Correlation vs. Causation in Economics

The Trap of the Omitted Variable

One of the primary reasons correlation is mistaken for causation is the presence of an omitted variable. This occurs when a third, unobserved variable influences both of the variables being studied, creating a "spurious" or misleading correlation.

Returning to our example of ice cream and sunburns: the omitted variable is "temperature" or "sunny weather." As the temperature rises, people buy more ice cream and they spend more time in the sun, leading to more sunburns. If we only look at a graph of ice cream vs. sunburns, we might incorrectly conclude that eating ice cream causes skin damage. In economic modeling, failing to account for these "lurking" variables can lead to disastrously incorrect conclusions about how a policy might work.

Reverse Causality

Even when a causal relationship does exist, economists must determine the direction of that relationship to avoid reverse causality. This happens when the researcher assumes variable $X$ causes $Y$, but in reality, $Y$ is causing $X$.

Consider a graph showing a correlation between police presence in a city and crime rates. A naive interpretation might suggest that increasing the number of police officers causes crime to rise. However, the reality is likely the reverse: high crime rates cause city governments to hire more police officers. In this scenario, the "effect" is actually the "cause."

Establishing Truth: The Controlled Experiment

How do economists move beyond mere observation to prove causality? The gold standard is the controlled experiment. In a controlled experiment, researchers take a sample of subjects and randomly assign them to two groups:

1. The Treatment Group: Receives the intervention or change being studied.
2. The Control Group: Does not receive the intervention and serves as a baseline.

By ensuring that the only significant difference between these two groups is the intervention itself, any difference in the outcome can be attributed to the cause being studied. While controlled experiments are easy in a laboratory, they are often difficult or unethical in the real world (for example, you cannot randomly assign people to live in poverty to study its effects). Therefore, economists use advanced statistical tools to mimic these conditions and "control" for omitted variables.

Summary Table for Graphical Interpretation

When analyzing economic graphs, use the following checklist to evaluate the relationship:

By applying these rigorous standards, economists can move from simply describing what the data looks like to explaining why the economy actually works.