Months Before AIG Went Bankrupt

A Story About Unexpected Risk

It was at the beginning of 2008, at our downtown office on Pine Street in Manhattan. I was a young quant portfolio manager at AIG and I was standing in the office of one of my mentors, a talented senior portfolio manager who was looking nervously at his screen.

“Close the door please,” he told me briskly.

We had a habit of informally catching up about once a week and chatting about all the good and bad of our company, the economy and everything else. This time, he was not smiling as he typically did. He kept looking at his screen and proceeded to say words that shook me:

“AIG is going to go bankrupt soon.”

With my back pressed against the door, I looked at him with disbelief. At the time, AIG was AAA rated, the largest insurance company in the world. Saying that AIG was going bankrupt felt pretty much on the same order as saying that the U.S. government will default on its debt.

“How could it be?” I asked.

He took a long pause, looked at me and said “Our derivatives risk is massively mis-priced.”

I asked whether he was alerting everyone. He said management thought he was being too negative and didn’t want him to spread rumors that might impact morale and growth.

I walked out of his office feeling like a rug had been pulled from under me. When you are exposed to information that is many standard deviations outside of the expected norm, the brain takes a while to adjust.

The feeling reminded me of an episode just 5 months prior to that, when the weekly return on popular quant strategies like size, value and momentum melted down multiple standard deviations. This was not supposed to happen, but it did, dragging down the largest hedge fund of the time, Global Alpha run by Goldman Sachs - an unimaginable event just a year before that.

Unexpected extreme outcomes often have devastating consequences, unless your risk management process has considered and prepared for them. The irony is that AIG’s long time CEO Maurice Greenberg had a very clear understanding of such ‘Black Swan’ risks long before Nassim Taleb published the theory.

In fact, I had a personal encounter with Greenberg’s focus on such risk in early 2005 when he asked our quant team to look at potential investment hedges against the worst hurricane the U.S. could possibly endure. He defined it as something like a ‘complete wipe out of all coastal cities of the eastern U.S.’ When this ‘special Greenberg project,’ came down to our department, everyone was smiling, thinking he was getting too paranoid. Nothing could hedge such a disaster, we assumed.

Nevertheless, I ran a bunch of simulations using various investments like cat bonds, gold and government bonds. The hedge that made the most sense to me at the time were options on VIX futures. But when we asked how much notional he would want to hedge, he said a 100 billion dollars. That was clearly not a feasible VIX options trade at the time, so we wrote up a report and sent it ‘upstairs.’

Although everyone treated these exercises with skepticism, it left an impression on me. I later heard my colleagues working on similar projects, hedging ‘epic bird-flu epidemics’, for example. Most of us would give up in these scenarios, but Greenberg wanted the firm to survive under all circumstances.

Unfortunately, this risk-focused culture was dismantled after Greenberg was ousted in 2005. (As an aside: he gave his televised goodbye address from the corporate communications broadcasting room a few steps away from my desk and I’ll never forget watching him walk out, surrounded by body guards and lawyers, with his ever-present steely smirk and sparkling eyes).

My ‘tribe of mentors’ (to borrow Tim Ferriss’ term) consisted of risk managers whose risk-management department was closed slightly prior to Greenberg’s departure.  They were an exceptionally bright team of risk-quants and took me under their wing for informal mentoring. Their outlooks were mostly conservative and bearish. They were skeptical of anything new, quickly could see flaws in any argument and were perfect for the risk-management roles.

After the department was closed, they were transferred to run various profit centers and investment strategies. It was one of these ex-risk-quants who warned me about the looming AIG bankruptcy. Back in the Greenberg days, this information would have made its way straight to his desk and been taken very seriously. Sadly, by 2008, management’s focus was all on the upside, while forecasters of downside risks were accused of pessimism.

Several months later, Bear Stearns had gone under, the financial world was falling apart, and Lehman Brothers and AIG were collapsing because of the ‘mis-priced derivatives risk.’

Seeing such unexpected events manifest in 2008 has forever impacted my way of seeing risk. I started looking at the investment models I was responsible for to see how they would fare in extremely unexpected events. This resulted in massive diversification in my models. In addition, by late 2008, I was looking at how often popular quantitative strategies crashed. For example, I had to go back 200 years back to see what the worst case looked like for the Price Momentum strategy. I discovered frequent and previously unseen crashes, just in time before Momentum proceeded to crash in March 2009 (See a detailed post here).

Understanding risk doesn’t mean you should be overly cautious, just that you should have a risk strategy in place to handle unexpected outcomes. The unexpected risk that I am discussing here is not always about the downside either – it can show up on the upside as well.

As the current bull market evolved, I started to see the same unexpected risk of missing the upside of the ‘market melt-up’ because of increased investor risk-aversion post 2008. Between 2009 and 2019, there were many instances when it was ‘clear to everybody’ that the expansion is over and that another 2008-like crash is right around the corner, and yet the market proceeded to go up, evolving into the longest bull run in U.S. history!

Early chatter about a massive upcoming market ‘melt up’ started to spread, when a few outlier super-optimists started to predict the market to double as early as 2015. Just like my mentor’s forecast of bankruptcy felt like an impossible outcome, so did these forecasts which arrived at what felt like the top of the market. By 2018, many respected mainstream players were talking about the melt up as the market continued to rally.

Missing the unexpected upside of the last decade has cost investors a huge amount of wealth left on the table both by bad market timing, over-diversification and the resulting low-return risk, worsening the unfunded liability problem for asset-owners. For example, on average 28,000 endowments and non-for-profits in the U.S. have earned only 3.8% per year since 2009! v.s. a 60/40’s return of 9.3% during the same time (See a detailed post here). You can still see this same downside bias today. Interestingly, Mr. Optimist is still holding on to his ‘melt up’ thesis even after the 2018 correction.

On the opposite end of the spectrum, last week’s post from Ray Dalio about broken capitalism is a good example of the downside Black Swan thinking. Reading it elicits a similar anti-optimistic, morale-dampening feeling to that evinced by AIG management when my mentor raised his concerns. Who wants to hear all the comparisons to the rise of nationalism and the Great Depression of the 1930’s? It seems far-fetched.

But the point of thinking about unexpected risk is not to spread fear and pessimism (or greed) but to cure our blindness about the really unlikely “what if’s.” Fear leads to bad investment decisions because it causes us to act emotionally and to over-estimate the probability of such rare events (i.e. sell off all investments, move to the woods, buy guns, water, canned food, and gold). The point is to recognize that such events are extremely rare, but not impossible, pointing to the need to create a plan and process for what to do before and during the times when such outcomes materialize.

And if you are in a position of power to prevent the bad outcomes, and not fall under the denial risk: listen to the super doomsday people (and the super optimists) and then design solutions to put a floor under the maximum worst case while allowing the system to participate in the upside.

When the new AIG CEO was appointed in the summer of 2008, he began a slow strategic review of AIG’s businesses. To many inside the company, it was obvious that our financial stability was threatened, and it was time for urgent radical action. As the strategic updates rolled in from management, the investment folks grew impatient for action. Eliciting an image of a little boy playing basketball who is panic-frozen because he just got a great pass right next to the hoop - We started joking: “Johnny, shoot the ball!” By the end of the summer, the government stepped in to bail the company out.

The takeaway about the unexpected risk is not to try to predict it, but to be prepared for it to show up, both on the downside and the upside. The investment process needs to be flexible, dynamic and anti-fragile enough to be able to withstand the very unexpected.

Which totally unexpected outcomes are you not seeing today that may be happening a few months from now?

PS. Thank you Craig Smith from for editorial review.