Financial and economic journalism

Explanations of daily changes in aggregate stock market indices are among the most ridiculous, speculative, and uncertain causal inferences made by journalists (other than that women tennis players, and only women tennis players, lose because they choke under pressure).

How do journalists know that a videotape appearance by OBL induced a swing of 00 points in a 500-stock index? To what trades did that explanation apply and to what degree? Can the journalist provide even one example of a trade for which that was the case? (“Yes, Lou, I moved out of IBM and into soybean futures this morning because I saw OBL take a walk on videotape.”) More that one example? Enough examples to explain the aggregate change? Compared to other reasons and strategies for buying and selling that day?

And financial journalists make these unsupported inferences every day the market is open.

The inference mixes levels of analysis: a few grand macro-events are mobilized to explain the aggregate result of thousands of different trading strategies, many of which are utterly indifferent to the events of that day. Many trades might well have been decided long before the news events of the immediate day. This method of explanation resembles astrology: macro-level astronomical events are used to explain and predict micro-level outcomes.

Note also that the effect is assumed to be a unique, single-day effect; and then the next day, some other news again produces a one-day effect. Just because financial journalists operate on one-day cycles does not mean that what they are trying to explain operates that way. A common error in analysis of evidence about human behavior is to assume that people think the same way the analyst, thinks. (This error also occurs in the analysis of surveys and polling data–as the analyst acts as if people are reasoning about politics as the analyst reasons about politics. Usually the error takes the form that the political belief systems of voters have the same coherence as the political beliefs systems of elites, an error exposed long ago by Philip Converse.)

There may be a self-fulfilling effect in partial operation here; journalism’s theory becomes a strategy applied by traders as they try to guess the day’s price moves. That is, the Keynes beauty contest analogy: not identifying the most beautiful but rather the one that will be thought to be the most beautiful. Many traders do watch the news flow to make the beauty contest guess. How many trades are based on that logic?

Editors should ask about every such inference: “How does the reporter know that this is the case?” And then publish the inference only if there is some kind of external evidence supporting the causal inference. The New York Times and The Wall Journal do not report the inferences of astrologers, so why these inferences?

These pseudo-explanations represent lazy journalism; the reporter looks at a few indices and at a few leading news stories on the computer screen–and then makes a grand causal inference. At times, this method provides an easy but false integration of the world news with the financial news of the day. When, and under what conditions, is idea that “news drives markets” true?

At a micro-level, this method of inference certainly has some leverage: a particular stock goes up or down in response to the day’s news about that particular stock. But it is a fallacy to extend routinely and automatically that method of explanation to performances of aggregate indices.

Here is a discussion of “Single Factor Analysis“