Early in October, when several women stepped forward in rapid order to accuse Republican presidential nominee Donald Trump of sexual assault, some conservatives cried foul on social media. They said the rapid string of events, so close to the presidential election, suggested a media conspiracy afoot to hit Trump at the time when it would most help his Democratic rival, Hillary Clinton.
There are lots of reasons those suspicions are unfounded. One of the biggest is economics. In a competitive news environment, reporters face overwhelming incentives not to “sit” on a potentially explosive campaign story. I said as much on Twitter on Oct. 12:
Again: The thing abt a 'conspiracy' like this is it would require near-impossible coordination among media orgs w/ big $ incentive to defect https://t.co/3H2ILVuxTJ
That tweet drew a thoughtful response from a smart economist, Josh Hendrickson, who is an assistant professor of economics at the University of Mississippi. He pointed out some economic arguments for why some reporters might be inclined to wait to publish some stories until Election Day draws closer. Then, based on our exchange, he built a theoretical model to predict when reporters might sit on a potential October Surprise — and for how long.
The model is complicated, and as you see, he and I both think it overrates how long a reporter might sit on a story. But it provides a great economist's-eye view on a subject that is particularly relevant this October. He and I chatted about his findings electronically; this is the slightly edited version of our discussion.
And spoiler: He agrees with my original Twitter point — the big stories will always come out as soon as reporters stumble onto them.
Tankersley: What do you assume about reporters, the value of scoops and the potential impact on an election — and what does your model end up telling us a reporter is likely to do if she has a scoop?
Hendrickson: So the basic idea is to think about the marginal value of the story. A story that ends up on the front page has some value. If I am a reporter and I have a story that I think is a front-page story, it necessarily knocks another story off the front page. So what I should care about is how much additional value I'm getting from this story in comparison to what would be there instead.
At the same time, this marginal value might be changing over time. Other news might make the story more or less valuable. So what reporters end up with is a trade-off. If they wait, the marginal value of the story could be higher, but the longer the reporter waits the more likely it is that they'll be preempted by another reporter. So the question is: How long should they wait?
What the reporter wants to do is optimally balance that trade-off. So the model basically implies that there is some threshold for the value of the story. If the current value is greater than that threshold, then you print it. However, if the current value is below the threshold, then the reporter should wait.
Tankersley: The idea being, a reporter might sit on a story for a while to allow its value to increase?
Hendrickson: Right. So in a campaign there is news that comes out each day. If the news cycle evolves in a way that makes your story more valuable to the public, then you'd prefer to print the story when it’s most valuable and is going to get the most attention.
Tankersley: But there's also a risk of that value evaporating, if one of your competitors beats you to a story, right?
Hendrickson: Right, so there are a couple of things that could happen. If you wait too long, then some other reporter could report the story before you do. Also, it is possible that you wait and the news cycle breaks the other direction and nobody cares about your story as much as they would have when you acquired the information.
Tankersley: Okay, so you have a range of possible future changes to the value of your story. Given those, what does your model predict a reporter in possession of a Surprise Scoop™ will do?
Hendrickson: So the reporter takes into account all of these factors when determining the threshold. If the story is really big and above the threshold, then they'll print the story right away. However, if the story is valuable, but below the threshold, then they will wait. In fact, the model suggests that the reporter is likely to wait a while before releasing the story, even if the value of the story isn't too far below the threshold.
One of the more interesting things is how things change when the threat of preemption changes. When it becomes more likely that the reporter will be “scooped' by another reporter, they lower their threshold. As a result, they are likely to publish sooner. But much of this incentive to publish sooner shows up in the immediate aftermath of acquiring the story.
Tankersley: So if the reporter sits on the story for a little while, she is more likely to sit on it for a long while?
Hendrickson: Yes, the probability that they publish initially becomes more sensitive to time and then becomes less sensitive to time. So initially the reporter is more likely to move early, but once a certain amount of time has gone by the reporter is more likely to wait.
Tankersley: You quantify this, right?
Hendrickson: Well I do some numerical examples, but it's not clear how accurate these are. Nonetheless, the simulations don't appear too sensitive to the parameters.
What I find is that if a reporter has a story on, say, July 1, the probability that they have published the story before October is 0.35 — 0.45 depending on the threat of preemption.
Tankersley: First, that's fascinating. Second, it seems really low to me. But I would suggest there's a reason for that: I think reporters are incredibly risk-averse. The consequences of being beaten on a big story — in the eyes of your bosses and your competitive peers — vastly outweigh any additional benefits to waiting to publish.
Hendrickson: I would say that my intuition is the same. The numbers seem low. I would suggest that the qualitative result is really what is important (i.e. the fact that reporters have an incentive to wait that doesn't rely on ideology or other biases or attempts to sway public opinion).
Tankersley: It also seems to me that your finding suggests anyone holding a true bombshell story (say, Trump's full tax returns) has overwhelming incentive to publish right away, right? The true October Surprises will always come out almost as soon as they are acquired?
Hendrickson: Right. A sufficiently big story will always come out right away. It's the stories that are not front-page stories but have the potential to be front-page stories that reporters will wait on.