Risk Consulting           Modeling & Analytics           About Us           Contact           Home
 
 

Value at Risk:  The Fall Guy  

Download PDF
share on Twitter Share on Facebook
 

The financial crisis of 2008 was a fascinating and trying time for market participants and even the most veteran traders will admit to being awestruck by the many unprecedented market events.  The magnitude of the losses faced by banks was enormous forcing many to shut down.  From 2008 to 2012, 465 banks were closed by the Federal Deposit Insurance Corporation (FDIC) in the United States.  This included Washington Mutual, the largest bank failure ever in the U.S.  The five years prior to 2008, ten banks were closed.  How could it all go so wrong?  Aren’t banks highly regulated?  Don’t they employ some of the smartest people and utilize sophisticated risk control systems?

Yes, banks are highly regulated in most of the world and for good reason.  First, they hold a unique place in the financial system in that bank failures can cause widespread financial panic.  A “bank run” is a situation where depositors lose confidence in the safety of their money and start to withdraw en masse.  Financial panics can have major ripple effects throughout the entire economy and so governments regulate banks to both prevent and manage these occurrences.  Second, banks are illiquid relative to their deposit base and rely on the continued confidence of their depositors to remain a going concern.  Even if most of their assets are sound (i.e. they are solvent), many of the assets are long term in nature and cannot easily be turned into cash without significant discounts and disruption to the economy.  Finally, banks operate using high financial leverage, hence their spectacular profitability in good times and devastating losses during economic downturns.  In Canada, the Office of the Superintendent of Financial Institutions (OSFI) requires banks to meet an assets-to-capital multiple of 20:1 along with other tests.  At this multiple, a 5% decrease in the value of the assets (loans, investments, etc.) wipes out all of the bank’s capital.  In 2011, the big six Canadian chartered banks had an average assets-to-capital multiple of 16.0.

Globally, the Basel Committee on Banking Supervision sets out international standards of capital adequacy for banks.  Under the Basel II Accord of 2004, Value at Risk (VaR) became the preferred risk measure on which to determine a bank’s exposure to market risk and thereby assess capital requirements.  The SEC followed suit in 2004 for the large commercial banks.  Prior to this, the banks were already using VaR to manage the risk around their trading books.

VaR is a modeling technique to measure the potential loss of a portfolio of assets at a certain probability level over a given time frame.  For example, a bank may report a VaR of $25 million at a 1% probability over a 10 day period.  This means that there is a 1% probability that the bank can lose at least $25 million over a 10 day period.  The 1% probability (confidence level) over a ten day period is the current standard under Basel 2.  Any confidence level and time frame can be used to calculate a VaR, but for comparison purposes and consistency it makes sense to use one set of parameters.  Also, to be most useful the confidence level should be low enough so as to occur fairly rarely.  There are numerous applications of VaR to commodity risk, such as quantifying the risk of a trading operation or estimating the capital reserves needed to support a particular business activity.

The main criticisms of VaR are as follows:

  1. It is subject to model risk, meaning that the user has some choice on how to calculate the model inputs thereby directly influencing the result.
  2. It is subject to estimation risk, meaning that there are possible errors in estimating the inputs.
  3. It relies on an arbitrary confidence level and holding period.
  4. It ignores the magnitude of losses beyond the chosen confidence level.
  5. It greatly underestimates the probability of tail events which are very large, market moves.  The 1987 stock market crash (a 20 standard deviation price drop) is a good example which under a normal distribution on which VaR is typically derived would not occur in the lifetime of the universe.

The last two criticisms are the most important, as all econometric models suffer (1), (2) and (3).  Ignoring the magnitude of losses beyond the confidence level is being addressed under Basel 3 by expected shortfall VaR, which adds all the losses beyond the confidence limit.  Underestimating extreme events can be improved by using a distribution which has fatter tails.  However, the difficulty of modeling tail events means that both scenario analysis and stress testing are needed to complement VaR, and stochastic modeling alone will not solve the extreme event risk.  Basel 2.5, included stressed VaR, which subjects VaR to a historic 12-month period of financial market stress.

Nicolas Taleb, author of The Black Swan (a very good read), in his testimony before Congress in 2009 states that VaR was “the engine responsible for the growth in leverage.”  He called for banning VaR, believing that it provides both regulators and managers with a false sense of security thereby fuelling excessive risk taking.  I would argue that the engine responsible for the growth in leverage was a prolonged low interest rate environment creating a real estate bubble on which new derivatives were created that were greatly mispriced (but that is another article).  Modeling financial markets will always be difficult for the simple reason that humans are involved operating with the powerful emotions of greed and fear.  We are not modeling the risk of a bridge failing due to wind loads and traffic weight where the properties of the materials are well known.

If you accept that regulation is the result of lobbying between the regulator and industry through a political system, then it is hard to imagine a capital level requirement that will ever reflect the full risk faced by financial institutions, irrespective of the risk measures used to determine capital adequacy.  The incentives for management of publicly traded banks encourages excessive risk taking due to the simple fact that bonuses are paid on good performance and there is little downside in performing poorly.  Referred to as the “blow up” trade, it involves collecting bonuses over a long time on a seemingly low risk position only to give back all of the gains and more during a very short period of time. This is also sometimes referred to as “picking up nickels in front of a steam roller”.  Even getting fired from large public corporations typically results in large termination packages for senior executives.  This asymmetry is a governance issue between shareholders and managers and really has little to do with the specifics of what risk measures are in place to determine capital requirements.  Hedge fund managers also have the same asymmetry as gains are shared with investors, but losses are not paid back.  Also, risk is largely hidden unlike quarterly earnings and so there is little incentive to manage it.  As Warren Buffett said, “it is only when the tide goes out that you can tell who’s been skinny dipping.”

VaR looks to be around for a while along with a host of other requirements for banks to manage their risk and assess a proper level of capital.  It’s a complex world and trying to find a single risk measure that does it all is bound to fail.

Iebeling Kaastra, CFA FRM

 

Gibson Capital Inc.
Calgary, Alberta, Canada

Email: info@gibsoncapital.ca


Copyright © Gibson Capital Inc. Commodity Risk Management, Modeling & Analytics

TURN RISK TO YOUR ADVANTAGE

Developing world class solutions for
commodity risk managers and traders since 1998.

 

Website by Media Eye Studios