This Is Not Science — It Is Witchcraft! Seeing How Bad Econometric Forecasting Is

As for forecasting, a look at the literature on exchange rate economics gives us an idea of how bad econometric forecasting is. We can predict precisely when a falling object will hit the ground and where a projectile will land, but we cannot predict with a reasonable level of confidence whether a currency will appreciate or depreciate on the announcement of unemployment data — that is, we cannot even predict the direction of change, let alone the magnitude of change. I must admit, however, that economists (or applied econometricians) are good at finding explanations for currency appreciation or depreciation in response to the announcement of unemployment data — only after the fact, of course. It all depends on what the researcher wants to prove and if that turns out not to be the case, an explanation is found for why it did not turn out to be the case, sometimes blaming the unwanted results on econometrics itself (such as the low power of the test). This is not science — it is witchcraft.

—Imad A. Moosa, Econometrics as a Con Art: Exposing the Limitations and Abuses of Econometrics (Cheltenham, UK: Edward Elgar Publishing, 2017), 52.

Econometrics Has Been a Success Only in the Limited Sense That It Can Be Used to Prove Almost Anything

Econometrics has been a success only in the limited sense that it can be used to prove almost anything. I have always challenged seminar presenters, saying that if they let me have their data I could turn their results upside down and come up with a different conclusion. Econometrics is very useful for those wanting to prove a pre-conceived belief or find results that support an ideologically driven hypothesis. Take, for example, Brexit, which had proponents and opponents. The empirical results produced by the opponents on the effect of Brexit on the British economy of leaving the EU are all over the place but ideological bias is conspicuous. For example, the Confederation of British Industry (2013), which is against Brexit, estimated the net benefit to Britain of EU membership to be in the region of 4 per cent to 5 per cent of GDP — that is, between £62 billion and £78 billion per year. Conversely, Congdon (2014) puts the cost of Britain’s membership of the EU at 10 per cent, attributing this cost to regulation and resource misallocation. Congdon’s estimates were prepared for the United Kingdom Independence Party (UKIP), which has a strong anti-Europe stance.

—Imad A. Moosa, Econometrics as a Con Art: Exposing the Limitations and Abuses of Econometrics (Cheltenham, UK: Edward Elgar Publishing, 2017), 18.