A second opinion

It is good to know that Lloyd Blankfein shares my opinion on many matters related to the New Depression. For those who need to be reminded is the chief executive of an investment bank on Wall Street by the name of Goldman Sachs.

He has made clear some of our shared opinions in an article in the FT. For those too lazy to read the FT, here is my summary of his points:

• He says

  “Risk management should not be entirely predicated on historical data.“

  I go further and say “Don’t rely entirely on mathematical risk models.” Those models are predicated on the assumption that markets and prices operate so similarly to the models that the models can be used a s a measure of value and risk. They can for most of the time, but sometimes markets break down and the models don’t work. It is better to have a risk management systems that work, perhaps imprecisely but all of the time than to have a system that is perfectly accurate for only 95% of the time and is completely useless the rest of the time. He says the industry needs to do more to stress test models. I say banking just needs more seat of the pants risk-averse common-sense decision makers.


• He says
  

  “too many financial institutions and investors simply outsourced their risk management”

  which is true enough but in a sense they always did. Unlike for example the lead bank in a syndicated loan or a bond underwriter, the rating agencies have no money at risk, but this is not the biggest issue. What is different now is that so much of the creditworthiness (and now, in the particular case of banks, the capital allocation) depends on the say-so of the rating agencies. In the past agencies would have rated a few well known companies, but nowadays they opinr on the structure and documentation of deals rather than company performance. It was relatively simple to take a view on the value of say a big oil company and its likelihood of defaulting in its debts. Can a rating agency really show the same certainty about the risks on the assets and documentation of a special purpose funding vehicle?
  
• He cites the

  “over-dependence on credit ratings coincided with the dilution of the coveted triple A rating”,

  which I mentioned a few weeks ago, but although he mentions that there were 64,000 AAA-rated structured deals but only 12 AAA corporates in 2008, he does not mention that this added to the pro-cyclicality effect of Basel II. When Basel II was first conceived there were very few AAA ratings and banks did very little business with AAA counterparties, so that there would be very little need for extra capital on a ratings downgrade of a AAA counterparty. The situation now is very different.


• He says that

  “size matters”,

  or that it is easy to forget that although the risk of loss on $5bn of high quality debt is the same as the risk on $50bn, the consequences of loss are far greater on the latter. Well blow me! He doesn’t miss a trick this guy. Some prudent banks have always set limits on types of asset they hold, counterparties and other risks. It seems some did not. Perhaps this is really an Enron-type problem, more connected with off-balance sheet items and the assumption that what you don’t see is not really there.


• He says that

  “many risk models incorrectly assumed that positions could be fully hedged”,

  which is the corollary of the assumption that risk models operate correctly at all times. The trouble is that hedging markets can breakdown just as easily as the market for the assets being traded, and there is no guarantee that the liquidity in the hedging market will be the same as the liquidity in the primary asset market.


• He says

  “risk models failed to capture the risk inherent in off-balance sheet activities, such as structured investment vehicles”,

  but I would put it differently. If there is any exposure to risk in an asset or liability then perhaps it shouldn’t be off-balance sheet. This is the Enron effect. A bank or a company does a trade, perhaps the sale of an asset that isn’t quite a full sale, although maybe we can convince the auditors it is close enough to a sale to warrant sale accounting treatment. Remember the Enron Nigerian barges? Maybe not, but that is what banks do all the time to remove assets from their balance sheet and perhaps to book gains on a sale. Only the problem is there are some residual risks associated with those assets, but if we speak nicely to the auditor he won’t mind them being there, and after all it’s only one deal. Except that single deal is repeated and gets the same sign off as before and then after a few more repetitions the auditor can’t change his opinion. In fact he probably goes to the accounting standards board to get his treatment approved, and then the business really takes off.


• He says there was

  "too much complexity",

  which made it hard to manage the growth of new instruments. I would turn the point around and say that there was too little regulation of new instruments which places some of the fault at the door of the regulators, but I would also put some fault at the door of the bankers who were profiting from the complexity knowing that there was inadequate regulatory understanding of the instruments being created.


• His last point is that

  “financial institutions did not account for asset values accurately enough”.

  I would differ saying that some of the instruments were not capable of sensible valuation and that even if some valuations were perfectly accurate at the time they were made, they should not have been relied on immediately thereafter. A balance sheet might measure the risk in some straightforward instruments, but it doesn't come close to articulating the risk in some instruments.

One area that he misses completely is the "we never saw it coming" line that is coming from so many mouths today. What they are really saying is "yeah, we knew about it but we shut it out of our minds". The bankers of today's modern finance rely heavily on trading and securitisation, and by and large they have the trader's mentality. An asset is a risk until it is hedged or sold (albeit that there might be some ongoing risk in the asset). A liquidity exposure is a risk until some funding is put in place, although the risk doesn't go away completely. It all happens on an item by item basis and the "market" is often assumed to be almost infinite, or at least to the exnt that putting another trade into the market doesn't absorb all liquidity.

When risk assessments were made in banks the one question that was not asked was "What happens if X market goes away?" because the likely answer was "it ain't gonna happen, but if it does we all go to hell in handcart, you, me, the Governor of the Bank of England and the whole economy". The risk was put in the very low risk category, not factored into any pricing models and generally ignored. On a deal by deal basis that made perfect sense because the loss on that deal was limited to the values related to that particular deal. On an aggregate basis the numbers were so huge that no bank would contemplate the losses resulting from any individual deal, so deals got done and the market consensus was that these sorts of things did not and would not happen. Which meant that the market grew and grew and the issues that were ignored were precisely those that occurred because nobody was worrying abut them.

That apart, Blankfein has some good points and should go far.