Current Events

Re: NY Fed's best forecast: "We can't forecast"

abskebabs — Wed, 30 Nov 2011 02:04:47 GMT

Where exactly does praxeology diverge from mainstream micro? In the case of price determination, we get demand schedules from maximising utility subject to a budget constraint, and supply schedules from maximising profit subject to a cost structure; then solve from the equilibrium condition. How does praxeology do it differently?

I'm so glad you asked! Buyers and sellers form valuations of the goods they would like to buy and sell w.r.t money, and given their expectations of future prices, these form the maximium buying prices and minimum selling prices they would be willing to buy or sell goods for. These form discrete demand and supply series, or as Rothbard called them, "schedules." Giiven that it is in the interests of all more competitive participants in a market to not needlessly outbid their competitors but to seek the best deal they can get, this explains why there is a tendency for them to prefer trading at the price between the bids of the marginal seller and marginal buyer (which is further narrowly determined if bids of the most competitve submarginal buyer and seller also lie within this range). It is not a theory of simultaneous determination; the price forms as a consequence of a competitive process in which submarginal buyers and sellers are gradually outbid/undersold.

For those buyers and sellers rendered submarginal, who however may have entered a market under the impression based on their expectations making insufficeint bids; this must lead to drastic revisions in their own plans of provision. But something very similarly, this must also have been the case for all supramarginal traders(all buyers with max. buying prices exceeding the clearing price), since if they knew in advance they could have bought at the clearing price, they could have allocated the savings made elsewhere, either to hold the money or to spend on other goods/services. Perfect anticipation would actually imply acute elasticity therefore for both buyers and sellers in the vicinity of the clearing price, before sharp inelasticity as soon as the price was reached. This also explains why price determination tends to be more rapid in markets populated by traders more accustomed to dealing with each other, and hence what their respective impressions of the going price may be.

Hence the Austrian theory of price does not require perfect anticipation of future prices or the demand/supply situation on a market to explain how these prices are formed as a result of a competitive process. This is most vividly the case for firms. They make factor bids based on expectations of future prices, not by solving an optiisation problem using "given" factor and product prices. Their plan revisions must come about not only as a consequences of failures of anticipations reflected in their profit-loss statements having conducted an entire production process, but also intermittently through the failure of buy/sell bids on each market.

Furthermore, a key principle underlying the Austrian price theory is diminishing marginal utillity (in purely ordinal form), explaining why max bids for further purchases of goods tend to drop for buyers (the opposite holds for sellers), and this informs the downward sloping nature of the demand series. This is not the case with neoclasical price theory. What is fundamental there is that indifference surfaces are convex and continuous (hence importantly implying a diminishing marginal rate of substitution), allowing for a unique tangency with a flat budget surface constraint. Positive, monotonic utillity and diminshing marginal utility are "nice" interpretive assumptions that play no fundamental role in this theory price determination (since the MRS will stay invariant given all functional transformations, not just monotonic, as a trivial consequence of he chain rule of calculus, inspite of what micro lecturers seem to inanely repeat). Consequently it loses much of its connection to economic interpretation. At best it's just a way to represent tradeoffs in an impossible situation of perfect anticipation, often assuming exchangers as infinitesimally minute price takers for convenience.

If you seriously want to learn how the Austrian theory of price works in greater detail than what I've very scantily said above, I recommend the following:

Man, Economy and State, Murray Rothbard, pp.103-161 (for an excellent, basic introduction)

Positive Theory of Capital, Eugen von Bohm Bawerk pp.193-234 (indispensable read), also pp.129-189 may help you begin to appreciate where we're coming from in terms of value theory (note of course BB was corrected later with regard to cardinal utility by Mises and Franz Cuhel).

I'm currently myself working on producing an essay that will compactly clarify and present in a slightly more "formal" manner this price theory, as well as discuss in greater detail its practical, analytical and philosophical advantages. Let me know if you'd like me to end it.

Re: NY Fed's best forecast: "We can't forecast"

ziragt — Tue, 29 Nov 2011 15:01:56 GMT

Aggretation is a necessary evil in that what you lose in detailed information gets made up for in managability. So far it's done remarkably well at explaining quantitative movements in variables ex post. Now what we need is to move away from calibration and make some ex ante predictions.

Being more tractable is not a good justification. As I said, there is no reason to believe that the economy acts like an individual at the aggregate level, whether this makes the math more tractable or not. We are making a very strong assumption, which may even be contradicted by our model (if the utility function does not have certain properties).

As for explanation, it's incredibly difficult to tell how well these simulations replicate the data. Yes, sometimes the model seems to follow the downturns and upturns in the data. Other times, it fails to do so. However, there are so many degrees of freedom and arbitrary calibrations in these models that fitting the data is not a very good way to judge the theory. Many RBC theorists seem to aknowledge this (in a paper on Argentina, Kydland even rejects the data, rather than his model).

Re: NY Fed's best forecast: "We can't forecast"

z1235 — Tue, 29 Nov 2011 15:00:56 GMT

Marginal Interest:
There was no risk in what you explained. You talked about bigger agents placing bets against you, which wasn't coherent. I interpreted it as "the government will keep printing money to prop up the housing market", and I addressed that. Was that an incorrect interpretation? If so, please explain it again.

I'll paraphrase one of your buddies: Markets can be (and usually are) "wrong" much longer than you can remain solvent. Here's an example of what happens to a hedge fund legend that tries to "fight the Fed":

...For those unfamiliar with Robertson, then here's what you need to know: He is the definition of a hedge fund legend. After attending the University of North Carolina, Robertson served as an officer in the US Navy and worked as a stockbroker for Kidder Peabody. He then founded and grew his (now defunct) hedge fund Tiger Management from $8 million at launch to over $22 billion in 1998 at its peak. And, as listed on our Tiger Cub biographies page, Tiger compounded a gross rate of 31.5% between 1980 and 2000. But, after losses of 4% in 1998 and 19% in 1999, Tiger shut down as the dot-com bubble expanded right in front of his eyes. He avoided what he deemed to be 'irrational investing.' The tech bubble would indeed be irrational investing, but his fund wouldn't be around to see it through...

Simply put, in a socialist (Fed controlled) economy there is no "right" and "wrong" - there is only betting with or against the Gosplan. Prices don't have to be going up forever for the correct predictors of a reversal (collapse) to get taken out to the cleaners before they are proven right. In trading/speculation, the timing, risk management, and staying power are everything. Having a printer of legal tender in one's basement allows one to literally paint price charts on the screens 99% of the time, which is more than enough to shake out every opposing bet (into insolvency) and force most participants to toe the line. Once the one-way avalanche has been started it takes a life of its own and they don't even have to push it anymore. And during the 1% of the time that markets get out of their control (such as in 2008) the risk of betting against them is that there would be no financial system and solvent counterparties left standing from which to collect your winnings, unless they crank up the printing presses and prove your bet wrong once again.

If you short a market at $100 thinking it should go to $20, and it proceeds to go to $200, then you would have lost all of your initial $100 principal, all your investors would have sued you (that is, the ones who wouldn't prefer you dead), your wife would have thrown you out on the street, and your bank account would have $0 in it -- even if the market fell to $20 after that, as you have correctly predicted. Do you understand this?

Who do you think is the largest buyer of US Treasuries today, and -- given what I just explained above -- which way would a smart market player bet?

Your questions truly baffle me. I feel bad but I will posit again -- as a kind favor to you because I like you, and not as an insult -- that you have been even more brainwashed at school than I previously thought.

It means the value of a portfolio sufficiently diversified to remove unsystematic risk will follow a random walk with a drift. That's pretty obvious, right. If there's some arbitrage opportunity, everybody will do it until it's not there anymore. So basically if I pick some stock index, like the S&P 500, my portfolio won't be able to consistently produce larger expected returns than a portfolio that's just composed of that index.

Have you stopped to think what the major component/contributor to this "drift" is? What is the main factor behind the econometric-derived "wisdoms" such as "60-40 stocks-bonds portfolio" and "buy and hold"? I could give you a hint, but it's better that you figure it out on your own. Let me ask you this: If all arbitrage is always and immediately eliminated, and there's no real $100 bills left on the pavement, why does an entrepreneur still get out of bed in the morning? Because he's stupid, and you're smart?

Re: NY Fed's best forecast: "We can't forecast"

Marginal Interest — Tue, 29 Nov 2011 09:09:37 GMT

z1235:

I just explained the risk. Which part did you not understand? Do you even know how market bets work and how markets operate?

There was no risk in what you explained. You talked about bigger agents placing bets against you, which wasn't coherent. I interpreted it as "the government will keep printing money to prop up the housing market", and I addressed that. Was that an incorrect interpretation? If so, please explain it again.

Here you go speaking in aggregates again. What is a "market trend" and how does one do "as well" as it's doing?

It means the value of a portfolio sufficiently diversified to remove unsystematic risk will follow a random walk with a drift. That's pretty obvious, right. If there's some arbitrage opportunity, everybody will do it until it's not there anymore. So basically if I pick some stock index, like the S&P 500, my portfolio won't be able to consistently produce larger expected returns than a portfolio that's just composed of that index.

Re: NY Fed's best forecast: "We can't forecast"

Marginal Interest — Tue, 29 Nov 2011 08:52:28 GMT

ziragt:

Well many of the assumptions are unfalsifiable, like that consumers maximise their expected utility. I'm not exactly sure how putting constraints on a model derived from that can be criticised when it's exactly praxeology... We find estimates of, say, labour supply elasticity, which if in every case of measurement happens to be negative would be good enough grounds to justify using upward sloping supply curves.

The microeconomic concepts macro models are built on are reasonably well tested, both econometrically and experimentally.

Taken together, these claims are false. An RBC model with rational expectations assumes that the economy acts as a maximizing agent and is in general equilibrium. Estimates of microeconomic magnitudes that seem plausible at the individual/industry level do not necessarily aggregate up. Therefore, If one believes that Walrasian models of general equilibrium tell us anything at all, then RBC models have no microeconomic justification except in highly implausible circumstances (such as homothetic and identical preferences and wealth among all consumers).

In other words, RBC models are merely based on faith that microeconomic processes will somehow "behave nicely" at the aggregate level, without any justification for this in the neoclassical system. At the very least, Austrian macroeconomics does not have this problem (in most variants).

You're not measuring a million individual elasticities, remember, you're measuring an aggregate. And the theory is trying to explain the movement of aggregate variables. Obviously it's absurd to try and predict the behaviour of millions of individual entities. Aggretation is a necessary evil in that what you lose in detailed information gets made up for in managability. So far it's done remarkably well at explaining quantitative movements in variables ex post. Now what we need is to move away from calibration and make some ex ante predictions.

I did see a couple of papers a while ago with RBC models with many representative agents with differing preferences. If I remember correctly, it added much more tedium to the model but not much more explanatory power.

Re: NY Fed's best forecast: "We can't forecast"

Marginal Interest — Tue, 29 Nov 2011 08:41:52 GMT

abskebabs,

Not sure what you mean in regard to "trades not at the equilibrium price". If trade is being conducted at a non-equilibrium price (in absence of frictions), then necessarily consumers aren't maximising their utility or producers aren't maximising profit....

I don't think the complaint that "price determination is implicitly assumed to explain it" is justified. The model you reference is something done in intermediate micro because it's easy. It's an easy way to introduce people to constrained optimisation, so it's not going to be extremely general. The price is not "given". For simplicity initial endowments are given. Price is determined endogenously, along with the quantity demanded, from the supply=demand equilibrium condition.

Where exactly does praxeology diverge from mainstream micro? In the case of price determination, we get demand schedules from maximising utility subject to a budget constraint, and supply schedules from maximising profit subject to a cost structure; then solve from the equilibrium condition. How does praxeology do it differently?

Re: NY Fed's best forecast: "We can't forecast"

z1235 — Tue, 29 Nov 2011 06:06:48 GMT

Marginal Interest:

z1235:
On the contrary, the risk of playing against an opponent who controls a printer of legal tender in his basement is extremely large. He can match whatever bet you make with an opposing bet of ten times your size. Plus, this system incentivizes his buddies (banks) -- who get first dibs to all newly created "capital" and control most other bets in this "free" market -- to bet in his direction. The risk/reward for this cabal is such that: (1) fail, everybody must bail us out lest ATM's stop giving out cash tomorrow morning, (2) win, well we win.

Are you sure you have the staying power (the capital needed to pay for your losses as prices are being pushed against you) to match theirs? Who do you think a smart trader would be betting with/against? Think again.

How does that risk exist? The only way it could possibly exist is if the ability to keep printing money would result in the real value of houses never coming down. That's absurd for two reasons: 1) It assumes a grotesque amount of money illusion; 2) if the prices never come down then it's not a bubble is it? So obviously you wouldn't short it.

I just explained the risk. Which part did you not understand? Do you even know how market bets work and how markets operate?

The best I can hope for, without constantly having my finger on the buy/sell buttons to react within seconds of new news becoming available, is to do only as well as the market trend.

Here you go speaking in aggregates again. What is a "market trend" and how does one do "as well" as it's doing?

Re: NY Fed's best forecast: "We can't forecast"

ziragt — Tue, 29 Nov 2011 00:38:44 GMT

Well many of the assumptions are unfalsifiable, like that consumers maximise their expected utility. I'm not exactly sure how putting constraints on a model derived from that can be criticised when it's exactly praxeology... We find estimates of, say, labour supply elasticity, which if in every case of measurement happens to be negative would be good enough grounds to justify using upward sloping supply curves.

The microeconomic concepts macro models are built on are reasonably well tested, both econometrically and experimentally.

Taken together, these claims are false. An RBC model with rational expectations assumes that the economy acts as a maximizing agent and is in general equilibrium. Estimates of microeconomic magnitudes that seem plausible at the individual/industry level do not necessarily aggregate up. Therefore, If one believes that Walrasian models of general equilibrium tell us anything at all, then RBC models have no microeconomic justification except in highly implausible circumstances (such as homothetic and identical preferences and wealth among all consumers).

In other words, RBC models are merely based on faith that microeconomic processes will somehow "behave nicely" at the aggregate level, without any justification for this in the neoclassical system. At the very least, Austrian macroeconomics does not have this problem (in most variants).

Re: NY Fed's best forecast: "We can't forecast"

onebornfree — Mon, 28 Nov 2011 23:44:51 GMT

Here is another good article I recently found on economic modeling [it is probably "old hat" to most regulars here] : "The Myth of the Model". by Max Borders at The Freeman.

Regards, onebornfree

Re: NY Fed's best forecast: "We can't forecast"

onebornfree — Mon, 28 Nov 2011 19:13:12 GMT

I don't know if this has been brought up before [probably has, somewhere !], but Doug French on Mises Daily recently posted a good article that goes to the heart of the debate in this thread, called" The Folly of Forecasting" . regards, onebornfree

Re: NY Fed's best forecast: "We can't forecast"

abskebabs — Mon, 28 Nov 2011 17:33:18 GMT

Example?

I can deduce the theory of price praxeologically (see Bohm Bawerk's theory of marginal pairs), through which we may explain the notions of "supply" and "demand", as consequences in explaining price determination. The complete theory allows for an can account for trades not at the equilibrium price. Application of the supply-demand "model" necessarily assumes all trade at the equilibroum point, unless dynamised using the Cobweb model. Something like the Cobweb Model makes very primitive assumptions about entrepreneurial expectations however, assuming that sellers simply react to yesterday's prices in their production for today in an automative manner, as pointed out by John Muth (to whom I think the credit should really go for Rational Expectations). Unfortunately, the assumption of rational expectations means that you have to make even stronger assumptions regarding people's niscience of the future to even explain something as simple as price formation and market clearing. These issues would all have been avoided if people within the profession had the abillity to think beyond the equilibrium Walrasian S-D model, or disequilibrium bifurcations of it, in which you also implicitly assume price determination to explain it given that demands are produced as solutions to a Lagrangian optimisation problem with price "already given" to each consumer as a parameter (for instance when put in the typical, Perfect Competition form).

Meanwhile, you can actually explain all instances of price formation with Bohm Bawerk's apparatus(as well as with Wicksteed and Fetter's variations along the same lines), as well as including the concept of expectations since all bids are based on them, but you don't need them to be anywhere near "exact" to explain price determination. Neither do you run into the nonsensical issue produced by the Cobweb Model, in which you don't have equilibrium convergence if demand and supply price elasticities around the equilibrium point are beyond a certain kind. These Walrasian price theories you can derive using reasoning that somehwat resembles praxeology, but my point is the errors produced are produced by the incorporation of bad assumptions and hence bad deductions based on them. They necessarily obscure the true explanation of price determination, inspite of the fact that one could gather data and pretend try to "test" and distinguish them.

With regard to income and substitution effects and attempts to distinguish them, I think I'll pass comment, aside of the and demand constancies often assumed in such an analysis. There are other issues too, like the fact that from a set of observations you might reckon e.g. demand is upward sloping, when what you're actually doing is finding different demands on a relatively stable "supply curve." Given that all you know are just the hypothesised intersections of 2 curves, it still seems to me that this kind of thing could never be determined by the data, rather, a prior udnerstanding is used to inform interpretation of the data always.

Finally, your final request, I think if carried out would probably lead to the complete (and deserved) discrediting of the vast majority of the economics profession. That's also why I don't think it'll happen.

Re: NY Fed's best forecast: "We can't forecast"

Marginal Interest — Mon, 28 Nov 2011 16:48:56 GMT

abskebabs:

There isn't a problem with using insights that are ultimately praxeological (although of an often bastardised form, using untenable assumptions where they cannot apply)

Example?

, but in thinking that you are letting the data speak for itself, distinguish between the theories you're utillising(which ultimately it can't) and impute forward relationships based on it (though as you and I have already noted, nobody in their heart of hearts really seems to believe this anymore).

Don't understand what you're saying here.

Perhaps I'll explain my view on econometrics: Obviously it can't give you truth. You can't figure everything out just by looking at some data in relation to other data. It's there to advance theory which has reached it's "praxeological limits". Logical deduction will only get you up to the point where you identify (going back to the same example) that there's an income effect and a substitution effect and consequences depending on which dominates. You need econometrics at that point to be able to say "this effect dominates" or "this effect is insignificant" and actually proceed with your logical deductions, conditional on that observation, so that your model can have more explanatory power. The other use, of course, is in falsifying unique predictions models make. Though (and this is my biggest complaint in regard to economics) there's not nearly enough drive to produce these predictions.

Re: NY Fed's best forecast: "We can't forecast"

abskebabs — Mon, 28 Nov 2011 16:33:25 GMT

Marginal Interest:
Well the problems of parameters being non-constant just due to the natural variation caused by "human complexity" is reflected in the fact that the estimators are given with a confidence interval whose width is proportional to the variance of the error term.

I wasn't referring to anything as trivial as that. The variance of error terms used to inform confidence intervals, is again an instance of assuming constancy of relationships in projecting forward past relationships to future data. Shocks inherently can never effect the conditional distributions in such forecasting models, as that very fact is their operative assumption.

There isn't a problem with using insights that are ultimately praxeological (although of an often bastardised form, using untenable assumptions where they cannot apply), but in thinking that you are letting the data speak for itself, distinguish between the theories you're utillising(which ultimately it can't) and impute forward relationships based on it (though as you and I have already noted, nobody in their heart of hearts really seems to believe this anymore).

Re: NY Fed's best forecast: "We can't forecast"

Marginal Interest — Mon, 28 Nov 2011 16:11:01 GMT

abskebabs:

I don't think what I said above has much specifically to do with the Lucas critique.

Well the problems of parameters being non-constant just due to the natural variation caused by "human complexity" is reflected in the fact that the estimators are given with a confidence interval whose width is proportional to the variance of the error term. So parameters which would change erratically in large magnitude would produce unusably large confidence intervals. So I think that problem isn't entirely valid. However the Lucas Critique remains a big issue with forecasting, so I f igured I'd mention it.

You claim these microeconomic theories are well tested, are you sure they're not just assuming what they are trying to prove? In order to "impute" utillity functions to consumers based on past purchase data you have to actually assume that their preferences were "constant." This is untestable, howver reasonable it might seem in a given scenario intuitively(which is usually how it is justified when push comes to shove, e.g. that desire for milk doesn't change each month).

Alternatively, similar problems are created if you try to find structural parameters from a simultaneous equations econometric model, e.g. of supply and demand. You have to input restrictions informed from economic theory in order to do this, but there is no way to actually test these restrictions themselves, leaving the frequent occurence that antagonistic and contradictory theories would simultaneously provide a good fit to the data (I guess I don't need to say who pointed this out again...). (In fact, I'm quite glad my econometrics lecturer was candid enough to admit this too).

However reasonable these assumptions of economic theory might be by "intuition" ( a word as definite as praxeology would probably make neoclassical economists want to run for the hills), they can't be tested or imputed from the data.

Well many of the assumptions are unfalsifiable, like that consumers maximise their expected utility. I'm not exactly sure how putting constraints on a model derived from that can be criticised when it's exactly praxeology... We find estimates of, say, labour supply elasticity, which if in every case of measurement happens to be negative would be good enough grounds to justify using upward sloping supply curves. Yes, the models used to form the estimates impose some theory on the data, however that's one of the great things about peer review. There are plenty of people out there who will check the sensitivity of your results to the model specification. Don't get me wrong, I by no means think econometrics is perfect. It faces amazingly difficult problems, especially with forecasting. But I don't think it's entirely useless.

Re: NY Fed's best forecast: "We can't forecast"

abskebabs — Mon, 28 Nov 2011 15:23:01 GMT

I don't think what I said above has much specifically to do with the Lucas critique. You claim these microeconomic theories are well tested, are you sure they're not just assuming what they are trying to prove? In order to "impute" utillity functions to consumers based on past purchase data you have to actually assume that their preferences were "constant." This is untestable, howver reasonable it might seem in a given scenario intuitively(which is usually how it is justified when push comes to shove, e.g. that desire for milk doesn't change each month).

Alternatively, similar problems are created if you try to find structural parameters from a simultaneous equations econometric model, e.g. of supply and demand. You have to input restrictions informed from economic theory in order to do this, but there is no way to actually test these restrictions themselves, leaving the frequent occurence that antagonistic and contradictory theories would simultaneously provide a good fit to the data (I guess I don't need to say who pointed this out again...). (In fact, I'm quite glad my econometrics lecturer was candid enough to admit this too).

However reasonable these assumptions of economic theory might be by "intuition" ( a word as definite as praxeology would probably make neoclassical economists want to run for the hills), they can't be tested or imputed from the data.