In yesterday’s online edition of BBC News, BBC Business Editor Simon Jack explains what an economic forecast is. The article expands on the heated debate on whether Brexit will make the UK poorer, so it is extremely important that the public is correctly informed about what economic forecasts are.
Brexit is a topic that inflames me as an economist, as a prospective British citizen, as well as an Italian living in the UK. I gave my own contribution to the Brexit debate before the referendum, but it is time that I add something more. This time it will not about the negative impact of Brexit, but on the very role of Economics at depicting scenarios and informing decisions.
One of the fundamental causes of the misinformed campaigns that led us to the dreadful Brexit referendum output has been the growing separation between the academic debate and the populist push for change. Academics are to blame, as they should have engaged with the debate in more approachable ways. Information practitioners and the media are also to blame, as conducting facts-checks has progressively been downgraded from public service to an opinion.
Well, I am really not happy about the way economic forecasts are being explained to the general public. It is not good enough, it is not correct, and it demeans the economic profession.
In his piece explaining economic forecasts, Simon Jack says:
In one respect a forecast is a bit like rolling two dice. The most common number you are likely to throw is a seven – that would be the central forecast. Now, you could get two or you could get 12. That doesn’t mean that the forecast of seven was wrong, it was just that a more unlikely outcome happened. Forecasts contain what is called a “distribution” – that is an upper and lower figure to show a range of possibilities.
So, it’s random, like dice-throwing?
No, a forecast is more than a roll of the dice, which is random. It’s a calculation based on experience, facts, judgement and statistics. On trade, for example, there are plenty of statistics which show the economic effects of more open trade between countries or more barriers to trade.
A forecast is an attempt to use past events to predict the future (not always with complete precision) and is seen as useful in showing a direction of travel – better for the economy or worse – given what we know historically. (BBC News, 28-11-18, 14:30hrs).
The article has been revised a few times over the course of the day, and some substantive changes have been brought in for the better. Much of the previous criticism I had it has been already addressed. However, there are still some imprecisions, and we need to clarify these for all.
So, are economic forecasts like dice-rolling or not?
Simon Jack says yes, and no. Let’s clarify this once and for all then. Economic forecasting involves much more than probability calculus. Computing chances and odds is an important part of what an economist does, but it is not the most important. To compute a forecast an economist will use probability calculus, combined with statistics, and also with knowledge and opinion on how the economy works (something Simon Jack barely mentions when referring to ‘judgements’). Statistics allow economists to observe and classify phenomena which happened in the past, and to use these phenomena to predict the future. In the process, probability calculus guarantees that the chance to make a mistake is under control and it is always quantified.
For instance, I could claim: I predict that income will grow by 1% on average over the next year, and I am confident at 95% that this estimate is correct within a ‘confidence interval’ with upper limit of 1.3% and a lower limit of 0.7%. In other words: on average I expect we will grow by 1% and, with a very high chance, my prediction states that we will grow no less than 0.7% and no more than 1.3%. There is very little dice-throwing in all this. I have collected time-series of GDP growth in the past, and related them to the growth of neighbour countries to control for trade effects. I checked the conditions in the labour market, the evolution of monetary and fiscal policy, data on exchange rates, and many other economic indicators, to come to this conclusion. Probability calculus only ensures that the chance of making mistakes is not too high.
More importantly, I had to rely on my own knowledge of the economic system to combine economic variables in a model that helps me to predict the future. Other economists might have different opinions, or different ‘judgements’ as Simon Jack says. That is, their economic models might be different from mine as their own experience, knowledge, and expectations of how the system works is different from mine.
Can economic forecasts be wrong?
It depends on how you define the term ‘wrong’. Economists are social scientists, and Economics is a social science, not a hard science. Economists use data, statistics and probability, but their analysis is also imbued of assumptions about how the economy works. As such, their judgement drives their results, and expecting them to be ‘right’ or ‘wrong’ is just a little too much to ask of them. Obviously, with tomorrow’s wisdom, we can find out whether an economic forecast matched the data at hand. However, it is very unlikely that economic models are wrong per se.
Our beliefs affect our assumptions, our assumptions combine with data, statistics, and probability calculus to generate a forecast. Economists’ mathematical models are generally correct, and the way they apply statistics and probability calculus to them is also correct. The real problem is assumptions: about how people interact, about what is more or less valuable, about the mechanisms that link variables to each other. This explains why forecasts from different institutions are generally different, as this also reflects the richness of opinions and views in the economic debate.
It is time to educate the public in a clearer, simpler, and transparent way. It is also time that academics engage with this mission more than what they have done in the past.
Very nice blog post Fabio. I completely agree with this – people often erroneously take economists’ point forecasts as evidence of how badly they do in volatile periods! The sad fact is that, although we SHOULD always output forecast intervals, for variables like GDP growth a 95% interval is often obscenely wide like -1% to +2%. This is then criticised for being uninformative, when it simply reflects the fact that economic data are quite noisy, which is in turn because variables like GDP are measuring highly complex interactions of all agents in the economy!
To some extent the Bank of England tries to lead the way with its fan chart forecasts of GDP growth and inflation. But how many members of the public read the Bank’s Inflation Report!? I agree that it is (at least partly) our job to make this information accessible and engaging.
Jack