Continued from Part 1
“No economist has yet claimed that economics can explain the Universe. The Universe remains, for now, the turf of physicists, whom most economists have for centuries been looking up to as their role models, in their desire to make their subject a true science.* But some economists have come close – they have claimed that economics is about ‘the world’.” – Economics The User’s Guide.
Economists often fall into the trap of believing the economy can be understood as universal laws of physics. The former likes to change, or, at the very least, be deceptive in the way it shows itself to economists. The latter can also be deceptive, but, once physicists make do with the deception, their job is complete.
With the exception of Quantum Mechanics, which likes to deal with the world of probabilities, classical physics is very deterministic. Scientific experiments within physics have set outcomes that can be recreated over and over again. This makes theory crafting incredibly easy as an idea will either be wrong due to fundamental realities or what I like to call deception. Deception is when the natural world is stubborn in revealing its true nature. In physics, this occurs when you lack the technology to make tools that will give you accurate results. Deception and fundamental reasoning go hand in hand. A theory in physics may contain the right fundamentals which are then falsely disproven by poor experimentation, or, alternatively, wrong fundamentals being falsely supported by poor experimentation.
But in economics, the natural world is much more deceptive. This is because the economy is a complex and dynamic system that is always evolving, and here, even the natural world is subject to change. Strangely, Economists do understand this principle, yet, they still try to emulate physics – at least partially. Where models in physics predict a set outcome, economic models reflect what is currently happening while also accounting for the probable possibilities of what could happen. This means that random events can completely derail our expectations and this is where extreme events such as COVID-19 or the financial crisis can alter the fundamentals supporting a model. When this occurs, the model falls apart and produces an error – the difference between the model output and the real world.
Here, the error in the model is ‘music’ which is where the error is its own dynamic and complex system that hasn’t been accounted for (because the system itself didn’t exist before the event triggering it). Economists then try to solve these errors by going back to the drawing board, pencilling in new functions and theories to account for the change in fundamentals and then, by implication, adapting their future range of possible outcomes.
While this does sound like quite a tight approach to modelling, there are two main problems with it which are interlinked. The first is that, because of the iterative process, there is a sunk cost associated with the assumptions and ideas supporting the models, which increases exponentially as the models get more and more complex. Here, economics shoots itself in the foot as our most accepted models are not built for engaging with fringe schools of thought which may provide insightful analysis at the expense of contradicting basic assumptions.
Second, I believe that there is a trade-off between predictability and the explanatory value of a model. When you make a model better suited to past and future data, you are incorporating errors – which are their own dynamic and complex systems – and trying to make them fit with your current model. Here, by assimilating additional errors to an existing model, you lose a lot of the value and lessons governing said error as you force it to fit within a range of assumptions – even when the logic explaining the error is mutually exclusive with the assumptions of your model. This merely solidifies the errors within the model, rather than scrutinising its faulty underlying assumptions. As you go deeper into these iterations, your model just becomes a really big probability exercise that has forgotten basic truths. In doing so, you are losing the explainability of your theories with the trade-off being predictability. And, when you lose explainability, your predictability gets even worse in the long run as you forget basic facts about the economy – a reality the Bank of England has fallen foul of in recent years.
Thus, it is clear that Economics needs a different approach to modelling, one akin to engineering. In engineering, problems change, and, as such, solutions must be adapted. Yet, that does not mean the solutions are cast away. Economists must learn from this principle as right now, models that do not hold in all cases are either tampered with or thrown away completely. A case example is Milton Friedman who had his theory of money growth ‘disproved’ and tampered with so it could be incorporated into a ‘popular consensus.’ If one makes an observation of the world around them and then works to create a model that fits that observation, why do we cast the model away when a new problem arises? Should an engineer approach a problem with a similar approach, he will find himself casting away his whole toolbox. Such an approach is hardly sustainable.
Yet, the issue is much worse than this as economists do not even have a toolbox, instead, they wish for a Swiss Army knife-style multi-tool. However, an engineer is bound to prefer a toolbox over a grand multi-tool. Should the multi-tool not have the solution to the engineer’s problem he must replace, rewire and reconstruct the entire mechanism as opposed to simply adding a new tool. By trying to oversimplify our economies into rigorous models we have instead made our methodology overcomplicated in the name of what? Empiricism? Forecasting? What use is forecasting if we do not have the explanatory value needed to solve our problems?
There is brilliance in the fact that the world is a place where the truth changes as humanity and society itself develop. What is less brilliant is that economics has so far failed to accommodate this principle. The academic conundrum of trying to imitate other sciences such as physics to become more respected has only made us more foolish. It is time for economics to go on its own accord so that it may more accurately represent the problems that it intends to solve. So, the economist reading this, I implore you to think wisely about the use of a model next time you encounter one, especially if it has been ‘disproven.’ Tools can wear, break down and become inefficient should the problem change, yet similarly, they can be repaired, restored and, more importantly, inspire and provide a framework for new solutions.