From the discussion in the article (especially the coin flipping example) it just sounds like distinguishing expected utility from expected value. This is a concept covered in beginner or intermediate undergraduate economics. I’m sure there’s more to it than that, maybe someone who knows can elaborate here?
I know some ergodic theory, but am not an economist. I cannot verify the veracity of the argument, but the argument itself is new, and not reducible to distinguishing between expected utility and expected value.
My comment is based on a Nature article by the author [1]. This is just based on a first reading, so my understanding may not be spot on.
The observation here is that the utility function in classical economics is not time-dependent. Time dependence is added-on by effects such as discounting. Discounting function itself may be personal - some investors are in a hurry, some are patient, this is usually taken into account.
In contrast, the authors introduce stochastic noise into the __rate of growth of the loss function___ (not the expected wealth). [Equation 7 in the Nature paper]. This makes the evolution of the growth rate function into a Brownian motion. The authors claim that this models individual wealth trajectories better.
This is as far as I have got. The underlying dynamical equations for the growth function are different, and the authors' claim is that their model predicts individual trajectories better than classical economics, and does not have to use psychology like behavioral economics.
“ But Ole Peters is no ordinary crank. A [non-economist] by training, his theory draws on research done in close collaboration with the late Nobel laureate Murray Gell-Mann, [who is also not an economist].”
Honestly this just reads like complaining about physics because your intro class didn’t take wind resistance or friction into account:
The title shows a huge lack of understanding of how human knowledge works. Never was "everything wrong" with contemporary theories on any field. If existing (applied) theories were 100% wrong then people wouldn't use them at all. New theories (even groundbreaking ones) usually build upon previous ones and/or become a more complete version of them, a superset. Thinking one can erase 300 years of thought on a field is veeery small minded.
“For most people, the series of bets is a disaster,” Collins wrote. “It looks good only on average, propped up by the extreme good luck” of a just a handful of players.
> I call it organized religion with equations, superstition. The only way to become free of superstition is through overcoming. But you need to study. I’ve always pissed off my academic colleagues and other economists who actually believe that is real science what they are doing.
> Our mathematical models of the weather can be judged by objective reality. If I am a meteorologist and come up with a prediction that tomorrow there is going to be a heatwave in Leicester square, all we have to do is to wait until tomorrow to see if I’m right or wrong. The weather will either confirm or junk my theories about it.
> Let’s say that I have the same kind of computer model and actual machine and data mining that the meteorologist does, but instead of using it to predict the weather I use it to predict the stock exchange. And suppose that I was somebody very highly respected as a predictor of stock exchange changes and let’s say that today, I were to predict that tomorrow is going to be a major crash in the stock exchange. There might be because I predicted it! In society and in the economy, our beliefs about the phenomenon under study are part of the phenomenon under study.
> ~ Yanis Varoufakis
Anyone who says otherwise is trying to convince you why the poor should exist.
It doesn't mean economics can't be scientific, only that, e.g., the act of measuring can't be separated from the measurand. Sound familiar?
Of course, the solution is pretty trivial now: commit the predictions to a blockchain or third party under seal, encrypted, whatever, and reveal it later. Or start applying quantum mechanical descriptions to economics.
This paper and discussion are sort of interesting because there's been analogous debates about ergodicity assumptions in psychology modeling growing over the last couple of decades. They started in the 90s and have become prominent topics of discussion in the last decade or so.
Almost no one really believes ergodicity assumptions truly hold anymore in those discussions. The catch is that neither necessarily does stationarity. That is, in the Nature Physics paper cited, something like the left side of Equation 1, although defineable, is not really estimable (because any function of time will change over time), so the right side, although not strictly equal to the left, becomes just as good of an estimate in many cases (because a system at one point in time practically speaking becomes a different system at a different point in time). The problem isn't in the expectations, it's in the uncertainties around those expectations -- or at least, in understanding when the uncertainty about the long-term time average of f is great enough that using the [cross-sectional] expectation value of f is better in some sense. Put another way, are there times when it makes sense to predict your future self better by looking to other persons, or are you always better off treating your situation as unique?
The discussions in the paper are generally about different types of issues, but they dovetail and coincide in certain ways. I'd have to reread the paper cited in the piece to get a better grasp on how. I'm not disagreeing with it; I think it's interesting, but wonder if it's falling into the trap itself is criticizing others for at some level.
Why not? Human trade is a natural phenomenon, and economics is the study of its structure and behavior. I don't see why that couldn't be a science.
Now if you mean to say that we're far from understanding it properly, or that there's little it has yet allowed us to apply for practical needs, maybe. But I'm not following the "it's not a science" bit.
Edit: Oh, or you are criticizing that the current practice isn't properly using the scientific method for it.
That the poor exist is not surprising at all. For almost all of the 200k years of human history, everyone was poor. It is the default human condition.
Only in the modern era have many people managed to escape poverty. The interesting empirical question is how and why the creation of wealth is possible, and how to foster greater prosperity.
That's an area of politics. You can create poverty by increasing living standards faster than people can increase their income. You can create poverty by allowing people to increase their income beyond someone else's income because poverty can always be relative. You can increase poverty by making it harder to get a job in the first place. There are so many reasons that have nothing with economics but politics.
Whether the coin flip game increases poverty or not is also purely about politics. If the one lucky guy cooperates with all the other players then everyone is better off. That cooperation can be implemented as a government policy.
After you say "That's an area of politics" you give two economic theories that reason poverty's production.
Is poverty an indicator of a better economy? Do the poor need to exist in better economies? What makes an economy better in the first place?
These are all questions studied by economists and all influence this "should poor exist" question. There is no reason to qualify economic questions by saying they're political; they just coincide and that is all.
>the branch of science devoted to the study of societies and the relationships among individuals within those societies. (wikipedia)
Which coming from a physics background I think of as not a proper science but lots of people study that stuff and you can get some results out of it. I think people get in trouble with economics if they start thinking it's an exact science like physics rather than a guess at what a bunch of people will do.
Suppose I come to you with a fair coin and say that I’ll give you $50 if it comes up heads and take $40 if it comes up tails. Good deal, right? Now I change the deal to 40% of your money with 5:4 odds and you can play as much as you like. You have $100 so it’s the same deal. There’s never a point where this looks like a bad deal, but you will eventually lose everything if you keep playing. On the other hand you will do well if you don’t adjust the amounts. There is a way to adjust the amounts safely, depending on a set acceptable probability of ruin (pr). Consider betting cash/log2(1/pr) at every turn. That reduces the gulf between the expectation of the log and the log of the expectation, but the probability of ruin still increases toward unity. As the timeline increases to infinity, the safe bet drops to zero.
As in almost all fields, the existing leaders will need to die off before correct theories have a chance to take hold.
In geology, it took many decades for absolutely compelling evidence for plate tectonics, for the catastrophic flood that shaped eastern Washington state, and for the meteorite that eliminated the non-avian dinosaurs (and every land animal bigger than a cat) to be taken seriously. (Geologists typically still insist it must have been a series of smaller floods.)
Historians have not come to grips with the very oldest Egyptian artifacts being, by far, the most advanced -- in some cases, well beyond what we could reproduce today without first inventing new methods.
Publications and trials in Alzheimers research still overwhelmingly favor the amyloids hypothesis, despite its utter, abject failure in every trial.
People still imagine that excess saturated fat causes heart disease, so that work to understand how eating meat really causes it gets minimal attention.
In psychology, for decades every publication was obliged to cast all results in terms of behaviorism.
In physiology, it is still required to evaluate every electromagnetic effect on biological activity purely in terms of heating.
Endless examples are available, in any field, not excluding physics.
Promoters of science like to congratulate themselves over known cases where the facts did, finally, win out, but are happy to neglect the necessarily more numerous cases where advocates were driven from the field, and never got vindication, so falsehoods are still taught.
This phenomenon is in large degree a product of the professionalization of science, and the system of journal reviews and, moreso, grants, controlled by the old guard in each field.
It is easy find Wikipedia articles being defended against new evidence by emeritus professors with plenty of time on their hands.
>Starting with $100, your bankroll increases 50% every time you flip heads. But if the coin lands on tails, you lose 40% of your total. Since you’re just as likely to flip heads as tails, it would appear that you should, on average, come out ahead if you played enough times because your potential payoff each time is greater than your potential loss. In economics jargon, the expected utility is positive, so one might assume that taking the bet is a no-brainer.
>Yet in real life, people routinely decline the bet. Paradoxes like these are often used to highlight irrationality or human bias in decision making. But to Peters, it’s simply because people understand it’s a bad deal.
In real life we would create an investment fund for this. We gather the money of many people into the fund and the fund managers would then play the game more than once. If you play it 4 times with two coin flips each the expected ROI is around 7%. However, there is something weird about this game. The expected ROI grows the more money you have. Playing the game 16 times with 4 coin flips would get you an even greater ROI. I don't think such an investment exists in reality because we would run into diminishing returns. Spending too much money will always decrease the ROI. People need a cheap car to drive to work and the car increases their earnings potential significantly but spending twice as much to buy a Tesla doesn't increase their salary beyond what the cheap car got them.
Perhaps I’m misunderstanding, but the outlier problem is simply a case of using an inappropriate type of average. If you take the median, you’ll get a far more reasonable answer
the median says not in 1 game, but let's say each game you start again with £100, and you go ten times, if you play that game over and over, you'll win overall
Bad reporting, just go read the paper directly. They busted the coin example.
I was super confused by it because it suffers from classic AM-GM inequality mistake. I doubted that Peters would make such a trivial mistake (he didn't). He talks about the utility of a single outcome of the game, not the expected value of 10 plays of 10 different games.
In the article: Using an arithmetic mean on percent-return doesn't make any sense. If you calculate the expected value and mistakenly use arithmetic mean you'll get +5%, but it's actually roughly -5% when calculated correctly with a geometric mean.
Which is why all their silly simulations come out negative.. none of which has anything to do with Peter's actual work.
reply