Rethinking Gender Pay Gap: Selection Bias Changes the Inequality Picture

The gender pay gap is often presented as a single number, a clear-cut measure of inequality between men and women. But new research from the European Central Bank suggests the reality is far more complex. In a recent study by economists Egshiglen Batbayar, Christoph Breunig, Peter Haan, and Boryana Ilieva, supported by data linked to institutions such as the Institute for Employment Research in Germany, the focus shifts from averages to the full wage spectrum. Their work reveals that the gap changes depending on where you look in the income ladder and who is included in the data.

Why Who Works Matters

A key insight of the study is that wage comparisons are not based on a random group of people. Most analyses focus only on those in full-time jobs, but the decision to work full-time is influenced by many factors. People with higher earning potential are more likely to be in full-time employment, while others may work part-time or stay out of the workforce.

This creates what researchers call "selection bias." In simple terms, the people we observe earning wages are already a filtered group. This matters because it can make the gender pay gap look smaller or larger than it actually is, depending on which group is being studied.

A New Way to Measure Inequality

To address this issue, the researchers developed a new method to better estimate wages across the entire population, not just those working full-time. Instead of relying on traditional models, they used early-career wages as a tool to understand long-term earning potential. These early wages reflect skills and ability but are less likely to influence whether someone is currently working.

This approach allows the researchers to correct for hidden biases and get a clearer picture of how men and women truly compare. It also avoids making overly strict assumptions about why people choose to work or not, making the results more realistic.

What the Data Reveals

Using detailed German labour market data, the study finds that both men and women in full-time jobs tend to be among the more capable earners. But the pattern differs across income levels.

For women, the biggest surprise is at the lower end of the wage scale. Women in low-paid full-time jobs are often a highly selected group, meaning they may have stronger skills or motivation than other women not working full-time. When this is taken into account, their estimated wages fall, revealing that the true gap between low-income men and women is larger than previously believed.

At the higher end, the picture changes. High-earning women are more representative of the broader population, so selection plays a smaller role. For men, however, the strongest selection appears among highly educated, high-income workers. After correcting for this, their wages drop more noticeably, which slightly reduces the gender gap at the top.

A More Nuanced Reality

The overall finding is that the gender pay gap is not the same everywhere. It tends to widen at the bottom of the wage distribution and narrow slightly at the top. This helps explain why earlier studies have often reached different conclusions depending on their methods.

The research also highlights the limits of simpler approaches that look only at observable factors like education or experience. These methods miss important hidden influences such as ambition, social constraints, or personal choices, which can affect both wages and employment decisions.

What It Means for Policy

For policymakers, the message is clear. Tackling the gender pay gap is not just about equal pay for equal work. It also involves understanding who is able to participate in the labour market and under what conditions.

Encouraging more inclusive employment opportunities, especially for women at lower income levels, could be key. At the same time, addressing structural barriers that limit full-time participation may help reduce inequality more effectively.

In the end, the study reminds us that the gender pay gap is not a single story but many overlapping ones. To truly understand it, we need to look beyond averages and consider the full picture of how people work, earn, and make choices in the labour market.