Johnny Elliot and Benjamin King
In August 2007 problems were emerging in the US sub-prime mortgage market. Rising numbers of borrowers were getting behind on their repayments, and some investors exposed to the mortgages were warning that they were difficult to value. But projected write-downs were small: less than half a percent of GDP. Just over a year later, Lehman Brothers had failed, the global financial system was on the brink of collapse and the world was plunged into recession. So how did a seemingly small corner of the US mortgage market unleash a global crisis? And what lessons did the turmoil of autumn 2008 reveal about the financial system?
When the first signs of trouble emerged, many economists considered them too small to be of macroeconomic significance. For example, one commentator urged a sense of perspective:
“The rate of loss in subprime mortgages keeps climbing. Perhaps it will double, maybe back to $67 billion. This is a large sum by absolute standards … but by the metrics of a large economy, it is nothing. Some smart, brave people will make a fortune buying in these days, and then we’ll all wonder what the scare was about.”
With hindsight, things played out very differently. So what did the crisis reveal?
Lesson 1: complex securitisation can be dangerous
Most analysis of the crisis points to the US sub-prime securitisation market. Securitisation allows a bank to turn some individual assets (e.g. mortgage loans) which they can’t trade, into securities which they can. So a bank bundles some loans up into a securitisation. Others can then buy a stake, and those stakes can be traded as securities. Some individual mortgages might default, but that risk is spread across all investors in the bundle. Often loans were packaged up into a legal structure called a ‘collateralised debt obligation’ (CDO).
CDOs were normally structured so that losses were not shared equally amongst all investors. Instead, they were split into ‘tranches’. The senior tranches were relatively safe because they got first dibs on the money that came back, while the more junior ones were riskier. The CDO took a book of moderately risky loans and created some very safe securities, some very risky ones, and some in between.
Then things got more complicated. Banks started to combine separate CDO securities into ‘CDO-squareds’. And as the supply of mortgages began to run out, ‘synthetic CDOs’ were created, using derivatives to link their value to existing issuances rather than buying them outright. Now many new securities could be created without any additional mortgages. For example, a $15m tranche of the Glacier Funding CDO 2006-4A was referenced by $85m of synthetic CDOs.
This all made it difficult to understand the underlying risk. And the valuation of tranches, especially the middle ones, hinges crucially on the correlations between those underlying assets – and these can be hard to estimate.
Complex products are not inherently dangerous. But two things went wrong.
First, too few investors understood the risks they had taken on, and in 2007-2008 there was a collective realisation that this was the case. The reaction of the chairman of one small bank in California may have been typical:
“I asked the CFO what the mechanical steps were in mortgage-backed securities, if a borrower defaulted, as part of a package security, what happens? And he couldn’t answer the question. And I told him to sell them, sell all of them, then, because we didn’t understand it.”
Second, it turned out the correlation assumptions were wrong. Pooling risks only works if those risks are independent. Investors thought middle tranches were reasonably safe, because models – used by banks and credit rating agencies – estimated that widespread correlated defaults were very unlikely. After all the US had never seen a synchronised, nationwide, housing downturn.
Lesson 2: it matters who holds risky assets
So who bought these securities? And how would they react if their prices fell?
In general, unleveraged investors – those investing their own, rather than borrowed money – are more resilient to price falls. A fall in price may be painful, but they can look through to the long term value and are unlikely to be forced to sell below a fair price.
But large sections of US securitisations were held by the opposite type of investors – those that were excessively leveraged and dependent on short-term funding.
Most of the leverage was supplied by collateralised lending, which allowed investors to borrow against the value of the securities. But as the market realised how risky the securities were, the ‘haircuts’ on the collateral increased, meaning that investors had to supply more and more collateral to maintain their positions. This meant many became forced sellers as they ran out of collateral.
Lesson 3: not only banks can run
A run on a bank is a familiar risk. By late 2007, there was a fresh example in the case of Northern Rock, now infamous for queues on British high streets. The intuition makes sense – banks have a limited supply of ready cash, so can’t pay if everyone withdraws at once. This can trigger a spiral of panic as confidence evaporates.
In 2008, the same structural vulnerabilities were present across the financial system, with ‘shadow banking’ entities investing in long-term assets while offering short-term deposits. Banks established off balance sheet funding vehicles that bought longer term assets (often securitised mortgage loans) while issuing short-term debt to fund their purchases. The debt was bought by money market mutual funds (MMMFs), who in turn offered immediate redemption to investors.
Over the course of 2008, as concerns mounted over the quality of the mortgage-backed securities, these markets began to break down. Off balance sheet vehicles were downgraded, and became unable to issue short-term debt. The pressures culminated in a run on MMMFs in October 2008.
Lesson 4: financial institutions are connected in multiple ways
In 2005-06, the correlation between bank credit default swap (CDS) premia looked like this:
Large bank CDS premia correlation
The big patches of green indicate that the CDS market thought that the riskiness of one bank wasn’t related to the riskiness of any of the other banks. Risk was thought to be a bank-specific issue.
After the crisis, the picture looked like this:
Large bank CDS premia correlation
The reddish patches show strong correlation – the market had realised that if something hurt one bank, it was probably bad news for the others.
During the crisis financial markets realised that banks had become riskier, as some had material exposures to sub-prime securitisations. But no one knew exactly who was exposed, and by how much. As a consequence, the interbank lending market – which underpins the liquidity of key financial markets – seized up. One measure of this is the rate banks paid to borrow from each other. Before 2007, banks were freely lending at tiny spreads over risk free rates. By late 2008, these spreads had jumped up to about 3%, more than 35 times the pre-crisis average.
A bank that could no longer cheaply borrow from other banks faced incentives to also stop lending, and to quickly sell assets to meet its own cash needs. A cycle began to take hold:
Ultimately, many banks exhausted this cycle before finally turning to governments for support – collectively requiring more than $400bn of liquidity and capital support from governments and central banks.
Lesson 5: liquidity, and prices, can drop – fast
Two charts illustrate the cumulative effect of some of these developments in financial markets over the course of 2008.
First, financial market liquidity. This measure is a composite of nine liquidity metrics, and captures a general collapse in liquidity across key markets – repo, FX, corporate bonds, equities – as uncertainty over the scale of sub-prime losses, and solvency of key financial intermediaries, took hold.
Second, prices. The price of mortgage securitisations fell sharply through 2007-2008, below that justified by fundamentals. But the wider collapse in liquidity also drove down the price of other unrelated assets, as investors sold assets at ‘fire sale’ prices to realise scarce liquidity. This points to the feedback loop – falling prices driving low liquidity, which forces investors to make further sales.
Lesson 6: the financial system can amplify stress
A core function of the financial system is to manage risk. When it works well, it can help the economy absorb shocks, by distributing risk to those that want to hold it. Before the crisis, many believed that the growing complexity and sophistication of the financial system – particularly securitisation markets – demonstrated that it was doing its job better than ever before.
By the time Lehman Brothers failed this belief had been shattered. Its collapse was a symptom of the problems in the financial system, not the cause – it was one of many banks over-exposed to CDOs and reliant on short-term funding. But it became part of the downward spiral – its failure dramatically hit confidence in other banks, further reducing market liquidity and prompting more forced sales. What started as trouble in one corner of the US housing market ended up as global crisis. Lehman Brothers is gone, but definitely not forgotten.
Johnny Elliott works in the Bank’s Capital Markets Division and Benjamin King works in the Bank’s Resilience Division
If you want to get in touch, please email us at firstname.lastname@example.org or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.