Last month I looked at Dr Ignaz Semmelweis’ struggles with “childbed fever” in nineteenth-century Vienna.

He investigated two rival theories: the psychological, according to which women died of fright when they saw a priest on his way to perform last rites; and the cadaveric matter theory, in which medical students were bringing something deadly into the maternity ward. Semmelweis tested each theory by seeing if its predictions proved accurate.

The psychological theory carried the prediction that keeping a priest off the ward would reduce the death rate. It did not. The theory had been “falsified”.

A consensus around a falsity is a very powerful barrier to truth

The cadaveric matter theory predicted that if students were made to wash their hands before going onto the ward, the death rate would fall. This proved to be the case. This theory was “confirmed” or “corroborated” or, in less cautious language, “proved”.

But watch out, there are hidden dangers here. There’s an important tacit assumption behind all this, namely that a correct theory will always produce accurate predictions, and an incorrect theory will always produce incorrect predictions. This is wrong.

Back to Vienna, it would have been perfectly reasonable for someone to put forward a theory that there was something toxic in the building materials used to build Ward 1.

What would this theory predict?

For one thing, relocating the ward to a new, non-toxic space would reduce maternal mortality.

OK, let’s say the hospital authorities agree to find a new space for the ward. If the new space happened to be on the other side of the hospital, a long way from the morgue, students would be far less likely to go straight from autopsy to maternity ward. What would happen next?

Maternal mortality rates would fall back towards normal levels… precisely what the theory predicted.

The theory would seem to have been corroborated; the science would be, as they say, settled.

No doubt vast sums would be spent pointlessly looking for the non-existent toxins in the walls of the old ward. The theory, however, would almost certainly remain in place, because having apparently solved such a major problem, it wouldn’t easily be let go.

Imagine the tortuous process required from this point to unearth the correct theory. A consensus around a falsity is a very powerful barrier to truth.

Economics is no different. If the predictions of a theory come true, we tend to accept that theory – and why shouldn’t we? What else is there to go on? Furthermore, is any policymaker or politician going to choose a theory based on, say, its internal logical consistency rather than the results it seems to produce? Of course not.

How about an example of a correct theory producing incorrect predictions – is that possible? It is.

Semmelweis’ work marked a crucial stage in the development of the germ theory of disease. Just like everybody else, he had no concept of microbes or bacteria.

With that in mind, imagine if the night before the new hand-washing regime began, a student happened to leave a contaminated item on the ward. Let’s say the item was so contaminated that it was able to act as a locus for infection. Then, despite the new cleansing routine, the students would be reinfected when they entered the ward. The maternal mortality rate wouldn’t go down after the implementation of Semmelweis’ procedures.

The cadaveric matter theory, though correct, would seem clearly to have been refuted. It predicted that if the students washed their hands before going onto the ward, the death rate would fall. The students did as they were told, but the death rate remained the same.

It’s always some extraneous circumstance that circumvents the predictions of a correct theory. And in the economy there’s a new extraneous circumstance pretty much every day. Sometimes they’re easy to identify as such, but sometimes they aren’t. And even when they’re easy, it can be politically expedient to pretend otherwise.

An analogy for the problem faced by economic policy would be this: imagine if, just after the cleansed students entered the ward, there was a structural collapse which resulted in the death of every patient. Do you think people would say: “See what all your hand-washing has achieved!”?

That would be ridiculous, wouldn’t it? Yet that’s exactly what happens all the time with economics. An extraneous event derails a policy and everyone cries from the rooftops “Look, the theory was wrong.”

In summation, a false theory can produce correct predictions, and a true theory can predict events which don’t come about. The trouble is that there’s absolutely no way to be certain. However, I’m not advocating some kind of epistemic nihilism, merely caution. Certainty can be very dangerous.

Peter Lawlor is the Principal Economic Advisor to 7Ridge Capital. He was formerly the Principal Economic Advisor to the German Stock Exchange (Deutsche Börse), and continues to act as an adviser to senior Wall St figures and political leaders. These are his own views and should not be imputed to any organisations with which he is, or has been, affiliated