For some reason, my use of commodity futures as predictors of future spot prices for commodities (e.g., soybeans) incites fire and fury from some Econbrowser readers. Hence, I want to cite another example of the use of futures.
The primary policy tool of the U.S. Federal Reserve is manipulation of the federal funds rate, an overnight interest rate on interbank loans that is quite sensitive to the total quantity of reserve deposits that are created by the Fed. The Chicago Board of Trade offers a futures contract whose payoff is based on the average value for the effective fed funds rate over all of the calendar days of a specified month.
A separate question from whether changes in futures prices are possible to predict is the question of how far in advance they give a useful estimate. One standard of comparison is the mean squared error, or the average squared difference between the implied futures forecast at a given date and what the actual fed funds rate turns out to be. A benchmark for comparison is the assumption that the fed funds rate itself follows a martingale, so that one’s forecast for the future value of the fed funds rate is always its current value. Such “no-change” forecasts have often proven to be very difficult to beat out-of-sample with financial data. The table below shows that, if you simply predicted that the fed funds rate isn’t going to change, you’d have a mean squared error of 389 basis points (that is, a standard deviation of about 20 basis points or 0.2%) predicting one month ahead and 2,522 basis points (50 basis-point standard deviation) predicting 3-months ahead. For comparison, the MSEs of the futures-derived forecasts are only a third as large.
The moral is, if you think the fed funds rate is going to do something over the next few months that differs from what is predicted by the futures prices, then think again.
So, here is another dimension in which futures are pretty good forecasts, even better than a random walk (martingale to be precise). How are the forecasts evaluated? The author uses “mean squared error” and “mean absolute error”, respectively:
MSE penalizes the square of the error, and is a consistent estimator of the error variance. Large errors are penalized — proportionately — more than small.
MAE penalizes the Euclidean distance in the error, and might be more appropriate if the population variance does not exist.
Note that in no case does a miss of a point estimate result in infinite penalty. In other words, if the futures predicts 801, and another prediction is for “not 801”, then the “not 801” does not win if the outcome is 800. Seems obvious, but that logic seems to elude some people.
In Chinn and Coibion (2014) [Google Scholar cites = 136], we use the ratios of (root) mean squared errors, of the futures against a random walk benchmark. This type of ratio is called a Theil U statistics. Chinn (1991) [Google Scholar cites = 96] deploys both (root) MSE and MAE statistics, following Meese and Rogoff (1983) [Google Scholar citations = 4803].
The author of the post on Fed funds futures forecasting, using the MSE metric, is …. Jim Hamilton. (Figured some people won’t trust the message coming from me, for whatever reasons, so am trying somebody universally trusted on the statistical theory).
By the way, futures — like forwards — don’t work so great for currencies.
An interesting post, whatever the underlying reasons for uploading it were.
But why would I trust Robert Plant more than Jimmy Page that something was more musically viable??
I tried to put a similar link up on this before, but it had some vulgarity attached to it. That’s not a complaint to anyone, it’s a bland statement. Now admittedly, I stretch the rules of this blog about as far as the good blog hosts can withstand. But also, never let it be said Uncle Moses doesn’t try to accommodate the gentlemanliness of our good blog hosts and meet them halfway, just as they, the good blog hosts, have tried to meet me a little past their own lines of personal good taste. With that aspiration in mind, I give you the following link— sans vulgarity
Now, some of you may feel comfortable voting for a man who expresses those views towards one of the most dark musty dungeon inhabiting warmongers this nation has ever known. If you’re comfortable voting for the b*stard, then have at it. NOT me.
Dick Cheney is a decent man???? I guess Biden is saying that if one can work with the Devil, one can work with anyone. Oh wait, Trump and McConnell are even more evil than the Devil!
Has anyone phoned up “Princeton”Kopits’ wife to make certain she gives him triple the dosage of anticoagulant medicine that he usually takes??
“Figured some people won’t trust the message coming from me, for whatever reasons”.
It is interesting that the “fire and fury” link went to yet another comment from CoRev accusing you of not being honest. Of course you were being honest but I guess when CoRev is faced with something that undercuts his undying love for Trump – he has to go off with his dishonest “fire and fury”. And yet yesterday CoRev actually claimed his never questioned your integrity – just your “objectivity”. I guess a requirement for being objective is to be a 24/7 Trump sycophant.
Of course Krugman has the perfect line – truth has a well known liberal bias.
Actually the Krugman is either “facts have a well known liberal bias” or this version:
“Evidence has a well known liberal bias”.
Check the link out as it has a picture of CoRev’s mentor – Stephen Moore!
I guess in their world facts and evidence are the opposite of “truth” so my original version of the quote might raise a few right wing objections!
Krugman’s oped includes this statement from Stephen Moore:
Recently Moore declared that the Fed is “filled with hundreds of economists who are worthless, who have the wrong model in their mind. They should all be, they should all be fired, and they should be replaced by good economists.”
Wow! Good economists in his view includes himself, Lawrence Kudlow, and Judy Shelton I guess. Good economists appear to be people who will say anything that their master (Trump) wants them to say. In other words, “good economists” are stupid political hacks who get paid for lying to the rest of us. I presume this is what CoRev means by being “objective”!
Menzie again tries to justify the WRONG THING! If I were teaching a class on ethics in economics I would use his year long series of articles as an example of how ethics failed. Why is this mentioned? Because of the multiple unrelated examples used over. This article again shows another: “For some reason, my use of commodity futures as predictors of future spot prices for commodities (e.g., soybeans) incites fire and fury from some Econbrowser readers.” Incites fire and fury was a link to a comment, but his response to it:
Menzie Chinn Post author
July 17, 2019 at 6:05 am
CoRev: Regarding accusations of my not properly attributing the data, do you remember you writing the following?
…Menzie has admitted he mis-attributed his full range of sources used, and has yet to provide ALL the data he used.
That wrongful accusation is something you have never apologized for, nor even admitted you misunderstood that FRED data was in that case BLS data….
That response was for a totally different article and series of comments. Is it deflection or obfuscation? Dunno. It clearly does not answer the question, and is a typical response of his when questioned.
A couple of frequently used adages in statistics and science is:
““Essentially, all models are wrong, but some models are useful.” – George Box”
““wrong” only comes into play if the model does not correctly answer the question that it claims to answer …” J. Michael Steele.
That is the fault with Menzie’s PREDICTION. It claimed a false precision.
The value of a model is in its predictive ability. How accurate and useful is its predictions for decision making. No one has questioned the value and efficacy of the Chinn and Coibion (2014) model. As it actually compared its predictive capabilities against other econometric models and available future prices. What has been questioned is the horrible phrasing and false precision implied in Menzie’s PREDICTION derived from it,
Soybean futures are remarkably good predictors of future spot prices of soybeans. Hence, my best guess of soybean prices one year from today is 872. …”
It was given on July 15 2018 without any reference to the uncertainty. Certainly the Model Run provided the necessary uncertainty values, but the implication of the PREDICTION was just the opposite. It is Standard Practice in many fields to show both probability and uncertainty values when they are available to evaluate the prediction’s quality or establish a range for evaluating the predictive value for decision making. Conversely it is a bad practice to NOT show them or to create a model which can not develop them. The Chinn and Coibion (2014) model did provide them:
“So, here is another dimension in which futures are pretty good forecasts, even better than a random walk (martingale to be precise). How are the forecasts evaluated? The author uses “mean squared error” and “mean absolute error”,…”
If Menzie believes this:
“MSE penalizes the square of the error, and is a consistent estimator of the error variance. Large errors are penalized — proportionately — more than small.”
” Note that in no case does a miss of a point estimate result in infinite penalty. In other words, if the futures predicts 801, and another prediction is for “not 801”, then the “not 801” does not win if the outcome is 800. Seems obvious, but that logic seems to elude some people.”
Then why ignore these values in a prediction?
Another serious problem with Menzie’s prediction was it mix and match use of data and bad formulation/phraseology of his PREDICTION. His model is based upon end of month spot prices. Yet his PREDICTION was for mid-July. His argument was the July 2019 soybean futures contract end date was July 12. That is true, but spot prices are also reported on a daily basis. His estimate was for a specific spot price on a specific day, a POINT ESTIMATE. Why digress from that point estimate to the July 12 contract end date? The July 12 price was higher than July 15, $9.13 vs $9.02, making his price error higher than his $8.72 best guess. Is it indicative of more sloppiness?
Did Menzie’s PREDICTION meet the Steele test? No! Being better at predicting than other models is not a test of its own predictive ability. Seems obvious, but that logic seems to elude some people. Did his model predict spot prices accurately? No! Is his PREDICTION complete enough to be used for low risk 12 month decision making? Probably not.
“Menzie again tries to justify the WRONG THING! If I were teaching a class on ethics in economics I would use his year long series of articles as an example of how ethics failed.”
First you call him a liar. Oh wait – you deny that as all you have done is to question his objectivity. Pretty rich for a serial Trump sycophant. Now you call our host unethical. Come on CoRev – we know your hero has zero ethics. And your ethical standards are beneath’s Trumps. But this parade of yours is getting really absurd. NO ONE here believes a word you write. Move on.
” If I were teaching a class on ethics in economics”
CoRev, you have no qualifications to teach a class on ethics, economics, or underwater basket weaving. You “might” be able to get a job teaching at an elementary school, but there are pretty strong requirements in my (very blue) state, so that’s probably right out too. Good luck in your teaching endeavors, oh right there are none. Just another load of junk from CoRev.
Now, now – I hear CoRev is the best basket weaver in his preK class!
Now now now, pgl, Menzie used the word “mean” twice in his heading and then in connection with his formulae. You must understand that CoRev is very fragile and cannot stand it when people are so “mean” to him.
We are mean to Trump too which is “very unfair” to the “nation” since Donald J. Trump is the nation!
“Did Menzie’s PREDICTION meet the Steele test?”
WTF do you mean but the “Steele test”? Oh yea – attacking Christopher Steele by the Republican sycophant yesterday rather than own up to all those documented acts of Obstruction of Justice as well as taking election aid from Putin. Yes – attack Mueller as well as our host simply because your hero is not only incompetent at economics, a blatant racist, a criminal, and a traitor! Make America White Again.
Maybe CoRev meant Danny Steele:
“Standardized tests are a measure of student achievement… but they do not measure a student’s heart, or their work ethic, or their kindness, or their creativity… or anything else that makes for successful adults.”
Let’s apply this to CoRev. Failure as a student. Only creativity seems to be defending Trump at all costs. Work ethic? Well he is working hard to spin the Trump doctrine. And he still is a failure at being an adult? Go figure!
CoRev: Let me be precise. You accused me of being inaccurate in sourcing of data in one instance, regarding the sourcing of one of the most frequently downloaded series on FRED (FRED series PAYEMS). I demonstrated you were wrong in your accusation. You have never admitted you were wrong on this count, nor did you ever apologize. That is pathetic.
Regarding soybeans, it took you several posts for you to understand the distinction between spot, near-month futures as proxy for spot, and specific futures contracts, and you have never admitted that confusion. Do I really need to compile and then post all the links to all the comments that demonstrate your sheer and utter confusion? Just say “yes”, and it will be so: you’ll get one whole post on the subject of how people constantly in the wrong refuse to admit error — and be gracious in admitting it.
He FINALLY did figure out that an inward shift of the supply curve does not lower prices. Progress I presume but even glaciers move faster!
I would not expect an apology from CoRev, Menzie. But then, I think you know that, a reasonable prediction as a point estimate, although he did admit that he had been wrong about expecting the tariffs to end soon. Some folks here are just not into apologizing.
I just wish CoRev would bother to read Fama (1970) but I guess that is too much to ask of the boy.
CoRev babbles on for pages ” . . . What has been questioned is the horrible phrasing and false precision implied in Menzie’s PREDICTION derived from it,. . . ”
CoRev, please stop trotting out this stupid lie. No one with half a brain cell considers a 12 month prediction of a spot commodity price based on futures prices to be accurate to a penny. No one. Thus, there is no false implied precision. Menzie assumes at least an average level of intelligence and intellectual honesty from his blog readers. He assumed wrong in your case. Your argument is silly and after repeating it many many times has convinced no one. Yet you persist in some sort of weird fetishization of “owning teh libs” stupidity. It only makes you look even weaker and dumber than you (probably) really are. You clearly are still confused about “FUTURES” prices and what they even are and event studies and how they can be used to isolate exogenous effects.
Well, the other theory is that you are paid to do this. I do wonder if this is all an attempt to goad Menzie into writing something inappropriate, or just spend too much time on this blog, and get in trouble with the University or federal funding agency.
Dave: Fortunately, it’s summer break, and I am on a 9 months contract, so I think I am relatively safe…. And I can always note that quoting and rebutting a statistics illiterate is useful as a teaching case for a basic stats course. Never thought I’d have to put up MSE and MAE formulas.
Now, it is possible WISGOP will get it into their head to FOIA-equivalent me for teaching (clearly socialist inspired) stats on my own time to non-university folk, etc. If they did it to Cronin, they can do it to me.
“Now, it is possible WISGOP will get it into their head to FOIA-equivalent me for teaching (clearly socialist inspired) stats on my own time to non-university folk, etc. If they did it to Cronin, they can do it to me.”
The Cronin(sp?) situation is partly what I was alluding too. See also Michael Mann. Of course, it seems you are 10 steps ahead of me on that.
Dave: Yes, you are right — Cronon.
I think you are well within the margin of safety (if I thought you were on the borderline I would even tone down my own comments). Very well within the margin of safety—-
However, playing devil’s advocate:
I dare you to find anything in professor Finkelstein’s writings that is factually false. He is meticulous in his citations (I don’t think you’ll find Finkelstein using “Quora” anywhere, for example) and makes sure to have multiple underpinnings, refortified contentions, and multiple corroborations to authenticate his writings and verbal arguments.
The problem is, he states facts that people don’t want to hear. Anyone who finds themselves with an affinity for reality and expounding on unpopular realities can find themselves on similar ground as Norman Finkelstein.
” the other theory is that you are paid to do this.”
If he is getting CoRev any more than the Federal minimum wage – someone is paying WAY TOO MUCH!
pgl: I concur. Too incompetent commenting to be paid-work.
Dave, of course “CoRev, please stop trotting out this stupid lie. No one with half a brain cell considers a 12 month prediction of a spot commodity price based on futures prices to be accurate to a penny. No one. ” Also no one expects an unconditional point estimate without defining uncertainty to be accurate to the penny, but that’s what Menzie provided. Can you explain why Menzie did so? Menzie won’t.
Some one of average intelligence and intellectual honesty would recognize the near impossibility of the implied precision and accuracy. Don’cha think? Why don’t you? Don’t you qualify?
No it was YOU that insisted forecasts do not require a confidence interval to which others corrected your stupid comments. There you go again – flat out lying about what the past discussions really said. It is OBVIOUS by now that you just make this up as you go with ZERO accountability for your own gibberish, stupidity, and serial dishonesty. Like I said before – you are a complete waste of time.
“no one expects an unconditional point estimate without defining uncertainty to be accurate to the penny, but that’s what Menzie provided.” “Can you explain why Menzie did so? ”
He did not do so. You are the only moron who fails to grasp the simple point. Everywhere and at all times, words written on the page are shorthand for full and complete meaning. Everywhere and at all times, one must interpret what is said/written in light of context and common general knowledge. If I e-mail my analyst to request his 1-year target price for XYZ company, and he writes back $8.00, it is plainly evident that there is a confidence interval associated with that number. If I am concerned about the size of the confidence interval, or would like more information regarding the certainty and/or base-case vs. optimistic vs. pessimistic scenarios and conditionalities thereof I will ask. I would also point out that this entire concept is embedded in the meaning of the term “estimate” which you use but fail to understand. Only CoRev is so blinded by his ignorant mind that he needs this spelled out.
Dave, exactly right! “If I am concerned about the size of the confidence interval, or would like more information regarding the certainty and/or base-case vs. optimistic vs. pessimistic scenarios and conditionalities thereof I will ask.” If you knew, absolutely, it was entirely likely that your analyst’s target would take an act of Got to be reached?
You’ve already admitted that the spot price was highly unlikely to be in error, when do you think a confidence level would have been provided? After more than 1 year it is still not provided. How long would you keep listening to this analyst’s advice?
Because, you and others don’t expect one, I do for those reasons I listed else where. Other fields use other and better standard practices.
Pgl, only you can take this: “Also no one expects an unconditional point estimate without defining uncertainty to be accurate to the penny, but that’s what Menzie provided.” and the year of comments to mean this: “No it was YOU that insisted forecasts do not require a confidence interval to which others corrected your stupid comments. ”
You really, really do amaze.
CoRev: Truly, I do not understand anything you are writing in this comment.
This is probably writing for the wind, except for perhaps Barkley Rosser, who has been very gracious. But the futures, which would be good indicators, can’t forecast policy for an Administration which does not have policy set, and may change it day by day. Lest this only seem to criticize the current Administration, let me point out that the Carter Administration had several radical changes in monetary policy (relative looseness, monetary aggregates targeting, monetary aggregates targeting plus credit controls, then back to monetary aggregates targeting) in the span of about a year. And, no, they were not all generated by the wicked Federal Reserve. My attempt to try to get a numerical version of what this Administration is doing on the level of tariffs doesn’t seem to be appreciated, but the people who are seriously investing in the market are either setting themselves up for disaster or doing something similar. How can you forecast an Administration perfectly that doesn’t know what it wants to do and is willing to pay the price?
Julian Silk: I agree that futures forecasts are unlikely to be particularly good forecasts under this administration wherein the market has difficulty disentangling Administration intent and actions (those are separate things, by the way, given sheer incompetence and fickleness at the top). But it’s hard to figure out what would be better. In general, futures should be inferior to random walk/martingale under the EMH. It just happens in the past, they have been superior at 12 months horizon for soybeans, given cost of carry and the possibility of a risk premium.
Thank you very much for a very gracious reply. By the way, I don’t disagree with you for soybeans. I will just say for the record that it is hard to imagine a market forecast procedure that would be better, so interest rate futures through some sort of orthogonalizing procedure, or foreign exchange values won’t do well, either. It would be a nice research project to see whether political betting markets, which might take account of projected reactions on the part of an administration, would add anything that is not currently present in the futures.
Julian Silk: Yes, I agree — interestingly the prediction markets I know of don’t get into things liek “will there be a trade agreement between US and China”; probably too hard to specify. In principle, futures markets should incorporate expectations regarding political outcomes, but I suspect the markets are somewhat “informationally segmented”, at least in short run.
Good point. The Efficient Markets Hypothesis is based on agents based their decisions on expectations which incorporate all publicly available information at the time. Now no one has a crystal ball with respect to WTF Trump is going to tweet over the next few weeks. Then again – one has to wonder how long any rational agent is going to take seriously Trump’s ongoing barrage of twitter nonsense.
I wanna say, as someone who considers themselves very open to different economic theories, even the more abstract ones, “martingale” has to be one of the most dumba$$ ones I have ever come across. In fact I would probably put it just barely beneath the Laffer curve in terms of asininity. And where it has added insight into any human activity, I would love to see it.
I wanna stress I am speaking in broad generalities here, not in any relation as to how the concept of “martingale” would either defend or excoriate donald trump.
[ Puts hands up blocking my facial area as I flinch my head back and grimace ]
Folks, you knew someone continually and desperately crying out for attention like me was going to get a name drop in the posts portion of the blog some time, didn’t you?? I mean do I really have to explain psychoanalysis to you people on top ALL of my other duties here?!?!?!?
Yes, Moses, “martingale” is an abstract mathematical concept from probability theory, essentially the generalized form of a random walk, which enters into some economic and financial theories such as the EMH. Is this what has you almost frothing at the mouth at it? Did you do badly on a test dealing with it when you were studying finance? Is this why you label it “dumba$$” all bolded?
Regarding Quora, which you are back to making silly remarks about, I would say it is not an acceptable source for an academic paper. But for use in blog discussions, it is not too bad. Again, many entries in it are pretty high quality by wide reputation, if not all.
Oh, and you consider yourself to have “duties” here? No, Moses, you do not. Grow up, boy.
@ Barkley Junior
How is Kamala Harris doing in that rural poll you quoted?? Has Harris “expanded” her 6 person lead over Biden you were so excited about when you quoted it??? Keep us updated how Kamala is doing in the latest Tweedledee and Tweedledum Poll, as everyone here is dying to know. That’s assuming you’re not too busy doing public relations for Quora social media.
Has the Phd Barkley Junior finally figured out the difference between a large population sample that is uniformly distributed and one he said was “skewed”?? I’d settle for that much progress from you.
Did you want to defend Nancy Pelosi handing over $4.6 Billion for donald trump’s ICE cages for children, no strings attached??
Rumor was, Pelosi wore a fancy pantsuit for her favorite fanboy Barkley Junior when she dropped the request for stronger protections for migrant children. It seems exaggerated talk, but due to her limited number of fans I’d say it’s possible she did wear the pantsuit for you Junior.
I hope nothing I, as you said recently “capitized” or “boldened” is annoying to you here Barkley Junior, that is if early stage dementia hasn’t effected your ability to read the English language (or rather, spell it).
Still feeling guilty over not apologizing and making up ridiculous excuses for not doing so, Moses? I cannot believe that you are back to referencing the skewed distribution matter again, one of the few things you have come up with that is dumber and more ignorant than having a fit about martingales. Next we shall have you declaring that nobody rational accepts irrational numbers.
You know, I wold actually like to get along with you, Moses. I think you are a well intentioned guy, and I do not think you consciously lie, unlike some others around here. But I fear that I have offended you deeply regarding very personal matters, with your idiosyncracies in that department on display yet again in this most recent post. But, really, I am apologizing now for having come down on you too hard regarding some of these personal matters that I think you are very sensitive about. I should have left well enough alone.
But, sigh, this is probably hopeless, and you have gotten yourself too far gone into hating me. That you keep bringing up things that you should be utterly ashamed to remotrly mention is a sign that this is pretty far gone, but I would prefer to put it all aside.
I would be willing to do so even if you do not apologize, although that would be a sign that maybe you really mean it. You do still owe me that apology, and no amount of ranting about the size of that poll in Iowa or any subsequent events in the prez race undoes that.
@ Barkley Junior
You had your chance for an apology. You passed on that apology. All you had to do was type out “Six” or “6” in answer to the question of what Harris’s 1% lead over Biden equated to in potential votes in that crap of a poll you were quoting. YOU passed on that apology. That’s your choice and that’s your right. It’s kind of like losing a bet. You get an offer of a place and a time to meet at the restaurant for a free meal. After that offer is extended twice, you don’t show up at the restaurant twice, you lose the right to that free meal. YOU chose not to type “6” or type “Six”, and therefor you lost your apology, which you would have gotten. For all posterity on this blog. Nope, not now.
You’re kinda like Bania on Seinfeld, only 10 times as annoying:
I am genuinely worried about you. Really.
BTW, regarding that poll you claimed did not exist and called me a liar for saying it did and you think was just terrible, I note that if 1 percent is six people, that means the full sample was probably around 600, which is not too bad for polls like that. Of course the polls have been all over the place, although in general Biden is leading in most of them.
Oh, and sorry, but you ain’t no Seinfeld.
I see you write “My attempt to try to get a numerical version of what this Administration is doing on the level of tariffs doesn’t seem to be appreciated, but the people who are seriously investing in the market are either setting themselves up for disaster or doing something similar. ”
I guess I don’t understand your remarks. Political uncertainty, and likely changes in policy, are definitely considered by those who trade futures products and thus set the prices/movements therein.
I also do not understand what “attempt” you are referring to, but perhaps I missed a post or two of yours.
I confess to not always being “gracious,” but appreciate that you feel that way.
EMH! CoRev tossed those initials in some rant just yesterday. Let’s see if he can bother to tell us what these initial even mean and how they were evenly remotely related to one of his more recent attacks on our host’s “objectivity” or “ethics”.
Maybe corev could bring in Rick stryker as a guest lecturer, as his position to permit lies as a method of furthering ones agenda would probably be a major topic in corevs course. They are both quite ethical boys.
CoRev’s lecture on ethics will air on the Comedy Channel between reruns of Beavis and Butthead and South Park.
Hey!!! Watch it, you’re on sacred ground Sir. I love those shows.
For years I have tried to educate those who will not be educated that alternative views do exist outside their cocoon of liberal looniness. Throughout those years I have been relentlessly attacked. I found an appropriate response.
Indeed, most of the time I’m ROTFL! 😉
Alternative views? Is that the same thing as KellyAnne Conway’s alternative facts?
Differences in views is fine. Serial stupidity and dishonesty is not.
CoRev Instead of ROTFL why don’t you try SICRAEB (sitting in chair reading an econ book)? The difference between a deterministic trend and a stochastic trend isn’t just a matter of taste, like choosing which font to use. There’s a right way and a wrong way to do statistical analyses. And I’m not aware of any ideologically based differences in how to compute loss functions. You could make a rather strained argument for preferring a one loss function over another on ideological grounds (e.g., linear loss versus quadratic loss), but ideology plays no role in how to compute those loss functions.
I have tried to educate those who will not be educated
Then start with educating yourself before you try educating your betters. This is an econ blog. If you want to be taken seriously, then show us that you deserve to be taken seriously. Try going to the effort of actually reading some econ textbooks and papers. If you want us to take you seriously about time series data, then educate yourself on the topic. If you want to be taken seriously about the science, then at the very least learn some elementary calculus; e.g., knowing that the first derivative of a constant is zero…a concept that seems to have escaped you on many occasions.
2slugs, that’s the rub: “If you want to be taken seriously, then show us that you deserve to be taken seriously. ” With the clown here no conservative will be taken seriously.
MDC, that’s funny!
It’s been a rough week for poor CoRev, so that might explain his apparent incivility. Not only has he been beaten to a pulp over the tariff and soybean issue, not to mention his failure to understand the basics of statistical loss functions and his cluelessness about stochastic trends, but it’s also been a rough week in his “home” field of climate science. The barycenter paper (much ballyhooed by climate change deniers) contained a schoolboy error that has forced the publisher to consider apologizing and pulling the paper. And then a Nature paper blew-up the Global Medieval Warming and Little Ice Age hypotheses and another Geoscience paper reinforced those same findings. And then there was that second heat wave across Europe. I feel for my poor sister who lives in Paris. And to top it all off, today’s GDP numbers are hardly comforting for Team Trump.
“And to top it all off, today’s GDP numbers are hardly comforting for Team Trump.”
Fixed investment fell a wee bit. Exports fell by a lot more. WINNING? NOT!
2slugs, this is a funny perception: “Not only has he been beaten to a pulp over the tariff and soybean issue, ” Even in weak moments some have admitted the truth of my concern over Menzie’s PREDICTION. AS for the statistics issues, not being a statistician, am I supposed to care? I don’t. Now the climate stories are the typical cover our back side responses to another paper shooting holes in the attempt to remove the MWP from the records.
As for the barycenter paper this has been so little accepted,I needed to look it up. Ballyhooed? Where? By whom?
Now the China 1/2 warming from UHI paper, another in a long line of similar studies is being touted. https://www.sciencedirect.com/science/article/pii/S092181811930102X?via%3Dihub “Detection of UHI bias in China climate network using Tmin and Tmax surface temperature divergence”
If true in China then it’s true every where else.
If you disbelieve that paper then explain the UHI impact I tried to show in my response to Moses ~ a week ago.
BTW, two European heat wave sandwiched a record/near record cold wave. Hear anything about it?
You know a very strange thing about you CoRev—or rather I should say the strange thing about the limited interaction between you and me on this blog?? many people in this world annoy me. They really aggravate the hell out of me. People like you, after an initial phase do not bother me hardly at all. Because I have met SO MANY people very similar to you in my life. As you might have guessed (or maybe not) I live in the south. Your kind is a dime a dozen south of the Mason-Dixon. So you know really the only lingering thought I have about people like you CoRev??? (vs others who say dumb things and are educated enough they should know better??? What it is you’re going to tell your grand-children 30 years from now when average temps have risen maybe 10 degrees. So instead of 90 Fahrenheit in “City X”t, it’s 100. Instead of 100 Fahrenheit in “City Y”, it’s 110 Fahrenheit. And sea levels have swallowed large portions of major coastal cities, and more animal and plant life has died, what will you tell them?? I would literally pay overpriced tickets to be a fly on the wall in that moment and hear you tell your grandkids what you claim to have said back in 2019. Something tells me it won’t be anywhere near to honest.
Maybe I’m wrong?? Maybe I’m being too cynical about the humans I have met and observed in my life?? Maybe….but I don’t think so.
Moses, why didn’t you answer my question re: the rural thermometer that ends up in an urban/near urban environment? You could have used that example instead or dreaming up another.
This comment is just absurd. “What it is you’re going to tell your grand-children 30 years from now when average temps have risen maybe 10 degrees. ” 10/30=.33 degrees per year or .33/4=.0833 degrees per quarter. We don’t have to wait 30 years. This year is just fine. I often use the Woodfortrees estimator tool. The author created this index: “I’ve created the WoodForTrees Temperature Index (WTI). This is created from the mean of HADCRUT4GL, GISTEMP, RSS and UAH, offset by their baseline differences. It covers only the time period where all four series are valid, so begins in 1979 and will only contain the latest month’s values when all four sources are in. It is updated from the master sources at 3am GMT/BST each night.”
That’s two surface station and two satellite datasets.
And NASA’s GISS (GISTEMP) is usually expected to be the most aggressive of the estimates. So what does the WTI and NASA GISS WORLD temperatures say for 2019 so far?
If you can not read a graph, the WTI trend is actually negative and the GISTEMP trend is nearly even. These are in centigrade and Fahrenheit so there is no exaggeration.
Where are you going to get that ole 10 degrees average increase in 30 years?
CoRev So where are the error bars in that little WoodForTrees toy software in your link? Also, is that your idea of adult level time series software???
2slugs, asks: “is that your idea of adult level time series software???” For the surface temperature data sets, yes! I’ve show you where to get them to do your own ADULT analysis. For the satellite data their calculated uncertainty is .04C/decade and for GISSTemp it is calculated at are near 0.05°C in the global annual mean for the last 50 years.
Since I used his tool, which doesn’t show them, I’d have to add them separately. For this example it isn’t necessary for showing Moses’ ignorance.
If you wish to re-graph them be my guest.
CoRev Since I used his tool, which doesn’t show them, I’d have to add them separately.
But his tool doesn’t show them. That’s my point. You criticized Menzie because you were too lazy to look up the loss functions in his paper, but you’re just fine with a tool that doesn’t show the error bars at all. My point was that you are a hypocrite.
I agree that Moses is likely wrong about a 10F degree warming in 30 years; however, he at least has the direction right. You keep on predicting global cooling any day now…and you’ve been predicting that for years and years and years. Moses’ 10F degree warming is entirely possible within the lifetimes of today’s young people, and I think that was his larger point.
BTW, that tool is nobody’s idea of adult time series software. Most of time series analysis is not about fitting trend lines; it’s about marshalling various diagnostic tests. It’s the kind of thing I would expect to see from someone who was a computer programmer with time on his hands but didn’t understand time series analysis beyond making a graph.
2slugs, then make your own adult analytical tool that develops a result. Temperature error bars are important because of the quality of the historical data.
I guess you missed the trend directions of the graph the simple tool provided when you claimed: “however, he at least has thedirection right. You keep on predicting global cooling any day now…and you’ve been predicting that for years and years and years.” NO, YOU AND HE DO NOT HAVE THE DIRECTION RIGHT! Not in the short term, and definitely not for this interglacial. http://jonova.s3.amazonaws.com/graphs/lappi/gisp-last-10000-new.png
It is this kind of ignorance that makes you make these kinds of assumptions: ” Moses’ 10F degree warming is entirely possible within the lifetimes of today’s young people, and I think that was his larger point.” That period coincides with those of the current model averages, but only using the worst case scenario(s).
Believing the models outputs shows you’ve never done any analysis of them. Others have: https://docs.house.gov/meetings/SY/SY00/20160202/104399/HHRG-114-SY00-Wstate-ChristyJ-20160202.pdf Is just one example of many since AR5’s release. Or we can compare a longer record: http://jonova.s3.amazonaws.com/graphs/lappi/Five_Myr_Climate_Change_Rev.jpg (The original paper is only summarized and requires registration.) For me the definitive graph is this: http://jonova.s3.amazonaws.com/graphs/lappi/gisp-last-10000-new.png
Accordingly, you assurance that Moses is probably wrong on his time frame, but your longer time frame is More, most, plausible (Barkley’s term), because YOU believe without any analysis. Blind belief trumps ????
On the very, very short term, we are cooling. On the longer term, 10K years of this interglacial, we are cooling. On the very, long term, 65M years, we are cooling. In the recorded history of that longer term interglacial record and coming out of one of the cooler periods we are warming. But, because some error filled models based upon this recent warming period, this planet is going to exceed warming in NEARLY all the long term records?
More blind belief.
2slugs & Moses, this is just down right embarrassing.
Trend: -0.024 ±0.205 °C/decade (2σ)
β=-0.0024143 σw=0.0025438 ν=16.202 σc=σw√ν=0.010239
From that hard RIGHT and SKEPTICAL site https://skepticalscience.com/trend.php
Data is UAH 6.0 years are 1997 through 2014 That’s 16 years.
To translate for Moses, since the previous super el Nino, in 1998, until just before the most recent 2015-16 super el Nino, Global Average temperatures actually went down.
Go play with the data sets and time frames and see how likely is that 10 degree increase is.
CoRev Yes, it is embarrassing. You should quit making a fool of yourself. To begin with, data starting in 1997 and running through 2014 is 18 years of data, not 16 years. Run out of fingers and toes? Once again you rely upon the piecing together of noisy and unreliable satellite data that measure temperature in a region where no one lives rather than land and sea measurements. And notice the huge differences between v5.6 and v6. And you complain about tiny revisions to ground temperature data? Seriously??? And why didn’t you show the full history of your beloved satellite data? It shows an increasing trend. Now let’s talk about that trend. Did either one of the toy software tools you used test for stationarity? No. They just assumed a linear deterministic trend. Hint: If you want to assume a deterministic trend, then all of the diagnostics tell you to use a quadratic trend and not a linear time trend. Oh wait…there are no diagnostics in your toy tools.
Look, there’s only so much information you can get out of a univariate data set. No one (except you) believes you can take a univariate data set and use it to predict far out into the future. If you want to use a univariate time series to model temperature change, then you should use something like an ARIMA (0,1,2) with constant. In that case the constant effectively subsumes all of the underlying physics, but the constant is less useful for predicting because GHG emissions are not rising at a constant rate. You can further improve the time series analysis by looking at the time series for the principal radiative forcings (which are individually nonstationary) and find the stationary cointegrating trend. That’s what Kaufmann, Kauppi and Stock did in their 2010 paper on stochastic trends in temperature data. But even that is limited and is primarily intended to understand data generating behavior more than it is a long run predictor. The high emission scenarios that predict ~10F degree warming by the end of the century are not time series models. Those projections are based on climate model simulations. That’s the actual reason they prefer to use the term “projection” rather than “prediction” or “forecast.” And that’s why they use the term “uncertainty” rather than “variability.”
Also, your desperation is on display for all to see when you fall back on having to use a climate crackpot’s link to a long discredited and irrelevant graphic from a long discredited book written by yet another climate crackpot. I’m referring to this howler:
2slugs, you have not yet successfully answered the question, How many times does a test for stationarity be run against the same expanding data set? Is the added data a magic bullet?
You’ve had a problem with the satellite data when it doesn’t fit your exaggerated extreme position. Do you remember these numbers from the RSS v3.3 Trend: -0.015 ±0.203 °C/decade (2σ)
β=-0.0014669 σw=0.0024578 ν=16.984 σc=σw√ν=0.010129? Or I could show you screen shot version of the data sets, before they ?reanalyzed? the sea surface temp data,
Or we could show the trend for the past 100 years of the most aggressive data set: Trend: 1.02 ±0.12 °C/century (2σ)
β=0.010169 σw=0.00017429 ν=11.898 σc=σw√ν=0.00060121
I ask again where you going to get your 10C at the end of this generation? You Are correct that the UAH satellite data show a positive trend, but how positive? Can your analyze how that 10F happens?
While you doing that analysis why don’t you analyze that graph you say that has been discredited. I guess Dr Alley and his single point ice cores are worse than tress N. Hemisphere rings reliant on tress the NAS said to not trust. Oh, and then abandons them coincidentally when they turn downward.
You might also consider that other China UHI study I cited originally. If 1/2 of the warming in China can be traced to UHI, why would it be different any where else?
Or for years you have been predicting a diminishing food supply due to GW, and yet we keep setting world records on production even as temps have supposedly increased. Supposedly because the only surface stations that are nearly perfectly sited, the USCRN, do not show warming.
Isn’t science interesting?
“his “home” field of climate science”
Ha! ROTFL indeed. CoRev, the resident clown.
Do forecasters generally include confidence levels when using futures to forecast spot prices or is the practice standard to merely state the expected price of the spot price along with MSE and MAE? In other words, are MSE and MAE the usual measures of error? I assume this is the case. When I read your link to Professor Hamilton’s futures comment, I don’t recall seeing confidence levels related to the spot forecast value, but I do recall seeing MSE and MAE reported.
A second thought, Econoday publishes weekly forecast summaries on many economic time series and includes the mean forecast and the range of the mean forecast by the forecast participants. Thus for some reporting purposes, it seems that confidence levels are not generally reported.
Econoday link: https://us.econoday.com/byweek.asp?cust=us
AS: For random walk forecasts, I don’t think it’s typical, nor for futures/forwards, because there is no sampling error due to estimation. For an ARIMA for instance, where one estimates coefficients, one can generate a prediction/confidence interval based on the variance-covariance matrix of the coefficient estimates, and the standard error of the estimated residual. In all cases, one can calculate an either in-sample or out-of-sample RMSE or MAE statistic, which are metrics for prediction accuracy. The former (RMSE) is (if I recall my econometrics) consistent with Classical inference *and* Normally distributed error term.
The intervals from surveys are different — those summarize the distribution of different forecasts, and are some complicated transformation of distributions of the individual forecasts.
It seems to me that your latest comment should end any reasonable objection to your soybean futures forecast.
As it is, it has been quite some time since there have been any reasonable objections to Menzie’s soybeans futures forecasts, if there ever were any at all. But that has not slowed down a certain individual so far. But, maybe now we shall hear no more out of him on this matter. It really has gone on way too long, but he is notoriously persistent with his distortions and errors.
AS, I’m not sure Menzie can be considered an unbiased source for your answer.
CoRev: It’s textbook. I challenge you to find a contradicting passage in a *published* textbook like Stock and Watson, Theil, Pindyck and Rubinfeld, Green. Otherwise, you should really stop digging. (By the way, these criteria have been discussed in several of my *peer reviewed* articles, including Cheung et al. (2005) (Google Scholar citations > 1000). I think I know what I’m writing.
Please “give -it up”. There are enough economics professors reading this blog that if there were a problem with Professor Chinn’s analysis, one would most likely have politely augmented his statement. Certainly, someone would have objected to the criteria in one of Professor Chinn’s peer reviewed articles.
Perhaps you could copy Professor Chinn’s answer to my question and see if an economics professor in your area may review the question and answer, maybe in exchange for lunch.
Save your disagreement for political or climate change arguments. If you in fact have some expertise in climate studies, you are totally undermining your legitimacy to discuss scientific concepts by objecting to a forecasting method that you may not have studied. My guess is that other than the experts who read this blog, not many readers have considered the ability of soybean futures contracts to forecast the spot price a year subsequent to the futures price. And thus, most non-expert readers were most likely not aware of the use of MSE and MAE to validate the accuracy of such forecasts as standard practice. As a non-expert, I certainly was not aware of the relationship of soybean futures to soybean spot prices a year subsequent.
At this point it seems like you are being disagreeable just to be disagreeable, perhaps due to past mocking or name calling. Put all that behind you and let’s move on. Again, save your arguments for something you claim to know.
Ah, but AS, CoRev “claims to know” about forecasting soybean futures prices, and his claims to know a lot about climate science got dragged in as part of his arguing that he knows so much about soybean markets, because weather and climate, and so on. And of course this also involves politics, because there was Menzie using his forecast to, gasp, criticize Trump’s trade policies, how dare he?
As it is, CoRev has assured us of his expertise by telling us how showed his grandson the tube that contains his diploma and also a wall that supposedly has some award (or maybe more than one) hanging on it, although maybe not since he later said he threw away his awards, something he now regrets.
It is hard to keep track of all the things he claims to know, but he has shown yet again his persistence in continuing to repeat nonsense over and over with minor variations, long after it has been shown to be nonsense.
CoRev has admitted he’s not a statistician nor an economist. That must mean he believes he IS entitled to his own opinions AND his own facts. Soybeans, climate, taxes, tariffs. Name it and he’s got plenty of his own facts (minus all those numbers and squiggly lines because, well, he’s not versed in……)
AS, I take your advice seriously. Yet, I point out Menzie’s reference, Cheung et al 2005, and then looked at it and the et al is: “Yin-Wong Cheung,*, Menzie D. Chinn,Antonio Garcia Pascual” to confirm my other concern: “…I’m not sure Menzie can be considered an unbiased source for your answer.”
As I’ve said I have no objection to his model but, to the false precision and accuracy implied in his 2018 PREDICTION and its wording. Economists are not known for the accuracy pf their predictions. How many actually accurately the Great Recession? Menzie’s argument was the form of his PREDICTION was the standard practice.
AS, you seem ready to accept this level of effort as a valid STANDARD PRACTICE for economics/economists. I am not.
This horse is dead, so I will give up commenting on it. In fact would have months ago had Menzie done so. Look at my definition on how I intended to had evaluate his PREDICTION.
CoRev: You don’t get it, do you? I’m using the same approach in Cheung et al. and *that* passed peer-review, plus being cited more than a 1000 times w/o anybody critiquing the statistical approach to forecast evaluation (I’m sure some people too issue with aspects of the general methodological approach).
As long as you continue to persist in this frankly stupid stance, I’m going to continue to call out your wrong statements (of which you continue to make more and more).
Menzie, just what would have been so difficult, wrong or otherwise problematic with showing either the MSE or MAE or both in your best guess conclusion?
I further note the Cheung et al paper is another paper comparing various models as was your own Chinn and Coibion (2014).
My complaint was to your choice to not show them ( your calculated MSE or MAE or any uncertainty levels) in a PREDICTION, giving a false impression of precision and accuracy. Succinctly 872 is not 902. Your PREDICTION failed an accuracy test because you chose to write an incomplete or deceptively accurate PREDICTION.
CoRev: Chinn and Coibion had the relevant MSE comparisons. As of July 2018, you can’t calculate the MSE through July 2019 because… well the realizations had not been realized. So one had to refer to the track record. Which was why I referred to Chinn and Coibion. Repeatedly. But you seemed not to pay attention … or understand.
Please continue to highlight your complete ineptitude with respect to statistics.
CoRev doesn’t get it? He’s posting on an econ blog arguing with economists while admitting he knows little about statistics or economics. He even chastises the host for not being objective, a comical insult meant to be serious.
In another life, he’s Elwood P. Suggins of Bellbrook, Ohio, the featured character in an old Jonathan WInters monologue. There, he’s funny. Here, just tiresome.
Menzie, is 872 not your expected/possible realization? If not, how did you get to that number?
CoRev shows his insanity by writing “Menzie, is 872 not your expected/possible realization? If not, how did you get to that number?”
CoRev, Menzie already explained “As of July 2018 . . . ” Please just stop.
With regard to Christy’s MTM “discrepancies” described in his 2016 congressional testimony, please see http://www.realclimate.org/index.php/archives/2016/05/comparing-models-to-the-satellite-datasets/ for a layman’s explanation of many (but not all) of the ways Christy mis-characterized the truth. It is quite telling that for all his shenaningans, he stuck with MTM data and ignored the lower troposphere, which for the purposes of sea level rise, droughts, etc. is all that matters. In a fair world, Christy would go to jail for lying to Congress and the world.
I will not even comment on your citation to the Journal of Quaternary Science Reviews.
Dave, this is directly from Christey’s testimony:
“In this testimony I shall focus on the temperature of the bulk atmospheric layer from the surface to about 50,000 ft. – a layer which is often called by its microwave profile name TMT (Temperature of Mid-Troposphere). This layer is particularly important because it captures the atmospheric region that is anticipated to warm rapidly and unambiguously if greenhouse theory is well-understood. As such, if the impact of extra greenhouse gases (GHGs) is to be detected, it should be detected here. In Fig. 1 I show an example from a climate model simulation (Canadian Climate Model run CanESM2_rcp45_r3i1p1) of the anticipated temperature change for the period 1979-2016.” (NOTE: Figure 1 was a graph of where the non-existent atmospheric hot spot was supposed to be in the TMT.)
Your Real Climate reference notes these problems:
” No model spread, inconsistent smoothing, no structural uncertainty in the satellite observations, weird baseline. ”
1) Model Spread He used the models’ data from IPCC AR5. The complaint used a different organization for that data, an apples to oranges comparison?
2) Inconsistent smoothing has to do with padding Which shows the beginning or ending data lost due to smoothing, a meaningless complaint. Either you show it or not. If you show it it is your decision as to how you want to represent it.
3) Structural uncertainty Are you forgetting my complaint of Menzie’s PREDICTION? These uncertainties are well documented and readily available. Surprisingly only during the 2015-16 super el Nino did these uncertainties cause the observation data to overlap the models.
4) Weird baseline The only weird baselines were those presented in the complaint of baselines, where the actual satellite anomaly baselines were obviously ignored. Another apples to oranges comparison.
But the reference’s bottom line was:
“…it became apparent that the forcings used in the models, particularly the solar and volcanic trends after 2000, imparted a warm bias in the models (up to 0.1ºC or so in the ensemble by 2012), which combined with the specific sequence of ENSO variability, explained most of the model-obs discrepancy in GMST. ”
TRANSLATION THE MODELS RUN HOT UP TO 0.1C) YUP! ENSO causes temperature change.
Climate science failed to meet the scientific test. Surprisingly, this Christy statement was ignored:
“The IPCC climate models performed best versus observations when they did not include extra GHGs and this result can be demonstrated with a statistical model as well….”