high-stakes exercise involves analysing banks' balance sheets, picking apart their procedures for assessing risk, and exposing them to hypothetical crises that have become all too real in parts of Europe, such as property market collapses and soaring unemployment.

It is a process that the ECB is keen to get right as it takes over the supervision of 130 eurozone banks later this year.

For Wilkinson, co-head of risk and finance at the ECB's favoured stress test consultant, Oliver Wyman, and an engineer by training, there are parallels with testing bridges.

"You look at historic stresses, how much traffic do you expect to go over it based on history ... what are the highest levels of wind you've ever experienced," he said. "Then you look at how the bridge has been constructed. You're trying to get to the potential point of failure in the bridge."

The problem for the investors, big depositors and taxpayers, who may have to make good any deficiencies in banks' balance sheets, is that the process of stress testing banks involves reams of data, differing definitions and a tight timetable.

That might mean a bigger bill for them.

"The worse the data, the higher the capital charge," said Stephen Smith, who leads accountancy firm KPMG's Asset Quality Review taskforce.

The Asset Quality Review is the ECB's review of eurozone banks' books, now under way ahead of this year's stress tests. The AQR assesses whether banks are currently adequately capitalised, while the stress test looks at whether they can withstand a future crisis.

Interviews with more than 20 officials, bankers and consultants involved with previous stress tests in Spain, Ireland, Slovenia and Greece - many of them speaking on condition of anonymity - suggest the ECB's health check will involve battles over basic definitions of loans, huge data requests and a painful process to verify the information.

THE BLAME GAME

Under pressure not to repeat the mistakes of previous stress tests in Europe and to uncover any nasty surprises on lenders' balance sheets before it takes over as the euro zone's financial supervisor in November, the ECB will be pressing banks for details about their loans, trades and risk models.

During Spain's stress test of its banks in 2012, one banker recalls that his institution, a mid-sized lender, had to send more than 500,000 individual pieces of data.

During tests in Greece, a local banker describes having to submit 180 fields of data for every personal loan being evaluated, and more than 100 for business loans. Hundreds of staff were involved.

With a tight timetable, disputes are expected between the consultants seeking the information and the banks providing it.

"When these things start they request all this data, they don't get it, immediately a blame game takes place," said one consultant who has worked on bank reviews in the euro zone. "They say we can't do the modelling because so and so hasn't provided the data."

The ECB's efforts to get banks to use a common definition for non-performing loans and loans where forbearance is being shown will introduce fresh problems, consultants and banks say, since data isn't already collected in that way.

Bankers say the stress testers couldn't possibly need all the data they are collating.

A European regulator admits that some of the data fields inevitably don't matter, but says you have to comb through a lot of useless data to find the ones that do.

"You're trying to find the numbers that will tell you what's going to happen in the future," he said.

When needs must, consultants can improvise. Several sources recall the drama when BlackRock employees found out that Ireland, whose banks they were testing in 2010, had no zip codes. In the end, that field had to remain unfilled.

'NO RIGHT WAY'

Consultants don't just take the data and set to work. They check its accuracy, including leaving their offices for on-site valuations.

Over 10,000 properties were visited and valued during the Spain's 2012 stress tests.

Some of the field trips have been enlightening. One consultant recalled sending someone to visit a cement factory who discovered it was in fact a pig farm.

Establishing the banks' current position is, of course, just half the battle. What stress testers are best known for is working out how banks will fare when the stress hits, and how much capital they need to cope with that 'adverse scenario', a feat that involves a high level of judgment.

Those judgments will have a major impact on the numbers the stress tests come out with.

"Is there a right way (to do a stress test)?" the European regulator said. "There isn't, really. There's a spectrum of approaches."

KMPG's Smith said while there is little leeway around the baseline economic scenarios, which use forecasts from the International Monetary Fund and the European Commission, there is discretion in forecasting elements such as future real estate prices and unemployment rates.

Translating those economic scenarios and constraints into future losses at banks, and ultimately into banks' need to raise more cash, is another minefield.

The Spanish banker said there were no issues with the loan loss models they encountered from Oliver Wyman, since their model is "well recognised and respected".

The broad thrust is that loan losses don't increase gradually, but instead rise exponentially, since rising unemployment and house price falls affect lots of borrowers at the same time.

What can be more divisive is predicting how much profit banks will make in the future, an important consideration since that profit cushions them against future losses and cuts the amount of capital they must raise.

"Here there was a bit of give and take - you have to try and convince them that the revenues will be as you say they will be," said the Spanish banker.

With so much discretion at play, James Longsdon, co-head of EMEA banks research at ratings agency Fitch, said there is a debate about the role and implications of stress testing, and the more fact-based Asset Quality Review is likely to prove more useful in the end.

"The AQR is the most important by a country mile," said Longsdon. "It genuinely is a re-assessment of solvency."

(Additional reporting by Sarah White in Madrid. Editing by Carmel Crimmins and Giles Elgood)