Monday, February 13, 2012

How to Detect Blather

Macroeconomists dislike criticism like anyone else, and especially hate the idea that their opinions are no better than anyone else. So, I especially liked this little quip from one of the most insecure economists blogging today, Brad DeLong, who was invited to speak on macro at some poli-sci class:

Professor Intro:Brad’s a great guy, we go way back ...he blogs...here’s Brad

J.Bradford DeLong: I do remember being horrified once when I was introduced by the Vice President of the Federal Reserve Bank of Kansas City, not as an economic historian or macroeconomist or the sometime Deputy Assistant Secretary of the Treasury for Economic Policy, but as the weblogger.

So I’m going to start by putting up what I’m calling equation 11 up on the board, simply to demonstrate I am a real economist and to say this equation 11 is going to be the centerpiece of Lawrence H. Summers and J. Bradford DeLong’s Fiscal Policy in a Depressed Economy, which is going to be presented at the Brookings institutiton on Friday March 23, and then published in the Spring issue of the Brookings papers on Economic Activity.

Intro: You are a serious scholar and a serious policymaker which I did omit...I apologize
Here's the equation 11 DeLong refers to, which contains 7 variables, a couple of which are straightforward (tax rate, the real growth rate), the rest could all be somewhere within the (-inf,inf) space that lets macro debates happen. Further, he's talking about rates of change, so there's an unspecified time dimension and these parameters are time-varying, having different short and long-term levels and the definition of 'long term' also probably varies over the business cycle.

He didn't actually discuss what the equation is, or how it adds to the debate, just that he, his credentials, Larry Summers and their initials, will be speaking at an Important Institution which will then publish research that contains as a centerpiece, an equation. In short, equation 11 is being used to prove his profundity in the most disingenuous way, because equations by themselves do not mean anything. That is, it's not like this equation has ever been empirically tested, let alone validated, and the history of economics is littered with empirically vacuous equations. Further, it's not like it is going into a refereed publication.

Many mediocre minds are impressed by famous colleagues, graduate degrees, Harvard, or equations. Those people aren't worth impressing. While such signals are correlated with good ideas, they are neither necessary nor sufficient for a good idea. When someone emphasizes these signals, however, that should lower their credibility among thoughtful people because it suggests bad faith, a preference towards pretentious irrelevancies.

I've seen this a lot, hiding behind equations. It often works because one does not want to sit down, understand what all the variables mean and how they are measured, etc. I just don't feel impressed at all by such equations because my Bayesian prior is they merely have a bad idea in equation form, so I don't give them the benefit of the doubt even when I'm lazy and have not enough time to evaluate their math.

13 comments:

W.C. Varones said...

Well said, sir!

Anonymous said...

He's making a joke and it wasn't funny. He's an economic historian and he has good things to say minus jokes.

Leo said...

Maybe you would be a better thinker if you were more insecure, Eric. A bit of humility might allow you to make the effort to understand what talented individuals are saying when they don't say what you want to hear. At any rate, focusing on the content rather than the form or the messenger is most likely a much better heuristic than the one you are proposing...

Steve in Greensboro said...

He's not much of an economist, but DeLong does know doughnuts!

Eric Falkenstein said...

Leo: Paradoxically, insecurity and humility are not positively correlated. An insecure person is less willing to expose his uncertainties to himself or others because he is afraid he might lose face, so he maintains beliefs well beyond their appropriate level of confidence.

Mercury said...

He wants to reserve the robes, holy books and authority for himself and the other high priests because he knows that if the alter is ever turned around and the liturgy translated from the Latin into the vernacular, it will be starkly obvious that it’s all just a bunch of hocus pocus.

Anonymous said...

He is not joking, he is dead serious about his analysis and the importance of "the equation"

http://delong.typepad.com/sdj/2012/01/thinking-aloud-about-fiscal-policy-in-a-depressed-economy.html

Patrick R. Sullivan said...

Maybe Brad's watched the old Hitchcock movie, 'Torn Curtain' too many times.

gpc31 said...

"Sir, (a+b^n) / n = x , hence God exists—reply!"

Euler could do this sort of thing tongue in cheek -- DeLong, not so much. Too self-important.

Why is it that when I watch and listen to Delong all I can think of is the Comic Book Guy on the Simpsons?

Jose said...

Too many people use equations (and technical jargon) for intimidation, signaling, or obfuscation and too many audiences let themselves be intimidated, impressed, or fooled by this. But I can't agree that "Those people aren't worth impressing."

Unfortunately, in many fields (academe, large bureaucratic organizations) your job depends critically on impressing people who have what the post calls "mediocre minds." Because of that, posturing becomes a job-critical skill and iterative selection guarantees that only those who impress get to decide the fate of others in those fields...

JCS

Tel said...

I can't help doing dimensional analysis when I see a new equation.

As you say, there's a time dimension involved and if you chase up other posts from Brad about eq.11 you see he uses regular % per annum for those rate factors (rd, g, ρ all have dimension of 1/time). I agree that those are not constant w.r.t. time, but that's OK because maybe eq.11 represents our best understanding for a given future horizon based on the information we have available (I'm trying to be generous here).

So that leaves the magic η factor and there's some explanation of where that comes from:

http://delong.typepad.com/sdj/2012/01/hysteresis-and-the-american-unemployment-problem.html

Note that the η factor must also be a rate with dimension of 1/time but here is Brad's derivation:

"Two years during which real GDP has stayed flat at 7% below our pre-2007 estimates of potential output have managed to push potential output down by 0.5%: that suggests a value for η of 0.5/(7 x 2) = 0.035."

The way I see it, this process delivers η as a unitless ratio instead of a per annum rate. If that's the case eq.11 *must* be broken, without any further study required.

I'd say it is also fraught with danger to use (rd - g) as a denominator, when you are subtracting two numbers that are quite close to each other (but both are highly debatable as to how you empirically determine what they are) and you get a denominator close to zero so a small error in either rd or g gives a large error in the answer.

J said...

My favorite teacher used to say after a lecture: Now I am going to clean the blackboard and fill it with meaningless equations, so the next class will know we have been learning engineering.

sell structured settlement said...

Paradoxically, insecurity and humility are not positively correlated. An insecure person is less willing to expose his uncertainties to himself or others because he is afraid he might lose face, so he maintains beliefs well beyond their appropriate level of confidence.