“All models are wrong, but some are useful”

How to use a wrong model correctly

“All models are wrong, but some are useful” – coined by the statistician, George Box, this aphorism is just as true in economics as it is in statistics. The economic crisis of 2007 left many asking the same question: what went wrong? In retrospect, multiple factors were at play but one key cause was the fallibility of trusted economic models. Institutions had developed extremely complicated investment vehicles and relied on arcane algorithms to calculate their investment risks. Uncertainties in the underlying assumptions within these models were overlooked and not fully understood. While these models were trying to simulate complex market forces, factoring-in fluctuations in asset prices and interest rates, they relied on assumptions about human and market behaviour that are almost impossible to accurately predict – people are not always rational and market competition is not always perfect.  

However, the purpose of an economic model is to reduce the complexity of a situation and enable us to derive key economic principles. For example Ricardo’s theory of comparative advantage describes two countries producing two separate goods and from this overly simplified scenario, he demonstrates how each country can increase its overall consumption if they engage in free trade by exporting the goods they have the comparative advantage in and importing the other.  

Problems arise when such principles are misconstrued and extrapolated, neglecting the assumptions they are based on. For example, economic theory suggests deregulation and privatisation produce strength in a mature economy, but introducing such policies in a country lacking a strong legal foundation can result in a kleptocracy run by a select few. In essence, economic models provide a powerful basis for decision-making but we must be aware of the assumptions, model mispecifications and the dynamic socio-economic and political environment of the real world. 

An interview with Professor Paolo Zaffaroni, Fellow Imperial Business Analytics with KPMG. Written by Ikram-Ul Haq (BSc Intercalated Management).