[ Pobierz całość w formacie PDF ]
.The lesson: Include liquidity risk as a key factor in financialmodels.The weight assigned to liquidity risk should increaseas leverage increases in the financial system.Mistake 6: Be Ready to Question the ExpertsAs AIG relied too much on the expertise of Professor Gorton, so manyfirms relied too much on the ratings of mortgage-backed securities andcollateralized debt obligations by the credit-rating agencies.Many finan-cial institutions held in their portfolios the triple A rated tranches ofmortgage-backed securities and collateralized debt obligations, while sell-ing lower rated tranches to investors.The risk of these triple A tranchesdefaulting was supposed to be 1 in 10,000 over a 10-year period, accord-ing to Moody s.In fact, by the end of 2008, 50 percent of the triple Atranches of mortgage-backed securities and almost 100 percent of tripleA tranches of collateralized debt obligations had partially defaulted.47Thomas Maheras, the head of fixed-income trading at Citigroup from2005 to 2007, was one of the people who put considerable faith in topratings on mortgage-backed securities and collateralized debt obligations. Credit Default Swaps and Mathematical Models 89According to published news reports, his team told SEC examiners inthe summer of 2007 that the probability of a default on Citigroup s sen-ior tranches of securitized subprime loans was so tiny that they excludedthem from their risk analysis. A slide used at a Citigroup meeting laterthat summer said these senior tranches were viewed by the rating agen-cies to have an extremely low probability of default (less than.01 percent). Around the same time, Maheras assured his colleagues that Citigroup would never lose a penny on these senior tranches.48Yet, ironically, these very same financial executives knew about themany conflicts of interest faced by the credit-rating agencies, the ratingsshopping that was taking place, and the magical transformation of sub-prime mortgages into triple-A rated securities.A bank sponsor wouldtake a group of subprime mortgages rated as junk bonds, put them intoa special purpose entity, and then create multiple tranches based on thecash flows from these lowly rated subprime mortgages.Then, in a mod-ern version of alchemy, many of these tranches would be rated triple Abased on the benefits of diversifi cation and default assumptions builtinto the models of the credit-rating agency.The lesson: Understand the limitations of your expert sadvice on modeling including the factors not covered by theexpert.Be especially skeptical of advice from experts likecredit-rating agencies with significant confl icts of interest.Mistake 7: Financial Incentives Lead to Rosy ProjectionsJust as the revenue of the rating agencies depended on their ability togive out the highest ratings to tranches of mortgage-backed securitiesand collateralized debt obligations, so also many of the employees of thefi nancial institutions had substantial compensation incentives to makeoptimistic projections with their own models.As the securitization ofprime mortgages accelerated, the modelers were under tremendouspressure from their bosses to approve as many deals as possible.Further,modelers and their bosses had limited time horizons because the finan-cial rewards for both were mainly in the form of annual bonuses thatdid not take into consideration actual losses in future years.At Citigroup, Thomas Maheras was one of the highest paid employ-ees, earning as much as $30 million per year.One member of the team 90 t oo bi g t o s a ve ?in charge of collateralized debt obligations at Citigroup said: I justthink senior managers got addicted to the revenues and arrogant aboutthe risks they were running.As long as you could grow revenues, youcould keep your bonus growing. 49 Similarly, the senior officials at AIGhad huge fi nancial incentives to support Professor Gorton s modelsdespite the fact they did not include liquidity risks, as they were aware.In December of 2007, Martin Sullivan, the CEO of AIG, told investorsconcerned about mounting losses in its CDS business that ProfessorGorton s models helped give AIG a very high level of comfort. 50 Mr.Sullivan s total compensation for 2007 was $13.9 million.51The best practice in financial firms is to award bonuses on perfor-mance measured over at least three years to encourage a consistentapproach over a longer term.It is too easy for someone to have spectacu-lar performance in one year simply because they happened to work inthe hottest corner of a rising market, or they took huge short-term risks.In addition, a portion of a large bonus should be deferred for one or twoyears, with a potential clawback if the deals from the bonus year later startto fall apart.(For more on compensation incentives, see Chapter 11)The lesson: Insulate your modelers from sales pressures asmuch as possible.Base yearly bonuses on multiyear performancewith a portion of the bonuses deferred for one or more years.Mistake 8: Don t Be So Impressed with TechnologyThere is a tendency among many people to be overly impressed by thetechnology behind models as well as the experts designing these models.For example, many outside directors and senior executives of majorbanks did not actually understand the financial derivatives used by theirtraders.Yet this technology was so dazzling that these directors and exec-utives assumed that these derivatives were helping to limit the overall riskof their banks.In fact, while derivatives reduced the risks of some banksthat used them, they dramatically increased the risks of other banks.To take a simple example, if a bank owned a volatile stock, it couldreduce that risk by buying a put option on that stock at its currentmarket price.The put option gives the bank the right to sell the stockat that price even if the price of that stock later plummets.By con-trast, the same bank would be increasing its risk if it sold a naked put Credit Default Swaps and Mathematical Models 91option on the same stock without owning the underlying stock.In thiscase, the put option is simply a bet that the stock s price will remainabove the strike price of the put option before it expires.If the stock sprice nosedives before the put expires, the bank will incur a large loss
[ Pobierz całość w formacie PDF ]
