Over the years, CFO has reported on several studies concluding that women outperform men at key aspects of finance and business.

One suggested that the stock market reacts more favorably to acquisition announcements and secondary equity offerings by companies with female finance chiefs. Another documented that women-run hedge funds earn higher returns.

A third research effort found that companies with a higher representation of women in senior management roles outperform other companies.

Now comes robust new academic research showing that companies with women serving as chief financial officers are significantly less likely to misreport financial results. And it’s not just a correlation; the study results strongly support causation, according to its authors.

To be published in the August issue of the Academy of Management Journal, the paper builds upon prior studies by other researchers indicating that:

  • An estimated 15% of public companies engage in accounting fraud annually, but fewer than 1% are caught doing so.
  • Women are, on average, more risk-averse than men, among other social-psychological attributes that differ by gender.
  • A 138-year-old mathematical principle called Benford’s Law reliably flags the likelihood of financial misreporting (which has been known since the mid-1990s) and correlates well with several accounting measures traditionally used to identify misstatements.

Under Benford’s Law, in naturally occurring collections of numbers, the leading digit (the first digit in any number) is “1” more often — approximately 30.1% of the time — than are the other digits. Each successive number has a lower probability of being a leading digit (17.6% for 2, 12.5% for 3, and so on, down to 4.6% for 9).

The principle applies to groups of numbers as diverse as those found in a newspaper (excepting page numbers), population data, and even the lengths of rivers (on any measurement scale) and street addresses. It does not apply when the set of numbers is artificially constrained, such as by setting a ceiling or defining the set as being all numbers between, say, 300 and 700.

When applied to line items in financial statements, Benford’s Law yields a “financial statement deviation” (FSD) score that indicates the degree to which the distribution of leading digits diverges from the expected distribution. The lower the FSD score, the lower the likelihood of financial misreporting.

The study authors used FSD scores so that they could study the entire population of public companies, rather than just the few that were caught fraudulently reporting their financials, says University of Alabama associate professor Sandra Mortal, a co-author of the study.

According to the research, companies with female CFOs had, on average, a 2.6% lower FSD score than those where men helm finance. Regression analyses tested the result against myriad variables to ensure a finding of causation.

“Given the increasing pressure on firms to have more female representation in top management positions, understanding how executive gender impacts corporate financial reporting can inform researchers, policy-makers, regulators, and the general public,” the study says.

The research covered 18,659 firm-year observations from 2006 to 2016 for 2,186 unique U.S.-based companies. The data set included 1,550 firm-year female CFO observations at 339 companies.

Is 2.6% a big difference?

“We did a back-of-the-envelope calculation of the overall cost of ‘cooking the books’ based on other research,” says University of Alabama associate professor Vishal Gupta, Mortal’s spouse and another co-author of the study.

“It didn’t make it into the final version of the paper,” he continues, “because there were a number of approximations in the calculation and, as you can imagine, peer reviewers and journal editors today are very mindful of what’s in [gender-focused] studies like this. But reducing fraud by even half a percent would provide a significant net benefit to corporate America.”

Of course, not all financial misreporting is the result of fraud; there are legitimate errors as well.

However, in cases where the SEC catches companies reporting fraudulent financials, their subsequent FSD scores plummet by an average of 6%.

Calculating FSD scores theoretically could provide regulators with an enforcement tool by identifying companies that are more likely to be engaging in fraud. “Our understanding is that the SEC is aware of it and has looked into it but has not gone public with whether they’re using it,” says Gupta.

The study also sought to measure whether high levels of external monitoring of companies, as represented by the levels of institutional ownership and analyst coverage, influence the relation between CFO gender and the likelihood of financial misreporting.

And that indeed was the case. Where institutional ownership was low (that is, companies in the lowest quintile of such ownership), companies with female CFOs had 6% lower FSD scores, on average. Where institutional ownership was high (the highest quintile), the difference shrunk to 1.6%.

The same trend was seen for analyst coverage — a 5.5% gap between men and women with low coverage narrowed to 1.6% with high coverage.

In addition to Gupta and Mortal, the study’s authors are Xiaohu Guo, a Ph.D. candidate at the school; Bidisha Chakrabarty of St. Louis University; and Daniel Turban of the University of Missouri.

, , , , , ,

3 responses to “Are Women CFOs Less Prone to Financial Misreporting?”

  1. We’re all for more women in senior roles and the raw statistics clearly show they are underrepresented. At the same time, citing statistical evidence of causation between financial misreporting and the sex of the CFO is a bit of a stretch where responsible journalism should require much deeper evaluation – of sample sizes, confounding influences and more that are excluded in this article. About 8% of over 18,000 observations were women CFO’s, so the differences in sample sizes alone between male and female were enormous. Might we need to wait until we have solved the underlying disparity, where we have a better representation of women, before we can conclude causation – that women CFO’s are less prone to misreporting? Premature conclusions are not helpful.

  2. Thank you for your comment, Jim. Here are responses to some of your points.

    First, I can hardly imagine a deeper evaluation of this question than the one performed by the professors, which was peer-reviewed and relied on dozens of regression analyses and numerous other tests to support the work’s integrity.

    Your quarrel with sample size puzzles me. As the article notes, there were 1,550 firm-year female CFO observations. That is way, way more than needed to support a finding of causation, as any research professional will tell you. The specific regressions and other tests were not reported in the article, but such rigor is a boilerplate aspect of any peer-reviewed research that supports a finding of causation.

    Second, all research, academic or otherwise, is subject to potential error. In this case, results are statistically significant at the 95% level.

    All peer-reviewed academic research will point out potential causes of error. Those pertaining to this research were unremarkable, frankly, and as we did not undertake to republish the entire research report on CFO.com, we omitted them.

    Of note, the “Limitations” section of the report acknowledges that FSD scores, the measure of financial statement misreporting used by the professors, do not distinguish between fraudulent and simply erroneous misreporting. The article points out that while not all financial misreporting is fraudulent, other studies have shown that the incidence of fraudulent reporting is extremely high.

    To your point, the article may have benefited from pointing out the research report’s statement that FSD scores are correlated with other measures of actual misreporting, such as earnings restatements and SEC Accounting and Enforcement Releases.

  3. Jim, causation has become a very important issue in organizational research (though, I am not saying it is addressed in every paper), and I am very glad you brought it up here. In the paper, we do several things to address causation, including instrumental variable analysis and exogenous shock. I request you to read the full paper (available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3387813), and if you still think we can do more about causation, we are very open to dig further. Regarding unequal sample size, as you note, there is high inequality between firms with women in senior roles (e.g., CFO) and firms without, and to address this problem, we followed best practices in the literature using Propensity Score Matched (PSM) analysis. As you may know, the purpose of PSM is to draw equivalent samples from two groups (in our case, firms with male and female CFOs). Of course, we are open to new statistical techniques that can help us address these issues in a more rigorous fashion, so suggestions are most welcome.

Leave a Reply

Your email address will not be published. Required fields are marked *