Financial innovation gets a bad press. In 2009, former Fed chairman Paul Volcker quipped that “the ATM has been the only useful innovation in banking for the past 20 years”. Sadly, the industry hasn’t helped itself. This week’s Net Interest looks at three cases of financial innovation gone wrong. In all three, the inventor came to rue his contribution to finance.
The Model
Dennis Weatherstone needed a number. He’d just been appointed chairman and chief executive officer of JPMorgan and was in the process of reorienting the bank away from traditional lending towards trading. One of his first acts in charge was lobbying the Federal Reserve to give his firm authority to trade and sell corporate stocks – making it the first bank-related securities firm with a full range of securities powers. By 1994, three-quarters of JPMorgan’s revenue came from investment-banking activities such as trading and securities issuance.
A currency trader by background, Weatherstone understood the risks inherent in such businesses. According to colleagues, he maintained “a steely insistence on evaluating the downside risk” of any trading decision. It was an insistence he imposed on the overall firm. Every afternoon, at 4.15pm New York time, JPMorgan held a treasury meeting to go through its various risk exposures. As risks proliferated, Weatherstone thought it would be useful for the risk management team to present a single number at the meeting, representing the amount of money the bank might lose over the next twenty-four hours. “At the end of the day, I want one number,” he instructed staff.
In 1990, JPMorgan introduced a new model, Value-at-Risk (VaR), to satisfy Weatherstone’s request. Volatility had long been used to measure fluctuations in a security’s price; Value-at-Risk took this further, using volatility as an input to estimate the minimum loss that might be expected on a day where the firm suffers large losses.
To illustrate, let’s say you own a portfolio of stocks worth $10,000. If the portfolio’s 99% daily Value-at-Risk is $200, it means that one day out of a hundred, you would expect to lose $200 or more; the other ninety-nine days, you would expect either to make money or suffer losses lower than $200.
The measure was a useful way for JPMorgan to keep track of firmwide risk and became the basis for risk budgets. Years later, JPMorgan would use it to measure risk on 2.1 million positions and 240,000 pricing series. But rather than keep it private, JPMorgan opened this valuable intellectual property to the world. In October 1994, it published full details of the model under the name Riskmetrics. Other banks and trading firms swiftly adopted it.
It’s unusual for banks to share proprietary knowledge that gives them an edge. JPMorgan wasn’t being altruistic; the bank’s motivation lay in regulation. Several years earlier, authorities had introduced the Basel Accords to harmonise capital treatment of credit risk. With market risk growing as a key component of a bank’s overall risk profile, regulators wanted to incorporate that, too. Jacques Longerstaey, one of the developers of the model, recalls (emphasis added):
“The regulators at the time were starting to think about moving on from the Basel I framework and impose capital requirements on trading books. The first incarnations were very simplistic and didn’t take into account any diversification, and therefore would have been very punitive for the bank. So Riskmetrics was more than a client tool, it was a way to popularise these methodologies so that all the banks would go to the regulators and ask to be allowed to use internal models to lower their capital requirements. This is one of those things that may have had positives and negatives.”
Two years after VaR entered the public domain, the Basel committee amended its rules to allow banks to use VaR models to calculate their trading book capital. Banks were required to set aside capital equivalent to the greater of the previous day’s VaR or the average VaR over the previous six days, multiplied by three. Importantly, banks were allowed to use their own VaR measures to compute capital requirements. By giving bankers and regulators confidence that trading risks were measurable and could be controlled, such risks were allowed to grow. The number demanded by Weatherstone went on to underpin large, complex balance sheets.
But VaR is no panacea. While good at quantifying the potential loss within its level of confidence, it gives no indication of the size of losses in the tail of the probability distribution outside the confidence interval. The one-in-a-hundred day event may be a lot more debilitating than the $200 loss in the example above. In addition, correlations between asset classes can be difficult to ascertain, particularly when banks begin to act in unison. The diversification benefits that VaR supposedly captures in a portfolio of different asset classes falls away when crisis hits and correlations surge.
In 2008, the year Weatherstone died, the complex balance sheets his number facilitated unravelled spectacularly. Citigroup took $32 billion of mark-to-market losses on assets that year, an order of magnitude greater than the $163 million of VaR it reported at the end of 2007. Value-at-Risk didn’t cause the crisis, but it certainly cultivated a false sense of security leading up to it.
“Dennis, you created a monster by asking for that one number,” says Jacques Longerstaey.
The Measure
At its peak, Libor underpinned hundreds of trillions of dollars of financial contracts globally. But its inventor didn’t start with such lofty ambitions.
In 1969, banker Minos Zombanakis was working on a problem at his office on Upper Brook Street in London’s Mayfair. Born in Crete and educated at Harvard, Zombanakis spent ten years in Rome before convincing his employer, Manufacturers Hanover Trust (now part of JPMorgan), to establish a presence in the United Kingdom. The eurodollar market – the vast pool of US dollars held by banks outside the US – was already well developed: the first eurobond had been issued in 1963 for Autostrade of Italy. But Zombanakis saw an opportunity to channel eurodollars into loans as well as bonds.
One of Zombanakis’ clients was the state of Iran. The country, under the leadership of the shah, needed a loan of $80 million. No single bank would lend that amount of money to a developing country with insufficient foreign currency reserves, so Zombanakis put together a syndicate. The problem was that with UK interest rates at 8% and inflation rising, banks did not want to commit to lending at a fixed rate for a long time.
Zombanakis came up with an innovative solution to offer a variable rate linked to banks’ funding costs. Banks in the syndicate would report their funding costs just before a loan-rollover date. The weighted average, rounded to the nearest eighth of a percentage point plus a spread for profit, became the price of the loan for the next period. Zombanakis called it the London interbank offered rate.
The Iranians were happy and the idea took off. By 1982, virtually all loans in the $46 billion syndicated-loan market used Libor to calculate the interest charged.
Soon the rate was adopted by bankers outside the loan market, looking for a measure of bank borrowing costs that was simple, fair and independent. In 1970, the first bond linked to Libor was issued, known as a floating-rate note. When derivatives tied to interest rates became popular, many used Libor as their benchmark.
As Libor became more central to the global financial system, pressure grew to codify the setting of the rate. In 1986, responsibility for administering it passed to the British Bankers Association (BBA). Each day, the BBA would go out and ask a panel of banks the rate at which they thought they could borrow from other banks across a series of maturities and currencies. At its height, Libor was published across ten different currencies and fifteen maturities. The base panel comprised 16 banks. The BBA would lop off the four highest rates and the four lowest rates and publish as its benchmark an average of the remaining eight rates.
While Libor had been used first in syndicated lending markets and then in other loan markets, its use really took off in 1997 when the Chicago Mercantile Exchange (CME) adopted it as the reference rate for eurodollar futures contracts. The rate became “the most important number in the world”, as it grew to provide the benchmark underpinning hundreds of trillions of dollars of financial contracts globally.
But it was still based on a survey. Unlike other benchmarks prevalent in financial markets, Libor didn’t reflect actual prices visible in a market. Rather, it reflected a hypothetical aggregate. Given the size of markets it went on to support, it wasn’t an ideal benchmark but once it had burrowed its way into those markets, the cost of switching became very high. Meanwhile, the BBA was heavily incentivised to sustain it: the organisation made money selling licenses allowing others to incorporate the benchmark into their products.
Perhaps reflecting its legacy as an informal average, or its home in London, where regulation was traditionally crafted around principles rather than rigid rules, Libor was loosely policed. As long as banks didn’t collude to move the rate in unison, they had a broad degree of discretion as to how they answered the Libor question. On any given day, there would have been a range of rates at which banks could borrow funds – they were free to set their rate within that range.1
Traders took advantage of that discretion to attempt to influence the outcome of Libor, and the lack of Chinese walls between rate-setters and traders made that easy. The size of banks’ derivatives books meant that even a single basis point shift in Libor could potentially be very profitable. In 2012, authorities spotted what was going on and came down heavily on them. Banks were fined a total of around $9 billion and 38 individual traders were prosecuted.
But it has since transpired that authorities themselves sought to influence Libor to stave off the stigma that a higher funding rate would create around banks in their jurisdictions during the 2008 financial crisis. In a new book, Rigged, journalist Andy Verity outlines the events.
An instruction from Downing Street to lower Libor, a measure of the price at which banks lend to each other, on October 29, 2008 wasn’t the only thing that no one was supposed to know about. It was much bigger than that… The evidence suggested it wasn’t just the UK government and Bank of England, but Banque de France, Banco de España, Banca d’Italia and other central banks and governments around the world, all co-ordinating to get Libor and Euribor down.
Libor didn’t survive its crisis of confidence. The BBA gave up responsibility for calculating Libor in 2014 and in June this year the panel will be disbanded altogether.
“When you start these things, you never know how they are going to end up, how they are going to be used,” said Minos Zombanakis in an interview shortly before his death. He died in 2019 but not before distancing himself from the market he created, calling it a “monster” and a “prostitution racket run by pimps”.
The Merger
For most of the twentieth century, banks and securities firms were kept separate. The Glass Steagall Act of 1933, passed at the height of the Great Depression, drove a wedge between the two industries. Companies had to choose between commercial banking – the business of taking deposits and making loans – and investment banking – the business of underwriting and dealing in securities.
Lenders like National City Bank were forced to dissolve their securities business, and securities firms like Lehman Brothers had to dissolve their depository business. JPMorgan elected to be a commercial bank, but a number of managers including Henry Sturgis Morgan (a grandson of J.P. Morgan) and Harold Stanley departed to set up Morgan Stanley.
Over time, market developments weakened the separation of the two business activities. Glass-Steagall did not prevent commercial banks from engaging in securities activities overseas and by the mid 1980s, US commercial banks such as Chase Manhattan, Citicorp and JPMorgan had built up thriving international securities operations. Glass-Steagall also carved out a category of exempt securities that commercial banks were permitted to deal in and it allowed them to engage in limited brokerage activities for banking customers; as time went on, those authorisations expanded. And when Glass-Steagall was drafted in 1933, futures markets were small, over-the-counter derivatives didn’t trade and currencies didn’t float. As those features changed, each of these asset classes became fair game for commercial banks.
These trends accelerated through the 1980s, and by the mid-1990s, many of the restrictions had been relaxed. The Federal Reserve was charged with ensuring compliance of Glass-Steagall and as part of that placed a limit on the share of revenue a bank could derive from underwriting and dealing in ineligible securities. Historically 10%, it was raised to 25% in December 1996.
But it was a merger that would lead to the ultimate repeal of the Glass-Steagall Act. In April 1998, Citibank, led by banker John Reed, and insurance and securities brokerage business Travelers Group, led by Sandy Weill, agreed to merge in the largest corporate combination to date.
Citicorp had a unique global position and strengths in credit cards, foreign exchange, private banking, derivatives, and relationships with multinational companies. Travelers Group had leading franchises in consumer finance, insurance and asset management, and – as owner of Salomon Brothers – in investment banking and capital markets, as well as an array of distribution platforms. “Undoubtedly, putting together such giants would propel us into a universe of our own,” Weill said. In particular, the merger would allow the group to make investment products such as stocks and bonds available to banking customers around the world.
The merger did not initially breach the 25% revenue test but to allow the combined business to grow unfettered, its architects knew that Glass-Steagall would need to be repealed. The new Citigroup pushed hard for legislative change. In the year the merger was announced, the company spent around $100 million on lobbying and public relations efforts. The following year, Congress passed the Financial Modernization Act, dismantling many of the restrictions of Glass-Steagall.
Citigroup went on to grow rapidly. Total assets increased from $800 billion in 1999 to $2.2 trillion by 2007. And then it all collapsed.
Even before the denouement with Citigroup forced to seek government aid, John Reed suffered buyer’s remorse. In April 2008, shortly after the failure of Bear Stearns, he told a reporter from the Financial Times that the merger of Citibank with Traveler’s Group that created Citigroup had been a mistake, saying:
“The stockholders have not benefited; the employees certainly have not benefited, and I don’t think the customers have benefited because our franchises are weaker than they have been.”
Testifying before the Senate Banking Committee in 2010, he went further. “There is no question that when we put Travelers and Citi together, we created a monster,” he said. In Reed’s view, the corporate culture that thrives in firms operating in capital markets, like Travelers did with Salomon Brothers, should not be allowed at depository institutions and that the banking and financial systems would be stronger if those functions were separated.
Citigroup hasn’t gone so far as to unwind the merger but it has spent the past 15 years slimming down and, under its fifth CEO since John Reed, now promotes a “simplification agenda”. Its balance sheet and Value-at-Risk are broadly at 2007 levels in spite of considerable asset price inflation along the way and today its Value-at-Risk “is conservatively calibrated to incorporate fat-tail scaling.”
The mid to late 1990s environment that nourished innovations that grew into monsters is no more. But the conditions that characterise it – lax regulation, misaligned incentives, narrow thinking – are timeless. In finance, as in other fields, innovations can and do go wrong. A new breed of monsters is always being grown.
The CME made this clear in a letter to the BBA: “A contributor panelist who can borrow ‘in reasonable market size’ at any one of a wide range of offered rates commits no falsehood if she bases her response to the daily Libor survey upon the lowest of these (or the highest, or any other arbitrary selection from among them).” For more on Libor, see last year’s piece, Freeing the Scapegoat.
Nice summary of each of these, a good read
Marc, thanks as always for the history and insights. On Libor, missed in it problems and warts is the effect of Goodhart’s law, which Marilyn Strathern summarizes as “When a measure becomes a target, it ceases to be a good measure.” Having worked in rates’ market through Libor’s twilight years, I can say it definitely had it problems but also the replacements our all-knowing monetary overlords are promoting do not fully replace the risk that Libor captured.