The Limits of Predictability
Over a year and a half ago, the economics blog Calculated Risk posted this handy chart:
Click for chartistic engorgement.
The author here is presenting a unique way to look at recessions. By centering each recession on the base of the employment trough and counting back to the past and forward to the future, we can easily see that most every post-WWII recession took about as long to recover from the base of the employment trough as it took to reach that ultimate low point. The most recent "recession," represented by the red line (and adjusted to account for an "artificial" hiring binge required by the 2010 census) had already hit its low. Time would only tell if this recession would or would not prove to be just like the others.
Shortly after this chart was posted and had made the intertubes rounds, I shared it as well, closing the post with an audacious claim: "We'll see what happens in, according to the graph, about 15 months. If we're back to pre-recession employment, I was wrong." Yes, in that post I predicted our most recent economic downturn would absolutely not resemble anything we'd seen in recent history.
To understand why I might have been in the ballpark with my admittedly audacious prediction, we have to understand the nature of science, especially when it comes to predictability. Let's consider an analogy to the economic downturns we call recessions in this simple graphic to the left. A simple pendulum swings to and fro with such predictable regularity that we use it to mechanically time all manner of things. With a simple formula, one can determine, for example, the length of the pendulum based upon the period of the swing.
Ah, but a Radiolab podcast some years ago called Limits of Science noted that this only works for a simple pendulum. Add just one joint between the lower-most weight and the upper-most pivot, and predictability becomes impossible. Check it out for yourself in the next video.
As you can see, in the first eleven seconds with the second pivot locked the pendulum behaves as predictably as any simple swinging object. Pull that lock, though, and the entire device becomes a dynamic and chaotic pattern generator with swings and gyrations that, again according to the Radiolab episode, prove impossible to predict.
In his book The Logic of Failure, Dietrich Dörner describes systems that become complex, perhaps too complex to be either modeled or managed by people:
If we want to solve problems effectively . . . we must keep in mind not only many features but also the influences among them. Complexity is the label we will give to the existence of many interdependent variables in a given system. The more variables and the greater their interdependence, the greater that system's complexity. . . .
A system of variables is "interrelated" if an action that affects or is meant to affect one part of the system will also always affect other parts of it.
(Dietrich Dörner, The Logic of Failure-strategic thinking for complex situations, Metropolitan Books, 1996, p. 38.)
With only two interdependent variables, the double pendulum is (oxymoronicly enough) a simple complex system, but complex nonetheless given the extremely interrelated nature of the swinging portions (what others have also called "tightly coupled"). The lower section's swing influences the upper section's swing, and vice versa. Each slight change in either magnifies the otherwise predictable period of swing in the same way, for example, that slight changes in the weather make predicting long-term weather, or (more to the point) like economic systems.
Again, I seem to be making an audacious claim. After all, didn't I just show a graph marking the descent and predictably symmetrical rise of employment during a recession and note its pendulum-like regularity? Ah, but those recessions were different than the most recent.
I'm going to detour a bit in this post to note another graph and what it might mean to the overall picture regarding recessions. The graph's author generated it to see what correlation there might be between the two indicators.
Data Bigger!
The conclusion: "As is readily apparent, the correlation is fairly weak over the entire timeframe (+0.36)." I'm going to audaciously claim that the grapher might have missed a variable or two that could clarify the situation he presented. First of all, note that the gas prices, unlike the population-adjusted miles driven, has not been slaved to any given time. Fixing the price in 1990 dollars, for example, might show more usable data. I doubt that, simply because I think that to show an actual correlation the graph would have to include a missing variable, one that fixing the price would only imply: The number of dollars in the economy.
You see, those three variables-dollars in the economy, miles driven (as a measure of economic productivity), and the price of gasoline-prove so tightly coupled that to eliminate any one of them would be to ignore the all-important bigger picture.
So, why doesn't someone include the bucks floating about and show this tightly-coupled interplay between gas use and money? This proves to be the real hurdle to understanding for a variety of reasons, most of which involve ignorance.
I'm not hurtling the I-word about here in a sophomoric attempt to smear my ideological opponents. Nope. In this case, one can make a strong argument that there is almost no agreement on what money is, how it is created today, or even how it can be destroyed. The number of Big Brained people who butt their big heads together over this topic has become a spectators sport of titan clashing without rules or anyone sufficiently versed in the "rules" of the conflict to render a decision either way.
With his book Debt-The First 5,000 Years, for example, David Graeber notes repeatedly that economists have put forth an evolution of money that has become locked in our heads as intractable legend: "First comes barter, then money; credit only develops later." (David Graeber, Debt: The First 5,000 Years, Melville House Printing, 2011, p. 21.) He then spends his enormous book tracing the various types of exchange with and without media such as money, all impeccably documented by social anthropologists and historians, only to reveal time and time again that the people supposedly responsible for thinking about money itself have gotten the development timeline exactly ass-backwards.
"The reason that people were ready for such a conversation [about the economy] was that the story everyone had been told for the last decade or so had just been revealed to be a colossal lie. There's really no nicer way to say it." (Graeber, ibid, p. 15.)
I won't belabor how the terms of credit have seeped into our brains and vocabularies (which I've already done), or harp on how money is created by banks yet again. Right now, I'll just note that the entire arena for discussing the issue is completely muddled to the point where economic Fauxbel Prize winners don't understand monetary creation. When our experts can't agree on reality, that reality can be well and truly described as completely confused, if not FUBARed. How then are any graph makers to extract the true nature of money from the mess?
Worse, the term "money" itself is nebulous to the point of blobbulisity and therefore impossible to define with adequate specificity. Steve Roth at The Angry Bear points out that too many learned voices on the money topic still use weasel words to describe what should be a settled science: "It’s not uncommon to find leading economists of all stripes . . . using vague, quasi-technical terms like 'moneyness' and 'money-like.' They don’t seem to have a tight technical definition that they can rely upon others to understand and use synonymously. . . ."
Imagine if physicists didn’t have a solid definition of energy - if they meant slightly (or wildly) different things when they referred to it, sometimes hewing to some vernacular usage, sometimes silently assuming various technical definitions. Or if one was never sure which definition they were using in any given discussion. Or if two physicists arguing were frequently using different implicit definitions, and often weren’t even aware of it. Or if they shifted their own (implicit) definitions within the course of a discussion, often even within a single sentence?
Physics discussions would be in the same kind of eternally inconclusive mess that economics is in, and has been in for centuries.
I won't go into the specifics of Roth's suggestion, other than to say he makes valid points that should be taken seriously (but probably won't be). I'll end his mention by repeating the biggest point of his post: "What is arguably the most important word in economics remains undefined or at best variously and inconsistently defined - and used."
It gets even worse. Years ago (I think), bradhicks highly recommended solarbird to those wishing to better understand our economy. She proved well-read and observant on specific economic topics and trends . . . until she dropped the topic entirely. She recently explained why:
I want to reinforce something that Jim Grant touches upon: the obfuscation of data, on a large scale.
There are a few reasons I've had less to say over the last couple of years in economics; some of it's just lack of time, some of it's lack of substantial change, and so on. But some of it is the inability to obtain valid market data. The financial and economics worlds have become so overwhelmingly clouded by specifically invisible non-market forces and moneys. . . .
And so many of the data sources are like that now. You have no idea where there's going to be a huge wodge of money thrown in at any time, and most of your analysis turns into trying to figure that out, rather than figuring out the nonpolitical economy.
And even all of that would be something one could manage, were it not all so intentionally opaque.
So that's why I'm not posting more. I'm pretty good at data analysis, but I need data I can trust to talk about what I'm seeing. And for the greater part - I no longer have it. Money supply? HAHAHAHAHA What the hell even is that anymore?
(Bold emphasis mine.)
So that thing we need to thoroughly contemplate the current trends is real; it has a reality. It just has a reality that is completely misunderstood by most and even apparently obfuscated officially. It's no wonder it isn't considered.
So where does that leave us? Simply put, we are left with inadequate tools-or even the mutually-agreed-upon definitions that would allow us to craft tools-to describe where we are now, over a year and a half after that first recession graph of employment was posted. Here's the most recent CR graph from last June:
Largenate the Continuing Recessionation!
As I noted at the beginning of this post, I was right. It's a bit difficult to notice, since this graph abandons the symmetrical positioning of the trough bottoms, cramming the rest of the recessions to the left of the graph and muddling the real original comparison; but it can be seen. From a bottom of an about 6.5% additional unemployment we have yet to climb half-way that distance, even though we are a full three months (as of June) past the recovery point a simple pendulum-esque recession would predict. The scariness of this graph has even percolated up to those who's job seems to be encouraging recovery (as in a return to our former normal) even though such a thing may not be technically possible.
This recession has an additional element or two contributing to and complicating the economic confusion, the most important of which I believe happened in 2005 and has not since reversed.
Embiggenation, por favor.
Without a reversal in this trend (which is unlikely in the extreme), the very first thing we will have to do is rethink how many of our current economic traditions we can carry with us into the future. Just getting to that point, though, is going to require abandoning the wishful thinking rife in economics reporting and education and focusing on hard definitions. Even with that major attitude adjustment, I doubt we will be able to predict exactly where our economy is heading; there are simply too many variables at play here, too many unknowns we as a species have not encountered in centuries. We can, though, start to explain why the scary numbers are still happening, why our world economy is gyrating wildly.
With that information better understood, we might at least get used to the wild ride.
X-Posted to talk_politics.
Click for chartistic engorgement.
The author here is presenting a unique way to look at recessions. By centering each recession on the base of the employment trough and counting back to the past and forward to the future, we can easily see that most every post-WWII recession took about as long to recover from the base of the employment trough as it took to reach that ultimate low point. The most recent "recession," represented by the red line (and adjusted to account for an "artificial" hiring binge required by the 2010 census) had already hit its low. Time would only tell if this recession would or would not prove to be just like the others.
Shortly after this chart was posted and had made the intertubes rounds, I shared it as well, closing the post with an audacious claim: "We'll see what happens in, according to the graph, about 15 months. If we're back to pre-recession employment, I was wrong." Yes, in that post I predicted our most recent economic downturn would absolutely not resemble anything we'd seen in recent history.
To understand why I might have been in the ballpark with my admittedly audacious prediction, we have to understand the nature of science, especially when it comes to predictability. Let's consider an analogy to the economic downturns we call recessions in this simple graphic to the left. A simple pendulum swings to and fro with such predictable regularity that we use it to mechanically time all manner of things. With a simple formula, one can determine, for example, the length of the pendulum based upon the period of the swing.
Ah, but a Radiolab podcast some years ago called Limits of Science noted that this only works for a simple pendulum. Add just one joint between the lower-most weight and the upper-most pivot, and predictability becomes impossible. Check it out for yourself in the next video.
Click to viewAs you can see, in the first eleven seconds with the second pivot locked the pendulum behaves as predictably as any simple swinging object. Pull that lock, though, and the entire device becomes a dynamic and chaotic pattern generator with swings and gyrations that, again according to the Radiolab episode, prove impossible to predict.
In his book The Logic of Failure, Dietrich Dörner describes systems that become complex, perhaps too complex to be either modeled or managed by people:
If we want to solve problems effectively . . . we must keep in mind not only many features but also the influences among them. Complexity is the label we will give to the existence of many interdependent variables in a given system. The more variables and the greater their interdependence, the greater that system's complexity. . . .
A system of variables is "interrelated" if an action that affects or is meant to affect one part of the system will also always affect other parts of it.
(Dietrich Dörner, The Logic of Failure-strategic thinking for complex situations, Metropolitan Books, 1996, p. 38.)
With only two interdependent variables, the double pendulum is (oxymoronicly enough) a simple complex system, but complex nonetheless given the extremely interrelated nature of the swinging portions (what others have also called "tightly coupled"). The lower section's swing influences the upper section's swing, and vice versa. Each slight change in either magnifies the otherwise predictable period of swing in the same way, for example, that slight changes in the weather make predicting long-term weather, or (more to the point) like economic systems.
Again, I seem to be making an audacious claim. After all, didn't I just show a graph marking the descent and predictably symmetrical rise of employment during a recession and note its pendulum-like regularity? Ah, but those recessions were different than the most recent.
I'm going to detour a bit in this post to note another graph and what it might mean to the overall picture regarding recessions. The graph's author generated it to see what correlation there might be between the two indicators.
Data Bigger!
The conclusion: "As is readily apparent, the correlation is fairly weak over the entire timeframe (+0.36)." I'm going to audaciously claim that the grapher might have missed a variable or two that could clarify the situation he presented. First of all, note that the gas prices, unlike the population-adjusted miles driven, has not been slaved to any given time. Fixing the price in 1990 dollars, for example, might show more usable data. I doubt that, simply because I think that to show an actual correlation the graph would have to include a missing variable, one that fixing the price would only imply: The number of dollars in the economy.
You see, those three variables-dollars in the economy, miles driven (as a measure of economic productivity), and the price of gasoline-prove so tightly coupled that to eliminate any one of them would be to ignore the all-important bigger picture.
So, why doesn't someone include the bucks floating about and show this tightly-coupled interplay between gas use and money? This proves to be the real hurdle to understanding for a variety of reasons, most of which involve ignorance.
I'm not hurtling the I-word about here in a sophomoric attempt to smear my ideological opponents. Nope. In this case, one can make a strong argument that there is almost no agreement on what money is, how it is created today, or even how it can be destroyed. The number of Big Brained people who butt their big heads together over this topic has become a spectators sport of titan clashing without rules or anyone sufficiently versed in the "rules" of the conflict to render a decision either way.
With his book Debt-The First 5,000 Years, for example, David Graeber notes repeatedly that economists have put forth an evolution of money that has become locked in our heads as intractable legend: "First comes barter, then money; credit only develops later." (David Graeber, Debt: The First 5,000 Years, Melville House Printing, 2011, p. 21.) He then spends his enormous book tracing the various types of exchange with and without media such as money, all impeccably documented by social anthropologists and historians, only to reveal time and time again that the people supposedly responsible for thinking about money itself have gotten the development timeline exactly ass-backwards.
"The reason that people were ready for such a conversation [about the economy] was that the story everyone had been told for the last decade or so had just been revealed to be a colossal lie. There's really no nicer way to say it." (Graeber, ibid, p. 15.)
I won't belabor how the terms of credit have seeped into our brains and vocabularies (which I've already done), or harp on how money is created by banks yet again. Right now, I'll just note that the entire arena for discussing the issue is completely muddled to the point where economic Fauxbel Prize winners don't understand monetary creation. When our experts can't agree on reality, that reality can be well and truly described as completely confused, if not FUBARed. How then are any graph makers to extract the true nature of money from the mess?
Worse, the term "money" itself is nebulous to the point of blobbulisity and therefore impossible to define with adequate specificity. Steve Roth at The Angry Bear points out that too many learned voices on the money topic still use weasel words to describe what should be a settled science: "It’s not uncommon to find leading economists of all stripes . . . using vague, quasi-technical terms like 'moneyness' and 'money-like.' They don’t seem to have a tight technical definition that they can rely upon others to understand and use synonymously. . . ."
Imagine if physicists didn’t have a solid definition of energy - if they meant slightly (or wildly) different things when they referred to it, sometimes hewing to some vernacular usage, sometimes silently assuming various technical definitions. Or if one was never sure which definition they were using in any given discussion. Or if two physicists arguing were frequently using different implicit definitions, and often weren’t even aware of it. Or if they shifted their own (implicit) definitions within the course of a discussion, often even within a single sentence?
Physics discussions would be in the same kind of eternally inconclusive mess that economics is in, and has been in for centuries.
I won't go into the specifics of Roth's suggestion, other than to say he makes valid points that should be taken seriously (but probably won't be). I'll end his mention by repeating the biggest point of his post: "What is arguably the most important word in economics remains undefined or at best variously and inconsistently defined - and used."
It gets even worse. Years ago (I think), bradhicks highly recommended solarbird to those wishing to better understand our economy. She proved well-read and observant on specific economic topics and trends . . . until she dropped the topic entirely. She recently explained why:
I want to reinforce something that Jim Grant touches upon: the obfuscation of data, on a large scale.
There are a few reasons I've had less to say over the last couple of years in economics; some of it's just lack of time, some of it's lack of substantial change, and so on. But some of it is the inability to obtain valid market data. The financial and economics worlds have become so overwhelmingly clouded by specifically invisible non-market forces and moneys. . . .
And so many of the data sources are like that now. You have no idea where there's going to be a huge wodge of money thrown in at any time, and most of your analysis turns into trying to figure that out, rather than figuring out the nonpolitical economy.
And even all of that would be something one could manage, were it not all so intentionally opaque.
So that's why I'm not posting more. I'm pretty good at data analysis, but I need data I can trust to talk about what I'm seeing. And for the greater part - I no longer have it. Money supply? HAHAHAHAHA What the hell even is that anymore?
(Bold emphasis mine.)
So that thing we need to thoroughly contemplate the current trends is real; it has a reality. It just has a reality that is completely misunderstood by most and even apparently obfuscated officially. It's no wonder it isn't considered.
So where does that leave us? Simply put, we are left with inadequate tools-or even the mutually-agreed-upon definitions that would allow us to craft tools-to describe where we are now, over a year and a half after that first recession graph of employment was posted. Here's the most recent CR graph from last June:
Largenate the Continuing Recessionation!
As I noted at the beginning of this post, I was right. It's a bit difficult to notice, since this graph abandons the symmetrical positioning of the trough bottoms, cramming the rest of the recessions to the left of the graph and muddling the real original comparison; but it can be seen. From a bottom of an about 6.5% additional unemployment we have yet to climb half-way that distance, even though we are a full three months (as of June) past the recovery point a simple pendulum-esque recession would predict. The scariness of this graph has even percolated up to those who's job seems to be encouraging recovery (as in a return to our former normal) even though such a thing may not be technically possible.
This recession has an additional element or two contributing to and complicating the economic confusion, the most important of which I believe happened in 2005 and has not since reversed.
Embiggenation, por favor.
Without a reversal in this trend (which is unlikely in the extreme), the very first thing we will have to do is rethink how many of our current economic traditions we can carry with us into the future. Just getting to that point, though, is going to require abandoning the wishful thinking rife in economics reporting and education and focusing on hard definitions. Even with that major attitude adjustment, I doubt we will be able to predict exactly where our economy is heading; there are simply too many variables at play here, too many unknowns we as a species have not encountered in centuries. We can, though, start to explain why the scary numbers are still happening, why our world economy is gyrating wildly.
With that information better understood, we might at least get used to the wild ride.
X-Posted to talk_politics.