The ostrich effect
The evidence is real and the argument about it is still running: how strong it is, how far it travels, or whether it repeats. Trust the direction, and hold the numbers loosely.
In 2009 Niklas Karlsson, George Loewenstein and Duane Seppi published a model of why anyone would refuse free information. Their argument was that looking does more than inform you. It amplifies the emotional impact of what you learn, and it resets the reference point you measure future news against. If the news is likely to be bad, looking makes it hurt more and hurt sooner. So not looking becomes a way to keep a loss on paper rather than in your chest.
They tested it against records from Scandinavian investment providers, including a Norwegian bank, a Swedish bank and the Swedish national pension fund. The pattern held. Investors logged in to check their portfolios more frequently when the market had risen, and less frequently when it was flat or falling. Nothing about the value of the information had changed. Only how it was likely to feel.
The scale question was settled in 2016. Sicherman, Loewenstein, Seppi and Utkus published Financial Attention in the Review of Financial Studies, built on 852 million daily observations of logins and trades from 1.1 million investors. Attention rose after positive market returns and fell after negative ones. It also fell when expected volatility was high, meaning people looked away precisely when the situation was least predictable. The behaviour turned out to be a stable individual trait rather than a mood, and it was more pronounced in men, in older investors, and, most uncomfortably, in investors with the largest balances. The people with the most at stake were the most selective about when they looked.
Then the contradiction. In 2014 Gherzi, Egan, Stewart, Haisley and Ayton ran the same question on a different population, active online investors rather than retirement savers, and found the reverse. Monitoring increased after both positive and negative daily returns. The effect persisted for logins that produced no trade and for weekend logins when markets were closed, so it was not about acting on the news. They called it the meerkat effect, hyper-vigilance rather than avoidance, and found that the personality trait of neuroticism moderated it.
So the honest statement is that the same behaviour runs in opposite directions in different groups, and nobody has cleanly established why. One plausible reading, and it is a reading rather than a finding, is that agency decides it. A pension saver who cannot meaningfully act on a bad day has nothing to gain from looking, so avoidance costs them little. An active trader who could act has a reason to look, and the news becomes information rather than only pain. If that is right, then the way to make anyone look at bad numbers is to give them something they can do about them.
The rest of the honest limits. The term itself is used inconsistently, and an earlier paper by Galai and Sade used it to mean a preference for assets whose losses are less visible, which is a related but different claim. At least one laboratory attempt to reproduce the effect in a simplified market did not find it. And the underlying mechanism, that bad news is avoided because bad news weighs more, is itself the subject of negativity bias.
The organisational version of this needs no decision. Nobody announces that the dashboard will stop being opened. The weekly review simply stops appearing in the calendar, the report that used to be circulated stops being circulated, and the number that was checked every Monday gets checked when someone remembers. Each individual instance has a reason attached, a busy week, a reorganisation, a more urgent priority, and the pattern underneath is that attention drifted away from the number at exactly the point the number turned bad.
The specific shapes are recognisable. Cohort retention that used to be reviewed monthly and now gets pulled when a board meeting demands it. The churn conversation that keeps being scheduled and moved. The client who has gone quiet and whom nobody has called, because calling would confirm it. The financial model that was accurate until the quarter it stopped being updated. In each case the information is available, free, and one click away, which is precisely the condition the research describes.
What makes it expensive is the timing. Avoidance is strongest when the news is worst, which is when the information is most valuable and when acting early matters most. A problem seen in month one is a decision. The same problem seen in month six is a crisis, and the difference between them was nobody looking.
Two practical moves follow, and the second is the one people skip. The first is to make looking automatic rather than voluntary, since a number that arrives in your inbox on a schedule cannot be un-checked, while a dashboard you have to open can quietly stop being opened. The second is to attach an action to the number. The research on active traders suggests people look at information they can do something with, so a metric nobody can influence will get avoided no matter how prominently it is displayed. If you want a team to keep watching a number through a bad quarter, the useful question is not how to make the number more visible. It is what they are allowed to do when it moves. The related failure of tracking what is easy to see rather than what matters is the McNamara fallacy.
McNamara measured everything that could be counted and lost the war, because the things that mattered could not be. The ostrich effect says people stop looking at their numbers at exactly the point the numbers turn bad. Both fail with numbers, from opposite directions: one looked too hard at the wrong things, the other stopped looking at the right ones. They meet on one point. The ostrich research on active traders found people keep watching numbers they can act on, and McNamara's dashboard was full of numbers nobody could act on. So a metric that captures what matters and that someone is allowed to do something about is the only kind that survives both. The rest either measure the wrong thing or quietly stop being opened.
Source: Karlsson, Loewenstein and Seppi, The ostrich effect: Selective attention to information, Journal of Risk and Uncertainty, volume 38, 2009, pages 95 to 115. The large-scale study: Sicherman, Loewenstein, Seppi and Utkus, Financial Attention, Review of Financial Studies, 2016, covering 852 million observations from 1.1 million investors. The contradiction: Gherzi, Egan, Stewart, Haisley and Ayton, The meerkat effect, Journal of Economic Behavior and Organization, volume 107, 2014, pages 512 to 526.
Matthew Syed, 2015
On the difference between industries that examine their failures and industries that look away from them, and what that difference costs. Aviation investigates every crash. Medicine, he argues, did not, and the gap is measured in lives.
Draw your own card. It does not take long, and it rewards taking your time.