Correlation is Not Causation

     It is one of the more vexing issues when considering any sort of problem or challenge. Just because something occurred does not mean that this something caused something else to happen. It might have but it will take considerable investigation to make that determination and it is very likely that other causal factors will emerge in the course of that investigation. A storm blows through and the roof of your shed flies off. It was the storm of course. Or was it? A deeper look shows the roof was not held on properly to begin with and then you forgot to close the door and the wind got in. All of these (and likely more) contributed. There are even weaker attempts to assign causation to correlation. The last time wind blew that roof off you had just put down seed on the lawn and now it has happened again – on the day you put seed down. Therefore, putting down seed makes storms come that blow the roof off your shed. You get the idea. On a much more serious note we are dealing with this conundrum when the pandemic is the topic.

At the start of the crisis in Europe the Swedish government elected to pursue a set of policies that differed considerably from those pursued by others in Europe – especially those other Scandinavian nations (Denmark, Norway and Finland). The result was that Sweden experienced a higher rate of infection, higher levels of hospitalization and fatalities. The instant assertion was that this decision to avoid a lockdown was the reason for the higher levels. It turns out that there are many other factors at work and any combination of these might be the reason for the higher levels. It is very likely that avoiding the lockdown played a role as well but the question is how much of one.

    This is only a partial list of the factors as determined by Joakim Book, Christian Bjornskov and Daniel Klein. Book is an Oxford grad and financial writer, Bjornskov is an economics professor at Aarhus University and Klein is an economist at George Mason University. Among the 15 additional factors to consider are relative population density of cities such as Stockholm, Copenhagen, Helsinki and Oslo. The Swedish capital is far denser than the other three. Factor number two is timing of a winter break. In Sweden, the breaks for the big cities are staggered so that Stockholm residents begin their break at the end of February and into early March. There were large numbers of Swedes from Stockholm visiting the Italian Alps at the precise moment that Italy was getting hit the hardest with the infection. Assuming that this is when there was exposure, the returning Swedes would be getting sick around the middle of the month. It was on March 12-16 that other nations were shutting down but if Sweden had followed suit it would already have been too late. The third factor was the size of the immigrant population. The percentage of people from West and North Africa is 9.8% in Sweden (3.0% in Finland, 5.0% in Denmark and 7.0% in Norway). The figures also show that people of African descent had a 50% greater chance of contracting the virus. This also played into the figures when it came to the elderly population. Sweden has had a higher rate of infection in the senior facilities than the others and there is much more dependence on the immigrant population to operate these facilities. These are the jobs the Swedes offer the migrants that come to Sweden.

    One of the factors highlighted by these authors has been the so-called “dry tinder” argument. Last year was an especially harsh year as far as the seasonal flu was concerned. All of Europe was hit hard and especially in some of these Scandinavian countries – much harder than in Sweden. More of the vulnerable population died in these nations in 2018 and 2019 than died in Sweden. The authors go on to detail a host of other factors that could have and likely did contribute to the differences in pandemic related deaths. This does not mean that the decision to remain unlocked was not a contributing factor as well. The point is that there are simply no simplistic answers or solutions to a challenge like this one or for any other challenge for that matter.

    There has long been an unwillingness to deal with complexity. We demand that everything be black and white and refuse to accept that real life is nothing but shades of grey. The pandemic is not going to be entirely dealt with by canceling events and keeping kids at home. It is not going to be entirely dealt with by having people wear flimsy masks. It will not be entirely dealt with through a vaccine – we know that even tried and true flu vaccines are not near 100% effective. It is a complex problem and it will demand complex responses. Masks may help a little and so might keeping kids out of school but there will also have to be acceptance that people will still fall ill and some will die. This forces a decision on balancing the costs of the cure against the cost of the disease and that is perhaps the most complex challenge of them all.

We recently developed a new product - Armada Strategic Intelligence System. It is a complex modeling of the economy made simple and accessible. The focus is on manufacturing and the forecasts we have developed have been in the 90% range and above as far as accuracy. See for yourself - we are offering a two month free trial - two issues of the ASIS, mid-month updates, podcasts and more. Drop me a note at [email protected]