sources high up in the brazilian financial sector informe me in incontrovertible terms that orthodox econometric models, while useful, are almost entirely useless.
a typical model will select a bunch of presumed relevant variables - say, previous market growth, consumer confidence and expected interest rates - and mix them up, based on their past or logical behaviour, to try to come up with either a prediction or an indication of what a certain decision, affecting a certain variable, will do to the others.
the funny thing is that these models aren’t, strictly speaking, built to work. although their results will be used to support managerial decisions, they are not, in the parlour of the model-maker, ‘fit for backtesting’. another way to say that is to say that their predictions are almost always almost entirely wrong. real world* economic variables have a sleep-disturbing inclination to misbehave, sometimes wildly, and it stands to reason that the more they deviate from the innocently assumed correlations they were suppose to be working upon, the less useful the model's conclusions will be. that is, the model is only useful as long as it predicts things that there's not much point in trying to predict them after all.
*bear with me here
the same thing happens, more or less, when you are going to travel and make a list, drawing some projections about what places you expect to visit and how much each activity should cost you - only to actually get there and throw the list shamelessly out the window as the trip takes on a life of itself and you can barely remember when was the last time you took a shower because you’ve been drunk for the last 3 days even though you lost your wallet on the cable car ride to the itinerant circus.
honest middle-managers will recognize a similar problem as the one that arises when they receive some gibberish order from above and have to convince their reasonable employees that - based on all available evidence and operational reality notwithstanding - all is well, the the CEO has a plan, and that their actions will not in fact be contributing to their own unemployment in the very near future. i have the good fortune of being acquainted with one particular manager who, in her rise to the most rarefied heights of the career ladder, managed to observe her own transformation from someone who had to force herself to believe the rationale of decisions imposed from the top to someone who has to come up with rational excuses to pursue semi-arbitrary strategic paths to feed her peers at the lower levels of the chain. in her own words, her job these days is to produce believable narratives (or models!) to ensure compliance to certain top-level decisions.
the sheer amount of managers a typical company needs to be able to get anything done should serve as evidence enough that humans are not very good at coordinating their behaviour even in modest scales. also, and a bit oddly, there’s reason to believe that the more humans to be coordinated, the more likely coordination is to get fucked up.
this issue, as we can see, is routinely tackled through the crafty expedient of pretending that it isn't there. even cognitive psychologists concede that pigeons, when given food by a machine at random intervals, will start adopting certain repetitive behaviours in the apparent belief that they can make a difference in how the machine works - assigning, to random phenomena, whatever goes for causality in a pigeon’s mind. (wasn’t there a tv show that did a similar thing with humans? can’t quite remember its name)
reality, as we all must know by now, is an unwieldy bitch. a lot of the time it is one random pigeon-feeder. it's a system whose basic purpose seems to be to generate massive ammounts of cognitive dissonance in humans, and one of the top three causes of anxiety in modern societies.
even though we seem to have figured out the workings of the occasional natural process - at least when they stick to the 3 usual dimensions and fit the periodic table - we can’t quite claim to have become masters of the universe just yet. all social processes are inherently unstable - not the less so because a great deal of this uncertainty comes from everyone trying to come up with arbitrary solutions to unsolvable problems, often by imitating each other.
so why do economists and other misguided creatures of the dark bother with modelling at all? out of desperation? just because they can? because all the cool kids are doing it? in other terms, what is the purpose of this mass-scale process of collective rationalization (beyond, of course, instituting symbolic reality as such)?
save it for another post to argue that The Economy is the embodiment of the social-as-such, entitling it to the blatant honour of capital initials at this time of night. suffice it to say that it's slightly less predictable than the weather, and that we display similar reactions to both - usually in the form of bland statements on the line of 'it seems to be quite pleasant today mate' followed by a terrible rush for dear cover when a storm suddenly materializes.
(this is a tropical country analogy, for which i make no excuse)
when not relying on misguided optimism, in one case and the other one of two strategies is usually employed: asking your mom for advice, or declining to leave the house at all.
as the expression goes, however, no battle plan survives contact with the enemy. in short, the more uncertain the trip, the more we need to rely on half-baked guidelines about how things are supposed to turn out. and the economy is a pretty uncertain trip, as i’m sure you will all agree.
one could even argue - if one wasn't so afraid of everybody arguing back - that the sort of economic model here presented serves little more than the purpose of supplying the vast array of clueless agents with a foundation for their collective pretending that 'everything is alright after all'.* keynes said so in 1937, but it is a message very difficult to listen to. even the more so, perhaps, because hearkening to it would undermine the very self-delusion on which the working of the system relies.
*bill bryson mentions the same phenomenon taking place in academic physics, where there seems to be a very slight possibility that the universe won't either collapse upon itself or expand until it cools down entirely. physicists call this possibility the goldilocks hypothesis.
the economist’s or the investor’s model, in this sense, is little more than rationalization for his actions, aimed at smoothing over the cognitive dissonance caused by having to make decisions about sleep-disturbing things that are for the most part entirely out of his control. this is in itself nothing new, and while economists brawl bitterly about how to properly account for this uncertainty, they mostly acknowledge it in some (more or less hypocritical) form.
however!
in view of the blatant disregard for common sense displayed by economic thinking (and often by common sense itself), we could say that there’s another, less widely accepted dimension to economic uncertainty. people in the human sciences often defend the startling position that consciousness has got it all wrong, that the so-called material world is not all it’s made up to be, and that language creates reality in a very fundamental way - a way to which us, beings subjected to language and not at all cruising comfortably above it as unbiased observers, are necessarily a little blind.
this potentially disconcerting claim echoes in a way keynes’ little quip about financial instability and the silly ways people try to deal with it by all looking at each other blankly until they can all fool themselves into whatever arbitrary arrangement most conveniently masks its own lack of concreteness.
complexity and uncertainty are spiritual sisters, and the reason why humanity had to wait 10 thousand years to finally come up with a god of confusion is frankly beyond me. reality is much closer to what douglas adams graciously refers to as the Whole Sort of General Mish Mash that to whatever fits into an economist’s model: it’s a messy jumble of somethings, and if we scarcely know what is going on at the subatomic level, we know even less about about collective human behaviour.
for tricia mcmillan, each possible arrangement of possible ways of looking at the mish-mash constitutes a particular instance of reality in the multiverse. this is, incidentally, a good way to support the argument (privately held, but tough to sell) that language creates reality. in a milder version of the douglas adams proposal, language creates reality by cutting it around and colouring the selected bits with meaning. whatever is said, is imposed upon the mess, and it’s easier to believe in the existence of words than of the things they refer to.
to make matters a bit worse, but also a bit more understandable, there's what dawkings here refers to as the queerness of middle world, and of biologically evolved perception. it's more or less the same concern that led niels bohr to claim, according to legend, that the physicist is the atom's way of looking at itself (this was at the time rutherford claimed that all science is either physics or stamp collecting. good times, good times).
to drive the point home with overwhelming certainty, let us consider the matter of tying and untying knots. as any experienced sailor knows, the ancient art of untying knots is all a matter of directional poking... no human mind can fully comprehend the vast complexity of a sufficiently fancy knot. to begin with, a knot is tied based on little heuristics and rules of custom, simple instructions like ‘pull your right hand over your left like so’ and ‘maybe you should bring an umbrella with you today, son’. once a knot is tied, it drifts beyond the realm of human understanding and your only hope of untying it is ignoring it in its complexity and selecting little sides to push, and little connections to unwind, and hope that each partial doesn’t end up causing an increase on the level of complexity on the rest of the jumble. it’s all a matter, in short, of framing, screening, and cutting it around on the symbolic sense. it’s an essentially metalinguistic activity. not to waste the analogy, let us mention that social reality is more or less equivalent to what happens to your earphones when left unattended for quite surprisingly not that long.
even the most skeptical reader will find it hard to deny that the world tends to be a bit confusing, and that we have to occasionally select through a couple of its less disconcerting bits and force them sheepishly into some semblance of sense. this should be argument enough to force academic economists to add metalinguistic uncertainty to their neatly taxonomized lists of instances of uncertainty in the world out there.
so, next time you feel justified to get mad at an economist’s doomed predictions or, if you’re into that sort of thing, your financial advisor’s unwise investment tips, kindly forgive them and remember that they are just little misguided observers, subjected to mostly self-inflicted power structures, and unable to keep tabs on their own imaginary creations. much like the rest of us. they don’t know much more than you. in fact, after a few years of college there’s a good chance he’ll be less confident about his predictions than someone who didn’t study economics at all, and as such is still under the spell of the economic not-to-be-named name-of-the-father. that this doesn’t look too good on a resumée might be the strongest of the reasons-d’etre for the continued existence of the whole profession.
a typical model will select a bunch of presumed relevant variables - say, previous market growth, consumer confidence and expected interest rates - and mix them up, based on their past or logical behaviour, to try to come up with either a prediction or an indication of what a certain decision, affecting a certain variable, will do to the others.
the funny thing is that these models aren’t, strictly speaking, built to work. although their results will be used to support managerial decisions, they are not, in the parlour of the model-maker, ‘fit for backtesting’. another way to say that is to say that their predictions are almost always almost entirely wrong. real world* economic variables have a sleep-disturbing inclination to misbehave, sometimes wildly, and it stands to reason that the more they deviate from the innocently assumed correlations they were suppose to be working upon, the less useful the model's conclusions will be. that is, the model is only useful as long as it predicts things that there's not much point in trying to predict them after all.
*bear with me here
the same thing happens, more or less, when you are going to travel and make a list, drawing some projections about what places you expect to visit and how much each activity should cost you - only to actually get there and throw the list shamelessly out the window as the trip takes on a life of itself and you can barely remember when was the last time you took a shower because you’ve been drunk for the last 3 days even though you lost your wallet on the cable car ride to the itinerant circus.
honest middle-managers will recognize a similar problem as the one that arises when they receive some gibberish order from above and have to convince their reasonable employees that - based on all available evidence and operational reality notwithstanding - all is well, the the CEO has a plan, and that their actions will not in fact be contributing to their own unemployment in the very near future. i have the good fortune of being acquainted with one particular manager who, in her rise to the most rarefied heights of the career ladder, managed to observe her own transformation from someone who had to force herself to believe the rationale of decisions imposed from the top to someone who has to come up with rational excuses to pursue semi-arbitrary strategic paths to feed her peers at the lower levels of the chain. in her own words, her job these days is to produce believable narratives (or models!) to ensure compliance to certain top-level decisions.
the sheer amount of managers a typical company needs to be able to get anything done should serve as evidence enough that humans are not very good at coordinating their behaviour even in modest scales. also, and a bit oddly, there’s reason to believe that the more humans to be coordinated, the more likely coordination is to get fucked up.
this issue, as we can see, is routinely tackled through the crafty expedient of pretending that it isn't there. even cognitive psychologists concede that pigeons, when given food by a machine at random intervals, will start adopting certain repetitive behaviours in the apparent belief that they can make a difference in how the machine works - assigning, to random phenomena, whatever goes for causality in a pigeon’s mind. (wasn’t there a tv show that did a similar thing with humans? can’t quite remember its name)
reality, as we all must know by now, is an unwieldy bitch. a lot of the time it is one random pigeon-feeder. it's a system whose basic purpose seems to be to generate massive ammounts of cognitive dissonance in humans, and one of the top three causes of anxiety in modern societies.
even though we seem to have figured out the workings of the occasional natural process - at least when they stick to the 3 usual dimensions and fit the periodic table - we can’t quite claim to have become masters of the universe just yet. all social processes are inherently unstable - not the less so because a great deal of this uncertainty comes from everyone trying to come up with arbitrary solutions to unsolvable problems, often by imitating each other.
so why do economists and other misguided creatures of the dark bother with modelling at all? out of desperation? just because they can? because all the cool kids are doing it? in other terms, what is the purpose of this mass-scale process of collective rationalization (beyond, of course, instituting symbolic reality as such)?
save it for another post to argue that The Economy is the embodiment of the social-as-such, entitling it to the blatant honour of capital initials at this time of night. suffice it to say that it's slightly less predictable than the weather, and that we display similar reactions to both - usually in the form of bland statements on the line of 'it seems to be quite pleasant today mate' followed by a terrible rush for dear cover when a storm suddenly materializes.
(this is a tropical country analogy, for which i make no excuse)
when not relying on misguided optimism, in one case and the other one of two strategies is usually employed: asking your mom for advice, or declining to leave the house at all.
as the expression goes, however, no battle plan survives contact with the enemy. in short, the more uncertain the trip, the more we need to rely on half-baked guidelines about how things are supposed to turn out. and the economy is a pretty uncertain trip, as i’m sure you will all agree.
one could even argue - if one wasn't so afraid of everybody arguing back - that the sort of economic model here presented serves little more than the purpose of supplying the vast array of clueless agents with a foundation for their collective pretending that 'everything is alright after all'.* keynes said so in 1937, but it is a message very difficult to listen to. even the more so, perhaps, because hearkening to it would undermine the very self-delusion on which the working of the system relies.
*bill bryson mentions the same phenomenon taking place in academic physics, where there seems to be a very slight possibility that the universe won't either collapse upon itself or expand until it cools down entirely. physicists call this possibility the goldilocks hypothesis.
the economist’s or the investor’s model, in this sense, is little more than rationalization for his actions, aimed at smoothing over the cognitive dissonance caused by having to make decisions about sleep-disturbing things that are for the most part entirely out of his control. this is in itself nothing new, and while economists brawl bitterly about how to properly account for this uncertainty, they mostly acknowledge it in some (more or less hypocritical) form.
however!
in view of the blatant disregard for common sense displayed by economic thinking (and often by common sense itself), we could say that there’s another, less widely accepted dimension to economic uncertainty. people in the human sciences often defend the startling position that consciousness has got it all wrong, that the so-called material world is not all it’s made up to be, and that language creates reality in a very fundamental way - a way to which us, beings subjected to language and not at all cruising comfortably above it as unbiased observers, are necessarily a little blind.
this potentially disconcerting claim echoes in a way keynes’ little quip about financial instability and the silly ways people try to deal with it by all looking at each other blankly until they can all fool themselves into whatever arbitrary arrangement most conveniently masks its own lack of concreteness.
complexity and uncertainty are spiritual sisters, and the reason why humanity had to wait 10 thousand years to finally come up with a god of confusion is frankly beyond me. reality is much closer to what douglas adams graciously refers to as the Whole Sort of General Mish Mash that to whatever fits into an economist’s model: it’s a messy jumble of somethings, and if we scarcely know what is going on at the subatomic level, we know even less about about collective human behaviour.
for tricia mcmillan, each possible arrangement of possible ways of looking at the mish-mash constitutes a particular instance of reality in the multiverse. this is, incidentally, a good way to support the argument (privately held, but tough to sell) that language creates reality. in a milder version of the douglas adams proposal, language creates reality by cutting it around and colouring the selected bits with meaning. whatever is said, is imposed upon the mess, and it’s easier to believe in the existence of words than of the things they refer to.
to make matters a bit worse, but also a bit more understandable, there's what dawkings here refers to as the queerness of middle world, and of biologically evolved perception. it's more or less the same concern that led niels bohr to claim, according to legend, that the physicist is the atom's way of looking at itself (this was at the time rutherford claimed that all science is either physics or stamp collecting. good times, good times).
to drive the point home with overwhelming certainty, let us consider the matter of tying and untying knots. as any experienced sailor knows, the ancient art of untying knots is all a matter of directional poking... no human mind can fully comprehend the vast complexity of a sufficiently fancy knot. to begin with, a knot is tied based on little heuristics and rules of custom, simple instructions like ‘pull your right hand over your left like so’ and ‘maybe you should bring an umbrella with you today, son’. once a knot is tied, it drifts beyond the realm of human understanding and your only hope of untying it is ignoring it in its complexity and selecting little sides to push, and little connections to unwind, and hope that each partial doesn’t end up causing an increase on the level of complexity on the rest of the jumble. it’s all a matter, in short, of framing, screening, and cutting it around on the symbolic sense. it’s an essentially metalinguistic activity. not to waste the analogy, let us mention that social reality is more or less equivalent to what happens to your earphones when left unattended for quite surprisingly not that long.
even the most skeptical reader will find it hard to deny that the world tends to be a bit confusing, and that we have to occasionally select through a couple of its less disconcerting bits and force them sheepishly into some semblance of sense. this should be argument enough to force academic economists to add metalinguistic uncertainty to their neatly taxonomized lists of instances of uncertainty in the world out there.
so, next time you feel justified to get mad at an economist’s doomed predictions or, if you’re into that sort of thing, your financial advisor’s unwise investment tips, kindly forgive them and remember that they are just little misguided observers, subjected to mostly self-inflicted power structures, and unable to keep tabs on their own imaginary creations. much like the rest of us. they don’t know much more than you. in fact, after a few years of college there’s a good chance he’ll be less confident about his predictions than someone who didn’t study economics at all, and as such is still under the spell of the economic not-to-be-named name-of-the-father. that this doesn’t look too good on a resumée might be the strongest of the reasons-d’etre for the continued existence of the whole profession.
No comments:
Post a Comment