I think it's better to think of it no as a complete rejection of reality, but recognizing the limitations of empirical methods. Mainstream economics today is arguably too focused on indicators, just collecting a bunch of data and finding correlations, then jumping to causation from that.
It is basically impossible to produce real empirical evidence in economics. You can look at data/correlations, and try to reason about cause and effect. But real empirical evidence requires you use the scientific method to test hypotheses (ideally done in a double-blind way), which is impossible for economists to do even if we had a national desire to do it, because the test itself would impact reality for its participants too much. It's similar to the "observer effect" in physics, but on a massive scale.
At best, I think you can create mathematical simulations built on certain assumptions about reality. Those simulations depend on how well the assumptions model reality, and very easy to miss an important detail.
Agreed. I'm all for experimental economics, but let's not exaggerate the generalizability of a result. Better to see it as a particular form of theory that needs to be backed up empirically in more real world contexts.
Right. In terms current thinking I can't speak for an entire field, and I should say from what I have seen economics maybe still follows the lead of hard sciences where coming up with a theoretical model and a beautiful equation on paper which is then strongly supported by empirical research is still sort of a holy grail.
But today, the availability of empirical real world data or a way to go get it if not yet collected is night and day different compared to 60 years ago when many of the old economic theories were being thought up and became popular. Go read the old papers there is no data just theory. Today, economic theories and formulas can often very quickly be tested with empirical research and data. And in my opinion current economics has turned toward empirical work having weight over theoretical work
How do you disaggregate? You do the best you can. And yes we do try to look at each part individually (not in a vacuum though). Because it turns out aggregating is worse. When you look at TFP the way you describe, you are not just aggregating technologies you are aggregating everything...aggregating all technologies and also whatever else is happening that K and L didn't capture. Who needs that, it doesn't tell you anything practical but that "other stuff is happening". Technology might turn out to be only <5% of TFP. There's enough evidence that TFP has too many "factors" in it to be lumped as just technology. Physicists didn't give up when they realized 95% of the universe is dark matter/dark energy that they don't understand, economics is following that lead, double down on the small stuff and hope to understand the big stuff a bit better and the best you can.
I looked at your bio, not sure if you use TFP in your work or its just a personal interest. I know a bit about about market and equity research. If you have access to the data I would use PCA over TFP, if you don't have this I would just proxy a series and risk off systematic variables and focus on key known numbers. Better to ignore fake data than listen to hurtful real data. Either way we are all always guessing.
I understand the argument and I accept the idea of natural experiments in principle. The most difficult thing with all of these is the ceteris paribus (holding all else constant), or as a second best, correcting for known confounding factors. This is in such stark contrast to a lab setting, where an adversarial scientist can regard a hypothesis as wrong and construct an experiment to prove it so, only to find out that the hypothesis isn't falsified. With statistics and models on the other hand it always devolves into discussions about variables, confounding factors and there's never a satisfying resolution, especially on controversial results.
The number of places for bias to creep in are way to many. I think the econometrics people have gone down a hopeless path, guided by wishful thinking and demand from people who want us to be able to make statements about things which we just cannot do.
The parts of economics that are valuable are not based on econometrics, but upon reasoning from first principles: Incentives, supply and demand, game theory etc. under which I would argue all of your examples fall.
Ok, I’m sympathetic to that POV. I’m grateful for the credibility revolution in econometrics, but it clearly can become an obsession. Prediction on its own can be worthwhile. I wouldn’t say causation is a myth - maybe rather that it’s overvalued.
There is no need to speak in general terms, there are very specific things that make economics a bogus science.
The biggest issue is that modern mainstream economics is modelling the world as Gaussian. There is no room for 'fat tails' in econometrics, for example, every time you see 'regression' mentioned somewhere -- this should raise alarm. The world is not-linear.
Same applies to other concepts mentioned above. This one issue of crude inefficient approximation (stochastic world rather than the world as a complex nonlinear system) is enough to undermine most of modern economics. This is because a lot of this so called 'science' consists of many very elaborate and intellectually beautiful rigorous constructs, which are unfortunately built on an obsolete base.
Linear models had their place, maybe, 20 year ago, because there was nothing better. But now we have a very solid body of knowledge on which to build further scientific advances, most of it achieved through experiments (with computers), and not though some cute 'mental experiments'.
So, let's not be too sensitive, there is science that is supported by experiments, and there is a kind of abstract sophism which is modern economics. The sooner the public sees that this emperor has no clothes, the better.
You can say that objective reality is dubious. But if you use that to subvert claims that survive testing, you're making a positive statement and you need a testable alternative. In any other case, what you're doing is using philosophy to push ideology by discrediting established facts.
I think we should separate "empiricism here is currently difficult" to "absolutely impossible"
Some fields of social science are undergoing internal turmoil (eg. Macroeconomics and some subfields of psychology). This is a good sign; old models are being discarded as new evidence and better methods pop up. In fact, I'm more concerned about sociology, where there wasn't a replication crisis.
Facts are being improved upon when you see crisis in replication. That's a sign of a healthy field of study.
As an econometrics/labor econ grad student, I can tell you I (and most of macro) could answer many questions definitively if we could do things we absolutely should not (eg. Run RCTs on real cities on things like the minimum wage). In microeconomics, we've resorted mainly to look for the effect of exogenous shocks on systems we care about (eg. The Muriel boatlift made a few studies on local unemployment/immigration due to the exogenous shocks nature of the event)
You could test your own theory against observations that calculations with real world data are very much a part of economics, but are just not part of this particular course.
Agreed, this is also a problem in economics (though even in economics there are some limited experiments that can be done). I think a key issue people overlook in this debate is the presence of a theory that makes sense. If you just plot data until you spot something that looks like a trend, or fit models until your F-test comes back significant or your R^2 is big enough, then yeah, you can't even make a good case for causation. But if you start with a reasonable theory that fits with existing knowledge, then you can at least make a case.
Yes as a theoretical construct the concept is totally fine. But speaking as someone who has heavy exposure to mainstream academic economics it’s usually taken way way further than that.
There’s a concept called the EMH and I think there’s a very strong tendency in Econ and pop culture to skip ahead and assume the hypothesis is basically true most of the time, often in the face of painfully obvious evidence to the contrary.
What I'm calling for is for economists to drop the pretense and misconception that using empirical methods for studying macroeconomics makes it scientific. And I'm suggesting that being good at math is not useful unless you're using useful data. In the study of logic, arguments can be considered valid if they are formally correct, but still unsound if their premises are false. Much in the same way, one can perform any number of valid mathematical transformations on data but still be left with unsound conclusions if those data were gathered incorrectly.
I am not saying that empiricism is inherently flawed, or that we should stop collecting economic data. And I I do not intend to advocate any particular school of economic thought here. All I'm advocating is that students be taught how to think critically about what they are being taught. So much of a modern economics education consists of looking at the changes in figures over time that very little is spent focused on a more general kind of reasoning.
The kind of reasoning I'm calling for is not easy to define. This is one of the tremendous advantages numbers have over argument in most minds. This kind of reasoning takes into account the notion that most of the information we obtain is not perfect or complete, and that many of our determinations are really judgment calls on what is more likely to be true. If empiricism is reasoning with your eyes, this is reasoning with your nose. It is a trained skill that allows you to recognize dubious premises and unspoken assumptions. When refined, it allows you to distill the essence of arguments down to a set of axioms that you can use to build a coherent model of the situation at hand. It is this theoretical side that allows you to understand how to construct experiments that test hypotheses, or whether that is even possible in each case.
To demonstrate the importance of gaining an understanding of the theory and rules behind something before testing it, I offer a parable:
The commissioner of the NFL once decided that teams were punting too much and he hired an econometrician (economic statistician) to study the situation and provide a solution to this problem. The econometrician applied his skills to the task at hand, aggregating data from several seasons to find correlations. He noted that there is an incredibly strong correlation between forth downs and punting, and he recommended that the commissioner ban fourth downs. In the next season, offenses were only given three downs. To the econometrician's surprise and the commissioner's chagrin, teams actually punted more frequently, as the fewer number of downs dramatically limited offensive opportunities.
The econometrician's misunderstanding was based on something rather obvious (if you understand American Football): a failure to separate correlation and causation due to an ignorance of the rules of the game. And compared to a global economy, football is a very simple game, with very simple rules. Applying reasoning to the example is very straightforward, but applying the same thing to a world of dynamic human behavior is much more subtle. Which is why students ought to be trained to question assumptions and sense where logic and math have separated themselves from the reality they are supposed to help us describe.
People will disagree about when things correlate to reality, and about what things make sense in parables. But almost anyone can learn to recognize when a number seems too specific, just like most decent coders learn to recognize "code smell." Just the other day, someone told me confidently that 65% of communication is non-verbal. Now, while I almost agree intuitively, I immediately asked where they heard that, and how someone could have arrived at that figure, which seemed oddly specific for something (communication) that I don't think is frequently quantitized. Every student of a soft science needs to have this skill strongly developed, or they will begin to take these kinds of things at face value.
I know research is hard, and these issues are fraught with complexities and challenges in acquiring data, but it still boggles my mind just how many macro economic theories don't have strong empirical data to support them.
There is a problem with the simple economic models that don't claim to predict real world behavior because they capture only one aspect of reality (everything else being equal...). They are not really scientific because they are not falsifiable. Math works and it seems to reflect reality, but we have no epistemic mechanism to verify this.
If we can do this eventually, then perhaps we can make our way to models that capture a larger fraction of the truth and can be tested against reality.
I think the point being made is that it's not straightforward to synthesize a plausible set of causes and remedies, despite the survey of 21 sources, and that economics as a science needs a bit of self-reflection in the light of this reality.
This is complete nonsense, economics is by far the strongest social science field in terms of empirical methodology. Pretty much all the modern causal inference methods were developed in econometrics. Try taking a look at the big econ journals (like Journal of Political Economy, American Economic Review, or Journal of Finance) and I assure you you won't find anything resembling arbitrary "opinions".
> most of the models used in Economics are based on abstract reasoning about "ideal" markets and actors (or more recently on game-theoretic ideas) instead of experimental data.
Essentially it's like pre-Enlightenment Natural Philiosophy, in that it's a discipline that claims to explain the way the world works, but actually explains who has the most sway in getting their ideas accepted.
> ...should provide ample data to develop "real" models of economic behavior against, and I think that many researchers actually already make use of this data.
I suspect reality will get about as much traction against idealogues of e.g. the Chicago School as actual climate science does agains useful idiots like Bjørn Lomborg. It's about providing views that happen to be useful to moneyed interests, not reality.
"To be clear, I am an empiricist and believe that the study of systems can only come about through careful experimentation, modeling, and formal mathematics. Mathematics, in my opinion, is the only objective measure by which we can analyse systems and trends."
I think we have to be a little careful about approaching economics as a problem to be solved. Logic, math, and empiricism have their limits (even if they are the best tools we have).
For one thing, empiricism relies on being able to control the variables. But that's hard to do, because others can always refute the results by introducing other variables and factors.
And (as you point out) the possibilities explode so quickly that it's very difficult to reason purely based on logic from base principles. And there are humans involved, which makes it even more difficult (if not impossible).
I'm not criticizing logic and empiricism as tools. But I think they can very easily give a false confidence in the answers you get. You have to realize how small a piece of the problem space your logic actually covers; and how difficult it is to distill real world events into a nice clean, unbiased data set.
reply