Three tips for breaking through bias and seeing evidence more clearly

This article is brought to you thanks to the collaboration of The European Sting with the World Economic Forum. Author: Alex Edmans, Professor of Finance, London Business School

---

Business has lost the public’s trust, on both sides of the Atlantic and Channel. As a result, regulators are implementing overdue reforms to ensure that business works for all of society. These include changes to pay practices and disclosure, new corporate governance and stewardship codes, and potential restrictions on share buybacks.

Advocates for reform aim to base their proposals on evidence. In a world where fake news abounds, surely this attention to evidence should be applauded?

Not necessarily. One of the most dangerous phrases is “research shows that …”. Given the huge range of available studies, you can almost always find research to show whatever you’d like. As I explained in a recent TED talk, What to Trust in a Post-Truth World, this is a particular issue due to confirmation bias – the temptation to accept evidence that supports your opinion, regardless of its quality. Since existing views on business are strong, flimsy studies can become influential if they support these views.

Here’s a few examples. A widely quoted statistic is that 91% of US profits went to share buybacks and dividends. According to the author: “[T]hat left very little for investments in productive capabilities or higher incomes for employees.” This statistic makes no sense. Profits are after deducting wages and intangible investments such as R&D and advertising. That’s like saying “the kids can’t have had much to eat because their plates are empty” – they’ve already eaten, which is why the plates are empty. But despite this basic mistake, this statistic is commonly cited, since it confirms common views on share buybacks.

In the 2016 House of Commons inquiry into corporate governance, a witness quoted a study which “found that firm productivity is negatively correlated with pay disparity between top executive and lower level employees”, referencing a 2010 draft. The finished version had actually been published three years before the inquiry. Having gone through peer review and tightened its methodology, it found the opposite: “Firm value and operating performance both increase with relative pay.” But despite a separate submission highlighting the error, the inquiry’s final report still referred to “clear academic evidence that high wage disparities within companies harm productivity”. The House of Commons Executive Pay report, released this week, quoted another unpublished study that makes a similar claim, but omitted the evidence presented at the prior inquiry.

Last year’s Corporate Governance Code consultation referred to “clear evidence that greater female representation in the boardroom and senior management has a positive impact on performance”. The “clear evidence” cited a single non-peer-reviewed paper, with elementary omissions such as not including dividends when calculating stock returns, controlling for other factors, or checking statistical significance. This does not mean that diversity initiatives are unwarranted, as I’ll shortly stress.

So what should we do? Faced with reams of available evidence, how do we know what to trust? One tip is to draw particularly from studies published in the most stringent peer-reviewed journals, to reduce the risk that a paper is biased or flawed. As the paper on pay disparity shows, peer review isn’t simply a rubber stamp, but can overturn a paper’s conclusions. Now peer review is far from perfect, because there’s a vast range in the quality of reviewing standards. But quality can easily be checked using a list of the most stringent journals, such as the Financial Times Top 50.

Every study starts off unpublished. So the second tip is, for an unpublished paper, to scrutinize the researchers’ credentials. Again, this can be easily done, for example by looking up their institution. This isn’t to be elitist, but to recognise the importance of evidence quality – just as we’d particularly trust the medical opinion of a leading hospital. Often, we latch on to a study because we like the findings, regardless of who wrote it.

The third tip is to place more weight on balanced opinions. There are two sides to almost every issue. But authors have the incentive to present only one and claim “clear evidence”, to give a more convincing message. Such one-sided views are likely misleading, and the reader should be less convinced by an author who isn’t secure enough to acknowledge the other side. The Norwegian Sovereign Wealth Fund’s position papers on corporate governance all include the arguments against their position. This gives confidence that they reached their stance after a careful consideration of both sides.

But evidence doesn’t mean that only one view is right. Even if we all agree on the price and characteristics of different cars, different people may buy different cars given their different preferences on price, fuel efficiency and safety. Similarly, even if higher pay ratios improve firm performance, this doesn’t mean they’re desirable. Some citizens may view income equality as more important than firm performance. Evidence only aims to put the facts on the table so that we can understand the trade-offs behind decisions.

And for many reforms, evidence isn’t even necessary. Gender diversity is desirable in its own right, and firms should pursue it even without evidence that it improves performance. It would be a sad world if the only reason firms increased gender diversity was to make money – and might hinder initiatives to promote other forms of diversity where there isn’t any evidence.

So evidence may be neither conclusive nor essential. But if it’s used, it should be of the highest quality.