Once upon a time there was a financial crisis. This caused much havoc in the world and it was agreed it would be advantageous not to have one again. So a bunch of people dissected the causes of the crisis and decided that these evil institutions called banks had been doing lots of terrible things. One of the things they had been doing was buying lots of assets and writing lots of derivative contracts which exposed them to potential future losses without having sufficient equity so that if losses occurred the banks wouldn’t go bankrupt. This was actually unexpected. 30 years earlier regulators had written a bunch of of complex rules to protect against exactly this eventuality. What had gone wrong? Banks found lots of clever ways to circumvent these rules.
So after the financial crisis, regulators came up with a solution – we’ll not just have a new and improved update of the complex rule-set from before, we’ll also have a really simple simultaneous leverage rule. This way, even if banks try to be naughty, they will always have enough equity to stay safe. We even know what the leverage ratio should be: it’s 3%!
And herein the problems start.
Problem 1. Why 3%? Why not 2%? Or 4%? Or even 5%? Usually when assessing the right amount of insurance someone needs, people talk about insuring against a 1 in a something year event. Buildings in San Francisco designed to withstand in a 1 in a 1000 year earthquake, flood defenses designed to withstand a 1 in 500 year flood. What event is a “3%” leverage ratio designed to defend against? A 1 in 500 year banking failure? A 1 in 100 year event? A 1 in 10 year crisis?
This leads me on to my second big issue. Regulators decided what the prudent leverage ratio is before they had even defined leverage. When banks used to do simple stuff like loan businesses money to buy machines, and loan people money to buy houses this was a simple calculation. You took the bank’s equity, totted up all the loans the bank had made to get its assets, divided the former by the latter and hey presto! You have a leverage ratio. Unfortunately, banks don’t operate like that anymore, because this thing called the repo market and the these other things called derivatives appeared and really messed up the whole totting up of assets bit.
So problem 2. Netting of matched repos, and netting of market-to-market paid for derivatives. To better explain, a bit of technical description is necessary. Imagine Bank A owes Bank B $100 for contract 1 and Bank B owes Bank A $80 for contract 2. Bank B says to Bank A you owe me $100 and I owe you $80 so lets say that net you owe me $20. All is fine until Bank A goes bankrupt. Bank A’s administrator now says to Bank B “you know that agreement you had with Bank A about netting money you owed each other? It’s not legally binding. I will collect your $100 and you can get some small percentage of your $80 some years down the line after a bunch of messy bankruptcy proceedings”. Bank A, instead of losing the $20 he was expecting, is now $100 dollars down. What does this mean for the leverage ratio? Bank B should recognise the full $100 Bank A owes it as an asset.
Except it’s not that clear. Banks recognised this issue and think they have a solution. They went away with their lawyers and drew up a whole bunch of new fancy contracts that mean that this time they really well be able to net the $80 from the $100. So in the future, at default, administrators will only be able to claim for $20. So whether or not you should allow netting of contracts is actually quite a complicated question. Your favoured treatment depending on how much you believe lawyer’s opinions today will stand up in the bankruptcy courts of tomorrow.*
Going back to the question at hand – what does this mean for the calculation of the leverage ratio? Well, a lot actually. Just looking at derivatives contracts, if you didn’t allow netting of the sums Banks owe to each other for gains and losses, JP Morgan’s reported balance sheet would be $3.7tn rather than $2.4tn – or a whopping 50% bigger**!
But that’s not even the point. The point is, at the start, you go around and you agree with all the banks how to calculate leverage ratios and what to do about repo and derivative netting etc AND THEN you say OK then the leverage ratio will be X. This way if you decide netting is fine maybe you have your leverage ratio at 4 or 5%. On the contrary if everyone decides netting isn’t fine than you have a lower leverage ratio of 3%. What you DON’T DO is declare what your leverage ratio is, give everyone an idea of how you are going calculate it and then spend the next year watching banks, lawyers and everyone else pick holes in your methodology.
The final issue I have with the leverage ratio*** is with the assumption ‘it can’t be gamed’. Again a little more explanation is needed. The problem with derivatives from a “let’s try and make simple regulatory rules” point of view is that they are what as known as ‘unfunded’. This means that I can enter into a derivative with someone else today and it costs me nothing****. So the new balance sheet entry is nothing. However, tomorrow that derivative can lose you $100. But if you lose $100 where do you get the money from to cover those losses?
Regulators have realised this is an issue. Losing $100 dollars and not setting aside a nice lump of equity to cover those potential losses is clearly not acceptable. The question is how much equity should you set aside? You could go and create a really fancy model of potential losses that factor in the volatility of the contract, the creditworthiness of the counterparty and a whole bunch of other stuff, but derivatives are complex and that would pretty much be relying on doing the same kind of things we’re trying to avoid in the first place. Alternatively you could have a nice simple model that says your derivative is of this type and you will therefore need to set this amount of equity aside. This has the benefit that everyone can understand it***** and it can’t be gamed. Except it can. The problem with simple models is that banks can look at your model and say, how can we design the riskiest thing possible that your model says is safe?
So a substantial part of your total asset base******, the denominator in your leverage ratio calculation, can be gamed! The thing you were trying not to do in the first place.
*Accountants also disagree on this question. American US GAAP thinks assets should be reported on a ‘Gross’ basis, whereas in Europe IFRS looks at derivative mark-to-market net.
** Source: http://www.fdic.gov/about/learn/board/hoenig/capitalizationratios.pdf Also basel actually calculates leverage in a slightly different way than accountants but the underlying change is likely to be of the correct magnitude. Banks also
*** I have others but they are less important.
**** This isn’t completely true since I may have to post collateral or margin etc but for some derivative counterparties trading OTC derivatives, particularly before the financial crisis this was the case.
***** If you are fairly technical person and have a decent understanding of derivatives pricing.
******This is exactly what happened pre financial crisis. Loans and bonds and other financial assets were put into buckets depending on how risky rating agencies thought they were. Banks were then told to set aside a fixed amount of equity for that loan / bond / financial asset depending on the bucket it was in. So what did banks do? They bought all the riskiest loans that just about managed to fit in the non risky bucket.