The news of JPMorgan’s increasing legal problems, coming at the same time as news that Japan’s Fukushima nuclear plant had begun leaking highly radioactive water more than two years after the original meltdown in 2011, inspired me to extend an analogy I made at the time.
The big banks were effectively factories, taking in assets such as loans or mortgages as raw material and processing them – often several times over – transforming them into structured products for investors. Instead of generating power in the way that Fukushima did, these factories generated money.
Like Fukushima, the banks concentrated risks in a very dangerous way. When the production lines for structured securities seized up from 2007 onwards the result was the spewing of financial radioactivity into the global economy and the near-collapse of numerous large financial institutions.
We know how things went wrong. The value-at-risk models used to measure risks inside these factories, and the VaR-based shareholder capital that was supposed to buffer the banks against the unexpected, proved woefully inadequate. That was by design, since if the models had been any good, the capital needed to support the securitisation factories would have made them unattractive to shareholders.
The fact that the bankers running the factories were incentivised by the revenues and bonuses they were making to keep using bad VaR models (and thus arbitrage the system), was really a failure of regulation, particularly the Basel rules covering the capital required for trading portfolios.
Regulators tried to fix these deficiencies with so-called Basel 2.5 rules which European banks implemented in 2011 and US banks last year. The results can now be seen in regulatory filings published by the banks.
Regulators didn’t have the guts to throw out VaR completely, but they did beef it up with something called stress VaR which basically forces a bank’s model to remember how bad things were in 2008. Bank filings from the second quarter of 2013 typically put the capital required for stress VaR at about three times the capital for standard VaR. However, to the degree that bankers have an incentive to keep the risks of financial innovations out of their model, it’s still the same bad old VaR.
Perhaps with this in mind, the regulators looked at what happened in 2007-8, in and particular, how the securitisation factories replicated traditional bank lending by warehousing a lot of credit risk. That’s why we now have the incremental risk charge (IRC) which allows for the fact that loans traded by banks sometimes default in addition to declining in price, which was supposedly captured by VaR.
The Fukushima-type financial radioactivity from 2008 is captured by the IRC, as well as two more Basel inventions, the specific risk charge and the comprehensive risk measure (CRM). They cover parts of the investment bank production line such as securitisation, credit derivatives and correlation trading – activities that caused tens of billions of dollars of losses at banks such as UBS, Deutsche Bank, Morgan Stanley, Royal Bank of Scotland, Merrill Lynch and Citigroup.
You get a feel for how important these charges are to chastened regulators when you see how much the charges actually bite. For example, almost half of Goldman’s and Bank of America‘s market risk capital comes from the specific risk charge. It’s also telling that firms such as Morgan Stanley or Deutsche Bank are selling off parts of their trading portfolios to try and reduce these charges.
The CRM is interesting for a couple of reasons. Firstly because it’s the one new charge that banks are allowed to model for themselves (Well, not completely because the Basel Committee spelled out the risks 
that the banks had to model). Secondly, the CRM cuts to the heart of the credit derivatives innovation bubble set in motion by JPMorgan with its 1997 BISTRO trade celebrated by Gillian Tett in her book Fool’s Gold.
Ironically, the bank with the biggest CRM capital requirement in 2013 happens to be JPMorgan, with a $2.5 billion charge that is more than the other four large US banks combined. It’s most likely that this is a result of the London Whale’s credit derivative trades that lost the bank $6.2 billion last year.
At first sight, this visibility seems like progress. Not only do Basel’s new market risk charges reveal the capital cost of pre-2008 activity still lurking in bank portfolios, but they show, like Fukushima, when the radioactivity starts spewing out all over again.
Unfortunately, by allowing banks to calculate the CRM themselves, Basel left a massive loophole in the system. Suppose that JPMorgan had reported its mammoth CRM charge in early 2012 when the Whale trades were implemented. The bank’s capital ratio would have declined and shareholders might have understood that they were being asked to take additional risk.
As we learned from the March 2013 US Senate report into the Whale fiasco, JPMorgan did something different. It did internally calculate a huge CRM number, and then its quants and traders promptly tried to game the model in order to reduce it down again, arguably deceiving shareholders.
So while Basel may have seemingly caught up with the financial crisis in its last round of tweaks, it left in the modelling incentives that serve as catnip for greedy traders. For bank shareholders, these incentives increase the risk of ‘unknown unknowns’ within banks – in other words, the most dangerous stuff of all.
Now, Basel does have a charge for ‘unknown unknowns’ – it’s called the operational risk charge, which covers Whale-style internal cock-ups and rogue employees as well as external threats like litigation costs or terrorist attacks.
European banks already have to allocate shareholder capital to operational risk under Basel rules. For example, at group level Deutsche Bank allocates €24.16 billion, while Barclays allocates £4.3 billion at group level and £1.9 billion within its investment bank, according to second-quarter filings. You can argue whether this is sufficient to cover things like mis-selling settlements and Libor fixing, but at least it’s there.
US banks on the other hand won’t have to allocate operational risk charges for another couple of years, when they implement Basel III rules. However we do have some inkling of what banks think the charges are likely to be. In its latest quarterly filing, JPMorgan said it would have to allocate an additional $14.1 billion of capital if Basel III was applied today. This includes operational risk as well as other stuff like new charges for derivative counterparty risk (of which JPMorgan has a lot).
Is this sufficient to cover potential Whale scenarios as well as litigation and the whole gamut of operational risks that JPMorgan faces? Given that the bank currently predicts its potential losses from litigation alone to be $6.8 billion, I suspect not.
Comparing JPMorgan’s current available capital (Tier 1 common) of $147 billion to the $127 billion that the bank says it needs under Basel III you have an effective solvency ratio of 115 percent (see Note 1). All it takes is JPMorgan’s own estimate to be off by $20 billion or more and the bank is insolvent on an economic basis (see Note 2).
You might think that $20 billion is an ample buffer but then again JPMorgan has already paid more than that in legal bills since 2008 – something that no analyst predicted would happen. Returning to the Fukushima analogy, $20 billion seems like an awfully thin defence against the leakage of financial radioactivity contained in a bank like JPMorgan today.
Note 1: My figure of $127 billion for JPMorgan’s required capital comes from dividing the bank’s reported Basel III risk-weighted assets of $1,587 billion by 12.5. Since RWAs for operational, market and some forms of credit risk are computed by multiplying a capital charge by 12.5 according to Basel rules, I argue that it is more informative to use the capital charge which is directly comparable to available capital, rather than divide available capital by RWAs to produce a misleading leverage ratio. That is also the approach followed by insurance companies that compute economic solvency ratios.
Note 2: By economically insolvent, I mean that the bank doesn’t have enough shareholder capital to cover the cost of a lot of adverse scenarios happening at once. Having a solvency ratio of less than 100 percent is equivalent to a Basel III core Tier 1 ratio of less than 8 percent, which of course is the reciprocal of 12.5.
Comments are closed.
The fact that the bankers running the factories were incentivised by the revenues and bonuses they were making to keep using bad VaR models (and thus arbitrage the system), was really a failure of regulation
I know I’ve said this before but I really can’t accept that this is a failure of regulation rather than management. The Basel rules are meant to deliver a capital charge at the group level in order to ensure that the whole (regulatory) entity has an adequate amount of capital, in aggregate, for its risks, in aggregate. Because they were designed for this purpose, disaggregating that total capital requirement into its component bits is generally not going to give you a usable number that can calculate the economic profitability of individual business lines (and therefore the bonuses that should be paid etc). It really looks like the capital should be additive and the rulebooks don’t make this anything like clear enough in my opinion, but it’s always recognised in practice by the supervisors that some of the model requirements will be too generous, some too harsh and that hopefully overall the pluses and minuses will cancel out.
If a bank chooses to make internal allocations and capital charges on the basis of regulatory capital requirements, then it’s basically allowing the regulatory rulebook to manage its business for it – this is a decision almost as dumb as pricing out all funding to the business units at flat LIBOR. If I was supervising a bank (and when I’m analysing them), divisional capital allocations on the basis of regcap requirements is a real red flag that the business of economic capital management (which is pretty much all there is to management of a bank) is not being handled right.
Dan, I think the evidence from the crisis supports my point. If regulation (specifically of trading book capital) didn’t fail, then why did Basel change the rules to reduce the role of internal VaR models? Why have regulators themselves said that Basel rules for trading books failed? The evidence suggests that the failed banks were indeed as “dumb” as you say, perhaps because their management saw weak Basel rules as a free option.
This is from the FSA report into the failure of RBS:
“The capital regime was most deficient, moreover, in respect of the trading books of the banks, where required capital for many instruments was estimated using value-at-risk (VaR) approaches…A regime which inadequately evaluated trading book risks was, therefore, fundamental to RBS’s failure. RBS was allowed by the existing regulations massively to increase its trading risk exposure counterbalanced only by a small increase in capital buffers available to absorb loss”.
This is from the 2008 shareholder report into the failures at UBS:
“MRC VaR methodologies relied on the AAA rating of the Super Senior positions…Until Q3 2007, the 5-year time series had demonstrated very low levels of volatility sensitivities. As a consequence, even unhedged Super Senior positions contributed little to VaR utilisation. Treatment under the “banking book” would have significantly changed the economics of the CDO desk business as this would have increased the required regulatory capital charges”.
As regards pricing funding at Libor, here’s the UBS report again:
“UBS, in pricing internal funding for the businesses, passed on its advantage in accessing funding in the market and the efficiencies provided by its centralised treasury and liquidity management process. This model resulted in significant funding being available to the businesses with prices within the ordinary external market spread (i.e. internal bid prices were always higher than the relevant London Inter-Bank Bid Rate (LIBID) and internal offer prices were always lower than relevant London Inter-Bank Offered Rate (LIBOR)).
…”¨At all relevant times prior to the onset of the liquidity crunch, the businesses with Subprime exposure in the IB had access to funding on this basis.”
Yes sorry, I suppose I should have expressed myself more clearly as I thought I was basically agreeing with you but reading it back reveals I didn’t actually say that. Unfortunately, the answer to your question is just what you imply – although it’s a really bad management practice, it is indeed totally ubiquitous! The better the bank, the more likely it is to have an economic capital model that is more or less totally separate from the regulatory capital model, but even the best banks have some desks that have their allocations based on regulatory VaR.
I think the issue is on the apportionment of blame though. To the extent that it’s the regulators’ job to keep on top of management, I suppose any failure of management is a failure of regulation, but there are some of the flaws with VaR (doesn’t work, doesn’t handle event risk, isn’t sub-additive) which look to me more like “the regulation is at fault and needs to be redrafted”, and some of the flaws (caused loading-up on CDO tranches, use of unrepresentative datasets) which look more like “the managers are at fault and need to run their business systems properly”.
So my answer to your question is that VaR was totally broken for all the reasons you say in your book and in this article, which is why the regulators had to (and have to) do something “fundamental” about trading book regulation. But the version of VaR that was enacted in the Basel 2 regulations was close to a decade behind the curve by 2007-8, and any bank that was using methodologies that were known to be inferior and to misestimate risks, can’t completely blame the regulators for their own shortcomings.
(I just brought the UBS LIBOR-funding issue up because it was so crazy – I remember hearing in something like 2001 that ABN AMRO priced all the funding to its investment bank from the treasury at flat LIBOR and being appalled then. The idea that a bulge bracket player was doing the same thing five years later – unbelievable).
Obviously it is a management error. The point is however, that most managers act absolutely rationally and are aximising their expected outcome. Banking salaries are basically call options, with limitée down side And unlimited upside. Increased risk taking directly increases their expected income.
Just stumbled into your blog from the DeLong one- read this- and immediately bought your book (DD). I am a Ph.D. candidate writing dissertation on income inequality, financial deregulation, and partisan politics from 1914 (the effective start of the Fed) to 2012. I use time-series (ECM) regression analysis as well as process-tracing in a combined methods analysis. Your article above is excellent for a non-technical political scientist (well- almost a political scientist). Since process-tracing is taking a case and tracking the decision-making from a known decision- your work will be invaluable. No real comment other than keep it up and I look forward to learning more from you.
This is the bank bearing the name of the man who undercut Tesla, the most shameful episode of hubris in American history, J P Morgans stance set the worlds technological progress and hence its economic progress back hundreds of years, here we see that name living up to its past legacy in continuance of its destructive policies