Acceptable risk

download this column in pdf

	I like this column
	869 persons like this column

About 2 weeks ago a party tent collapsed at the DickyWoodstock pop festival in Steenwijkerwold due to a local tornado[1]. With the drama of Pukkelpop in Hasselt at the top of our minds you fear many fatalities, but fortunately there were none. About 13 people got seriously injured and needed treatment in hospital, but again, it was not as bad as it first seemed to be. The next day 11 of those were discharged and sent home. After the skies cleared, questions were asked whether the festival should not have been aborted given the approaching thunderstorms. The organization replied that it was known that bad weather was coming, but there was no indication at all that it would be this severe. The tornado was extremely local, like on Pukkelpop in 2011. According to local government,  if the tent had been 100 meters away nobody would have been hurt. Police did not see any need for further investigation. Apparently, such extreme events are seen as bad luck that happens once in a while. In other words, an acceptable risk.

Completely different was the message of last Wednesday evening. Becaise of approaching extreme weather, the NS (dutch railway operator) announced an adjusted train service. In case the perfect storm would really strike, it would be easier to contain problems if there were less trains running. Otherwise, any problem would propagate very fast through the network. As the NS has seen quite a number of the train service collapsing after bad weather it seems fair to state their expertise in this field. This was a risk they did not want to be exposed to. As the summer months are not the busiest in terms of traffic and NS did not decide on a reduced service in the summer months like they did in previous years, it seemed a reasonable measure. However, the weather turned out not to be as extreme as expected, and the measure would not have been needed. As they say, hindsight is always 20/20.

The problem can be approached from a different viewpoint though. And it would not be the Netherlands if there was nobody willing to do so. Regarding the NS, there is ROVER to rely on for criticizing the NS. In this case, they were in disagreement with NS from the start[2]. They used 3 arguments to support their position. The first was that the service level was adjusted way too quickly if something bad threatened to happen, and that escape route should be blocked. The second was that bad weather was a foreseeable event and the system should be able to cope with it. The third argument was that it did not matter how much traffic there was, if a tree would block the railway, nothing could pass it and the system would collapse in the end. A cynic might add that the only reason to reduce the services was that it meant that less trains could be late so that the bonus of top management would not be put  in jeopardy[3].

To return to the arguments of ROVER, on first sight they seem reasonable. But a more thorough analysis reveals some flaws in their arguments. That a measure has been used often in recent history (perhaps even while not needed) does not mean anything for the actual situation (reasoning from the perspective of Wednesday afternoon). The second argument also does not hold. It is fine if people have the opinion that the system should be able to deal with extreme weather. But experience shows that the system cannot cope (at least, not now) and given experience on what the system can deal with, any opinion on what the system should be able to deal with is irrelevant from a risk management perspective. Facts prevail over opinions. It is like legally forbidding bad weather, utter nonsense. But the third argument makes sense. It the probability of a system crash does not depend on the amount of traffic, the measure is completely redundant, except perhaps with regard to the time it would take to get all trains on the position where they are needed the next day. But that is an interest of the NS and not the interest of the passenger. ROVER does have a point. But now I am interpreting. I do not know whether ROVER had this intention. Given the dissatisfaction with the performance of NS in the past years,   ROVER may have been using arguments without a full analysis. 

Yet, my sympathy is with NS in this case. A small, announced and predictable nuisance does not seem a terrible sacrifice to limit the potential for a system crash. If NS would have accepted the risk by not reducing the services and the system would have crashed because of bad weather, ROVER no doubt would have been the first to condemn the NS because of their inability to deal with it. And suppose that bad weather would have struck but the system would not have crashed because of the reduced service level. Then somebody could claim it was not bad weather as the system did not crash, so that the measure was unnecessary. To summarize, in dealing with uncertain situations the risk manager is always wrong. 

How should a risk manager deal with this? Perhaps the most important lesson is to accept that with hindsight virtually always other options would have been better. A second lesson is to realize that risk management is a balancing act between interests. Whatever choice is made, there will be a group that will be worse of that others, and they may protest that in advance. The decision therefore is not perfect both before and after the event, or even downright wrong in the eyes of some parties. But if that is the case, there is no need to consider those opinions anymore. Time could be spent fully on potential effects and their probabilities, plus the cost of potential measures. But even then the decision on the acceptability of a risk is like taking on an 800 pound gorilla with bare hands. As risk manager you should get a risk premium.

<< back to overview