I saw an interesting comment threat over at Money Illusion which I could use elaboration on-from whoever can provide it. Frances had this comment on Sumner's style.. Actually Mark Sadowski quoted her-she had made it over at PragCap initially:
"Frances, I hope you don’t mind if I share this comment made by you at PragCap:
http://www.themoneyillusion.com/?p=26468#comment-326694
Neither Mark or She elaborates here-I haven't read the PragCap piece yet, true-but I wonder what this means? I appreciate it being a 'virtue'-as long as issues that truly aren't material ot the big picutre arent abstracted for. Mark do you believe he abstracts nothing that really isn't needed to get at the big picture?
I'd be interested for either Mark or Frances to elaborate what they mean here. Frances also makes this comment about Morgan Warstler:
"Mark, Yes indeed. Being able to see the “big picture” is definitely a virtue in macroeconomics. Though we also need people who understand compositional issues, or we get fallacies of division and the like. Really my disagreement with Morgan Warstler about his GI idea is about compositional issues, game theory and stuff like that – I don’t think he “gets” microeconomics."
"Frances, I hope you don’t mind if I share this comment made by you at PragCap:
“Typically, Sumner simply avoids the problem by ignoring the endogeneity and focusing on the exogenous “envelope” which constrains endogenous monetary base creation and interest rates. He’s very “macro”….sometimes I wonder if he realizes that woods are made of trees.”
"I found that hysterically funny, because it’s true. It used to irritate me how Scott would run roughshod over “compositional” issues. But I’ve come to realize that it’s one of his virtues."
http://www.themoneyillusion.com/?p=26468#comment-326694
Neither Mark or She elaborates here-I haven't read the PragCap piece yet, true-but I wonder what this means? I appreciate it being a 'virtue'-as long as issues that truly aren't material ot the big picutre arent abstracted for. Mark do you believe he abstracts nothing that really isn't needed to get at the big picture?
I'd be interested for either Mark or Frances to elaborate what they mean here. Frances also makes this comment about Morgan Warstler:
"Mark, Yes indeed. Being able to see the “big picture” is definitely a virtue in macroeconomics. Though we also need people who understand compositional issues, or we get fallacies of division and the like. Really my disagreement with Morgan Warstler about his GI idea is about compositional issues, game theory and stuff like that – I don’t think he “gets” microeconomics."
I think Morgan is missing something somewhere too in admittedly thought provoking proposals. What is is though that she thinks he misses in microeconomics?
P.S. I think I kind of get what she's getting at: I make the same point when Sumner gloats about the sequester having no impact last year: even if you believe it had no effect that showed up on the Macro numbers-and I'm not convinced that this is true in any case-there are some very bad Micro effects-if you mean like millions of Americans getting furloughed to whom it is small comfort if NGDP is claimed not to have fallen.
P.S.S Yes, I did have Tom Brown in the title yet I never got to him. Well, he had some good comments in the thread as well, plus, I was thinking he might be able to explain some of this to me. Here he is:
"Frances, I look at Beckworth’s use of the word “exogenous” there in one paragraph and compare it to a previous conversation I had with Nick Rowe regarding Beckworth’s use of “exogenous” in a previous post of his. Take a look when you get a chance:"
http://pragcap.com/crickets/comment-page-1#comment-171571.
P.S.S Yes, I did have Tom Brown in the title yet I never got to him. Well, he had some good comments in the thread as well, plus, I was thinking he might be able to explain some of this to me. Here he is:
"Frances, I look at Beckworth’s use of the word “exogenous” there in one paragraph and compare it to a previous conversation I had with Nick Rowe regarding Beckworth’s use of “exogenous” in a previous post of his. Take a look when you get a chance:"
http://pragcap.com/crickets/comment-page-1#comment-171571.
"Mark do you believe he abstracts nothing that really isn't needed to get at the big picture?"
ReplyDeleteI think that in macroeconomics people frequently make claims which are subject to the fallacy of composition. The economy is not just the sum of its parts. The contemporary compulsion to microfound has generally resulted in a poorer understanding of how the macroeconomy works, mostly because people draw the wrong conclusions about the direction of causality, and the nature of the effect.
I've also wondered why absolutely no one thinks microeconomic models would benefit from "macrofounding", when intuitively it is obvious that microeconomic phenomenon are highly impacted by the business cycle.
Well I certainly agree with that-I don't see why Macro must be the hadmaiden of Macro. Though again, as I've said before, Macro is basically Keynesian economics and Micro is how they saw the economy prior to Keynes-methodlogical individualism.
DeleteMike, I don't really know what is meant by "compositional issues." That's the 1st time I've seen that term. I'm guessing it has to do with "components" as in zoomed in a little bit more than looking at broad aggregates (the forest).
ReplyDeleteAs for the rest, I'm starting to form this idea in my head (which could be completely wrong, since I've never cracked a book on this subject) that endogeneity and exogeneity are:
1. Complements of each other
2. Are both a matter of degrees, and we might even be able to assign a number to each of them on the interval [0,1]. So that if epsilon = degree of exogeneity of one variable wrt another, then 1-epsilon is the endogeneity of that same variable wrt the other same variable.
I could be way off there, but I ran a simple proposed method of measuring this degree of exogeneity/endogeneity past David Glasner, and he didn't laugh at me (and yes, he did respond). Similarly Frances Coppola was at least OK with my verbal description there.
This isn't quite perfect as an analogy, but imagine we have just two values we're looking at: they could be overnight rate (FFR) or monetary base (MB) and the other might be inflation rate (IR). Now the following isn't exactly true of MB and IR (you need to manipulate things a little to get them in the format I'll describe), so say we've done a manipulation and we have two related variables y (related to MB) and x (related to IR). Then if x and y are jointly Gaussian, and we looked at a bunch of sample points in time taken at a fixed sample period T, you might be able to plot x vs y and have it look like this:
http://support.sas.com/documentation/cdl/en/statug/65328/HTML/default/images/mcmcex20transcatter.png
Forget the axis labels on that chart: just say x is horizontal and y is vertical.
So they are not fully independent nor dependent. If they were fully independent, the ellipsoid would be lined up perfectly with its major axis along the x-axis or along the y-axis. If they were fully dependent (just one degree of freedom) all the points would lie on a perfect line, not a fuzzy ellipsoid. The degree that y can be considered a function of x can be measured by the magnitude of their correlation coefficient, rho. This magnitude ranges from 0 to 1, where rho itself ranges from -1 to 1 (negative correlations are downward sloping).
So y has a variance, and so does x, but the variance of y given x is much smaller than the full variance of y. In fact it's var(y|x) = var(y)(1-rho^2). Also the mean of y is different than the mean of y given x: mean(y|x) = mean(y) + rho*(x-mean(x))*sqrt(variance(y)/variance(x))
http://en.wikipedia.org/wiki/Multivariate_normal_distribution#Conditional_distributions
So I'd bet that in this case (with MB and IR) we'd have 0 < |rho| < 1
... some degree of MB endogeneity/exogeneity wrt an inflation target (and vice versa), not fully endogenous nor exogenous ("fully exogenous" in this case just means fully jointly exogenous, because one variables is always assumed held to be the target, so that one is always exogenous, and further we assume that whatever we achieved was actually the target we were after.
So, that's my current way of thinking about it... using this jointly Gaussian model. And like I say to even make that work you have to do some transformations on the original data, and it helps if you have a constant small sample period, etc. I'll dig up my comment to Glasner if you're really interested (I go into more detail).
But in the end the "degree of endogeneity" or "exogeneity" isn't as important (probably!) as the raw conditional variance of y and conditional mean of y (both wrt x), for example.
So I think *maybe* "forest fro the trees" Sumner in terms of my jointly Gaussian random variable analogy is just looking at the mean and variance of the semi-dependent variable (y), notices that it's a function of 1) being conditioned on x and 2) the value of x, and declares y to be "exogenous"... not worrying about the "small stuff" (i.e. that y bounces around about this new x-determined mean with a new smaller variance).
DeleteFrances Coppola used the term "envelope" as in x sets the envelope for y. Beckworth liked her description.. My "envelope" has fuzzy edges.
So again I caution you, that I don't know if my analogy is close ... it makes all kinds of simplifying assumptions. But I think it *might* be a more useful mental image that just a binary on or off endogeneity or exogeneity.
Did any of that help at all?
As an example of two variables which should be nearly fully jointly "exogenous" (meaning |rho| = 1) we could have x (representing P) and y (representing IR) where P is the price level. Knowing one should basically tell you the other (only one degree of freedom between them). That' not exactly true because you need to know at least one sample P(t0) along with IR(t) to completely know P(t) at all the other sample times, t[n]. Do you see what I'm saying?
In this case the only transformation required to make this work should be a log since IR is generally a geometric growth rate rather than a linear one (i.e. it's measured in % / year rather than dollars / year).
I guess another way you could look at it is if you had some kind of normalized functions x and y:
x = f(y, other parameters and variables)
y = g(x, other parameters and variables)
in general, then it's like we instead hold x constant, and then factor y:
y = u(x)*epsilon + v(other variables, all of which are independent of x)*(1-epsilon)
with epsilon on [0,1]. Then as epsilon approaches 1, y approaches being fully exogenous along with x (y is completely determined by x). As epsilon goes to 0, then y approaches being fully endogenous (not a function of x at all). Again, just an analogy. A deterministic one this time.
Guys it's not that complicated...
ReplyDeleteSumner's thing is that Printed money expands base. Full Stop.
then lots of people ask questions, I was one of them for a LONG TIME, about well how does that money get into the economy.
And Sumner say Hot Potato Effect.
So imagine 2 party system, you have at-bill and I have cash to buy it. You are about to sell.
Fed swoops in and buys it.
Now you have cash, I have cash and Fed has the T-Bill.
That's the end of Sumner's composition. Because now YOU AND I both have cash are are trying to buy the next t-bill, or if no t-bills are left,t he next thing closest to one..
And then fed swoops in and buys that one too!
Now ANOTHER guy has cash, you have cash, and I have cash, and we are wlll looking for the next t-bill, and guess what?
The point is eventually, one of us, buys something other than a T-Bill. We move into another risk category into another financial asset class.
And this drives Cullen and Frances who WANT DESPERATELY TO TALK ABOUT BANKS ONLY LENDING WHEN THYE SEE A GOOD LOAN TO MAKE.
You see how Sumner's hot potato runs over top of their entire construct?
-----
Saxie, the idea I don't get Micro is hysterical. Macro was a "new" thing for me compared to Micro.
But Frances committed to doing a post on GI/CYB very soon, so we'll see.
Yeah I don't know. To me Microeconomics is for conservatives anyway. To me Macroeconomics is Keynesian econ and Micro is what we had before that.
ReplyDeleteMike your little thumbnail blogger image reminds me of this:
Deletehttps://worldoftanks.com/dcont/fb/wows_forum_art/kilroy/kilroywas_here.jpg
LOL. I'll take it reminding of Killroy. You mean my new image?
DeleteYeah the new one, because it cuts off your mouth right under your nose.... leaving you with a Charlie Chaplin type mustache. We see too much of your nose for you to look like this guy:
Deletehttp://images6.fanpop.com/image/photos/33100000/Wilson-home-improvement-tv-show-33144922-1024-768.png
If you look at the top right it has this same picture and there it doesn't cut off my mouth. It's only in the comments it does that.
DeleteAnyway, I can always count on you to notice everything-LOL.
That's how I knew you didn't actually have a Charlie Chaplin mustache. :D
DeleteTest to see if my new blogger pic appears.
ReplyDelete