Risky Business

In the state-controlled Straits Times today, the newly reinstated CEO of Temasek Holdings announced a greater than $40b loss in Temasek’s total portfolio for the last financial year. Temasek Holdings, keeper and executor of a significant portion of our nation’s wealth, reported an annual VaR of $40b which was lost amidst the financial tsunami which originated in the States and swept through financial markets across the Pacific and beyond.

Supporters of Temasek Holdings will argue that Temasek is but a victim of the excesses, greed and irresponsibility of Wall Street. They will claim that the loss is but collateral damage, a price Singaporeans have to pay to stay invested in the markets. The truth is that Temasek waded waist-deep into the financial abyss that is Merril Lynch and UBS with amazingly little foresight. They should rightly be entirely responsible for its tremendous losses instead of trying to shift the blame to the financial maelstrom of the past year.

There is plenty of chagrin and widespread criticism of the stupidity of Temasek’s investments thus far within the blogosphere. I do not wish to add to this growing community my opinion and disappointment of the collective idiocy of what many consider the best and brightest minds in Singapore. Instead I shall delve deeper into the intricacies of the risk numbers quoted in the Straits Times today and demonstrate that either Temasek Holdings is taking on excessive risks with what is essential the people’s money or that risk controls within Temasek are not only ineffective but downright meaningless.

The CEO of Temasek Holdings reported an annual VaR of $40b for the last financial year. She went on to explain the VaR figure implies there is a 16% chance that the portfolio will drop by $40b in the period. This is rubbish!! Either she has no idea what she is talking about or she thinks people who read the papers have not enough knowledge to differentiate nonsense from news. Let me explain what the VaR figure is actually implying.


Value-at-Risk simply tells us the maximum amount of loss possible at a predetermined level of certainty. There is a level of certainty associated with VaR figures. Statisticians call this the “confidence interval”. Typical conventional VaR models adopt a 95%-99% confidence interval. That is to say the model computes the maximum loss that will not be exceeded 95%-99% of all the possible economic/financial conditions. Only in the extreme circumstances of that 1%-5% probability, will the losses exceed the VaR figure derived from the model. These extreme events will occur in the tiny left-hand tail of the probability distribution diagram shown above. The exact derivation of the VaR number may be immensely complicated and will involve analysis of serial and matrix correlations of the portfolio constituents. The VaR model may also embed other statistical and stochastic models in its simulation. However it is not necessary to fully comprehend the underlying mechanics used to generate the VaR number. Once the final figure is derived, the significance and application of that statistic is simple and intuitive.

Let us apply the statistic to the case of Temasek. It is reasonable to assume that Temasek, like all major financial institutions, adopts a similar VaR model like the one I described above. Therefore, it is entirely wrong to claim that there is a 16% chance that the $40b loss will be incurred. In fact, the probability will be closer to 1%-5% instead of 16%!! Why then is a high probability of 16% mooted in the news? Let me provide a reasonable explanation. What Temasek is implying when they claim 16% probability of loss, is that the investments they made is not really of a high-risk nature. They are saying that the portfolio value have declined because the broad market have turned south by roughly 35%. The losses are kind of in line with the broad market decline and thus ordinary and non-excessive. This is definitely not what the VaR figure and I will explain why.

With a VaR figure of $40b, with a reasonably assumed confidence interval of 95%-99%, Temasek has in fact taken on excessive risk as compared to the broad market investor. For the losses to hit the $40b VaR mark, the applicable market decline with respect to Temasek’s investments is closer to 45%-49% instead of the broad market mean of 35%. In other words, Temasek, the main custodian of the collective wealth of our people, have punted and lost 10%-14% more than the already huge losses of the common broad market investor over the last financial year!! Temasek was essentially placing huge bets with the peoples’ money and these bets are effectively 50% more risky than the common broad market investments. Where else in the world can you find such careless and irresponsible financial management that goes literally unchallenged and unaccountable for but in sunny wonderful Singapore?

There is another angle to this issue. It may be possible for the VaR of $40b and loss probability of 16% to be correct. But this will lead to an equally startling revelation as the one above. For the risk numbers to be accurate, Temasek will have to be running their VaR models at a 1 standard deviation level. There is probably no respectable financial institution in this world that has an equivalently lax risk control system! A VaR figure generated at a 1-sigma level will be close to being a completely useless and meaningless statistic. At 1-sigma level, the fluctuations of portfolio value over the year will exceed the critical VaR threshold more than 65% of the time!! What good is a limit if the limit is breached more than 65% of the time? To make the VaR more useful and informative, the model has to increase its confidence level. In that case the VaR number Temasek should be quoting may in the region of $100b, 2-3 times that of $40b! The current portfolio value of Temasek is around $190b. Imagine you have a CEO, making investments that can possibly lose more than 50% of the total company’s value if conditions turn south, who doesn’t even understand the basics of risk management!! I shudder to think of the consequences.

Moreover, the regular review (be it monthly/quarterly/semi-annually) of the company’s portfolio should have alerted the relevant authorities within Temasek. Even if we accept the ridiculously low standards of a 1-sigma $40b VaR model, the progressively deteriorating credit conditions over the past year would have push the VaR figure way past the $40b mark. Why wasn’t there any significant move by Temasek to bring the VaR down to the original $40b threshold? Surely the potential losses from capital depreciation far outweigh the transaction costs involved in the hedge. It should be the duty of a SWF to safe-keep the nation’s reserves and not take unnecessary risks, isn’t it?

Two images flashed across my mind as I think about the reasons for such inactivity. The first is that of a nervous punter that stubbornly held on too long to a losing position in the hope that the markets will turn and he will not have to face the consequences of a margin call. The second is that of an oblivious executive playing golf in his country club and chatting happily with his other well-heeled counterparts. How many people see the same images as I do? Surely there ought to be some consequences for such careless and irresponsible behavior.

There may be some higher and more sophisticated reasons as to the actions and decisions of Temasek. A humble commoner like me may not be able to comprehend. But please at least take the effort to come up with a semblance of a more reasonable excuse for a job poorly done instead of the constant jibberish we see in the news. The truth of the matter is that the business of Temasek is indeed a very risky one and we need to know more about the truth behind the facade of control and prudence which Temasek clearly lacks