Brexit and the Art of Economic Forecasting

One alarming  consequence of the British electorate's vote to leave the EU has been the sheer uncertainty regarding the process of leaving and the economic impact. Questions that are still unsatisfactorily answered include the precise pathway to negotiating an exit, the various "partnerships" with the EU that could be constructed and the economic impact of these different possibilities.

The state of play in the days after the referendum could be summed up by a senior Tory MP and Brexit supporter who admitted "there was no plan", even amongst those actively campaigning for Leave.

The question of economic forecasting is particularly interesting, particularly in the light of the confident assertion made by Michael Gove that the British people "have had enough of experts" during the referendum campaign. Two questions are raised by this, do the "experts" share a solid forecast for the short and medium-term and, if there isn't a shared forecast, why not?

Image Credit

What do experts agree will happen to the economy?

The short answer to the first question is that there is a broad consensus but little agreement on details, scope or methods. There are some areas of agreement amongst economists, the largest and more important being that there will be strong negative consequences for both British and EU economies as a result of the British exit. This is broadly shared amongst the IMF, large commercial parties such as HSBC, the UK Treasury and many independent economists, including the majority of UK academic economists.

However, beyond this broad prediction, there appears to be little agreement. The precise figures are fiercely contested, as expected, but the very logic behind the forecasts are also at odds with each other. The IFS (Institute for Fiscal Studies) modelled the effect on domestic austerity policy, the Treasury's figures are based on large assumptions and little to no State intervention and one economics Professor foresaw unpredictable chaos, terming it a "clusterfuck".

To try and get a better understanding of why there is so much vagueness around Brexit economic forecasts we can turn to someone like Nate Silver, a statistician who specialises in both sports forecasting and political election forecasting. His claim to fame, correctly predicting 49 of 50 states in the 2008 US Presidential Election, was based on a deep understanding not only of statistical calculation and interpretation but also of human variables and qualitative factors.

Sorting the Signal from the Noise

Silver notes that economic forecasts are frequently, and sometimes immensely, off target, ranging from the failure to consistently forecast relatively minor indicators, such as house prices, to missing warning signs of large crises such as the 2008 financial crash. This track record is starkly at odds with the vast amount of raw data and economic indicators that are regularly processed and analysed by various sophisticated economic models.

Several reasons why economic forecasts are notoriously unreliable are put forward in Silver's 2013 book, The Signal and the Noise. We can summarise those most relevant to Brexit forecasting below. 

  1. Modern complex systems
  2. Feedback loops
  3. Causation and correlation
Economies today are immensely complex systems, involving millions of individual decision-makers and vast sums of capital deeply interconnected with local, regional and global financial and economic systems. It is worth stressing that the disconnect between a straightforward economic model (for example a simple supply/demand pricing model) and a real-world economy is not just an increasing amount of economic actors or systems, but rather often an entirely different approach is required to make sense of the real-world economy.

Complexity theory, sometimes also called Complex Systems Theory, arose because of the recognition that traditional modelling and simulation simply didn't scale up well to modern economies, non-equilibrium physics and other complex systems. 
In the context of Brexit, it is clear that three interlinking systems are the subject of forecasts, the national economy of Britain, the connected but individual economies of the EU nations and global financial and economic networks (sometimes referred to as the "world economy"). To speak of the "EU" economy is misleadingly simple, and straightforward traditional modelling (for example simply trying to factor in rebates and grants) will fail to provide an accurate indication of the future.

Adding to this complexity, feedback loops are an overlooked part of economic forecasting in the public discourse. Feedback loops (as shown below) are a recognition of the interconnected nature of complex systems. A feedback loop is created whenever the output of one activity adds to the input of that same activity when it is repeated, sometimes intensifying and reinforcing the cycle and at other times causing a result different to the original intent driving the original input.

Feedback Loop: image source

Feedback loops can distort economic indicators and lead to a misreading of economic activity. Silver gives the example of rising house prices, usually an indicator of positive economic health, being distorted by government policies aimed at maintaining real estate growth. Since property sales are a recurrent cyclical activity, feedback loops can magnify market interference, result in artificial inflation and cause non-complex economic modelling to give false readings of the health and direction of the economy.

The difficulty of discerning between causation and correlation further muddies the waters. With a mass of raw data and economic indicators, human intervention and design is still required to interpret and establish probable causality.

The example given by Silver is that of the Super Bowl. Between 1967 to 1997, an NFL team winning the Super Bowl resulted in stock market gains while a AFL team winning lead to losses, holding true for 28 years out of the 30 sampled. 

The statistical likelihood of this relationship is 4,700,000 to 1, yet very few people would argue that this is evidence of a causal link between the Super Bowl and the stock market, nor would many people base economic policy upon such a link. The existence of correlated events, even if extremely unlikely, is simply not a solid foundation for economic forecasts.

In an immensely complex system such as a modern national economy, identifying between causation and correlation is a difficult task. Modelling economies can involve tens of thousands of variables and a huge range of economic indicators, this alone provides a considerable challenge. When unexpected feedback loops and human behaviour (i.e. non-rational economic behaviour) is taken into account, this becomes even more daunting.

Forecasting Brexit

It is clear that pretending to be able to give detailed, accurate forecasts of Britian's exit from the EU is extremely unlikely. The lack of clear forward planning by the Leave campaign, the range of exit options available and the sheer complexity of the economies in question render economic forecasting more of an art than a precise science.

However, the variables involved in forecasting economic performance should be divorced from economic dangers, which are clear, present and endorsed by the majority of experts in the field. Reading post-Brexit forecasts as potential situations may have more value than expecting detailed accurate modelling for an inherently uncertain future.


1 comment:

  1. Your style is very unique in comparison to other folks I've read stuff from. Thanks for posting when you have the opportunity, Guess I will just book mark this page. outlook 365 login