8 ms·
> That is, most economic discussions are based on terms that aren't meaningful in any practical way. I see a lot of economists poking at the metrics with their
by chimeracoder 15y ago
> That is, most economic discussions are based on terms that aren't meaningful in any practical way. I see a lot of economists poking at the metrics with their pencils and talking theory while waving their arms around wildly.
Actually, the problem is that people infer terms to mean things that they don't mean, oftentimes because the terminology is (regrettably) overloaded, even when the technical meaning is very precise and limited. Understandably, people interpret these terms literally, which sometimes leads to conclusions that seem paradoxical or impractical[1].
Remember that economics is really just the study of choices: choices in the small (microeconomics) or choices in the large (macroeconomics). Choices may involve money, but money is not inherent in economics - it just so happens that there are many advantages to a monetary system, which is why most economies develop some form of currency, and which is also why many (but not all!) economists deal with money. Even though it uses similar terminology, macroeconomics is fundamentally different[2] from microeconomics at its core. One is the study of individual decisions on a small scale, and the other is the study of the non-random patterns of many independent decisions.
GDP is not the perfect measure of what it's meant to gauge (macroeconomic economic 'health', which is itself a problematic term). GDP systematically overestimates the health of oil-backed economies with huge wealth disparities, for example. That said, there's no metric that's much better for the general case - all have similar systematic flaws in some way.
Furthermore, GDP has the advantage that it's completely unambiguously defined. That doesn't mean it can be measured perfectly, but having an unambiguous definition is a key step. When you start talking about these contrived indices (the 'happiness' index, the 'corruption' index, etc.), you get the problem that there's little underlying justification for the way that the model is constructed, so even if the end results are useful for a particular domain-specific application, they don't generalize well. GDP, on the other hand, can be derived quite easily from first principles, so it generalizes very nicely. Furthermore, GDP happens to correlate very well with most things that we would classify as 'good' (quality of life, corruption, happiness, etc.), so it's a nice proxy for these intangible 'goods' (no pun intended), which are more difficult to measure precisely.
So back to your question: maybe the measurements themselves are rigged - that just gets at the question of who is actually doing the analysis. But the construction of the model is no more 'rigged' than any other model. You just have to be careful to know that, like all models, its application is still very limited, even if it's one that we happen to use quite frequently.
As George Box says, 'All models are wrong, but some models are useful'. GDP is 'wrong', but it also happens to be by far the most useful way of describing the patterns choices on a large scale.
[1] This is by no means specific to economics: it happens in statistics as well, and (to a lesser extent) in computer science.
[2] That is to say, they aren't mutually exclusive, but they are based on completely separate assumptions, so principles from the small do not necessarily carry over into the large scale.
- DanielBMarkham 15y agoI agree with everything you say, but I believe you may have actually obfuscated the discussion rather than advancing it. The problem, ironically enough, is in the terms involved. My comment was about the discussion around economics, not economics itself. Of course mathematical models are more or less useful and all of that, but if the same English term is used in multiple arguments where it has a slightly different meaning each time? One columnist uses GDP as a stand-in for "size of the economy". Another uses it for something more akin to cash flows. That's not a problem with models, that's a problem with common usage of terms. Secondly, and related, is the use of the term "useful" There's a lot of wiggle room in "useful". You could drive a truck through it. :) Useful to whom? In what context? There's a larger and deeper discussion about the Wittgenstein-ian problems with networks of terms -- I believe it may be more accurate to say that GDP is a structured observation, not a metric. It's a subtle difference, but I believe it has a huge impact on the study of economics in general. But hey, like I said, I'm just an observer.
- Willwhatley 15y agoWithout meaning to be tedious, Daniel do you mind explaining what a metric (a measurement?) is, what a structured observation is, and (if not immediately clear) add a sentence or two on what the difference between the two amounts and why they might be confused? (I didn't get very far into reading Wittgenstein.)
- DanielBMarkham 15y agoI'll try, but I think I really need to hash this out on my blog a few times to nail it. The problem here is that you are building a conceptual model of economics before you measure anything. It's backwards. If I buy a candy bar, am I trading an hour of my wages for food? I'm using money, but how did I get the money? Am I trading an equal amount of gold that represents the perceived valued I provided the rest of the economy? What is the real definition of the thing taking place here? I'd wager that there is no universal definition that applies to all economic transactions. It's something more like cellular automata with multiple dimensions, not a causality chain built of little blocks. You want to measure something, and you're immediately knee-deep into abstract terms and aggregate numbers. So you come up with an abstract model for measuring aggregation before you even begin to measure anything. It's never going to work. You're never going to be able to construct a reproducible model with high fidelity like that. So if you want to talk GDP, you can say you observed this amount of cash moving around. I'm okay with that. But if you say the economy grew or some such, it doesn't work. You are presupposing the thing you are trying to describe. Each transaction involving that measurement had it's own value and definition. You're taking an aggregate of a fudge of a bunch of smoke. At the end you get a number, sure, and you can take it and plug it in to various models and get something kinda/sorta useful. It's heuristics for mathematicians, not science (as we understand it) If I'm actually on to something here and am not just being a crackpot, the implications could be that economics as a science can probably be only understood in a causal way by really complex computer models very carefully and rigorously constructed.