I have just discovered them, but their record is impressive. FiveThirtyEight are fine, but they are at least as good.
They had predicted a Hilary 57-43 win a week out-they don't actually call these predictions but benchmarks, but it was very close to nailing it. She actually finished slightly better at 58-42.
Currently they have her winning in Pennsylvania next week, 55-45.
http://www.benchmarkpolitics.com/2016/04/pennsylvania-final-county-benchmarks.html?m=1
A poll came out yesterday which had her up by 13.
A 10 point win with 189 pledged delegates up for grabs would be good stuff. It would get her 104 delegates-she already has 1930 delegates so it would push her over 2000 to 2034. Remember she needs 2383 to win.
Benchmark Politics also has her winning Maryland 66-34.
"Reran the numbers in Maryland and save for a couple changes in some counties, they are still the same. 66%-34% Clinton."
https://twitter.com/benchmarkpol/status/722873287416524801
That would get her about 63 delegates to 32 for The Bern.
Add that to 2034 and we get 2097. Again, 2383 for the win-this number includes SDs. That leaves her at 88 percent of the way there.
BM also has her up 55-45 in Connecticut.
https://twitter.com/benchmarkpol/status/722875270563139584
This gives her 30 of the 55 CT delegates bringing her grand total to 2127.
It has her up 60-40 in Delaware.
"Very little polling on DE, so this may change, but 60% Clinton - 40% Sanders in Delaware."
https://twitter.com/benchmarkpol/status/722878118537474048
By the way: BM puts no stock in a poll yesterday which showed it a 45-38 Hillary lead as it was by Graves Marketing-a fraudulent company. They had the one NY poll which showed Bernie within six points. In Delaware they sampled African Americans at 5 percent when they are 22 percent of the population.
Delaware is not worth a ton of delegates but that would get her about 13 of the state's 21 delegates. Adding that brings her to 2140 out of the required 2383.
This leads Rhode Island which is Bernie's best shot at a win as it's another New England state. Not as good for Bernie though:
1. Considerably more diverse than some other NE states.
2. It's a 'semi closed' primary.
BM has her up 53-47, which would get her about 13 of the 24 delegates. Not a big haul either way, but this would bring her to 2152 out of 2383. Think about that. That would leave her at 90 percent of what she needs.
Bernie talks about California as where he's somehow going to make up for losing all these other states. Strange calculus. One state will make up for losing a bunch of states in a landslide. The idea that this could ever happen is implausible to say the least.
But think of it this way. CA has 475 pledged delegates at stake. Which means this: she could lose CA and that would still give her the 231 delegates she would need after next week-where we project she has 2152-to get to 2383.
Again, assuming she loses CA-which she's leading by about 10 points according to the averages-she could still get the delegates needed to get to 2383. That's because with 475 delegates at stake there's no way even if he won she wouldn't still get way over 200 delegates.
And this is ignoring all the other primaries between April 26 and June 7 in California as well as the other June 7 primaries. That same day there is also a NJ primary that she probably wins by about 10 points as well worth 126 total delegates. So she'll probably get about 70 there as well.
We haven't even talked about the fact that there remain 200 super delegates remaining, most of who will go to her between now and then.
As for the Bernie claim that she has to get to 2383 without the aid of SDs, this is false. This would be a brand new standard. Obama won with the aid of SDs as has every Dem nominee going back to 1984 when Bernie's manager, Tad Devine, invented them.
But I will tell you. Based on this analysis, she is going to have way more than 2383 total delegates. There are 1651 total delegates left up for grabs and if she can just split them she this would put her way over 2400. In truth she will do much better than split them.
They had predicted a Hilary 57-43 win a week out-they don't actually call these predictions but benchmarks, but it was very close to nailing it. She actually finished slightly better at 58-42.
Currently they have her winning in Pennsylvania next week, 55-45.
http://www.benchmarkpolitics.com/2016/04/pennsylvania-final-county-benchmarks.html?m=1
A poll came out yesterday which had her up by 13.
A 10 point win with 189 pledged delegates up for grabs would be good stuff. It would get her 104 delegates-she already has 1930 delegates so it would push her over 2000 to 2034. Remember she needs 2383 to win.
Benchmark Politics also has her winning Maryland 66-34.
"Reran the numbers in Maryland and save for a couple changes in some counties, they are still the same. 66%-34% Clinton."
https://twitter.com/benchmarkpol/status/722873287416524801
That would get her about 63 delegates to 32 for The Bern.
Add that to 2034 and we get 2097. Again, 2383 for the win-this number includes SDs. That leaves her at 88 percent of the way there.
BM also has her up 55-45 in Connecticut.
https://twitter.com/benchmarkpol/status/722875270563139584
This gives her 30 of the 55 CT delegates bringing her grand total to 2127.
It has her up 60-40 in Delaware.
"Very little polling on DE, so this may change, but 60% Clinton - 40% Sanders in Delaware."
https://twitter.com/benchmarkpol/status/722878118537474048
By the way: BM puts no stock in a poll yesterday which showed it a 45-38 Hillary lead as it was by Graves Marketing-a fraudulent company. They had the one NY poll which showed Bernie within six points. In Delaware they sampled African Americans at 5 percent when they are 22 percent of the population.
Delaware is not worth a ton of delegates but that would get her about 13 of the state's 21 delegates. Adding that brings her to 2140 out of the required 2383.
This leads Rhode Island which is Bernie's best shot at a win as it's another New England state. Not as good for Bernie though:
1. Considerably more diverse than some other NE states.
2. It's a 'semi closed' primary.
BM has her up 53-47, which would get her about 13 of the 24 delegates. Not a big haul either way, but this would bring her to 2152 out of 2383. Think about that. That would leave her at 90 percent of what she needs.
Bernie talks about California as where he's somehow going to make up for losing all these other states. Strange calculus. One state will make up for losing a bunch of states in a landslide. The idea that this could ever happen is implausible to say the least.
But think of it this way. CA has 475 pledged delegates at stake. Which means this: she could lose CA and that would still give her the 231 delegates she would need after next week-where we project she has 2152-to get to 2383.
Again, assuming she loses CA-which she's leading by about 10 points according to the averages-she could still get the delegates needed to get to 2383. That's because with 475 delegates at stake there's no way even if he won she wouldn't still get way over 200 delegates.
And this is ignoring all the other primaries between April 26 and June 7 in California as well as the other June 7 primaries. That same day there is also a NJ primary that she probably wins by about 10 points as well worth 126 total delegates. So she'll probably get about 70 there as well.
We haven't even talked about the fact that there remain 200 super delegates remaining, most of who will go to her between now and then.
As for the Bernie claim that she has to get to 2383 without the aid of SDs, this is false. This would be a brand new standard. Obama won with the aid of SDs as has every Dem nominee going back to 1984 when Bernie's manager, Tad Devine, invented them.
But I will tell you. Based on this analysis, she is going to have way more than 2383 total delegates. There are 1651 total delegates left up for grabs and if she can just split them she this would put her way over 2400. In truth she will do much better than split them.
Wow Mike!... You're on top of this, as usual. I got emails from several of my old high school friends (still living here in California), and it actually seems my emails to them about a month back (when the "bring your guns to Cleveland petition hoax" had just come out) convinced at least a few to vote Trump in the primary!... I guess it's county by county as to whether you need to re-register... so one friend (a lawyer) took the "plunge" (literally like plunging into cold water he said) and re-registered because it wasn't clear to him if they'd actually let him have a Republican ballot in his county (he's been a life-long Democrat). I hope there's enough of us left who are NOT sabotaging the Republican primary to push HRC over the edge. I *think* I'm currently registered Republican, but I'm not actually sure. Also, it seems to me that in Santa Barbara county they didn't give me any trouble in giving me the ballot of my choice last time.
ReplyDeleteI've been looking at your posts (but not as frequently), as I've been absorbed in a fun project recently. It all started with Jason's posts in March having to do with Godley & Lavoie's stock-flow consistent models (SFC models). He got into a bit of a spat with some of the post-Keynesian types, that I never fully understood, but it set me off on doing a long series of posts on the simplest of the SFC models that G&L present in their book (SIM). This led me to corresponding with Roger Sparks (who's also interested in that), and eventually let to me "building" a circuit simulation that is equivalent (in some sense) so the SIM model (in the spirit of Jason's later post containing a comparison of it to a simple resistive-inductive-capacitive (RLC) circuit (actually it's more like just an RL or and RC circuit)).
So that's where I've been all this time! I think I've pretty much done my final iteration of the "circuit" (I found a pretty cool online circuit simulator). If you want to try it out, just click on this link, and then click on the switch at the top-center of the circuit you see to get it going... and then after a few seconds it'll fire up and start running. If you want to know what you're looking at, here's a screen shot with annotations (you can add labels directly to the circuit simulation, but there's a bug which screws up the resulting link, so don't use them). The information for the circuit is actually stored right in the link... it's pretty cool!
And if you're really a glutton for punishment, here's an associated post I did on it with more information.
OK, I knew you had gotten into something-I thought it might have been work. What you are actually doing sounds like more fun.
ReplyDeleteSo you are going to vote for Trump in CA?
As a NYer I'm still celebrating Hillary and Trump's big wins.
As I suspected Nanute is for Bernie so we put up a friendly wager back in March. The one who's candidate wins takes out the one with the losing candidate to lunch at Popeye's.
Next Wednesday, at noon lunch is on me.
As for all the talk of Stock-Flow consistent models I have only a vague idea of what that means. I know it's an idea that the post Keynesians believe in-as opposed to the mainstream econ folks.
Sumner would have no time for it, that much I know.
What would be a fairly simple definition of being stock-flow consistent?
I will check out your posts with great interest. Even if I only dimly understand what a 'circuit simulator' is.
Ok, I just picked at Jason' original post and it's not hard to see why he kicked the PK Hornet's Nest: he claimed to have discovered a fatal flaw. LOL
ReplyDeleteIt took me a long time to see where he was coming from with that... and in the end I'm not sure I agree. But nevertheless, it was an interesting detour for me.
DeleteI was just checking out your post. Holy BeJesus. I feel like trying to read a dissertation in Latin without subtitles! LOL.
ReplyDeleteNot a dig at you but more me for not being great with such a formal analysis!
I will finish it-later...
Maybe I should finish Smith first and then you as that was the order of the conversation.
I make it sound way more complicated than it is. That's what I've been struggling with is how to be accurate, explain a couple of subtle things that confuse people, and yet be clear.
DeleteHere's a previous version of the circuit which runs faster and is more satisfying to try (it looks lot simpler too... more in line with what Jason had in mind in his RLC circuits post I think):
http://tinyurl.com/j9rzyvk
Again, click on the switch at the left to apply a voltage from the battery, and you're off and running. The little yellow dots moving along the wires represent the current and the color represents the voltage: green = + voltage, gray = ground (0 v) and red is negative.
And here's a decoder chart so you know what you're looking at.
The "scopes" (little graphs) across the bottom are from Godley & Lavoie's SIM model: G, Y, T and H. It's not a complete model because it's missing YD and C (that's what my more complex circuit fills in, among other things).
G = government spending per period ($20 in this case)
Y = GDP
T = taxes collected by the government per period
H = total cash in the economy
It's a "chartalist" model, so it's a given that people accept the government's cash for money, and they simply fire up the printing presses when they want to buy something. Then they shred it again when they collect taxes. There's only government script (cash): there are no electronic deposits or banks. It's a very simple model!
G is represented by the voltage of the battery you switch in to the circuit (you can edit the voltage on the fly... you can't mess up the model because it's actually stored in the characters of the link).
Y, T and H are modeled as currents (amperes).
The circuit is an RL circuit (just resistance and an inductor (L)). The inductor represents the store of money (H)... the only "stock" in the model. It's H that determines the model's dynamics (how long before it reaches equilibrium).
G, Y and T are all "flows" (they are dollars per period).
OK, you're lucky you got the short version. =)
There's more info about this simpler model in my post (actually a page) here. Originally it was just an idea for Roger Sparks... but then it metastasized into so much more! Lol.
I started out trying to explain a mechanical analogy, with springs and "dashpots"... but it was easier with a circuit IMO... plus that website allows you to easily build it and provide someone with a link to an interactive simulation, so the can see it in action. (which I thought was very cool!).
The "10" in SIM10 means I've done 10 posts on that stupid model! Lol... some of the early ones have an embedded interactive spreadsheet implementation (rather than a circuit). Here's an example of one of those. Again, it's there for people to play with... you can't mess up the original..., and if you mess up your own, you just refresh the webpage and you're back to where you started.
DeleteRoger Sparks has been diving in with spreadsheets and circuit models of his own. That's part of the fun for me: it's something I can hopefully help another interested party to understand. Putting your economic ideas down in the form of a definite model really helps you be much more clear about things. I can't say I've learned any more about economics though... perhaps a wee bit (but only that) about SFC models.
I'm afraid young John Handley is still WAY ahead of me there!
(John Handley is way ahead in terms of economics... he's not personally a fan of SFC models... he has his own models)
DeleteThe Post Keynesians and MMTers have some interesting ideas but they are sort of like the Berners in getting very dismissive of those who don't immediately see the light.
ReplyDeleteThis comment has been removed by the author.
ReplyDelete"I make it sound way more complicated than it is. That's what I've been struggling with is how to be accurate, explain a couple of subtle things that confuse people, and yet be clear."
ReplyDeleteOk, let's put it this way. What are these subtle things you are trying to clarify?
My hangup is I'm very hung up on the 'So what' of a thing. I need the big picture thing we're driving at to really make sense of a thing.
Like you are trying to put together a model-am I right so far? I get it that in economics-and physics, etc-there are models meant to test theories and evidence, etc.
What is the model for? What is it trying to demonstrate?
Excellent questions: The model comes from Godlley & Lavoie: it's a "simple" model of the macro economy, and they describe it in chapter 3 of their book.
DeleteJason chose to focus on that particular model to criticize SFC models (which are associated with post-Keynesian economics (PKE)).
I didn't understand his criticism, but I thought Ramanan might, so I emailed him. This caused a HUGE explosion of comments on Jason's blog, with a number of PKE types and associated sympathizers commenting there (I think he had well over 200 comments on that post).
Mostly Jason was catching flack from the PKE crowd. That was his 2nd post on PKE. He followed up with about three or four more, and actually turned on comment moderation for a time (it's off now).
So naturally I wanted to see what all the fuss was about, which led me to recreate Jason and G&L's results. I did see a potential problem in the way the way the sampling time was changed in the model (assuming there's an underlying continuous time model). I think the issue has been put to rest now.
But anyway, this SIM model really is simple, but the subtleties come up in the way you try to interpret an underlying continuous time model. I will get into that if you like (it's there in my blog posts)... but I'm not sure it's totally worth getting into.
The important thing is that I made a continuous time model -- one of an infinite number of possible ones which match G&L's discrete time model at each sample time. To accomplish that the right way (even though the continuous time model is not unique), I integrate results over each sample period with integrator circuits, while the model itself generates instantaneous rates.
The integrator circuits (on the upper left in the circuit) are like accountants: they add up all the activity for the variable of interest (in this case Y = GDP), over each "sample period" (which you could take to be a year for example, so it would be annual GDP).
The subtleties all have to do with doing that accounting correctly. This confused me at first as well... but I eventually sorted it all out.
In spite of all that, the model itself REALLY is very simple! Lol... you just have to trust me on that. It's not very realistic... but it demonstrates to G&L readers (in their book) how such models work. They go on to build much more sophisticated ones later (or so I understand).
Nick Edmonds builds them on his blog on a quasi-regular basis, and so does Brian Romanchuk (you can google either of those guys to see their blogs). Nick actually provides the formulas, so you can recreate them (which I did in great detail for one of his more sophisticated models about two years ago -- though with very little understanding of what the model actually represented... I was way more interested in implementing the model in an interactive way on a web page).
Brian Romanchuk (in response to Jason's March posts on the subject) has done about 3 or 4 posts on the subject himself, specifically focusing on SIM in at least two of them.
tl;dr version:
DeleteI guess G&L's model is trying to show:
1. How SFC models work and how they're constructed
2. How an economy reaches a steady state (kind of like an equilibrium) in a simple model like that.
In particular the government starts spending $20 per period (after everything was previously "dead" at 0), and after some "adjustment time" (about 10 periods or so), the economy reaches a steady state, with about $80 of cash in circulation, taxes coming in equal to government spending (so no deficits), and an annual GDP of $100 (there are also consumption and disposable income measures (C and YD) which also reach stead state values of $80 per period).
But other than that, I don't know what it's supposed to show! =)