Macroeconomics is tough for students and laymen: the professors and textbooks sound like nonsense after you've been a fed a lifetime of macrogruel from the legacy media.
I've got a running list of real world complications that make understanding macroeconomics hard. I've just picked up another one from Brian S. Wesbury.
Westbury is a chief economist for financial firm. My guess is that he spends a lot of time on the TV and radio presenting his positions.
The problem he's observed is that the mainstream media just about requires that every interview piece about the macroeconomy provide both a pro and a con.
That's admirable as a principle.
But ... it's useless as a way to approach issues that aren't subjective and that aren't split 50/50 to begin with.
Broadly, the U.S. economy is in good shape about 75% of the time, and bad shape the other quarter of the time. This is a good approximation to the amount of time we spend in expansion and contraction.
Clearly there's going to be a problem if a mainstream media outlet insists on presenting the economy as a 50/50 split.
The difference between 50% and 75% is big, but doesn't sound too problematic - after all, people can have different opinions, can't they?
Not so fast. A little math shows that this leads to a huge bias towards inappropriate presentation of the economy.
Let's take the highly stylized example of a news talk show hosting two experts to provide opinions about macroeconomic issues. I'm going to assume that the issues follow the 3/4 and 1/4 breakdown mentioned above. I'm also going to assume that the experts have that breakdown on issues as well. But, initially I'm going to assume that the experts are in fact naive: they match up those probabilities, but they don't have any special skill at labeling a good issue properly (statistically, the probabilities are independent).
So, the talking head drops a new issue on the discussants, and asks them to comment on it. Because, each expert has a 3 in 4 chance of labeling it as a good issue, they have a 9 in 16 chance of agreeing that it's good (3/4 times 3/4). Extending the arithmetic, they have a 1 in 16 chance of both labeling it bad, and a 6 in 16 chance of disagreeing.
But, recall that 3/4 of the time the new issue is in fact a good thing. So, we need to multiply the earlier numbers by this fraction. We'll do the same thing for the 1/4 of the time that a bad issue is raised, and we are left with 64 possibilities that break this way:
- 27 times a good issue is properly labeled by both experts as a good issue
- 18 times a good issue yields a split opinion
- 3 times a good issue is mislabeled as bad by both experts
- 9 times a bad issue is mislabeled as good by both experts
- 6 times a bad issue yields a split opinion
- 1 time a bad issue is properly labeled as bad by both experts
This looks really bad: the experts agree on the correct label less than half the time. I'll take care of that shortly.
First, let me point out two more subtle issue that is at the heart of Westbury's argument: 1) the experts are going to provide more mixed or bad evaluations of good issues than they will of bad ones (21 to 15), and 2) they are far more likely to offer a split opinion on a good issue than a bad one. Note that both of these follow directly from the fact that good issues are more likely than bad ones.
Anyway, recall that I said this is a naive model. I assumed that the experts opinions are independent of the true nature of the issue. But, that can't be correct: they're experts precisely because they are better at attaching the proper labels. Without going through more math, the upshot is that choosing experts neutrally is going to increase the numbers in the first and last bullet point, and decrease the numbers in the middle.
Yet, this is precisely what the mainstream media does not do.
Instead, they go out and look for experts with differing opinions. This means that the media is intentionally decreasing the probability of any of the four bullet points where the experts agree, in favor of the two where they offer a split opinion. But, since good issues are more common than bad, split opinions will be more common over good issues as well - even though a neutral or random selection of true experts would be unlikely to yield such a result.
Is it any wonder than a nation full of leisurely people who can afford to pay to have someone to redo their closets to hold more stuff have unfounded angst about the economy?
Read Brian's whole piece entitled "Fair and Unbalanced" in the August 9 issue of The Wall Street Journal.
Comments