Extreme Risk: Endogenous risk

A data-driven risk management process can not capture endogenous risk, only exogenous risk.

The following is taken from my book Illusion of Control

Risk can either be endogenous or exogenous.

The distinguishing feature of all serious financial crises and extreme tail events is that they gather momentum from the endogenous responses of the market participants themselves, like a tropical storm over a warm sea gains more energy as it develops. As financial conditions worsen, the market's willingness to bear risk evaporates because the market participants stop behaving independently and start acting as one. They should not do so, as there is little profit and a lot of risk in just following the crowd, but they can be forced to by circumstances.

A dictionary definition of the term endogenous refers to outcomes having an internal cause or origin. How an infectious disease spreads through a population is endogenous to the nature of that same population. If we always keep a safe distance between ourselves and our fellow countrymen, we will not get infected, but if we choose to live cheek by jowl with other people, our chance of infection is high. Your chance of getting a cold is endogenous to your behavior and those around you. One reason why taking the New York Subway can be hazardous to one's health. Why social distancing was so important in the Covid-19 crisis.

The opposite of endogenous is exogenous, where outcomes have an external cause or origin. When an asteroid hit the Gulf of Mexico 65 million years ago, wiping out the dinosaurs, that was an exogenous shock. There was certainly nothing the dinosaurs did that caused their demise.

Suppose I wake up this morning and see on the BBC website there is a 50% chance of rain. If I then decide to carry an umbrella when leaving my house, my doing so has no bearing on the probability of rain. The risk is exogenous.

Suppose, instead, I wake up this morning and see on the BBC website that there is some negative economic news about the United Kingdom, and in response decide to buy a put option on the pound sterling — I profit if the pound weakens. My actions make it more likely the pound will fall. Not by a lot, mind you, a tiny, tiny amount. But tiny is not zero — there is an endogenous effect.

I would not be doing anything wrong. On the contrary, I am behaving prudently, hedging risk. Another ingredient is needed, some mechanism coordinating the actions of many people so when we act in as one in the market, our impact is strong. The chance of that happening is endogenous risk. All that is needed to turn some shock into a crisis is for a sufficient number of people to think like me (all wanting to prudently protect themselves) for the currency to crash. It is the self-preservation instinct of human beings that so often is the catalyst for crises.

No crisis is purely endogenous or exogenous, they always are a combination of an initial exogenous shock followed by an endogenous response. The same initial exogenous shock can one day whimper out into nothing, and the next blow up into a global crisis. When Covid-19 infected the first person in Wuhan, all the subsequent events could have been averted if that person had behaved differently. But nobody knew at the time. The reason why the exogenous shocks blow up into a crisis is that they prey on hidden vulnerabilities no one knows about until it is all too late.

A common view about crisis is that they arrive from the outside, just like the asteroid hitting Wall Street. That is not true. The main driver of crises is endogenous risk, underpinned by the system's hidden mechanisms, like sell-on-loss trading rules. Unfortunately, there is too much tendency to focus on the triggers of crises, not the mechanism. The problem is there is a tiny number of mechanisms but an almost infinite number of triggers. They are visible, but not all that important. Some are even human.

We confuse the trigger with the underlying cause. In the 1914 crisis, the assassination of Archduke Ferdinand, and in 1763, the end of the Seven Years' War. In both cases, a crisis was inevitable: it was not a question of if, but when.

We should not focus on the trigger but on the underlying vulnerability, the mechanisms that cause crises. If one starts getting too concerned about the triggers, we risk what my skiing instructor once called paralysis by analysis. If one thinks about skiing down a one kilometer 45° slope, one may conclude it is best to take a lift down. Paralysis by analysis. It is the same with triggers because there are so many of them; if you start searching for and protecting yourself against all potential trigger events, that will end up being all you do.

Focusing on the triggers can lead policymakers seriously astray. But that is often what they end up doing. The press reports on the triggers, the public demands someone is blamed. Meanwhile those responsible get off scot free. Our attention should be focused on the mechanisms, and the financial authorities and hedge fund managers should know that.

Crises take (almost) everybody by surprise. It could not be any other way, because if many people anticipate a crisis, they will get out of the market — either preventing the crisis or causing it to happen immediately.

However, even if we are surprised, crises don't happen randomly. They are the culmination of risky behavior over years and decades. Few people have said it better than the former head of the Bank for International Settlements, :

The received wisdom is that risk increases in recessions and falls in booms. In contrast, it may be more helpful to think of risk as increasing during upswings, as financial imbalances build up, and materializing in recessions.

In a paper I wrote with Hyun Song Shin and my LSE colleague Jean-Pierre Zigrand, we applied endogenous risk to the measurability of risk, classifying risk into two categories, perceived risk, what the riskometer measures and actual risk, the underlying hidden risk that is undetected until it suddenly materializes in a crisis.

Riskometers, by definition, can only measure perceived risk because they are based on using fluctuations in market prices to infer the riskiness of the underlying assets. If the fluctuations are low, so will the risk measurements. If the market is confident, prices are rising, fluctuations are low, and perceived risk low.

The use of riskometers based on perceived risk is one reason why bubbles happen. Market prices get increasingly disconnected from the underlying reality, and the risk of a significant correction (actual risk) steadily increases. Suppose some investors buy an asset. Consequently, the price of it increases slightly, which in turn encourages more people to buy, making prices rise a little more. The result is a series of small price increases, which, when translated to price fluctuations, imply volatility is falling.

If we then use a common device to measure risk adjusted returns, the Sharpe Ratio, proposed by the Nobel laureate William F. Sharpe, the asset looks better and better. Prices increase steadily, perceived risk falls — a fantastic investment, steady profits at low risk. Money for nothing, as the 1980s Dire Straits song would have it. A bubble born by the positive feedback between rising prices and falling perceived risk.

Prices continue to rise faster and faster until some trigger event causes them to crash down. Because everybody wants to sell before the prices drop, they all rush to the exits, and the prices crash immediately. When things are good, we are optimistic and buy, which endogenously increases prices — the bubble feeds on itself. This eventually goes into reverse, and negative news prays on falling prices, with the markets spiraling downwards much faster than they went up. Yet again, prices go up the escalator and down the lift or elevator if in America.

It is only after the prices have crashed that perceived risk increases. By then it is too late.

So, what happens to actual risk through all of that? It increases along with the bubble. After all, it captures the fundamental risk, the risk of a market crash, hence falling when the bubble deflates.

I often wish someone could have convinced the financial regulators that the risk after 2008 was much lower than the risk before and they should have been encouraging risk taking and not discouraging it.

Endogenous risk doesn't usually leave much of a footprint on the financial markets. While it lurks in the dark corners of the system, most of the time, all we see is exogenous risk. When markets appear to be efficient, with prices that follow random walks, as in Burton Malkiel's fantastic book A Random Walk Down Wall Street. But the good times don't last. Endogenous risk will show its face when some trigger pushes us to act in concert with each other, like the pedestrians on the Millennium Bridge. Price movements will be amplified — bubbles and crashes.

The spirals of coordinated selling that are unleashed by endogenous risk are normally held back by the inherent stabilizing forces in the markets — the arbitrageurs, the hedge funds, the sovereign wealth funds, the Warren Buffets, the Soroses. They step to the plate in crises to buy cheap assets, putting a floor under prices. An excellent expression of this is from Baron Rothschild in the 18th century, who supposedly said: "Buy when there's blood in the streets, even if the blood is your own." The self interest of the speculators benefits society, as in Adam Smith's classic statement, "It is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own interest."

What lets endogenous risk off its leash is decisions and policies that serve to harmonize market participants' behavior. Rules that prevent them from putting a floor under the markets by buying all the assets that are so undervalued because of the crisis.

The riskometer puts the contrast between endogenous and exogenous risk in the sharpest relief. Almost all methods for measuring risk are based on the assumption that risk is exogenous, because that is the easiest way to deal with risk. All one has to do is to collect some daily historical financial data — market prices, credit default swap spreads, interest rates, trading volumes — and feed them into a riskometer. Nothing wrong with that if all we care about is exogenous risk — short term fluctuations.

If, however, we care about tail risk, banks failing, and crises, we have no choice but to find some way to measure endogenous risk. That is not easy. Large losses and crises happen because of risk everybody missed, so identifying endogenous risk before it is too late is like searching for a needle in a haystack, except that we don't even know how the needle looks like. We only know that the needle is there when we stick our hand into the haystack, and it gets pricked.

Of course, after a crisis, everybody knows what went wrong, and we prevent a repeat — closing the barn door after the horse has bolted. Meanwhile, the forces of the next crisis start gathering strength somewhere else where nobody is looking.

The 2008 crisis is an excellent example of how we missed all the warning signs. While I suspect many people thought financial products based on the American housing market were dodgy and opted not to buy them, a few profited from it. The book and movie "The Big Short" by Michael Lewis tells the story of a few plucky players who did exactly that. Meanwhile, no government authority had any inkling of the danger lurking right under their noses.

The lesson from 2008 is that the buildup of endogenous risk happens almost wholly out of sight. Eventually, the vulnerabilities hit the hidden trigger — risk got a little bit too high on a certain day when the markets had little tolerance for it — and it all blew up.

The challenge for investors and the financial authorities is that while endogenous risk is everpresent, it cannot easily be measured. As endogenous and exogenous risk move in opposite directions, the riskometers get it wrong in all states of nature, reporting too little risk before a crisis and too high after, just like the ECB's systemic risk dashboard.

Even though endogenous risk gives a useful description of the world, it is difficult to work with in practice. It cannot readily be quantified or incorporated into riskometers or put onto risk dashboards. There is no box for endogenour risk on these pages. The general practice is to assume risk is exogenous, that it arrives from the outside the financial system, impacting on the system but not being affected by it. Like an asteroid might hit Wall Street. All my numbers here only capture exogenous risk.

As a practical matter, it may be perfectly OK to assume most risk is exogenous. It just depends on our objectives. Anyone concerned with day-to-day risks, or risks of frequent small losses, would not be very far off by focusing on exogenous risk. It is only those who care about tail risk — the infrequent large outcomes that cause significant losses, the failure of banks, and financial crises — who should focus on endogenous risk.


Extreme risk

Daily market risk forecasts and analysis
© All rights reserved, Jon Danielsson,