I stumbled upon this article on the Scientific American web site and though I’d share it since it resonated with me.
1 Like
Like the great George Soros once said, “All models are false but some are useful.”
While the models are wrong, it’s still important to understand them because these models still guide policymakers’ decisions so will help you clue you in how they will act and allow you to adjust your trading approach accordingly.
It’s kind of like how meteorologists depend on various forecasting models to predict whether it will rain. If their models are forecasting rain, then you can assume that your local news TV station will accept this information and announce rain to their audience. Then you can determine whether to bring along an umbrella (or not). 
Most of the models include assumtions based on past performance and if is something goes wrong, the model is no longer reliable. Another important point is that it is impossible to predict unexpected event that could influence substantially the general situation. For example, prices for most of the commoditiies depend also on weather conditions. If, for instance, the weather would be cold enough, countries could increase the purchases of energy sources while wheat and corn prices could increase due to the decrease in harvest.
These are main reasons why particular assumtions made upon certain model could be wrong. Each basic model itself could cease to perform well just due to the substantial changes in underlying processes. Internal relations in all spheres became more and more complicated nowadays, so that numerous factors influence the final outcome.
To my mind, it would be interesting to watch the development of models using AI-based technologies and neural networks since they offer new opportunities due to the possibility of big data processing. This potentially colud help people to find different corellations that havn`t been spotted before.
Cause-effect relationships tend to be vague in economics because there’s a million factors alongside them.