The way that markets are reported affects the way they are perceived, and the way that investors behave. So can the way journalists report the markets be measured? Are we doing a good job?
Historically, it appears that we are not. This week, Diego Garcia of the University of Colorado at Boulder presented a paper to a conference on journalism and markets at Columbia Business School. Entitled The Kinks of Financial Journalism, it shows that journalists are biased — collectively we are more negative about market falls than positive about market rises.
The methodology involved building a database of market reports published by the New York Times and the Wall Street Journal between 1905 and 2005, assigning a score to the choice of words on a scale from positive to negative, and then seeing how positivity and negativity of coverage correlated with moves in the Dow Jones Industrial Average. (Few other benchmarks were available a century ago).
In theory, the relationship should be almost linear. As the market performance on any day improved, so the language in the next day’s paper would grow more positive. And equally, as the market fell more, so the coverage would grow more negative.
But the relationship turned out to have a kink. After a while, extra increases for the market made little or no difference to the positivity of journalists’ prose; but as falls grew worse, the language grew ever more negative.
This implicit bias towards negativity was common to both papers and consistent over time. In recent decades, market columns have carried the byline of a particular writer, and all writers show the same negative bias.
So, for more than a century, financial journalists have had a measurable bias towards negativity. (I am well aware from my own inbox that many level this charge at me, but it looks as though I am not alone).
Further, Mr Garcia’s research also showed that bias became more pronounced with the passage of time. When comparing language with market performance two, three and four days before, all discernible effects from an “up” day vanished on day two. But even after four days had passed, the effect of a bad day was still evident in more negative language.
This contradicts earlier theories. Mr Garcia launched his research after reading Robert Shiller’s Irrational Exuberance, which brilliantly diagnosed the 1990s dotcom bubble. He had expected to see the media exaggerating average days, and hyping stocks on very strong days, as much as they overdid gloom in bad times.
Over history, this does not happen. Personally, without having done detailed research, I suspect that the dotcom bubble of the late 1990s, when Mr Shiller was writing, may have been an exception. Much US media coverage of markets, and pop culture references, became wildly bullish and uncritical. Personal finance magazines flourished, and always trumpeted bullish calls on their covers. Maybe this was one of many ways in which the dotcom bubble was an exception to historical norms.
How should we explain the negative bias? Mr Garcia offered two economic possibilities. Either journalists are demand-led, telling investors what they want to read, or supply-led and trying to “hype” to sell newspapers. I do not think either explains the “kink”.
Personally (and other FT journalists might well disagree), I would offer two different explanations. I was not surprised by the finding, and I expect that an examination of the FT over the years would have found the same thing.
First, we have asymmetrical incentives, a variation on what is known to behavioural psychologists as “loss aversion”. The great Israeli psychologists Daniel Kahneman and Amos Tversky found that investors will be much more upset by a loss than they will by missing out on a gain of exactly the same magnitude. Our horror of a loss is a predictable human irrationality.
Applied to journalists, we are far more scared of encouraging readers to buy and ushering them into a loss, than we are of urging them to be cautious, and leading them to miss out on a gain. I hate the fact that I have often been incorrectly bearish over the past eight years. But it is far more important that I was correctly bearish ahead of the disasters of 2008.
For another example, a journalist who tells their readers to buy Enron will be lampooned and vilified. (In the UK, journalists often call this the “Private Eye Effect”, as you will appear in the satirical magazine Private Eye, which mercilessly documents bad journalism.)
But I once, during the bubble, wrote a column claiming that Amazon was insanely overpriced. A month later, I apologised after Amazon’s share price had nonetheless doubled again. But nobody complained. I had merely been prudent — even though steering people away from 100 per cent gain (in a month) is financially as bad as steering them towards a 100 per cent loss in Enron.
This asymmetry creates a bias in favour of caution and negativity.
It is our job to try to overcome that bias, but it is there. It is precisely analogous to investors’ well-documented bias in favour of avoiding losses.
A second factor is that journalists tend to perceive themselves, and to be perceived, as watchdogs. We are supposed to be sceptical, and to be the public’s first line of defence against people in the industry trying to oversell them things. Again, this leads to asymmetric negativity.
To be clear, neither I nor any of my colleagues have conscious biases, and it is our job to overcome any biases of which we are aware. But that is my best explanation for the phenomenon. Feedback welcome.