/facepalm
you obviously have very little understanding of the foundations of science in general.
Measurement of what? Measurement requires a rule against which to measure. If we say a wire is 4ft. long, that means you could lay four 1ft. rulers end-to-end beside the wire and the ends of the wire and the joined rulers would be congruent. So, measurement is comparison of a specimen to a rule. What is the rule? What would a employment market unhampered by “gender bias” look like? 50% men, 50% women? That wouldn’t make sense unless women gave up reproduction, an act which is considerably more taxing for the female than it is for the male. Without a rule to compare against, there is no such thing as measurement, just collections of specimens.
Clayton -
So you go from trying to determine whether the wage gap even exists to trying to figure out its causes. One step at a time.
Yeah, so not all time series are non-stationary. Plus that problem doesn’t arise in cross-sectional data. And, as it says there in the first couple of paragraphs of the wiki article, sometimes we can transform non-stationary data into stationary data (similar to the way we can transform homoscedastic or autocorrelated data into satisfying the assumptions we need for BLUE estimators).
Well if you’ve specified your model correctly such that the error isn’t correlated with the regressors, then the relationship is either causal or reverse-causal (but no confounding variables). And, where possible, you make the variation in the regressor you’re interested in exogenous. Normally you can do that by controlling the experiment. With econ you mostly have to find natural experiments. And that gives you causality (at the blah% significance level).
That article just has a problem with probability as a use of subjective feeling. That’s not the same as saying “the true mean will fall into 95% of the intervals we construct this way” or “given that the distribution is normal (from the central limit theorem), there is a 2.5% chance that the variable will be greater than 1.96”. But yeah, there are limits to econometrics. Making accurate predictions is more a limit of the economic model you’re using, but I get your point. But people shouldn’t go around discarding the whole field as useless. It’s a very powerful tool for testing your models already, but it also has frontiers which need to be pushed.
Thank you so much for all of the insight in this thread. Heres a recent response from my debating opponent. Id appreciate any insight, and would be curious to find out how you guys would respond to it.
"I’m not sure i understand your point. Do you reject that observational data and statistical analysis can be used to make inferences about our world? If that’s the case, our dispute is a mathematical one.
Just because human behavior changes does not mean the science is any less empirical. Thats because we are using data to draw conclusions about what those changes were, and what may have caused those changes. Lastly, statistical and predictive models, are based on probabilities, so deviations can occur, but the greater the distance from the predicted mean, the lower the probability of occurrence. I’m not sure if that explains in part your disagreement. Quite frankly its 4am and I’m exhausted too. but yah lets pick this up later. "
Clayton-
It’s not really clear what your argument is. You seem to be saying that in order for statistics of any kind to make sense, we would need to have a complete counterfactual. If this were the case, as I stated above, then we can know very little about the social world.
In that case, will Austrians stop making proclamations about how important ABCT is? After all, it’s merely a tendency among many others. Its effects on such a complex, nonstationary process such as the economy can’t be certain.
/facepalm to you to. Why don’t you enlighten me with your immense knowlege of the foundations of science. Or maybe you’d just like to insult my intelligence some more? I don’t really mind either way.
Methodology is constantly talked about and critiqued in the mainstream.
It’s been a while since I’ve done any labour economics, but I’m pretty sure you can just measure it the same way you measure monopoly power. Lerner measure, Herfindahl-Hirschman index. For the deviation from the natural rate of interest, you use the rate at which underlying inflation is accelerating. Counter-factual prices aren’t necessary.
I’m not sure whether those examples of indices help your case for empirical work. After all, they impose quite a bit of theory on the data. Most Austrians would not agree that firm size leads to monopsony power, although I don’t know what the relevent literature says.
Well the Lerner measure is markup over marginal cost (since marginal cost is hard to measure often average cost gets used). And that’s a straight up measure of market power. I’m guessing for labour monopsony you’d use the wage and average product instead of price and average cost.
Mises is not just disputing the use of probability to describe a subjective feeling… he’s saying that there is no sense in which you can make probabilistic statements about something like the likelihood of FDR winning an election in 1944 because it’s a completely unique event. Class probability only makes sense when you can speak of the ratio of the event of interest to the class of all events under consideration.
The application of probability theory and the methods of statistics to historical data is of no use in predicting the future of human action. This sort of speculation is the role of the entrepreneur, not the economist. It is the entrepreneur who looks at the available data (prices, inventories, production facilities and other industrial information) and speculates about the future and then acts on these speculations. We have no idea whether the entrepreneur has correctly or incorrectly speculated… only time will tell.
The economist, on the other hand, is studying the manifestations of voluntary exchange in human action. Why do people engage in industry? What is production and consumption and their relation? What are prices and how do they arise? Could we all run out of work to do and starve? And so on. The answers to these questions are predictions only in the sense that “the sun will rise in the morning” is a prediction.
The proper use of probability theory and statistical methods is in regard to economic history. Why have things turned out as they have? Why did people used to do X, Y and Z? This is the domain of economic history (which I believe Mises termed “thymology” but I’m not 100% sure on his terminology).
Clayton -
I’m not really sure how we got on to probability here. I disagree with mises, but that’s kind of irrelevant. I haven’t been defending the use of probability in assessing the likelihood of FDR winning the election, I’ve been defending econometrics as a means of falsifying models.
Please take a look at Salerno’s article. He explains the whole issue much better than I could.
Clayton -
All I got from that article was the attitude “I’m just going to reduce everything to unfalsifiable premises which I made up”. This is my fundamental problem: a priori reasoning can only give you knowledge of an abstract world. In the world of, say, mathematics, I can just make up my own rules. I can construct a set with some elements I say are in it, I can define an operation on those elements, define some assumptions about the reflexivity and transitivity of equality. I can make up whatever rules I want. The great thing being, if I start with true things, and I only filter those through other true things, I’ll end up with something that’s objectively true. That’s pretty cool. Unfortunately, I just made up the rules. If I want something that’s going to accurately model reality, I can’t just make up my own rules. I have to observe reality. I have to figure out what reality’s rules are, then I can apply logic to it. That’s empiricism. You need to observe reality in order to make statements about reality. So for your model to be in any way relevant to this world, it needs to be compared with reality. Stastistics is just the tool we use to make these comparisons, because it rigorously gives us a way to filter out all the shit that just happens due to random variation, makes sure we don’t incorporate biases due to omitted variables or confounding variables, and all that jazz. This is fundamentally what econometrics is. The stuff where it tries to overreach its bounds, like saying “if we lower taxes by 10% growth will increase by 4%”, is not representative of econometrics as a whole. Yet it seems people in the Austrian school have latched on to that stuff and are criticising it based on something even mainstream economists know to take with a grain of salt.
Fortunately, some Austrians do use statistical analysis, such as Peter Klein and Nicolai Foss. Their work on organizations represents an area that greatly benefits from both theory and econometric analysis.
Again, many areas of interest in the social world require empirical analysis. Do more police lead to more or less crime? Do cops racially dicriminate at traffic stops? Does the stock price at an IPO depend on the experience of the underwriter? Do 3 strikes laws reduce crime? Did Mexico’s failure to reform its banking system in the 1980s lead to stagnation in growth, or was it something else? Was the 1990s R&D caused by an increase in the supply of external finance, or an increased demand for finance? (All of these are addressed in empirical papers)
We can say call those who attempt to answer these questions are historians, but that does not change the fact that econometrics can help us to understand these phenomena and the social world. A priori theory, whatever its merits, is of little help.
I think you’ve read it too hastily, then.
I have some reservations with some of the a priori methods of some Austrians (particularly the misapplication of “impossibility of the contrary” arguments). However, it is true that we have a priori knowledge about what it is to be human (to have appetites, to seek to satisfy them, and so on). We can formulate this knowledge into axiomatic propositions about which there can be no more doubt than about the axioms of any natural science. We can use these axioms as the starting point for deductive arguments which, like the deductive arguments of geometry or calculus, give rise to useful, non-obvious theorems which, of course, were already contained within the axioms.
That’s a revisionist description of empiricism. In the extreme, empiricism entails the denial of classes, that is, that knowledge is comprised solely of particular facts or tautologies and that any concept of classes emerging from the particulars is meaningless babble symptomatic of the feverish imagination of a slightly-evolved ape’s brain. Empiricism is positivism applied to natural philosophy (science), not just the idea that you have to do experiments and perform observations to get knowledge about the physical world.
The assumptions of praxeology are not just “made up.” They are, in fact, true. People do in fact prefer more goods to less. People do in fact prefer goods sooner rather than later. People do in fact seek to alleviate felt uneasiness. People do in fact utilize the means available to them to attain their most urgent ends. I know all these things because I am a typical specimen of homo sapiens, because I experience these aspects of myself and because I can easily see the causal role that each of these things plays in the fact that I am alive and experiencing the world to begin with; just like all the other homo sapiens on Earth.
To question my knowledge in this regard is not rigorous, it’s just unbridled skepticism for skepticism’s sake. I can also question whether the world is “really real” but pursuing such questions is fruitless. I am no less certain that people prefer more goods to less than I am that the sun rises and sets every day. It is a rudimentary part of my experience of the world.
There’s a more fundamental disagreement at stake over the very nature of knowledge. The statistical approach to economics is motivated by an errant epistemology that is nihilistic about the capacity of science to reason about the subjective motives of human beings. The practice of science is objective but science may study the subjective and may make assumptions about the subjective in order to study human behavior action. It’s a stunning contradiction that the very same epistemological tradition which seeks to objectivize the subjective in neuroscience (via the so-called “neural correlates of consciousness”) is repulsed at the thought of assuming certain things about the subjective in order to draw conclusions about human nature and human behavior.
Clayton -
Praxeology creates truth by manipulating definitions and adding unstated premises. People prefer more goods to less because we have defined economic goods as scarce things we want (to achieve our goals). People prefer goods sooner because we have ruled out counter-examples, such as ice in summer versus ice in winter, as being different goods. Implicit assumptions are then smuggled in to make conclusions: Labor supply is never backward bending. Giffen goods cannot occur. Learning occurs at a rapid rate. Re-switching of techniques never happens.
Let’s talk about the Central Limit Theorem. According to Wikipedia:
Some questions: How large a number is “sufficiently large”? How does one know whether the mean and/or the variance is finite? How does one know whether a given (or presumed?) random variable is independent? How does one even know how many random variables are at work?
Wikipedia goes on to state:
More questions: Which variant is correct? Does it depend on the context? If so, how?
Finally there’s this:
So again, how does one know whether the random variables under consideration are necessarily either 1) independent and identically distributed or 2) have specific types of dependence?
My understanding of praxeology - and, by extension, the Austrian School of Economics - is that it assumes at least two fundamental assumptions about people, namely 1) that they have some capability of action (i.e. conscious behavior) and 2) they are always self-interested. Combine those together and you get another proposition: to the extent that they can act, all people always act in their self-interest. To me, these propositions seem self-evident. The question then is what else can be derived from these propositions.
This is the heart of the issue and this is why the mainstream prefers to choose methods which are agnostic regarding the motivations of individuals, particularly of those associated with the State and those in its orbit. If your methodology is democratic vis-a-vis the purity and nobility of human motives and intentions, then you cannot escape the conclusion that Senators, Supreme Court judges, police commissioners, priests and so on are motivated by roughly the same kinds of things that motivate speculators, day-traders, hedge-fund managers, CEOs, used-car salesmen and even strip-club owners. But we’re not supposed to think about such “abstract” and “sociological” issues because we’re scientists and scientists think about real, hard problems like indifference curves.
The prevailing paradigm of human action (among the masses) is essentially DC Comics - good guys do heroic things to stop the bad guys who are constantly plotting to take over the world (for unspecified reasons) and either destroy all humanity (again, for unspecified reasons) or even the very Universe itself. The heroes always act out of sheer altruism and the villains always act out of a kind of altruistic malice - they are not seeking to live off the productive efforts of humanity - after all, they are more like spirit beings than physical beings, who do not even require sustenance.
This is not very far from the propagandized picture of the world promoted by the Elites. Police, firemen and soldiers act out of purely altruistic motives to deliver us from the plots of evil terrorists who simply hate us for what we are. The terrorists do not want to subjugate us so they can live at our expense… they are ascetics who have no use for material wealth. It is childish, magical thinking. The long-faced PhDs who stand behind mainstream theory are no greater assurance in its seriousness than the learned men of three or four centuries ago who gave expert testimony to the Holy Office during its Inquisition.
Clayton -
I’m beginning to think that praxeology is just a systematic way of pointing out the obvious, and it was developed precisely because in economics people have such a hard time keeping a grasp on the obvious.