The concept Var (Value at Risk) has been all over the news the past few weeks after JP Morgan announced it had lost $2 Billion on a trade, partially as a result of their VaR model miscalculating things. We even covered it here. But how many of us really understand what VaR is? Is there a real-world example for the rest of us? We think we might have one…
Last night, we had the opportunity to take several clients from Spain out to dinner. The 9 PM reservation was early for our Spaniard friends, if late for us. As an aside, we took our guests from Spain to an Italian restaurant in an area of Chicago called Ukrainian Village, but that’s a topic for another time. Anyway, we drove to dinner, and found a parking spot half a block away from the restaurant. Bingo.
Now, parking in Chicago works on a pay box system, where you put a credit card into the machine and get a small voucher/ticket to place in your window. Parking is enforced up until 9pm, at which point no voucher is needed. We pulled into the parking spot at 8:52 PM, and would have required a $0.25 or so payment to be in the spot for the next 8 minutes, after which the spot was free.
But, being folks who don’t shy away from risk, a little voice inside our heads was suddenly saying – there’s no way they are going to be checking windshields in this neighborhood for paid parking vouchers in the 8 remaining minutes until 9 PM – save your $0.25.
And thus, right there on Chicago Ave. at 8 minutes to 9 PM, we did a quick Value at Risk calculation. The internal risk model was telling us there was essentially a 0% chance of having to spend any money for that parking space – or, in VaR parlance, we had a 99.99% confidence level that we would not lose more than the $0.25 we were “earning” by not buying the voucher. So, we put on the “trade,” risking $50 to make $0.25, getting no voucher and headed into dinner.
You know where this story is going. At 10:45 or so, we came out of the restaurant to find a ticket on the car. That’s right – surprise, surprise – the true odds of someone checking windshields in the last 8 minutes of the night were actually a bit higher than we had calculated. The VaR model had failed, and we ended up losing 200 times what we tried to earn – losing $50 because we thought we had a 100% chance of making $0.25.
Now, if this were an exact analogy, we probably would not have the $50 to cover the parking ticket (despite having “earned” the $0.25 thousands of times before without a ticket), and would have had to be “bailed out” by a friend who loaned us the money at essentially 0% with no timetable to repay.
This is what is happening at the too big to fail banks. They are risking many times what they hope to gain based on the logic that there is a 99% chance they won’t lose what they are risking. But just like the parking attendant happening upon the car in the last few minutes of the night, the thing that isn’t supposed to happen ends up happening a bit more frequently than the statistics tell us it should – causing losses they usually aren’t prepared to handle.