The precautionary principle has a dark side
Updated on 16 September 2023
In a risk-averse society the Precautionary Principle (PP) – “better safe than sorry” – has enjoyed wide popularity. It has been enshrined in constitutions, and it has been the mainstay of a pro-active climate change policy.
Economists distinguish between “risk” and “uncertainty”.
Risks refer to stochastic events: we know they will occur, we just don’t know when – death being the most common event. Actuarial sciences began with life insurance. People realized that it was a good thing to insure themselves against the vagaries of death. Life insurance – as road accident and other insurances – are very useful tools: for the individual they replace a large payment, which has an uncertain due-date, with a sequence of affordable small payments. Everybody gains – well unless the underlying actuarial data change between the time the insurance was set up and the time it comes due. In 1945, when Social Security was set up in the USA (and then in other Western countries), life expectancy was 67 years. Retiring at age 65 meant that the state only had to pay Social Security for a couple of years on average. Pay as you go seemed reasonable. Meanwhile life expectancy has moved to over 80 years. The state has to pay for 15 or more years. The scheme is going bust.
Uncertainties are events we know nothing about we can set in numbers. They are often rare events, to rare to quantify. When a market share crash of 10-25% in one day happens once in a life time, it is difficult to make provisions for it. Even worse, the long “lead time” leads to complacency: “It won’t happen on my watch”. Politicians are so prone to think. Uncertainty peters out into hunches and prophecies, or just plain ignorance. How to deal with them? Donald Rumsfeld spoke of “unknown unknowns” – they blew up in his face, did they not?
The PP is expected to deal mainly with uncertainties. It is a rule of thumb, a heuristic, and reflects a more than human wariness of taking big risks. If you don’t know how thick the ice sheet is under your feet, you will not gladly walk over the frozen lake. Makes sense, does it not?
Yes – and no. To illustrate my point, I’ll take an example from Japan. In 1923 the Big Kantô Earthquake struck Tokyo and Yokohama. Many died in the ruins, most in the ensuing fires. 1923 was shortly after Japan had annexed Korea (in 1905). Many Korean workers had moved to Japan to work in the factories, driven by poverty. They were certainly no fans of Japan, but they went about making a life just above the poverty line. They gathered among themselves, and were mostly recognizable by the accent. As for many Japanese, they despised the Koreans – be it from xenophobia or guilt.
After the earthquake struck, the rumor started in a village that Koreans were setting fires and throwing bombs. One can trace the spread of the rumor, and it spread wide and fast. Vigilante groups emerged spontaneously in self-defense – the police being overwhelmed by the many tasks. After all was said and done, 2613 Koreans had been killed – about 2% of all dead from the quake.
Under the PP the authorities began to warn the population of possible Korean misdeeds. After all, who could be sure that some Koreans were not burning and bombing? By so doing they fanned the xenophobic fire. Messages of anecdotic (if uncertain) significance percolating out of Tokyo were taken as synecdoche – our minds tend to equate the occasional with the typical, and we readily turn single events into statistics. The press chimed in, adding “plausible color” in form of interviews of witnesses. Having copy to fill, these testimonies were invented, but they were published anyway: se non è vero è ben trovato – even though it may not be true, they ring well and true.
We have here a case of “path-dependent outcome”. Rumor reinforces rumor, and mass hysteria develops. Even dousing the mania with calls for self-control becomes counterproductive. The press was advised by the authorities that “most Koreans were law-abiding” – leaving it open that some might not. In doubt, you kill the suspicious one in front of you who is carrying a tinned can of food he saved from the ruins of his dwelling – after all, it could be a bomb. The PM chimed in with an appeal to observe the Koreans closely, strictly to report their doings, but not to act – lest the image of Japan abroad be tarnished. This is counsel of perfection: the police had other fish to fry than disperse its efforts in checking rumors. Once the rumor had started, the mania had to take its course, and very little could be done to slow it down or avoid the ensuing pogroms.
The geologic situation under Tokyo showed an example in risk taking also. In 1915 many lesser tremors occurred. Were they precursors of the “big one”? The geological profession split over the issue, and hard infighting ensued. (The same situation prevailed at l’Aquila in Italy, in 2009, when swarms of tremors preceded the main event.) In both cases the risk was denied by the authorities – lest people be frightened unnecessarily. Alas, the authorities took their own medicine, and did little or nothing. When the “big one” arrived, they were caught unprepared.
In Tokyo in particular, where lots of experience had been gathered in past major quakes, the iron rule under the Tokugawa shoguns had been quietly forgotten – no one was allowed to flee with belongings. Not only do belongings crowd the narrow streets and make movement of people difficult; embers falling on them set them on fire, and this mightily facilitated the emergence of fire storms. 38’000 people at Hifukushô burned or were trampled to death because the mass of belongings littering this huge open space burned readily and engulfed the crowd, which had felt safe there.
The line between risk and uncertainty is never drawn univocally, and a lot can be learned by studying risks closely, and applying rational arguments to it. One classic example of how to apply the PP badly is its selective use. Nuclear power plants have been shown unsafe in earthquake situations. Fine – show them to be resistant, or close them down. But so are dams. Not only can they collapse, but catastrophic landslides can push huge amounts of water over the dam. This is how over 1200 people died at Longarone in Italy in 1964. And the landslide was not even triggered by a quake: geologists had simply failed to assess the risk properly. in the 2006 earthwuake in Pakistan most people died from landslides.
PP in such a setting has its limited role for “worst case scenarios”. Too often, however, it is an excuse for laziness, muddled thinking, and facile policies. PP should be thoroughly tested, before it is used – including its possible collateral role in triggering manias.
 Akira YOSHIMOURA (2010) : Le grand tremblement de terre du Kantô. Actes Sud, Arles, 285 pp.)