Debating GDP Dogma With An Economist

For a general discussion of topics relating broadly to power and political economy (e.g., capital-as-power, Marxism, neo-classical economics, institutionalism).

Moderator: sanha926

Debating GDP Dogma With An Economist

Postby bfix » Sat Mar 30, 2019 7:43 pm

In a recent published paper entitled "The Aggregation Problem", I point out the severe problems in the measurement of real GDP. In the paper, I demonstrate how changing prices create uncertainty in the measure of real GDP growth. Since the US government does not acknowledge this uncertainty, I argue that the official measure of real GDP hides the problem.

Recently an economist contacted me about this paper. The economist took issue with my claims. Specifically, the economist argued that "Fischer" indexes solve the base year problem and thus remove the uncertainty.

For those who are not familiar, a Fischer Index is a way of calculating inflation using a moving base year. Instead of choosing a single base year and fixing prices, the Fischer index "chains" many base years together. It is essentially a moving average.

In the series of posts that follow, I present our exchange. The debate did not end well. Initially, I tried to argue on scientific grounds. I wanted the economist to provide objective criteria for showing that chain-weighted GDP was the "correct" measure of output.

It turned out the economist found this insulting. GDP, the economist insisted, was based on normative values. Thus notions of "correct" were meaningless. Instead, what mattered was that the chosen measure lived up to the chosen philosophical ideals.

At this point, I thought the debate was over. The economist had basically conceded my points: GDP is based on normative decisions. Different decisions will lead to different measures, therefore there is uncertainty.

But it turns out that I was wrong. Instead, the economist used a kind of bait and switch. The economist admitted that the real GDP was normative, but would only accept criticism from within the normative framework. In other words, questioning the normative framework was out of bounds as legitimate criticism.

The debate ended in an unsatisfying way. But I think it nicely illuminates how dogmas work.

References
Fix, B. (2019). The Aggregation Problem: Implications for Ecological and Biophysical Economics. BioPhysical Economics and Resource Quality, 4(1), 1-15. doi.org/10.1007/s41247-018-0051-6

Preprint: https://osf.io/preprints/socarxiv/tfwju/
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby bfix » Sat Mar 30, 2019 7:46 pm

Economist's Message 1
***************************
“The US government has imposed an official way of *hiding* the [base-year] problem.”

Right. If you use a Laspeyres index, growth in the index is sensitive to the base year. But that’s not true for a Fischer index, which *by design* accounts for changes in base.

I suppose it’s a bit like the way that though I struggle to talk with my cat, I “hide” the problem by talking with my wife instead?

I agree that there is a fundamental problem with measuring growth as production patterns change, but why would choosing an index which attempts to get closer to the truth of things be “hiding” the problem rather than attempting to deal with it?


My Response
***************

Hi -------,

How do you define closer to the truth?

To know if something is closer to the truth, you need to first define what the correct change in production is. I'm saying we cannot do this objectively because our unit is unstable. So any way of dealing with the base year problem is by definition arbitrary. It hides the real uncertainty that exists. In honest disciplines, scientists report uncertainty. They do not hide it.
Last edited by bfix on Sun Mar 31, 2019 12:45 pm, edited 2 times in total.
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby bfix » Sat Mar 30, 2019 7:48 pm

Economist's Message 2
***************************
Well, in a sense that is my question for you.

Real GDP is, in principle, supposed to be an index of production. So there is a problem of aggregation. One way to aggregate is to pick one set of prices (dollars, tons, cubic feet, what have you) and weigh production of component products by those prices. If, between period 1 and period 2, the production of each component doubles, then total production has doubled, and this method of indexing will reflect that exactly— no matter what set of prices we choose, or indeed if we choose numbers at random. We can say at the chosen prices, production doubled between period 1 and period 2.

The problem for this kind of index comes in when the growth in production across components is nonuniform.

So let’s say just for arguments sake we choose period as weights
Period 1 we produce 100 apples sold for $1 each and 1 orange for $100.
Period 2 we produce 1 apple for $100 and 100 oranges for $1 each.

So Y1=$200, Y2=$10,100 at period 1 prices.
And Y1=$10,100, Y2=$200 at period 2 prices.

So has production grown by a factor of 5050 or shrunk by a factor of 5050? We know neither is accurate, because we cannot be producing more than a factor of 100 more than before; nor can we be producing less than a factor 100 less. Now, a Fisher index would tell us production is changed by a factor of only 1 (I.e., not at all), which may not be “truth” because we still have an indexing problem, but it sure as hell makes a lot more sense— given the dollar basis for our weights. It helps correct for the problem of shifting basis.

Now maybe you don’t like the spending biased revealed preference implication of price weighting. (Averaging component growth rates by expenditure shares is not obviously appropriate from a welfare perspective, for example.) But moving to a Fisher index doesn’t hide *that* at all.

Now, what do *you* mean that moving to a Fisher index hides the base year problem rather than serves to correct for it?



My Response
***************
You have provided a great example, so I will use it to make my point. You write:

"So let’s say just for arguments sake we choose period as weights
Period 1 we produce 100 apples sold for $1 each and 1 orange for $100.
Period 2 we produce 1 apple for $100 and 100 oranges for $1 each.

So Y1 = $200, Y2 = $10,100 at period 1 prices.
And Y1 = $10,100, Y2 = $200 at period 2 prices."

You then state: "So has production grown by a factor of 5050 or shrunk by a factor of 5050? We know neither is accurate because we cannot be producing more than a factor of 100 more than before"

First, there is a math error here. According to year 1 prices, production has grown by a factor 50.5. According to year 2 prices, it has shrunk by a factor of 50.5. So your statement that "neither is accurate because we cannot be producing more than a factor of 100 more" is incorrect. Both are accurate.

Let's backtrack to review what is going on here. We are trying to aggregate two different commodities. To do this, we need a common unit of analysis. This will allow us to compare apples and oranges. The unit assigns a weight to how we "convert" apples into oranges (and vice versa) so that we can add them together.

There is a huge range for this weighting, and this will affect our measure of output growth. In the limit that oranges carry no weight at all, output has shrunk by a factor of 100. In the limit that apples carry no weight at all, output has grown by a factor of 100. So this is the range of possible measures for the growth of output.

The point of a unit is to FIX the weight between different commodities so that there is no ambiguity in measurement. So if we used mass, we might find that the average orange weighs 2 times as much as the average apple. So let suppose apples are 1 pound and oranges are 2 pounds. This gives

Y1 = 102 lb, Y2 = 201 lb

So in terms of mass, output grew by about 100%. Of course, there will be some uncertainty here because of variation in the mass of apples/oranges, and from the measurement error in our mass scales. But this uncertainty will be small.

So lets go back to price. Once we choose price as our dimension, we have a different problem. Now our UNIT is unstable. The prices of different commodities change in divergent ways over time. This is fundamental and irreducible uncertainty in our unit of analysis. Depending on the year we choose prices, we get different measurements for output growth.

The point is that we are stuck with the measurements that are returned to us. We do not get to choose which ones are correct. ALL OF THEM ARE CORRECT. In science, when we have different measurements for the same phenomena, we acknowledge that there is uncertainty in our measure. We report this uncertainty openly.

Back to your example. We have two conflicting measures of output. One measure says output grew by a factor of 50.5. The other says it shrank by a factor of 50.5. This means we have uncertainty in our measure.

To frame the problem, suppose these were measurements for the growth of bacteria in a petri dish. The scientist doing these measurements would report them as follows: the average estimate is zero growth, but the uncertainty is plus or minus a factor of 50.

Anyone reading this reported measurement would conclude that it is useless and that the scientist needs to redo the experiment and take more accurate measurements.

But in economics, there is nothing to redo. Our measurement instruments are not the problem ... it is our unit of analysis. The only way to remove the uncertainty is if price change suddenly became uniform.

So yes, chain-weighting hides the problem. It does not (and cannot) remove the uncertainty in the measurement of output. It picks out of thin air a particular method for measuring average price change. This is like our scientist arbitrarily picking ONE measurement of bacteria growth, when there is actually a range of estimates. We do not get to do this in science.
Last edited by bfix on Sun Mar 31, 2019 12:40 pm, edited 1 time in total.
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby bfix » Sat Mar 30, 2019 7:51 pm

Economist's Message 3
***************************
Let’s see if I can make this even more explicit. GDP— philosophically— values relative market prices as measures of (productive) value. And yes, these measures vary over time. From a philosophical standpoint this is a feature and not a bug. Therefore, a measure of value consistent with the philosophy must be flexible enough to handle changing prices. By contrast, the Laspeyres index measure of GDP implicitly takes base-year prices as the value of a particular unit. Meaning it *fails* over time to reflect the underlying philosophical value. It is not an uncertain measurement of the very specific philosophical value, but a *mis-measurement* thereof. This for the very reason that— philosophically— the base period prices do not reflect values in later periods. Likewise, the prices of a later period do not reflect the philosophical value of production in earlier periods.

Using a Fisher index (or, really, any chaining strategy) instead of Laspeyres simply aims to bring the realized values in line with the philosophical ideal by taking into account the changes in relative prices so fundamental to the philosophy. That’s the opposite of hiding the choice of values.

Again, this doesn’t make market prices the appropriate basis for value. But that’s not our point of disagreement, either.




My Response
***************

Your write: "GDP — philosophically — values relative market prices as measures of (productive) value. And yes, these measures vary over time. From a philosophical standpoint this is a feature and not a bug."

I agree that price change is a feature of society. In this sense it is not a bug. It is important that we scientifically study changes in prices --- specifically the differential change in price of different commodities (see the work of Nitzan and Bichler below). When we do this, price is the FOCUS of analysis. It is not taken to reveal "productive value".

You correctly identify an important duality in economic theory. GDP assumes that relative prices indicate "productive value". It is not price change per se that is the problem. The problem is when we try to apply this duality to a system with changing prices. Permit me to take a winding route to make my argument.

Science is predicated on the idea that there is a "real world". But we can only observe the "real world" through measurement. This means there is no absolute standard by which we can judge if a measurement correctly represents the "real world". All we have our measurements. So what are we to do when measurements conflict with one another?

This was a major source of argument among early empirical scientists. There are really only three possibilities. Either we assume that:

A. NONE of the measurements represent the "real world".
B. ONE of the measurements represents the "real world"
C. ALL of the measurements represent the "real world"

Option A was never really considered, because it leads to nihilism. I means that measurement says nothing about the world, so why bother. Option B was quickly rejected because there are no criteria to tell which measurement represents the "real world". All we have are our measurements. So empirical scientists decided on option C. We accept that ALL measurements represent the "real world". But this comes with a bitter pill ... it means that there is uncertainty in our knowledge, bound by uncertainty in measurement.

So let's apply this to a simple example, and show how it relates to the measure of GDP. Suppose you want to measure your height using a meter stick. As an empirical scientist, you assume that your measurements give you knowledge about the real world, so you accept the following duality

measured height = real height

But unbeknownst to you, your meter stick changes in length over time. You repeatedly measure your height, and each time you get a different value. So what do you do? One option would be to pick one of your measurements and assume it represents "real height". But there are no objective criteria for doing this, so we must reject this option. This means we must assume that ALL of our measurements represent your real height. The consequence is that their is uncertainty in our knowledge of your height.

One thing we can do is measure the central tendency of the different measurements. But this by itself is not enough. First, there are many different ways to take the average --- arithmetic mean, geometric mean, root mean square, etc. Each one will be different, and there is no way to know which is "correct". Second, if we report only the mean of our measurements, we are hiding information about the underlying uncertainty. This is a mortal sin in science.

So back to GDP. As you observe, GDP is based on a duality. Economists propose that prices are a measure of "productive value".

relative prices = productive value

So here, "productive value" is the element of the real world that we want to measure. It is not observable directly. Instead, it can only be revealed through prices. So we measure the growth of productive value using prices. But like our changing meter stick, prices change over time. This means we get different results for growth depending on the year we choose prices. This is analogous to getting different values for height from our changing meter stick.

So what are we to do? One option is to choose a year in which prices are "correct". But there are no objective criteria for doing this. Worse still, it means rejecting our duality. It means that only SOMETIMES do prices represent productive value. This is a slippery slope. Occam's razor would suggest we just assume that prices NEVER represent productive value. But this means rejecting the very premise of GDP. (This, by the way, is what I think we should do).

The other option is to maintain our duality. We insist that prices ALWAYS reveal productive value. This means that we must assume that every base year gives a "correct" measure of real GDP growth. This leads to inherent uncertainty in our measurement. So what should we do?

As above, one option is to simply take the mean. This is basically what chain-weighting GDP does. But there are many ways to take the mean. Why should we believe any single one of them reveals reality, and the rest are WRONG. What criteria do we have to tell this? None.

This is why chain-weighting is an official way of hiding the problem inherent in GDP. It hides the uncertainty. Like the uncertainty in height cased by a changing meter stick, we need to report the uncertainty in GDP. When we do not, we are hiding information. We are pretending that our metric reveals the "actual" growth of the economy ... that it gives a true statement about reality. But it does no such thing. Instead, it is the measurement UNCERTAINTY that gives the true picture of our knowledge. It shows us that, despite what economists claim, we know very little about the growth of "productive value". Uncertainty is backbone of scientific skepticism. False certainty is the backbone of religion.

At a very minimum, I would like to see uncertainty in GDP reported in official documents. But I doubt this will ever happen. It would seriously undermine the credibility of the government and the economics profession. We should be under no illusion that economists practice science.

But ultimately I reject the very premise of GDP. I don't think prices have anything to do with "productive value". Moreover, I think "productive value" is a hopelessly subjective concept. We have no business trying to measure it scientifically.

If we want to do science, then we need to reject GDP. If we want to measure the scale of the economy, I have argued that energy use is one good metric. The laws of thermodynamics mean that energy use is vital to the functioning of society, and may have to do with complexity. At the very least, the science of energetics means that aggregating energy consumption can be done with objectivity.

References

Fix, B. (2015). Rethinking Economic Growth Theory from a Biophysical Perspective (C. Hall, ed.). New York: Springer.

Nitzan, J., & Bichler, S. (2009). Capital as Power: A Study of Order and Creorder. New York: Routledge.
Last edited by bfix on Sun Mar 31, 2019 1:36 pm, edited 1 time in total.
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby Ikonoclast » Sat Mar 30, 2019 10:15 pm

I read your paper, "The Aggregation Problem". I was very impressed with it from a scientific and philosophical point of view. The logic seemed impeccable to me. I say this as a layperson in economic, scientific and philosophical matters, albeit one reasonably well-read in all three disciplines.

I have given up reading and paying attention to orthodox economics. It is a profoundly dishonest discipline in scientific and philosophical terms. At best it is a failed research project. That's the most charitable thing one can say about it. At worst, it has now degenerated into an extended exercise in intellectual dishonesty.

But just as the Church once was "right" and Galileo was "wrong", so today those who challenge orthodox economics are deemed "wrong" by the instituted powers.

The comprehensive refutation of orthodox economics simply awaits its imminent collapse from factors it deems extraneous; namely feedbacks from the real systems of the biosphere and feedbacks from real human bodies and brains. The need then, at that crisis juncture, will be to have more objective and more humane theories ready for deployment to try to save the earth's ecosystems and eco-services: and thence to save civilization and humanity themselves. This is if it is still possible after the horrendous natural system damage wrought by orthodox, capitalist economics.
Ikonoclast
 
Posts: 28
Joined: Sun Dec 15, 2013 4:47 pm

Re: Debating GDP Dogma With An Economist

Postby bfix » Sun Mar 31, 2019 12:54 pm

Economist's Message 4
***************************
This exchange has been so far— in my view— completely useless. I have yet to gain the slightest confidence that you understand what I’m driving at.

"I agree that price change is a feature of society. In this sense it is not a bug"


No, no, no. I don’t mean that it is a feature of society— I mean that it’s a feature of (chained) GDP as constructed that it involves “measuring sticks” that change over time. Precisely because relative values change over time.

"But unbeknownst to you, your meter stick changes in length over time."


What is the metaphor here? Orthodox economists *understand* their meter sticks are changing over time. That’s the whole point of chaining!

"So here, "productive value" is the element of the real world that we want to measure. It is not observable directly. Instead, it can only be revealed through prices. So we measure the growth of productive value using prices."


Absolutely not. Orthodox economists *choose* relative prices as a proxy for relative productive value. That’s normative.

"But like our changing meter stick, prices change over time. This means we get different results for growth depending on the year we choose prices. This is analogous to getting different values for height from our changing meter stick."


Right! That’s *precisely* why we chain!

"So what are we to do?"


Chain!

"The other option is to maintain our duality. We insist that prices ALWAYS reveal productive value. This means that we must assume that every base year gives a "correct" measure of real GDP growth. This leads to inherent uncertainty in our measurement. So what should we do?"


This is obviously silly. Like pretending Einstein didn’t produce a theory of relativity. We should chain.

"As above, one option is to simply take the mean. This is basically what chain-weighting GDP does. But there are many ways to take the mean. Why should we believe any single one of them reveals reality, and the rest are WRONG. What criteria do we have to tell this? None."


I am not sure you understand how chaining works. It doesn’t even require averaging. You can chain index using only Laspeyres growth. And yes, the choice of mean (or even if to average) matters… a little. The point is to chain together short enough periods that the weights don’t much change, and therefore the choice of mean also doesn’t much matter.

"This is why chain-weighting is an official way of hiding the problem inherent in GDP. It hides the uncertainty. Like the uncertainty in height cased by a changing meter stick, we need to report the uncertainty in GDP."


Your “uncertainty” is no such thing, and to the extent the problem you describe actually does exist, chaining helps correct for almost all of it.

"If we want to do science, then we need to reject GDP. If we want to measure the scale of the economy, I have argued that energy use is one good metric."


Ugh. So our economy shrinks insofar as it becomes more energy efficient? I’ll take a pass.

In short, taking each year’s prices and quantities as given, any “uncertainty” remaining after making the particular choice of Fisher indexing is almost certainly small when compared to the uncertainty in the surveys which provide the prices and quantities to begin with. And those— barring a systemic bias— should cancel out over time.

But what you are calling “uncertain” is quite certain, because we can and do observe how the measuring stick is changing over time!




My Response
****************
To reply to your latest comments, I'd like to summarize your arguments. Correct me if I'm wrong, but I think these are your main points:

1. Economists are aware that prices change over time.
2. Price change can be objectively measured.
3. Economists have developed price indexes that are designed to deal with changing prices (Fisher indexes, etc). These indexes deal with the base-year problem, and so get rid of uncertainty in the calculation of real GDP.

I agree with point 1. Yes, economists are aware that prices change over time. This is why they try to adjust for inflation.

Whether point 2 is valid depends on how we define price change. If we simply mean the nominal change in the price of a commodity, then yes, price change can be measured. But this is not how economists measure price change. Economists want to differentiate between a "pure" change in price, and a change in price that reflects a change in "productive value" (i.e. quality). So take computers. When economists construct price indexes for computers, most of the change in indexed price is due to their measurement for quality change. Since this measurement is inherently subjective, it negates the objective measurement of price change. So no, price change (in the sense meant by economists) cannot be objectively measured. See Jonathan Nitzan's article for an illuminating discussion of this problem.

http://bnarchives.yorku.ca/137/

On point 3, I agree that economists have indexes that "deal" with the base-year problem. These indexes take as inputs changing prices over time, and they return a unique value for the price index. So in a mathematical sense, there is no uncertainty in these functions. As such, they return an unambiguous value for the growth of real GDP.

The problem is that there are an infinite number of functions that could do the same thing. They each take changing prices as inputs and return an unique price index. But each assigns different weights to the price change of different commodities. The challenge is to show that your chosen method is the CORRECT one. This means showing that ALL other methods return the WRONG price index, and thus the wrong measurement for real GDP.

In your first email, you wrote that chained indexes get "closer to the truth of things". But how do we know this? You have not given any objective criteria by which we can judge if one index is better or worse than any other. Instead, you are arguing by decree. You are simply stating that the Fisher index is the correct measure of inflation.

To be fair to you, most economists do the same thing. The government simply decrees an official measure of inflation and that leads to a unique value for the growth of real of GDP. In a sense, this is fine. Anyone is allowed to decree that they want to measure growth a certain way. But it needs to be explicitly stated that this is a measurement based on a subjective decision. When you acknowledge this, you do not get to claim that your index gets "closer to the truth". It is a normative claim, and nothing more. It ranks with the Pope's normative decrees about the holy trinity. Its dogma, not science.

Science is about the search for OBJECTIVE knowledge. If a measure gets "closer to the truth", this means it can be measured objectively by anyone. And ideally, there should be independent ways of confirming a measurement.

As an example, consider the Hubble constant, which quantifies how fast the universe is expanding. Astronomers know they have a problem with this measurement because their independent measures disagree. If they use the microwave background, they get one answer. But if they use super nova, they get a different answer. Which one is correct? No one knows. This is how real science is done.

https://astronomynow.com/2018/07/13/cos ... -constant/

So how do you convince me that the Fisher index is the "correct" way of measuring real GDP? The easiest way would be to go out and measure the growth of "productive value" independently of prices. If you found the same value as the official measure of real GDP, then you would kill my arguments.

The problem, of course, is that it impossible to do this. I know of no way to measure productive value without defining it by prices. This takes you right back into the thick of the problem. You have defined "productive value" using an unstable unit. There is no way out of the mess, other than to report the uncertainty openly. Or even better, recognized that measuring growth using prices is a dead end.
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby bfix » Sun Mar 31, 2019 1:00 pm

Economist's Message 5
***************************

It is entirely possible I am failing to make myself clear— rather than your failing to comprehend a clear message. It is exactly because your claims appear to be so outrageous, I thought it better to seek clarification privately rather than call you out publicly. However, your responses have so far largely reinforced my initial impression.

How can I be so immovable after so much feedback? Because nothing of substance that you have added has come as news to me. I keep nodding my head and thinking either “yes, I agree, and if you think this is a point of dispute I must again try to clarify” or “yes, I already understood that to be your position"

To reply to your latest comments, I'd like to summarize your arguments. Correct me if I'm wrong, but I think these are your main points:

1. Economists are aware that prices change over time.
2. Price change can be objectively measured.
3. Economists have developed price indexes that are designed to deal with changing prices (Fisher indexes, etc). These indexes deal with the base-year problem, and so get rid of uncertainty in the calculation of real GDP.


I think this is an unhelpful summary. However, in the interest of helping sort out your thoughts, I will proceed.

"I agree with point 1. Yes, economists are aware that prices change over time. This is why they try to adjust for inflation."


Good. If we could not agree on that, then this would be a truly useless discussion. Can we also agree then, that the economists at BEA, say, are deeply aware of the base-year problem?

Whether point 2 is valid depends on how we define price change. If we simply mean the nominal change in the price of a commodity, then yes, price change can be measured. But this is not how economists measure price change. Economists want to differentiate between a "pure" change in price, and a change in price that reflects a change in "productive value" (i.e. quality). So take computers. When economists construct price indexes for computers, most of the change in indexed price is due to their measurement for quality change. Since this measurement is inherently subjective, it negates the objective measurement of price change. So no, price change (in the sense meant by economists) cannot be objectively measured. See Jonathan Nitzan's article for an illuminating discussion of this problem.


Yes and no.

The reason for quality adjustment is that not everything is a commodity in the strict sense. A personal computer of 1986 and a personal computer of 2016 are *not* the same. Because Apple IIe’s with 256K RAM are not currently in production, we cannot observe a price for currently-produced Apple IIe’s with 256K RAM. Clearly, a similar computer today selling for the same nominal price is a *very* different thing. So YES there is uncertainty in our estimate of prices because we *absolultely* need *some* way to account for the fact that the items are not identical! I meant for this to fall under the umbrella of "the uncertainty in the surveys which provide the prices and quantities” (email, 3/27 @ 6:15pm)

But, as I stated in that same email "taking each year’s prices and quantities as given, any ‘uncertainty’ [meaning, relating to the base-year problem which we were discussing].., is almost certainly small” in comparison.

In short, while I understand your concern regarding this point, I don’t understand why you think it is relevant to the base-year problem, specifically.

"On point 3, I agree that economists have indexes that "deal" with the base-year problem. These indexes take as inputs changing prices over time, and they return a unique value for the price index. So in a mathematical sense, there is no uncertainty in these functions. As such, they return an unambiguous value for the growth of real GDP.

The problem is that there are an infinite number of functions that could do the same thing. They each take changing prices as inputs and return an unique price index. But each assigns different weights to the price change of different commodities. The challenge is to show that your chosen method is the CORRECT one. This means showing that ALL other methods return the WRONG price index, and thus the wrong measurement for real GDP."


No! Absolutely not!

1. Recognizing that there are an infinite number of ways to aggregate does NOT mean that the measurement for real GDP is *uncertain*. GDP is an index of the value of output measured at market prices (or, to head you off on this point… as near to market prices as possible— e.g., owner’s equivalent rent)

2. The challenge— for construction of GDP measures— is to reflect that as accurately as possible. To that end, chaining pretty unambiguously improves upon simple base-year indices. As I wrote previously "The point is to chain together short enough periods that the weights don’t much change, and therefore the choice of mean also doesn’t much matter.” (email, 3/27 @ 5:24pm)

3. This does *not* mean, however, that GDP is the “CORRECT” measure of *output.” The choice of GDP as a measure of output is normative— "Orthodox economists *choose* relative prices as a proxy for relative productive value.” (email, 3/27 @ 5:24)

4. Given the normative choices in GDP as a measure of output, chained Fisher is clearly superior to base-year Laspeyres indexing. I’m not saying it corrects completely for the problem and therefore the measure is perfect. I’m saying the point of chaining is to address a particular problem.

In your first email, you wrote that chained indexes get "closer to the truth of things". But how do we know this? You have not given any objective criteria by which we can judge if one index is better or worse than any other. Instead, you are arguing by decree.


This is pure insult. I’m talking about chaining GDP versus simple base-year. YOU said that GDP is uncertain because of the base-year problem. That problem is known. The POINT of chaining is to correct for this *exact* shortcoming in methodology. Do you understand this?

It’s like we have two “yardsticks”— one 36” and the other 37” and we have a philosophical reason for measuring things in 36” units. If something is just as long as one stick but not the other, it’s not uncertain how many yards long the something is. One of the “yardsticks" more accurately measures things in our desired 36” yard!

(Again, exhaustedly, that doesn’t mean we *should* be measuring things in units of 36” as opposed to, say, kilograms. But given that choice, it is *clearly* superior to use a 36” yardstick.)

Likewise, we *know* that base-year indices do not accurately reflect productive value— given the orthodox price-based normative choice for value. Given the normative choice, we *know* exactly why base-year indices are— over time— a bad method for aggregation. So we use a *better* method. Meaning one which is more consistent with the *normative choice*. I have no idea where I’m losing you, here.

As I wrote previously "Using a Fisher index (or, really, any chaining strategy) instead of Laspeyres simply aims to bring the realized values in line with the philosophical ideal by taking into account the changes in relative prices so fundamental to the philosophy.” (email, 3/26 @ 2:14pm)

You are simply stating that the Fisher index is the correct measure of inflation.


This is a terrible misrepresentation, and very much an erosion of trust.

First, I have stated repeatedly that the choice of aggregation is based on a normative judgement.

"Now maybe you don’t like the spending biased revealed preference implication of price weighting. (Averaging component growth rates by expenditure shares is not obviously appropriate from a welfare perspective, for example.)” (email, 3/25 @ 6:28pm)

"Again, this doesn’t make market prices the appropriate basis for value.” (email, 3/26 @ 2:14pm)

"Given* the particular (philosophical) values implicit in the entire exercise, chaining is necessary in order to produce an index which better *reflects* those assumed values.” (email, 3/26 @ 2:16pm)

"Orthodox economists *choose* relative prices as a proxy for relative productive value. That’s normative.” (email, 3/27 @ 5:24pm)

Second, I have stated that Fisher indices are not the only way to correct for the base-year problem, even given the normative choice. My argument is only that given the normative choice, chaining with a Fisher index is an improvement in that the method more accurately reflects that normative choice. I never said it was the only way to do so.

"Using a Fisher index (or, really, any chaining strategy) instead of Laspeyres simply aims to bring the realized values in line with the philosophical ideal by taking into account the changes in relative prices so fundamental to the philosophy.” (email, 3/26 @ 2:14pm)

"You can chain index using only Laspeyres growth. And yes, the choice of mean (or even if to average) matters… a little. The point is to chain together short enough periods that the weights don’t much change, and therefore the choice of mean also doesn’t much matter.” (email, 3/26 @ 5:24pm)

So I’d appreciate it, that if you think I’m saying something which contradicts what I’ve pretty explicitly stated, that you be kind enough ask for clarification rather than assuming you understand my argument fully and assign to me the *least* generous interpretation of my words. It’s exhausting to keep repeating myself and *still* have you misrepresent my position so obviously.

Anyone is allowed to decree that they want to measure growth a certain way. But it needs to be explicitly stated that this is a measurement based on a subjective decision. When you acknowledge this, you do not get to claim that your index gets "closer to the truth". It is a normative claim, and nothing more


In case it’s not yet clear, we agree regarding the normativity inherent in GDP as a measure of output. But your use of “closer to the truth” is stripped from the context I provided. I have striven mightily to explain that the “truth” I mean here is that the measure should be constructed as consistently as possible with the *normative* choice—NOT that the measurement yields the “truth” of output.

"This is how real science is done."


This is social science, and the normative choices are fundamental.

But this astronomical example is— even apart from that-- a terrible metaphor for our point of departure. A better metaphor is

https://www.sciencemag.org/news/2012/02 ... no-results

Scientists were trying to measure neutrino speeds, but it turned out the equipment was bad. They didn’t say, well, there is uncertainty in the measurement. We don’t say that neutrinos can be observed to move faster than light if we use this machine that doesn’t really do what we want— in theory— for it to do. We simply accept the measurements made with better equipment.




My Response
****************

I think we need to reflect on our debate so far. You initially wrote me to question my claim that chain-weighting hides uncertainty in the measurement of real GDP. Your exact query was:"why would choosing an index which attempts to get closer to the truth of things be “hiding” the problem rather than attempting to deal with it?"

My interpretation of your query depends crucially on your wording. You used the words "closer to the truth of things". I interpret "truth" as a scientific assertion. "Truth" means a a statement about the real world that is objectively verifiable. Thus, I read your query as saying this: chain-weighted GDP gets closer to objectively measuring the growth of the real economy.

Clearly I disagree with this assertion, as our exchange has shown. I think chain-weighting is an arbitrary decision, informed by economists subjective ideals. As such, official metrics of real GDP growth say nothing about the "real" (objectively verifiable) growth of economic output. When we take objective measurement seriously, we find enormous uncertainty in the growth of real GDP. Official metrics hide this problem.

Based on my interpretation of you query, I have repeatedly asked you to provide objective criteria that justify your claim that chain-weighted GDP gets "closer to the truth of things". Your response has been to express exasperation that I would even ask for such objective criteria. Instead, you redefine the word "truth". By "truth" you mean "the measure should be constructed as consistently as possible with the *normative* choice".

This is an odd definition of "truth". It means that any measurement that is consistent with our philosophical ideals is "true". That's a pretty low bar. For instance, suppose we are medieval members of the clergy. Our "philosophical ideal" is that the earth is the center of the universe. Because it is consistent with our philosophical ideals, the Ptolemaic (geocentric) model of the universe "gets closer to the truth". Copernicus and Galileo deserve their place in hell.

As far as I can tell, your arguments mirror this logic. You argue that economists believe that prices reflect productive potential. Given this normative (i.e. subjective) ideal, economists choose chain-weighted GDP as the most appropriate measure of economic growth. Because this measure most closely matches economists' normative ideals, it "gets closer to the truth".

This position is a devastating critique of real GDP. It means real GDP makes no objective claims about the real world. Instead, its purpose is to merely to reflect the subjective ideals held by economists. This is exactly what I argued in "The Aggregation Problem".

It means that most of the field of macroeconomics is a waste of time. There is no need for production functions or elaborate regressions to explain the growth of real GDP. Why? Because there is nothing to explain. Real GDP makes no objective claims about the growth of the economy. Instead, it is a subjective metric that is meant only to reflect economists' philosophical ideals.

The problem is that I see immense cognitive dissonance in your arguments. You acknowledge the subjective element of real GDP measurement. But then you use the language of science to talk about it. You use words like "truth", "unambiguously improves", "exactly", "clearly superior", "do not accurately reflect", "mis-measurement", and "correct for". This is the language of objectivity. We have no business using it when we are talking about a normative measure informed by subjective ideals. The best we can say is that this measure is consistent with our ideals. If economists talked about real GDP using this soft language, no one would take the measure seriously. And rightly so. The allure of real GDP is maintained only by constantly giving it the cloak of objectivity.

To conclude, we could have saved ourselves a lot of writing if you had expressed your position accurately in the first email. You should have written:

Chain-weighting does not "hide" the base-year problem, since there is nothing to hide. Real GDP is a normative metric. As such, its purpose is only to reflect the philosophical ideals of economists. It is not meant to measure any objective facts about the real world. Since chain-weighting better reflects economists' ideals, I consider it an improvement. It gets closer to what economists intended when they defined real GDP.
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby bfix » Sun Mar 31, 2019 1:03 pm

Economist's Message 6
***************************

Okay, you’re disturbingly hung up on one word- which in retrospect may have been a poor choice on account of a lack of mutual understanding regarding a number of issues. Now, I have spent days bending over backwards to bridge the gap here, and you’re still obsessing over my use of the word based on your original understanding of what I meant by it.

So I’m going to try to step through this carefully by breaking it down into small steps and you can tell me at what step I lose you.

1. We have to make a normative choice of what constitutes output. Does it include housework? Rent?

2. *Given the choice of definition of output* we have any number of objective measures respecting aggregate output. We can measure the energy consumption involved in production. We can measure the mass or volume of goods produced. We can total the dollars exchanged for goods and services. Thus, we can produce objective indices of the energy consumption involved in production of output. We can produce objective indices of the mass of output produced. We can produce objective indices of the volume of output produced. We can produce objective indices of the total dollars exchanged for goods and services produced. Again, *given* the choice of definition for output.

3. Note that— regardless of the definition of output-- none of these measures are indices of output. Rather, they are indices of certain characteristics of output. (Just as we don’t measure the proton, but the mass and volume and charge and spin of the proton. There is no objective measure of proton.)

4. None of these objective indices are entirely defensible as measures of output. If we take mass as our measure of output, then services do not count toward our index of output. If we take volume, then producing smaller phones reduces our index of output. And so on.

5. Expanding further on the second example in item 4, it is a matter of intuition that if production of every type of good and every type of service is unchanged from one period to the next that a "good” measure of output would also be unchanged. Call this the "item-5 problem"

6. So, given any nontrivial definition of output, there is no “objective” measure of output per se. We cannot construct some instrument which will reveal the objective “truth” about growth in output. We can seek merely to perfect instruments to measure objectively the growth in particular characteristics of output.

7. GDP is— by construction— the realized exchange value of all an economy’s currently-produced goods and services defined as output. In the US, the exchange value is denominated in USD. Some of the transactions are non-market and require imputation of the dollar value they *would* have fetched in the market. In this sense among others, GDP is objectively but imperfectly measured.

8. Consistent with item 5, we don’t care for GDP as a proxy for output itself. In part— but hardly exclusively— this is because even if production of every single type of good or service is unchanged, the dollars for which that production is exchanged may increase. An objective increase, perhaps, but also not intuitive as a measure of output.

9. We can fix this "item-5 problem" with GDP as a proxy for output by howsoever fixing the prices of all goods and services in the measure— irrespective of when they were produced.

10. The fix of item 9 creates its own problem— that growth in the measure depends of the choice of prices employed. Call this base-price problem the “item-10 problem”

11. If we select prices from (as well as possible) a consistent survey of the economy at a certain time, then the “item-10 problem” becomes the more specific base-period problem. Call this the “item-11 problem”

12. Now, let us back up a bit. Let’s also make a normative judgement that the relative price of goods and services reflect “real” relative value of each item’s marginal product. So we can pick absolutely any good or service as a basis for the calculation. By declaring that item as the basis, we find that GDP divided by the price of that good is “real” in any period— in that it is consistent with the normative choice. This “real” GDP is the (normative) value of total output relative to the (normative) value of one additional unit of production of the base unit. Unfortunately, this creates a problem in the sense that the growth of “real” GDP varies according to the choice of base item. Again this is due to relative prices changing over time. Call this the “item-12 problem”

13. Fortunately, no matter how fast relative prices are growing, then over a small enough time period, growth in “real” GDP over the period is insensitive to the choice of prices so long as the relative prices reasonably apply sometime in the period.

14. So we may— for example— pick not prices at the start of period, nor those at the end of period, but pick average price of each item throughout the period. We can compute the rate of growth in “real” GDP over the period holding constant those prices. And we can do the same for the next period using the average next-period prices of each item. Etc. We can simply chain together these small-period growth rates to produce a long-period index of “real” GDP which much better estimates changes in the relative value of production over time. (That is, we have adjusted our measure to work around the item-12 problem.)

15. It is not an improvement because it better reflects growth in *output* because item 6. That is, it doesn’t better reflect “objective” growth in output because that’s a nonsense proposition.

16. It is an improvement because it better reflects growth in output as valued by observed prices. That is, it better reflects the normative choice of value.

17. The “uncertainty” in real GDP you describe arising from the base-year problem is not uncertainty in the sense of not knowing what the length of our ruler is. Not is it uncertainty in the sense of not knowing what output is. It is uncertainty regarding whomever’s choice of values (as reflected by the choice of base year.)

18. Chaining does nothing to hide this. In a very true sense it merely reflects a different normative choice. If anything, then, it *adds* to your “uncertainty” rather than hides it.

19. Alternatively, one could— normatively— select any specific base year prices as the “true” measure of value just as one could select relative prices. And there would be zero “uncertainty” in the measure because such a choice would not be subject to the base-period problem. A Laspeyres index of output value based on the price of that base period would do a great job of reflecting that choice.

20. So yes, none of the base-year indices reflect the normative choices implicit in real market-valuation of GDP. This is not “uncertainty”, but a collection of inferior estimates.

21: In short, if one wants to estimate real market value of output, then chaining is important. If one wants to estimate growth in output as opposed to growth in some observable characteristic of output, then one is up a creek. That is what I meant by “the truth of things”— that the chained measure better reflects growth in real market value of output. The “truth” of output growth is… nonsense.

22: Yes, economists are sloppy about this in the sense that GDP and output are often used interchangeably. I’m certainly guilty of this. But that is where the normative choice is hidden— not in the chaining.

I look forward to understanding at which step we part ways.


My Response
***************

Ok, I think we are finally getting somewhere. I agree with almost everything you said in you last email. As I read it, it reiterates most of the issues that I raised in "The Aggregation Problem". I think the point of contention is the word "uncertainty".

I think you interpret this word as "statistical" or "instrument" uncertainty. So when we take small samples or use an instrument that is not very good, there is large uncertainty. This uncertainty is inherent in our instruments and the mathematics of randomness.

In the context of GDP (or any form of aggregation) I am using the word "uncertainty" differently. Its not about our instruments or randomness. It is about OURSELVES. I am talking about the role of the observer in determining the measure. When we aggregate, we make choices. Each time we make a subjective choice, we open a "possibility space" for what our measurement could have been, had we made a different choice. I am referring to this possibility space as "uncertainty".

Your point of disagreement is that you do not see "uncertainty" here. You look at the decisions that are made. You see that they are rational, and driven by a goal. These decisions collapse the possibility space, leaving an unambiguous measure. So take real GDP.
You see that economist have laid out criteria that they want to meet. At every step of the way, they make decisions according to these criteria. This leads them to calculate real GDP using chain-weighted price indices. So once all these decisions have been made, there is no uncertainty in the measure --- in the sense of instrument or statistical error. All the underlying inputs are known with precision. If we use the word "uncertainty" in this instrumental sense, then I agree with you. Real GDP has very little instrumental or statistical uncertainty.

But I am using the word "uncertain" in a different way. I am looking at the possibility space of aggregation decisions. So I look at real GDP and see immense "uncertainty". Why? Because I do not hold the philosophical ideals that economists do. I do not believe that prices reflect productive value. I see no reason to measure growth in a way that is closest to the moving base year. These to me are arbitrary decisions. Instead, I stand back and look at the whole possibility space of decisions that go into calculating real GDP. When I do so, it seems immense and unwieldy. It includes the base-year problem, but also boundary issues and quality change issues. When I look at this possibility space, I see immense uncertainty. And I see no compelling scientific criteria for collapsing this space.

So the point of contention between us is that you are comfortable with the normative decisions that go into calculating real GDP. These decisions collapse the possibility space, and you are fine with that. You look at economists' goals and see that they have made decisions that are appropriate, given their goals. Therefore there is no uncertainty in real GDP.

But I am not comfortable with these decisions. Unlike you, I don't think social science should be normative. POLITICS should be normative. SCIENCE (of any kind) should be objective. I see normative aggregation choices are part of politics. As such, they need to be framed as a political process. These are decisions that need to be deliberated and resolved democratically. When we do this, the normative decisions are out in the open, and it is clear that they are based on subjective preferences.

So suppose a group of people decide that they need a way to value their activity. They sit down and make the decision democratically. Everyone agrees that it is appropriate to equate relative prices with productive value. And everyone agrees that they want a measure that is closest to the moving base-year price index. Voila, they have democratically decided to measure real GDP the way economists do. I have no problem with this, since the decisions are out in the open. And they are made politically ... they do not mascaraed as science.

This is very different from the way we do things now. In our world, the government has decided to value the economy, but has outsourced the task to a small cabal of economists. These economists hold ideologically charged views, but often do not recognize them as such. Based on their (dubious) theory of income, economists equate price with productivity. They then use their theories to make a series of choices that define how real GDP will be measured. These decisions are not debated among the general public, or advertised openly. Instead, they are mostly hidden in math. The official measure is published, and most people interpret it as an objective measure of output.

And we can see why. The BEA calls real GDP a "quantity index of output". This is extremely misleading. They should call it an "index of the value of the economy given the assumption that prices reflect utility, and that our measure should be closest to the moving base-year average". Then they need to add the caveat: "here are all the assumptions we used to calculate this metric. If you do not agree with them, then the measure has no meaning for you". Do you see how this would lead most people to be skeptical of real GDP? They would see a host of political decisions that were never debated democratically.

The alternative is to separate science and politics. As scientists, we have no business making normative decisions. These are tasks for democracy. As scientists, our job is to be objective. This means being agnostic about methods. Unless we have compelling scientific reasons for a certain aggregation choice, we need to investigate the consequences of ALL POSSIBLE CHOICES. I call this possibility space "uncertainty". If you don't like that word, call it something else.

The most objective metrics are the ones where the possibility space is collapsed for us by scientific laws. So take energy use. There are many conceivable ways we could weight different types of energy against one another. But the laws of thermodynamics tell us that all but one are WRONG. This is what I love about science. It tells us that certain choices cannot be considered. You are free to believe that coal and natural gas have the same energy content per unit of mass. But you are WRONG.

The science of energetics collapses the possibility space of our measurement. As a result, the aggregate measure of energy has very little "uncertainty". But there is still some. If we are measuring primary energy, we need to make choices about renewable energy. Renewables have no primary (fossil fuel) energy component, so we have to assign it. We usually do so by looking at the efficiency of thermo-electric power plants. We divide the electricity output of renewables by this efficiency, and impute a primary-energy equivalence for renewables. But like prices, thermo-electric efficiency changes over time. So what values do we use?

Our choices here open a possibility space for our measure. Thankfully, renewables are a marginal source of energy. So when we consider all possible aggregation methods, the range of possibility is still small. This is not the case with real GDP. Here, the possibility space is enormous.

My view is that science should be objective. This means that, as scientists, we do not get to choose the method that we "like". This is left for our role as citizens. If no laws of nature can be used to eliminate decisions, scientists need to calculate the possibility space of all possible methods. From this space, democratic institutions are free to choose what they like. But as scientists, we recognize the arbitrary nature of these decisions. As scientists, we recognize that the most objective measures have the smallest possibility space. The task of science is to find laws and principles that collapse this space, leaving a single objective measure. The task of politics is to make value-informed choices.

See the difference?
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby bfix » Sun Mar 31, 2019 1:07 pm

Economist's Message 7
**************************

Thanks.

I spent a long time breaking this down and have no plan whatsoever to read this in depth until you point out exactly where along the numbered line of reasoning we depart.


My Response
***************
This is a disappointing response, as I think my previous email outlines exactly why I disagree with you. I think the usefulness of our correspondence has come to an end. Nonetheless, I will give you one more assessment of what is going on.

You have adopted a normative framework and used it to create a series of goals. These goals inform decisions about aggregation that logically lead to your chosen measure for real GDP. Every single decision from 9 onward is based on your normative vision. I appreciate that you have made these decisions explicit. But this does not mean I have to engage with them.

You see, the problem is that you are only willing to accept arguments from within your normative framework. You want me to accept your normative vision and then debate the logic of your choices from within. But I am under no obligation to do this. Instead, I am REJECTING YOUR NORMATIVE VISION. This means that your chain of argumentation is nullified. I do not accept the goals that you are proposing or the philosophy on which these goals are based.

To try to make this explicit, you are trying to get a heretic to debate the interpretation of the holy trinity from within church dogma. But this doesn't work. The heretic rejects the dogma. The heretic thinks there is no god. If the church admits that its dogma is normative (as you have done), then it has no way to rebut the heretic. There is only a stake and a fire.

This is why I find cognitive dissonance in your arguments. You admit that your reasoning is based on a normative framework. But you insist that we engage in arguments from within this framework. You don't seem to realize that when you admit something is normative, you must admit that the whole framework can be rejected. That is what I am doing. I am rejecting your whole framework.

If you had insisted that your framework was scientific, then I would be happy to debate its merits empirically. But because you have admitted the normative nature of your claims, I don't have to debate your logic. I simply reject your normative assertions as political ideology that I don't agree with.

Since you seem unable or unwilling to accept this reality, I see no point in continuing this discussion. It has given me a glimpse into how economists think, and has been fruitful in helping me understand how dogmas are perpetuated. I hope that at some point in the future you are open-minded enough to consider my arguments seriously.
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Re: Debating GDP Dogma With An Economist

Postby bfix » Sun Mar 31, 2019 1:17 pm

Economist's Message 8
***************************
Whatever. If you can’t be bothered to fulfill such a modest request, then forget it. It’s just stupid for me to try and read anything you say if you insist on responding to such an amazingly simple reply by ignoring utterly my query.

I’ll just stick you in the “cranks” folder.

Thanks.


My Response
****************
Since the economist had resorted to ad hominem attacks and seemed to have no intention of engaging with my arguments, I choose not to respond to this last message.
User avatar
bfix
 
Posts: 14
Joined: Mon Dec 29, 2014 11:03 am

Next

Return to Political Economy

Who is online

Users browsing this forum: No registered users and 2 guests

cron