Economists often appear in public debates, disagreeing on the same topic as if they are speaking two completely different languages. One study claims inflation is driven by excess demand, another points to supply shocks. Some economists argue that higher interest rates slow the economy, while others suggest they merely follow economic downturns rather than cause them. Perhaps the most puzzling is that these disagreements continue when economists analyse the same data. The numbers are identical, the time periods overlap, yet the conclusions still differ. How can experts trained in the same field reach such different answers from the same evidence?
One reason lies in the misconception that data speaks for itself. Economic data is often treated as something objective and untouched, as if economists simply observe reality and record it. In practice, there is no such thing as raw data. Every dataset reflects a series of choices: what to measure, how to measure it, which observations to include, and which to exclude. Inflation is measured by tracking price changes of a fixed “basket” of common goods, unemployment is defined by whether people are working or actively searching for a job, productivity is measured as the ratio of output to inputs. Before any model is estimated, the data has already been constructed, and different choices in data can lead economists down very different paths.
Even when economists agree on the composition of data, they may still be trying to answer different questions. One researcher could be interested in short-term effects, while another focuses on long-run trends. Some look at average outcomes, while others look at effects across different groups. For example, if a firm decides to reduce the working week from 40 hours to 32 hours. One economist might look at the headcount of employees and conclude that employment is unchanged, while another focuses on the total hours worked and argues that the overall labour input has dropped.
Economists rely on models to make their conclusions. These models are not mirrors of the economy but lenses through which it is viewed. By design, they simplify reality, highlighting some relationships while ignoring others. A model that assumes linear effects will tell a different story from one that allows for asymmetries. When economists choose such different lenses, it is hardly surprising that the picture they see does not always look the same.
Another source of disagreement comes from the question of causality. The economy is not a laboratory where variables can be isolated and controlled at will. When two things move together, it is often not clear which causes the other, or whether they are both driven by something else. For example, interest rates tend to rise before recessions, but central banks often raise rates precisely because they expect the economy to slow down. Economists develop strategies to clarify these relationships, and the assumptions behind those strategies often determine the conclusions they make.
Economies are shaped by institutions, expectations and feedback loops that change over time. Unlike controlled experiments, economic relationships rarely remain stable for long. When new data arrives or circumstances change, previous patterns can weaken or disappear entirely. Different conclusions are not evidence of failure, but of a system that resists simple answers. Economic claims should be read with this in mind. The data may be the same, but the lenses through which it is viewed rarely are.