Risk assessment: like taking a look in the crystal ball
Can pandemics and wars be predicted? No, says ZHAW data scientist Pasquale Cirillo. He knows the mistakes made by those who can’t get by without forecasts and the traps that they fall into. And he knows what the data nonetheless reveals.
Not only did the virus spread at a dizzying speed at the start of the corona crisis – soon people could scarcely keep pace with all the graphs and tables that were being published. “It was amusing, but also somewhat scary to see how everyone was suddenly predicting the course of the pandemic”, says Pasquale Cirillo, Professor of Data Science at ZHAW. It almost seemed as if a whole swathe of self-appointed statistics experts had come into being overnight. And their prime qualification was that they had worked with Excel at some point.
“Individual risks cannot be equated with systemic threats.”
The researcher explains that the ever-increasing number of forecasts fell into two categories. In the first category, the risk was downplayed right from the start. Here, people liked to stress that Covid was no more serious than the flu, and the probability of dying from the virus was less than the probability of dying by falling off a ladder. “The problem with comparisons like this is that individual risks cannot be equated with systemic threats.” To give an example: if the woman in front of me at the Migros supermarket till is eaten by a shark the day after tomorrow, that is terrible for her but doesn’t change anything in my life. If she harbours a highly infectious pathogen, however, it's virtually impossible to estimate the consequences for me and all the others in the supermarket queue. In the second category, the meagre database sufficed to extrapolate figures straight to the apocalypse. It would only be a matter of weeks, it was claimed, before everyone in the world was infected by the virus.
“No one engaged in serious science makes point predictions”, says Cirillo, whose area of expertise is research into extremely rare occurrences which have a major impact. Their very definition means it is difficult to say when they will occur and what course they will take. “But, precisely in a crisis, that’s not what people want to hear.” In the early stage of the pandemic, experts were therefore under immense pressure to issue precise recommendations as rapidly as possible. People did not want to acknowledge the fact that experts can also be uncertain and unable to agree with one another.
Pandemics are unpredictable phenomena. They have repeatedly cost the lives of millions of people in the course of past millennia, yet “in most cases we were lucky and the chain of infection broke off long before a global crisis had developed”, explains the scientist from the ZHAW Institute of Business Information Technology.
“From a statistical point of view, there is unfortunately no indication that we are less inclined to violence than previously and that wars have become less likely.”
War is also one of the rare events that shake up the world for a long time yet defy prediction. How many of us had until recently sincerely believed that the deployment of bombs and tanks was a thing of the past – in Europe at least – that conflicts between countries are resolved around the negotiating table today and that we might perhaps even have become fundamentally better people? “I had naturally hoped that that was the case too”, the scientist emphasises, but from a statistical point of view there is no indication that we have become any less inclined to violence than in the past and that wars have become less likely. “When we analyse data, we must not allow ourselves to be guided by either hopes or fears.”
Together with Nassim Nicholas Taleb, risk and chance researcher and author of the international non-fiction bestseller “The Black Swan”, Cirillo recently drew up a list of major conflicts in the past two thousand years and studied these. Long periods of peace were not uncommon in the past either, the researchers found in their study. Examples are the Pax Romana and the Pax Britannica. “Two hundred or even three hundred years can elapse between two wars.” The fact that we in Europe have not experienced any armed conflict for seventy years – although that is not quite correct, Cirillo notes, since the Yugoslavian war was waged virtually on our doorstep at the start of the 1990s – might seem like an eternity to us. But that’s not the case.
“No one engaged in serious science makes point predictions.”
“I can’t predict when the next crisis will come, Cirillo says, but “what I do know is that there will definitely be one.” And something else that risk experts like Cirillo know is that, however rare such events are, they frequently display surprisingly stable characteristics. Future wars and pandemics are not likely to differ fundamentally from those that went before them. Or, as the Italian scientist puts it in a different study: if a comet is heading straight for Earth, we should not wait until the impact has occurred, saying that we want to find out more about the specimen in question. Precisely this, however, is what happened at the start of the pandemic. In many cases, decisive days and weeks were lost before the virus was brought to a halt, since it was felt that not enough data was available. But a lack of comprehensive information is typical of crises of this kind and is no reason for not taking action – quite on the contrary.
While we can’t prevent crises, we could at least cushion their impact more effectively. Good risk management, however, frequently takes time and is costly. The longer it has been since an event occurred, the more difficult it becomes to maintain preventive strategies and measures and continue justifying these, since at some stage no one will remember the rare event that triggered them. “We’ve seen this most recently in countries which had cut back on the number of beds in their intensive care units”, says Cirillo. Beds that would have been urgently needed over the past few years. Moreover, it is part of human nature to belittle even the biggest of risks once it has been accompanying us for a while. Just as children in war zones start playing between bombed-out buildings after a certain time, we all get used to the dangers around us. Science calls this risk homeostasis. You could also say that we simply want to live our lives.
To begin with, it has to be said that there is also no crystal ball that will predict the next financial crisis for us. There are naturally signs that have generally emerged prior to previous crises – and should also serve as an indication of crises to come, says Jörg Osterrieder, expert for Artificial Intelligence and Finance at ZHAW. These include banks suddenly being overgenerous in granting risky loans, or the value of shares or government bonds being far too high in relation to the underlying values. “But even these developments don’t say precisely when the next crisis will come.”
”Banks are doubtless better capitalised today and in a better position to absorb losses.”
Following the global crisis almost fifteen years ago, however, legislators and governments made great efforts to avoid future financial crises, or at least to mitigate their potential consequences. “Legal provisions governing the way banks deal with risks have been in place since the 1970s – that’s what is known as Basel I to Basel III today. “In recent years, however, the provisions have been extended to cover many more risks and products”, says the professor at the ZHAW School of Engineering. The regulations set out both quantitative and qualitative measures. Banks now have to examine the risks of an investment in greater detail and hold enough capital to offset any losses. This essentially means having sufficient equity. Financial institutions are also required to be more transparent toward the public and disclose key risk figures on a regular basis.
Osterrieder does not want to go as far as to say that these measures are the reason why we have so far been spared another global financial crisis. “Banks are doubtless better capitalised today and in a better position to absorb losses”, says the scientist summing up the situation. There are fundamentally a whole range of risks for a financial crisis, whether these be political and macroeconomic factors or suddenly emerging global changes.