We're documenting things with tiny, tiny, tiny Ms
Over three decades, accounting research has quietly traded meaning for volume — a field once anchored in the practical concerns of auditors and executives now finds itself adrift in oceans of data, surfacing findings so small they barely disturb the surface of real-world understanding. John Hand, a professor at the University of Chicago Booth School of Business, has traced this drift through more than 200 studies, documenting a fifty-fold collapse in explanatory power even as datasets have grown sixty-fold. The paradox is a familiar one in the age of abundance: more information has produced less insight, and statistical significance has become a substitute for meaning rather than a measure of it. The question Hand leaves open is whether a discipline can recover its purpose once it has learned to mistake precision for truth.
- Accounting research, meant to serve the real decisions of CFOs and auditors, now routinely produces findings so statistically microscopic they explain almost nothing about how businesses actually behave.
- The sixty-fold expansion of available datasets since 1995 has made it dangerously easy to achieve statistical significance — a threshold journals reward — without achieving anything economically meaningful.
- Hand calls this 'over-krilling': researchers trawling vast data oceans to surface effects so tiny they carry no practical weight, while the field's most prestigious journals continue to publish them.
- A telling symptom has emerged — papers with the weakest findings tend to be the longest, their footnotes and appendices expanding in inverse proportion to the importance of what they actually found.
- Proposed reforms aim to reanchor the field: mandatory disclosure of explanatory power alongside statistical significance, and a deliberate return to manual data collection to force researchers back toward questions worth asking.
John Hand arrived at the University of Chicago Booth School of Business in 2017 carrying decades of experience in accounting research — and a growing unease. After reading more than 200 articles published across the field's four most prestigious journals between 1995 and 2024, he reached a troubling conclusion: accounting research has lost its grip on the real world.
The evidence is numerical and stark. Variables that once explained a meaningful share of real-world outcomes now explain nearly fifty times less. This is not a story of declining rigor — it is a story of abundance gone wrong. Since 1995, the datasets available to researchers have grown sixty-fold, making statistical significance easy to achieve and easy to mistake for importance. Hand calls the result 'over-krilling': sifting through oceans of data to surface effects so small they are, in any practical sense, invisible.
The irony cuts deep. Accounting research exists to serve practitioners — the auditors, executives, and financial officers who need to understand what actually drives outcomes in business. Instead, Hand observes, the field has become absorbed in documenting effects so minor they would never register in a boardroom or an audit. He also noticed that the weakest papers tended to be the longest, their elaborate scaffolding of footnotes and appendices perhaps compensating for the thinness of what lay at the center.
Hand's proposed remedies are both structural and cultural. Journals should require researchers to report explanatory power — not just whether a variable clears the significance threshold, but how much of the real world it actually accounts for. And researchers should be nudged back toward manual data collection, whose very slowness demands deliberation and forces the question: does this variable matter to anyone making a real decision?
Without such corrections, Hand warns, the field's findings could become so small as to be functionally irrelevant within a generation. The discipline now faces a choice between the comfort of statistical significance and the harder work of genuine meaning.
John Hand has spent the last few years reading through accounting research papers the way a doctor reads test results—looking for signs of health or disease. What he found troubled him. The professor, who moved to the University of Chicago Booth School of Business in 2017 after decades at the University of North Carolina, examined more than 200 articles published in the field's four most prestigious journals between 1995 and 2024. His conclusion was stark: the explanatory power of accounting research has collapsed.
The numbers tell the story. In the mid-1990s, when a researcher identified a key variable and tested it, that variable explained a meaningful slice of what was actually happening in the real world. Today, the explanatory power of those same kinds of variables is nearly fifty times smaller. This is not because researchers are asking worse questions or conducting sloppier work. It is because they are drowning in data.
The datasets available to accounting researchers have grown sixty-fold since 1995. Digital collection has made information cheap and abundant. This abundance, Hand argues, has become a trap. When you have access to millions of observations, you can find statistical significance almost anywhere if you look hard enough. A variable might pass the threshold of statistical significance—the t-statistic that academic journals use to decide what gets published—while explaining almost nothing about the real world. Hand calls this phenomenon "over-krilling," borrowing from the image of whales sifting through vast oceans to find tiny organisms. Researchers are sifting through oceans of data to find microscopic effects.
The irony is sharp. Accounting research is supposed to serve accountants, auditors, chief financial officers, and other business practitioners. These are people who care about factors that matter—what Hand calls things with "a capital M, or at least a medium-sized M." Instead, he observes, the field has become obsessed with documenting "tiny, tiny, tiny Ms." A finding that explains one-hundredth of one percent of variation might be statistically significant. It is not useful.
Hand noticed something else while reading these papers: the ones with the weakest findings tended to be the longest. They contained more footnotes, longer appendixes, more careful annotation. He suspects this is not coincidental. When your actual result is small, you spend more time building the scaffolding around it, perhaps hoping to convince readers that the finding matters more than it does.
The solution, Hand believes, requires changing how research gets done. Journals should require that papers report the explanatory power of their key variables, not just whether those variables are statistically significant. More radically, researchers should be encouraged to collect data manually rather than simply downloading massive datasets. The slowness of manual collection forces deliberation. It makes you think about what you are actually measuring and why it might matter to someone running a business.
Without intervention, Hand warns, accounting research risks becoming irrelevant. If the field continues on its current trajectory, in thirty years its findings might explain one-thousandth of one percent of the outcomes that practitioners actually care about. "This isn't a smart direction for us to be heading," he says. The question now is whether the field will listen.
Notable Quotes
The real world cares about stuff that matters with a capital M, or at least a medium-sized M, yet we in accounting seem to be documenting things with tiny, tiny, tiny Ms.— John Hand, University of Chicago Booth School of Business
If the work we do continues to decrease in explanatory power such that our scholarly findings in 30 years' time explain just one thousandth of 1 percent of what's happening to the outcomes that people care about, we risk becoming wholly irrelevant to business.— John Hand
The Hearth Conversation Another angle on the story
Why does it matter if a finding is statistically significant but explains almost nothing?
Because statistical significance just means you probably didn't find it by accident. It doesn't mean it matters. A variable might be real but so small that a business person would never care about it or act on it.
So the problem is the data itself?
Not exactly. The problem is what abundance does to human judgment. When data was scarce, researchers had to be thoughtful about what questions to ask. Now they can ask a thousand questions and find answers to all of them, even the ones that don't matter.
And the longer papers with more footnotes—you're saying researchers are hiding weak results?
Not hiding, exactly. But when your finding is small, you need more words to make it seem important. It's a natural human response to abundance.
What would manual data collection actually change?
It would slow everything down. You'd collect fifty observations instead of five million. That forces you to choose your observations carefully, to think about what you're measuring. You can't just fish through an ocean.
Do you think accounting professors will actually do this?
That's the real question. It means publishing fewer papers, taking longer on each one, accepting that some questions can't be answered with the data you can reasonably collect. That's hard to ask of people whose careers depend on publication counts.