When it Comes to Finance, ‘Normal’ Data is Surprisingly Unusual

Business researchers often depend on assumptions to interpret their findings. However, when these assumptions prove to be inaccurate, it can lead to significant complications—something that can occur more frequently than they might expect. This was evident in a recent study that examined financial data from approximately a thousand major US companies.

A prevalent assumption in data analysis is that the data will conform to a normal distribution, commonly referred to as the bell curve in statistics. If you’ve ever viewed a chart displaying people’s heights, you’ve encountered this curve: most individuals fall near the average, with fewer at the extremes. Its symmetrical and predictable nature is often taken for granted in research.

But what occurs when real-world data deviates from this orderly curve?

As professors specializing in business, we investigated financial data from public US companies—metrics such as firm market value, market share, total assets, and similar financial measures and ratios. Researchers frequently analyze this data to gain insights into company operations and decision-making processes.

Our findings indicate that the data rarely adheres to the bell curve. In certain instances, we identified extreme outliers, like a few large firms being thousands of times larger than their smaller counterparts. Moreover, we observed distributions that are “right-skewed,” meaning that the data points are concentrated on the left side of the chart. Essentially, while most values are on the lower end, a handful of exceptionally high numbers inflate the average. This observation aligns with the fact that financial metrics tend to be non-negative; for instance, you won’t encounter a company reporting a negative number of employees.

Why it matters

If business researchers base their conclusions on flawed assumptions, their insights—such as those regarding the factors driving company value—may be misleading. These inaccuracies can have broad repercussions, affecting business decisions, investor approaches, or even public policy.

For instance, consider stock returns. If research assumes that those returns follow a normal distribution, yet they are actually skewed or filled with outliers, the conclusions drawn may be erroneous. Investors relying on that research could be misled.

Researchers are aware that their work has real-world implications, which is why they often dedicate years to refining a study, gathering feedback, and revising the article before it undergoes peer review and publication. However, if they neglect to verify whether their data follows a normal distribution, they risk overlooking a significant flaw. This could undermine even well-structured studies.

Given this, we encourage researchers to ponder: Do I truly grasp the statistical methods I’m employing? Am I assessing my assumptions—or merely taking them for granted?

What still isn’t known

Despite the critical role of data assumptions, many studies neglect to report tests for normality. Consequently, it remains uncertain how many findings in finance and accounting research are built on unstable statistical foundations. Further efforts are needed to grasp how prevalent these issues are and to promote best practices in testing and addressing them.

While not every researcher needs to be a statistician, anyone working with data would be prudent to ask: How normal is it, really?

D. Brian Blank, Associate Professor of Finance, Mississippi State University and Gary F. Templeton, Professor of Management Information Systems, West Virginia University.

This article is republished from The Conversation under a Creative Commons license. Read the original article.