The perennial predicament of planning for low probability high impact events

Not so easy

This isn’t another blame-game, or “I knew it all along!” article. It's about an observation that sometimes, long term planning is hard, and different places handle the dilemma differently.

Arguably we’re in the middle of a low probability high impact event – but that said, probability is a function of the time over which things are sampled. One or more global pandemics within any given century is actually a quite likely occurrence. That’s why it has long topped the league tables of such high impact events. The probability of it happening in any given year though is low, and perhaps more pertinently, so is the probability of it happening in any given electoral term for any of the world’s democracies. Perhaps even more pertinently, the phenomenon can go a generation or two sometimes without surfacing, so it might not always be very prominent in living memory, excepting perhaps our eldest.

New coronavirus has all our attention at the minute, but the family of low probability high impact events is a wider issue worthy of contemplation. There are many examples of these events, and professionals of all hues – geoscientists, meteorologists, engineers, and pathologists, are dealing with them all the time. 

Shaking the foundations

Earthquakes are an obvious example. San Francisco, Anchorage, Mexico City, Tokyo, Wellington, Istanbul – these are cities that invest in planning for large earthquakes, and it makes sense. Large earthquakes are known to occur in these places with regularity and very big events are possible. That means the case for allocating spend on defence against earthquakes is very logical and is easy to defend. But let’s tweak that theme a little and turn it into the earthquake threat for London and New York. These are places where earthquakes are not impossible, but we know from the geological and historical seismicity record that they are far less frequent, and events there are unlikely to be as big. Yet if such an event was to occur, they could still be high impact, even for a moderate size. A big magnitude 5, or a magnitude 6, close to these venerable cities, where much of the infrastructure is old, and was never really designed with earthquakes in mind, could cause a lot of damage. But it might be centuries before such an event occurs, or it could occur tomorrow. So how much money do we invest now in preparing for such events?

Such questions are not easy to answer. Sometimes just having a think about it, a plan, is enough. Is that always the case though? When does an event become sufficiently high probability and sufficiently high impact to justify real infrastructure changes and investment from an already burdened taxpayer, providing more effective protection?

Shut down temporarily, or spend up permanently?

There are other examples that help illustrate the problem. Consider snowfall in the UK. Every time the UK has a major disruptive snowfall, especially the more populous areas of England, there are cat-calls from the tabloid press and mockery from snowier countries about how the country is floundering in what for many would be a relatively routine event. Yet it’s really not quite as simple as the lazy stupid unpreparedness it is made out to be. Do we really want all of our local councils to be investing in highly expensive snow clearing machinery for events that might happen just a few times each decade?  Are we truly prepared to pay the extra rates and taxes to fund ubiquitous snow clearing equipment, for events that are only rarely disruptive? Or does it actually make much more sense just to accept every decade there will be a few down days for this reason? The equation is very different for cites like Moscow or Chicago, that regularly get heavy snowfalls every winter – there the justification for investment in such equipment is obvious. The problem is not one of technical do-ability, but one of economic justification.

Machine matters

Another facet is how much investment and effort we put into designing our machines for every eventuality. Many risks can be dealt with feasibly in a technical sense – but they virtually always introduce extra costs. How much money do we invest in a car design to make it safe in a crash? A lot actually, because we know these are events that happen with alarming regularity. Moving it up a notch, how much do we spend on our aircraft to make them fail-safe? Again, actually quite a lot – air travel has, current situation excepted, been so ubiquitous that we all want to feel safe both in aircraft and under them. Yet can we really spend inordinate amounts on every accident eventuality? This is not as easy a question as we might imagine. It’s a testimony to how much we do spend on such things that air travel has become as safe as it has – but I’m guessing there are all sorts of things that could be done, that aren’t done, and which are instead treated as an acceptable risk by ourselves in order to keep air travel cheaper. 

An interesting illustration is the concept of bird strikes and two engine passenger jet aircraft. Virgin Airlines for some time made quite a thing of fairly uniquely, only having four engined aircraft – both its 747’s and A340’s. It didn’t last as the economics of four-engined aircraft became harder to sustain amidst rising fuel costs. The safety arguments didn’t change; it would still in theory be safer to land through a flock of geese in a four engine jet than a two engine one – but the chances of an event happening where it would actually make a difference were (presumably) felt to be so low as to not warrant the extra costs involved. At least that is the decision that was made across the industry, and most of us board two-engined jet aircraft of any reputable airline without too much concern. The chances of two engines going out simultaneously due to bird strike in such a way that a third and fourth would survive to save the day, are so small as to not warrant undue worry. That is not the same as having a negligible chance of occurrence. It could happen.

COVID context

What implications do these planning questions have for something like our current COVID nightmare? How much money do we spend preparing for an event, in any one given year, for something that might only occur two or three times a century? We now know all too fully how the impact can be devastating. No-one would argue today that it’s not a high impact event. I’m not suggesting I know the answer to this difficult question, but one thing is sure, and that is - clearly we have all have to keep our guard up and keep asking such questions. The challenge is always to think in longer time scales than that of the political cycles we build into our democracies. Amidst the pressures that every government experiences to divert cash to the loudest shouters, to hold that resolve and keep some set aside for ongoing spend - spend that prepares for the highest impact low probability contingencies. Easier said than done. 

What proportion of income and for which contingencies is not trivial to work out - there are many, but scientists, mathematicians and models exist to allow it. The talent is out there if the question is asked. It will always involve uncertain assumptions and estimates of probabilities and impact, but it is better than nothing. Some events, let's be clear, will swamp any amount of planning. We are not so omnipotent a species that we can prepare meaningfully for every conceivable catastrophe. That’s not an excuse for not trying, especially those ones that have a good chance of reduced impact when we do prepare. For all the competition for government funds in normal times, it has been striking over the past weeks how much funding can be made available when emergencies arise. Not just in one nation, but in many. It does inevitably beg the question of whether adequate spend is being allocated preemptively on protection. 

Clearly a lot of questions will be very rightly asked about these events in the years to come concerning what could have been done better. The differences in preparedness and reaction style of various countries will be put under the microscope again and again. Predictions of pandemic havoc made before the event will be dredged up and many will shout loudly that they, they told us so.  Fair enough in many cases, but it is worth making the observation that in the world we live in, whatever disaster befalls us, it's likely someone predicted it some time ago. 

The question that is most relevant though, is less “did someone say it would happen” than the question “what probability and cost was assigned to this event beforehand, and were those figures correct?”. That is a much more meaningful question in that it allows a real analysis of whether correct judgements were made and acted upon. If no numerical estimates were made, just a qualitative “this might happen”, then it is harder to argue that a genuine warning was given. Lots of things might happen, but to guide a government’s spending and decision making, a cost and a probability needs to be attached. If that is then ignored, then any criticism has a much stronger traction than otherwise.

Different places are well, different.

Who can fail to notice today that Germany had five times the number of intensive care beds per capita going into this crisis than the UK.  It’s notable that the US has nearly the same number (per capita) of intensive care beds as Germany, near the top of the league table, but the funding of them and access to them is structured very, very differently. Who can fail to notice that nationwide lock-downs are being given very different treatments around the world, with countries openly making radically different calculations about the priority of short and medium term economic impacts. It is a different juggling of the threats from coronavirus and the perceived threats from social/economic disorder, neither of which, it has to be said, are trivial, and both of which have costs in lives.

Clearly administrations do things differently, and for better or for worse, when all this is over they will all be engaged in a lot of explanation to justify the decisions they made, both before and during the event. I’m not here to encourage recriminations but I am interested in this notion that different countries have very different attitudes to long term "insurance" planning against such events and investment in protective measures. Perhaps the differing nature of the various economies and their demographics has a role in influencing that. 

There may be a very real sense in which a country’s economic “character” has strengths that empower fantastic dynamism and rapid economic response, but which de-prioritises setting aside funds for events that in any one year might be considered low probability. It may be genuinely difficult to both have that cake and eat it. 

The wider question

It is a safe prediction that here in the UK, after the dust has settled on the worst of these Covidian events, (whether that is two months or two years from now), it will be suggested that we weren’t as prepared as we should have been. It's hard not to feel health facilities were allowed to descend from earlier levels that should, at the very least, have been maintained. It would be a shame though, after all this, if we didn’t widen the remit of the question a bit further. Let's ask the bigger question so that it deals not just with pandemic, but with a generic strategy for all those events that in any given year are in the low-probability high-impact genre. Estimating the appropriate levels of funding to spend on them in preparation being a key objective. 

These are I suppose questions for later. For now, it is too late to think about it in our current situation, and we are all compelled to do the best with what we have, and to protect as best we can those most vulnerable. Our leaders, whatever their hue, whatever nation they lead, and whatever our view of them, face unenviable decisions. Unless they are made of stone, they are also feeling intensely the very human tragedies occurring around us all. 

Questions will be asked afterwards, painful ones, but I think it is also helpful to sit up and say, without any presumption of knowing any of the answers, that planning for these kinds of events is a difficult problem. Hindsight is a wonderful thing. When the dust does settle, thinking more widely about this genre of event, and not just pandemics, will be worthwhile. For if we do get through this test, it might not be a pandemic that challenges us next.