Explaining the Parable of the Thermostat
At first I was getting really frustrated with Craw and then Gene in the comments of this post. I couldn’t believe they were accusing me of misrepresenting Scott Sumner’s views; I thought I was almost literally quoting him, with the exception of translating the discussion into the world of a household thermostat (which of course was the analogy he had used to take me out to the woodshed).
But now I totally see what is happening here:
1) OBVIOUSLY Scott wins the debate in the analogy as he set it up. He puts ridiculous arguments into my mouth, and he puts obviously correct arguments into his own mouth, when it comes to a discussion of household thermostats. In that setting, I look ridiculous and Scott looks obviously correct. In fairness, Scott explicitly said that I don’t hold the view about thermostats. But the implication is that I do hold analogous views when it comes to monetary policy, whereas Scott still holds the obviously correct view.
2) But no no no; Scott has not adequately translated our respective monetary views into the world of the household thermostat. That was why I wrote my own post. I was saying (though perhaps I should have been clearer) that if you want to take the monetary discussion into the world of thermostats, my dialog was a lot closer to everyone’s (including Bernanke, Krugman, etc.) views than Scott’s dialog.
[Now one aside before I begin: As Tel pointed out, it actually doesn’t make much sense to talk about thermostats because a thermostat keeps a constant temperature. Rather, it would make more sense if Scott had talked about a furnace with settings from 1 to 10, indicating the power level at which it blew hot air into the house.]
For example, in the thermostat world, Scott says:
Bob Murphy’s theory of temperature would be that when people are frequently turning up the thermostat, you can expect houses to be relatively warm. And when people are frequently turning on the AC, you can expect houses to be cool. The Sumner theory is that when people are most frequently turning up the thermostats, houses will be relatively cool, even though that action makes them warmer. Bob Murphy’s theory is that houses are relatively warm in the winter, because people frequently turn up their thermostats in the winter. Sumner’s theory is that houses will be relatively cool in the winter, despite the fact that people turn up the thermostat more frequently in the winter, and despite the fact that turning up the thermostat does in fact make houses warmer, ceteris paribus. Bob Murphy will claim that Sumner contradicts himself on house temperatures.
No no no, that is totally wrong, if we’re using this as an analogy for monetary policy. Watch: If I translate the “Sumner theory” above into a theory about monetary policy, it would mean this: “The Sumner theory is that when central bankers are most frequently cutting interest rates, economies will be operating below full employment and price inflation will be below target.”
Yes that’s right, but who doesn’t believe that? That’s why the central bankers keep cutting.
Where Sumner differs from other people is that he says ultralow interest rates are a sign that money has been too tight. And that the Fed’s actions since 2008 have been the tightest monetary policy since the Hoover Administration.
Those type of statements aren’t analogous to the “Sumner theory” above. Instead, they would be analogous to someone saying, “A lot of people assume that when the homeowner has continually pushed up the thermostat until it finally hits 80–the maximum setting–that he is engaged in a deliberate policy of warming. But actually, this is usually a sign that the homeowner has engaged in a cooling policy.”
Ben Bernanke, Paul Krugman, Brad DeLong, and 90% of the economics profession think that the outside got really cold, that Ben Bernanke pushed the thermostat up to 80 in a deliberate WARMING policy, but gosh shucks it wasn’t enough to offset the exogenous outside forces of nature.
That is precisely the view that Sumner wants to blow up; it’s why he started blogging. Sumner thinks that we should DEFINE the stance of heating policy not with reference to the objective actions the homeowner takes (like pushing the thermostat up to 80, or running higher utility bills in one month than in the previous decade), but rather by looking at the temperature of the house compared to the desired temperature. If the homeowner realizes that his actions will not make the house as warm as he wants it to be, then by definition he has engaged in a cooling policy–even if he pushed the thermostat up to 80.
I am saying that’s a nutty approach, which no normal person would adopt when it comes to houses and thermostats. Amongst its problems, it assumes the very thing under dispute. If our economy doesn’t in fact need looser money in order to recover, then Scott’s proposal (to make “tight money” mean the same thing as “an economy the NGDP of which does not grow sufficiently”) would make it very hard to realize he had misdiagnosed the problem.
Likewise, if we’ve let the furnace blast for a month straight, paying exorbitant bills, and yet we’re still shivering with sweaters on, it’s not very helpful for someone to define that situation as “inadequate furnace exertion.” Maybe there are a bunch of windows open, or maybe we all have pneumonia, or maybe there’s a crazy weather pattern and the optimal thing would be to turn off the furnace and go to Florida for a week. But to literally define any undesirably cold house as “inadequate furnace exertion” would be crazy.
“Instead, they would be analogous to someone saying, “A lot of people assume that when the homeowner has continually pushed up the thermostat until it finally hits 80–the maximum setting–that he is engaged in a deliberate policy of warming. But actually, this is usually a sign that the homeowner has engaged in a cooling policy.””
-I’m honestly not sure whether Scott means “has” or “is”. On his position on EZ 2011, “has” is the best interpretation of him, on his position on 2008, “is” would be the best interpretation of him.
“Instead, they would be analogous to someone saying, “A lot of people assume that when the homeowner has continually pushed up the thermostat until it finally hits 80–the maximum setting–that he is engaged in a deliberate policy of warming. But actually, this is usually a sign that the homeowner has engaged in a cooling policy.””
-I’m honestly not sure whether Scott means “has” or “is”. On his position on EZ 2011, “has” is the best interpretation of him, on his position on 2008, “is” would be the best interpretation of him.
Yeah thanks for the mention. Probably worth noting that an update appeared on Sumner’s blog:
I take that to mean that Sumner admits the thermostat analogy was never all that hot (hey!), and at least one of his regular readers was kind enough to explain that.
Now if you do want to go talking about a furnace with settings from 1 to 10, indicating the power level at which it blew hot air into the house, that implies the person twiddling the power setting there is actually doing the job of the thermostat (keeping the temperature constant). To some extent (and this is an analogy) the Federal Reserve with things like inflation targeting is also doing the job of a monetary thermostat, and personally I’m cool with that (hey!) but if you are going to do the job of a thermostat, then that’s exactly ONE job… no dual mandate.
But you all know my views on that by now I’m sure.
It miiiight help if economists took a deep breath and reviewed the history and conventions of cybernetics, since it’s older and all of this material has been covered a number of times already over the last hundred years.
https://en.wikipedia.org/wiki/Cybernetics
As a bonus, there’s an existing nomenclature and conventions relating to how to discuss the somewhat confusing terminology in control theory. By using existing terminology you save time, expand the number of people who have a clue what you are one about and get access to a useful body of existing knowledge.
Right, so the “desired temperature” is called the “set point”. That’s just an expression of a preference.
So “looking at the temperature of the house compared to the desired temperature” is known as the “error term”. That is to say, the difference between the “set point” and the measured value. This does require that you can actually measure the thing you want to control (and you should never presume that you can). A positive error term means that you would like the house to be hotter (i.e. it’s too cold right now) while a negative error term means you would like the house to be colder (i.e. it’s too hot right now).
Where you finally decide to adjust those furnace settings (from 1 to 10, indicating the power level at which it blew hot air into the house) is known by the name “control effort”… that is the exertion you impose in an attempt to bend the system to your will (deliberately “effort” has undefined units, because the physical units may depend on the specific case at hand, furnace power in the case of heating, but maybe mechanical force in the case of a servo system).
How you relate the “error term” to the “control effort” is the core of your controller design. There are a lot of available controllers to choose from, start with the basic deadband controller (i.e. how a simple electromechanical thermostat works) and then do the PID controller before you master any others. Seriously, people are still arguing about the nuances of controller design, it’s a whole art and science to itself.
What Sumner seems to be saying, is don’t judge the set point by looking at the control effort, especially if you have no idea how the controller works. Yeah, fair point… he could have just come out and said that rather than stumbling around with a badly explained thermostat analogy.
And that helps with the basic concepts:
https://en.wikipedia.org/wiki/File:Feedback_loop_with_descriptions.svg
Nice exposition, Tel. Somehow, despite being somewhat familiar with control theory and having implemented a toy PID controller in software, I had been certain up to this point that Cybernetics referred to man-machine interfaces, like Google Glass or The Terminator. I blame the movie Terminator 2 for this misunderstanding.
Upon reflection, it is the term “cyborg” which let me astray.
ANd possibly the Sirius Cybernetics Corporation from Hitchhikers Guide to the Galaxy.
“The Sirius Cybernetics Corporation is the primary manufacturer and supplier of androids, robots and autonomic assistants for the known universe. They are known for their catchy jingles and catchphrases, supplied by their Marketing Department.
They are not, however, known for the quality of their products.”
Just sleep under the stars around a camp fire, don’t bother with any house. If you feel cold, stack up a bigger camp fire. No problem.
When I was learning control theory, the lecturer used to say, “You can do anything, but you cannot change your mini minor into a formula one.”
What he meant by that was that fine tuning your control algorithm is a fine thing to do, but ultimately you will still be limited by the maximum control effort you have available. If you furnace has 5000 kW rating on a scale of 1 to 10 then you simply do not have the option of turning it up to 11 no matter how much that might assist your control strategy. I recognize that Americans probably don’t know what I mean by 5000 kW, and I can’t remember the appropriate conversion to your more usual Furlong/Firkin/Fortnight (FFF) system of units, also I’m only guessing what a regular household furnace is rated at being too lazy to look it up.
Economists understand this concept with the “zero lower bound”… you hit that limit of available control effort, and that’s all you get (other than QE of course, but let’s not go there). There’s also “by gum we seem to have hit the debt ceiling once again” but perhaps that’s a slightly different thing.
I think you are wrong Tel
https://www.google.com/search?q=amp+wth+setting+of+11&ie=utf-8&oe=utf-8
wrong link:
https://www.youtube.com/watch?v=KOO5S4vxi0o
Bob,
Suppose the thermostat in a house is set to 80; which scenario is more likely?
1) The air conditioner is on. It’s summer and it’s hotter outside, but the owner is wearing lighter clothing and wants to save money so he’s only cooling down the house to 80.
2) The heating is on. It’s actually 43 degrees inside the house, but for some reason setting the thermostat to 80 has caused the furnace to knock a hole in the wall, so all the heat is escaping.
Perfect.
I think this is simpler than it is being made out to be. Josiah’s implication is exactly right. It is more likely the owner has been using the AC to cool. If you just see 80 the smart guess is that the owner is cooling not heating. So any argument that cites the setting, 80, as if it implied it was set at a warm value out of a desire to be warm or cause warming has to come equipped with a reason to reject the inference that the AC was on.
As for the rest, do bloggers get paid by the word?
Craw wrote:
As for the rest, do bloggers get paid by the word?
No actually I’m putting this up for free, since some people find it useful. If you don’t like it, you know what to do.
Yeah. I’m over-paid too.
I get 0c for each word and so far that hasn’t added up to much, but I’m hoping that I can make it up with sufficient volume.
Like the National Change Bank!
🙂
Anyone remember this?:
Time to spread the wealth around, dahling!
http://www.theblaze.com/blog/2011/03/03/time-to-spread-the-wealth-around-dahling/
“The unpaid HuffPo bloggers are revolting!”
😀
Here is my take (I think you need a broken thermostat for it to make sense):
The inside temperature is the health of the economy
The outside temperature is external economic conditions
The central heating system is monetary policy
The thermostat is interest rates (its broken so while turning it up always increases the temperature, and turning it down decreases it , there is no direct correlation between what it is set to and the actual temperature in the house).
Bernanke/Krugman etc: When it gets really cold outside, even when the thermostat is at its max level the house will still be cold and you will have to start burning the furniture.
Sumner: What we really need is someone smart in charge of the heating system and it will be easy to maintain the temperature in the house. Don’t even worry about what the thermostat says. Just keep adjusting the amount of gas fed to the boiler and there is no limit to how much heat you can generate. If, when it gets cold outside , you don’t increase the gas flow in time and you rely just on the thermostat, don’t be surprised if you end up shivering with the thermostat on max later on.
Bob: You don’t need a central heating system at all. Just leave everyone to their own devises and they will keep the house temperature at the optimal level. And actually its only cold outside becasue we have central heating anyway. If you just keep increasing the gas when it get cold outside, you going to get huge gas bills and eventful blow the whole house up. And BTW: There is no single optimal inside temperature anyway. I wrote my theses on that.
To finish the analogy:
Sumner is arguing that as the internal house temperature is still below optimal it would be crazy to turn the thermostat down as that would decrease the internal temperature more.
Bob accuses Sumner of inconsistency because as he thinks it the gas supply that controls temperature not the thermostats he should not talk of the thermostat changes at all.
Sumner responds by pointing out that actually Bob agrees with Bernanke/Krugman that the fact that the thermostat is at its max level and the gas supply is high compared to the summer that we have a “high heat policy” – and he is equally wrong as the house is still a bit cold.
“That is precisely the view that Sumner wants to blow up; it’s why he started blogging. Sumner thinks that we should DEFINE the stance of heating policy not with reference to the objective actions the homeowner takes (like pushing the thermostat up to 80, or running higher utility bills in one month than in the previous decade), but rather by looking at the temperature of the house compared to the desired temperature.”
OK, the problem is making a lot more sense to me, now.
You have the right position on this, but Sumner doesn’t realize that you’re talking past each other.
Sumner seems to think, if I can rephrase what you’ve said, that, no matter what action is taken by the Fed, it’s the fact that the money supply is less than the planners want that makes it “tight” in his view, and which makes any action – whether increasing or decreasing the supply of money – a tight monetary policy.
I agree with you that that’s just weird.
Austrians are saying that the policy – the action itself – is either tighter or looser, regardless of the money supply. More injections of unbacked notes is a loose policy. Removing unbacked notes is a tight policy.
And the flaw with Sumner’s thesis is that if monetary stimulus is weak or impotent those applying the stimulus will never know it. They will keep adding more on the faith that until the economy responds according to theory then more monetary stimulus must be added. This blind faith in theory reminds me of the medical knowledge of the 17th century. Bloodletting was believed to be a cure. So if a little bloodletting did not cure the patient then more bloodletting was applied! Then the patient died. But rather than question the efficacy of bloodletting the doctors theorized that the disease was the problem.
“Likewise, if we’ve let the furnace blast for a month straight, paying exorbitant bills, and yet we’re still shivering with sweaters on, it’s not very helpful for someone to define that situation as “inadequate furnace exertion.” Maybe there are a bunch of windows open, or maybe we all have pneumonia, or maybe there’s a crazy weather pattern and the optimal thing would be to turn off the furnace and go to Florida for a week. But to literally define any undesirably cold house as “inadequate furnace exertion” would be crazy.”
Bingo, especially if and when the furnace overexertion, at the hands of inept technocrats who have no idea what the setting should be, weakens and destroys the integrity of the house being able to retain heat.
I’m with you 100% on this Bob, and “guest’s” comment above is spot-on. Regarding “talking past each other,” sometimes I imagine Scott Sumner at a dinner party, saying (to paraphrase the old joke), “But enough about NGDP, what do you think about NGDP?”
I can’t help thinking a lot of economists who love math probably failed diff eq and discrete math. This totally looks more like kids playing swords at the piss pot than making sense of serious subject matter. A case where Mises would have definitely frowned on the use of metaphors.