The Default Line: THE INSIDE STORY OF PEOPLE, BANKS AND ENTIRE NATIONS ON THE EDGE (32 page)

Taylor ruefully remembers sitting in marketing meetings and realising that his colleagues assumed that the only way cheap two-year fixed-rate mortgages were going to make a profit was if the clients stayed with the product for six months after the two-year term, and went on to standard variable rate (SVR). ‘That’s where they’d claw some back. The insanity, though, is that they fucking wrote to them after eighteen months and said you can get a new product at this lower level. So I was sitting there going “I know this will make a loss.”’

So cheap money channelled into the UK mortgage market for profitless mortgages. The winners were the investment bankers who designed the deals. Northern Rock could book up-front profits on essentially loss-making mortgages because of the arrangement fees. In the familiar pattern of the age, bonuses would be paid on day one to all the bankers, even on business that was clearly going to become unprofitable, even within three years. At the Bank of England, many viewed Northern Rock as a glorified hedge fund. Property owners gained. Politicians who care principally about the illusory wealth accruing to the property-owning class gained. Newspapers gained from adverts placed in their personal finance pages. But taxpayers were to lose. And younger people were to lose out big time, becoming the sucker last buyers at the top of the market.

Could it have been stopped? In theory, it might have gone on forever. Or at least until Northern Rock was funding all the mortgages in Britain. The beauty of securitisation was that the funding was trapped for decades. Once an American or Asian or African investor had bought into Granite, the Rock’s customer was funded for ever. The problem in 2007 was that Northern Rock had the largest ever pipeline of mortgages to fund. Net residential lending had been 42 per cent higher in the first quarter of 2007 than the year before. And the Rock was between securitisations, with the biggest gap of funding to find. It had a monumental short-term funding gap. Additionally, neither the Rock nor its competitors playing a similar game had worked out that the end-investor in these securitisations was the money market. In August 2007 the medium-term money market basically stopped. No roadshow to the far corners of the planet would have made a difference. Even so, if the Rock had set aside a few billion pounds reserve (a so-called ‘warehousing facility’) as insurance it could have ridden out the credit crunch, at the expense of higher costs. Adam Applegarth told MPs later that Northern Rock’s wide funding base meant he thought such a reserve would not be necessary.

Northern Rock spread the cushion against unexpected losses thinner and thinner. At the end, the ratio of its assets to its equity base was 58:1, the highest in Europe. But the Rock was compliant with all laws and financial regulations at the time.

The UK regulator, the Financial Services Authority, signed off Northern Rock as one of only four lower-risk banks that only needed a full risk assessment every three years. Northern Rock was deemed so safe that it was singled out for special treatment amongst the thirty-eight major UK banks: Northern Rock did not have to be subject to the FSA’s ‘Risk Management Programme’. As a result, the FSA stopped even bothering with Northern Rock. The top five banks were inspected just under once a week between 2005 and 2007 under the ‘close and continuous’ regulatory framework. Slightly smaller banks were inspected once a fortnight. Northern Rock was not inspected at all in the whole of 2005, once in the whole of 2006, and on only three separate days in 2007. So the regulation of the small bank growing into Britain’s biggest mortgage provider – inspected only four times in the three years leading up to its collapse – was neither close nor continuous.

It was problems with liquidity that ultimately felled Northern Rock. But at the main regulatory meetings in 2006 the regulators were largely uninterested. There
was
a meeting between the FSA and Northern Rock on liquidity in April 2007. The FSA signed off the Rock as having passed its stress test on securitisation. This assumed that the funding market never shuts. Indeed, according to a Rock insider, ‘There was no concept of the market shutting.’

The FSA itself says in its internal report into the collapse of Northern Rock that its assumption was slightly different. The FSA’s ‘approach reflected a presumption that, in the event of a crisis like that experienced in August 2007, general market liquidity provided by the Bank of England would be increased and in extremis, liquidity would be provided for systemically important institutions’. This presumption was included in Rule 11.1.19G of the FSA’s handbook (in 2009 it was deleted).

Was Northern Rock systemically important? Did the then governor of the Bank of England consider Northern Rock systemically important? It is clear that Northern Rock and the FSA initially believed that funding markets would never close; even if they did, the Bank of England would be there as the UK’s lender of last resort. Northern Rock’s mortgages, profits and bonuses were all built on the backstop of the Bank of England. With my own eyes, on the first day of the Northern Rock run, I saw one of Britain’s top regulators struggling with the reality that the glory days of light-touch regulation were over. ‘We’ve got to allow innovation,’ he repeatedly insisted, his face contorted with fear. It was as if he could not quite bring himself to utter the unspoken question: ‘Don’t we?’

Those that work inside Northern Rock point out that holders of Granite bonds have not lost even a penny (although any investors who sold at the bottom of the market did lose their shirts). Credit quality held up sufficiently, even on the high-risk ‘Together’ mortgages. The technology worked, and then it worked for the taxpayer when Northern Rock was eventually nationalised. Indeed, one insider points to the fact that securitisation has helped the entire banking system survive after the crash by enabling banks to cash in assets at central banks. ‘It’s how I sleep at night,’ said one Rock insider. Others might argue that it made the British state a one-way all-in bet on sustaining high property prices. A great crash has failed to materialise so far, but only at the cost of the destruction of savings and a stagnant zombie economy. House prices remain high, and repossessions remain low. Some of the same people who created Granite now look after it for the taxpayer at UK Asset Resolution. The end result of this process was that many first-time house-buyers have ended up with a huge burden of debt, funded profitlessly in the medium term, but upon which bankers in Wall Street, London and Newcastle extracted huge day-one profits, and a taxpayer bailout that is set to cost £2 billion in losses.

All that from a bank that only existed independently for a decade, a bank that thought it had found the elixir of modern banking – funding tens of billions of pounds of mortgages without the need for real capital, because its credit risks had been engineered away. The FSA was so convinced of the efficacy of this elixir that it barely regulated the bank at all. It was a mistake, admitted the now defunct regulator, but the FSA insisted that it was a one-off error. Yet the credit fault line seen so clearly in Newcastle extended in one form or another across the entire global banking system.

California is used to tremors. The 1906 earthquake in San Francisco is the most famous of many. Oldrich Vasicek remembers the 1989 quake, when buildings started to make faces and whole panes of glass crashed down onto the street. ‘You can see the temperament of different people,’ he says. ‘Some rushed out of the buildings crazily, some hid under the table. I myself am a complete fatalist.’ He recalls a meeting of the loan committee at Wells Fargo HQ on Montgomery Street in the mid-1970s, just a few years after he was forced out of Czechoslovakia following the 1968 Soviet invasion. At Wells Fargo they were trying to push through a hike of 3 per cent in interest rates on car loans. A big quake came, the room on the sixteenth floor of the skyscraper swayed, floating on a bed of sand. And when the aftershocks stopped, the committee reconvened. Perhaps disturbed by the interjection of plate tectonics, the hawks on the committee backed down, and the car loan rates were frozen at 13 per cent.

But in the engineering of a Montgomery Street skyscraper, or even in the construction of the Del Monte mill that survived the 1906 quake, there are lessons to be learnt about probability and risk. Earthquake losses have fat tails – one might say obese tails. This refers to the curve of probabilities of various losses from an earthquake.

The most famous version of such a chart is the bell curve, which represents the so-called normal distribution (also known as the Gaussian distribution). Such a curve, in the shape of a bell, is the core of modern statistics. The bell-shaped curve illustrates how certain kinds of variables (such as human height, weight and arm length) are predictably and evenly dispersed around an average. Variance around this average is called the standard deviation or sigma. The further one gets away from this average, up or down, taller or shorter, heavier or lighter, longer or smaller, the less likely is that outcome, until it basically becomes impossible. In a normal distribution, the thin tail tapers away exponentially to zero as the event rapidly becomes more improbable. Earthquakes, it turns out, are far from normally distributed. The upper tail is fat.

If earthquakes
were
normally distributed, it would be possible to calculate definitively the probability of significantly larger-than-average tremors. The probability of a so-called four-sigma catastrophe (which means four standard statistical deviations away from the average) would be just 0.003 or 1 in 33,000. If that were the case, the people of San Francisco could rest a little easier and might even risk working or living in a warehouse without a steel-reinforced roof. In reality, the fat tail means that the four-sigma catastrophe is fifty-one times more likely than a normal distribution suggests. A 2008 Harvard University study used US Geological Survey data to show that deaths from earthquakes have fat tails. The highest number of deaths, around 283,100, was from a 9.0 magnitude earthquake off the west coast of Sumatra in 2004. This was 3.5 times the death toll of the next worst earthquake, in Pakistan in 2005. ‘This is the characteristic of a distribution with fat tails; events in the far-right of the distribution can be really large,’ says the Harvard study.

What things are actually normally distributed? Economics and finance? It turns out, they are more like earthquakes than physics. Andrew Haldane of the Bank of England has calculated that, assuming normality, an economic catastrophe, such as a three-sigma fall in the stock market, would occur once every sixty-four years, while a three-sigma fall in GDP would only come about every 800 years. (It should be noted that the GDP measure was only invented in the 1930s.) But in reality, such falls occur every eight years in the markets and only about once a century for the economy. Normal distributions occur only in spheres such as the dynamics of gas molecules, the physical characteristics of large populations, and long runs of coin-tossing or dice-throwing.

‘Where there is interaction,’ Haldane argues, ‘there is non-normality. But risks in real-world systems are no game. They can wreak havoc, from earthquakes and power outages, to depressions and financial crises. Failing to recognise those tail events – being fooled by randomness – risks catastrophic policy error.’

But – no surprise – normal distributions are hard-wired into economics and quantitative financial modelling. The towering example of this is value-at-risk (VaR), the measure used by banks and regulators to assess risk on their trading books, and to set limits on traders. VaR is supposed to tell a bank and its regulators how much a trading portfolio will make on 99 per cent or sometimes 95 per cent of trading days. Remember that at the time Northern Rock and its competitors were going crazy, credit risk was migrating off balance sheets and out of the regulated loan book and into the trading book. The regulatory dam was a flawed set of equations rooted in normal distributions, and the notion that a limited history of past pricing patterns could be extrapolated into the future. This method of assessing risk was developed by J. P. Morgan. Quantitative risk managers would be the first to admit the flaws in the measure, or indeed the flaws of boiling risk down to one number. But global regulators demanded banks calculate, and sure enough the familiar pattern of misinterpreting, gaming and reverse engineering formulae was quickly applied to VaR.

The
Financial Times
quoted Goldman Sachs’ chief financial officer during the 2007 credit crunch as saying that twenty-five standard deviation moves were happening several days in a row. To put that in context, he was suggesting that occurrences that his financial model suggested would only happen once in a period of many trillions of lifetimes of the universe, were actually happening every day.

The ‘fatal flaw’ of VaR, as Haldane argues, is that it is silent about the tail risk. A trader could be given a so-called 99 per cent VaR limit of $10 million, but VaR would be blind to the trader’s construction of a portfolio that gave a 1 per cent chance of a $1 billion loss. J. P. Morgan itself discovered in May 2012 that the ‘London Whale’ corporate credit portfolio that was assessed with a 95 per cent VaR of $67 million in early 2012 had lost them $2 billion within weeks. In its February 2008 annual results, RBS calculated a 95 per cent VaR on its trading book at £45.7 million. The disastrous purchase of the toxic asset-laden ABN Amro had increased that measure by just £6 million. A footnote did warn: ‘VaR using a 95 per cent confidence level does not reflect the extent of potential losses beyond that percentile.’ And sure enough, just a few months later, the losses on the trading book in 2008 topped £12 billion. A basic problem is that past trading performance is no guide to the future. VaR models were routinely specified to assume that the very recent past is the best guide to the future. Before long VaR came to be seen, quite incorrectly, as an upper-end assessment of likely losses. In fact, the VaR measure essentially slices off the tails full of catastrophic risk. So it is useless as a means of managing catastrophe risk. It is a little like having an emergency parachute that opens swiftly – except when you jump out of a plane. VaR was never designed for this.