Wednesday, October 31, 2018

Scratch That

I made a mistake. And it's a good thing.

Following up on my Running on Empty post, I wanted to give a more finely-grained analysis of climate costs relative to GDP growth, so I returned to my sources to see who their sources were and how they did their calculations. Watson et al., compiled their estimate from National Centers for Environmental Information, "U.S. Billion Dollar Weather and Climate Disasters" and Paulina Jaramillo and Nicholas Muller, "Air pollution emissions and damages from energy production in the U.S. 2002-2011."

Jacobson, Delucchi et al. presented their estimates in two tables, broken down by country. I had taken what I thought was 15% of their global climate cost estimate but it turns out that my number was roughly double theirs. I may have taken their high estimate instead of the mean. Anyway their actual estimate was around $4 trillion in 2013 dollars. But that is not the mistake I am concerned with here.

The mistake came to light after I had all the carefully checked numbers, recalculated in 2012 dollars and I compared annual GDP growth to annual health and climate costs. Climate costs exceeded growth in 13 of the 28 years from 1990 to 2017. Over the entire period those costs offset 95% of reported real GDP growth. Or so the numbers claimed.

Then I tried a different tack. I subtracted estimated annual climate costs from GDP and then calculated annual growth. To my surprise the resulting annual growth was only somewhat lower. O.K., I thought, I'm starting from a much lower total in the base year and the year-to-year number reflects only the the annual increase in climate costs and not the annual cost.

So which calculation, then, is the "right" one? Year-to-year growth minus annual cost or Year-to-year growth of (GDP minus climate cost)? Neither! Both calculations rely on double counting of the sort that Irving Fisher warned about 112 years ago.

The problem is also illustrated by the relationship between GDP and Net Domestic Product. Annual growth of Net Domestic Product and Gross Domestic Product track each other pretty closely:

Meanwhile, look at what has happened to depreciation. It has nearly doubled as a percentage of GDP, from around 9% in the late 1940s to almost 17% in recent years.

Something that doesn't show up too well in the graph above is that most of the growth in depreciation relative to GDP has occurred in the last 48 years. So, doubling every half century we could have an economy in 125 years that runs entirely on depreciation! But that would be O.K. because GDP would be so much BIGGER then. Isn't that right, Professor Nordhaus? Professor Solow? Professor Krugman?

I don't have the solution to this computational problem, other than to point out that it is the inevitable consequence of a poorly-conceived metaphor for the "measurement" of heterogeneous good and services. The monetary yardstick is made of a highly elastic material and correction for changes in the cost of a basket of consumer goods (CPI) does not begin to address the real issue. GDP and GDP growth has become the increasingly opaque lens through which we view society and "the economy." It is a cracked, scratched, smudged, distorting lens that may not even enable us to tell whether what we view through it is upside up or upside down.


spencer said...

I was aware that depreciation on the new high tech equipment was much more tan for traditional capital equipment and a major reason that net capital formation was much weaker now than historically.

But I did not think depreciation was as large as you have here.

I found this much lower level of capital stock growth was a major reason growth of the net real capital sstock and productivity was slowing on a long term basis

Sandwichman said...

On the one hand, a lower level of capital stock growth...

On the other hand, an explosion of household net worth...

Decoupling the stock from the flow is a weird trick.

Blissex said...

Dear Sandwichman welcome to the branch of Rabbit Hole Studies that is national statistics. :-)

There are many issues with using "GDP"... GDP as originally and properly defined was a vector of physical quantities. Instead both nominal GDP and "real" GDP are indices computed from that vector and a vector of prices, plus in the USA a vector of "hedonic" adjustments.
However even in the actual GDP as vector of physical quantities there is a lot of double counting; for example most of education or research is not final output, because most of it is done to increase production, so either it does increase production, and then it is an intermediate cost, or it does not, and then it cannot be added to GDP regardless.

Another problem is that often people talk GDP, but actually mean GDI, and a large part of GDI arises from rentierism rather than production, something that causes complications in reconciling GDP to GDI.

Blissex said...

«Economists know also that their proper business is political agenda pushing and NOT science. So, they deliberately try to keep everything in the swamp of vagueness and inconclusiveness ... National Statistics is NOT a Rabbit Hole but has been deliberately made a jib-door through which failed/fake economists vanish when their proto-scientific fool’s gold is put to the acid test.»

Dear E.K-H, unfortunately I very strongly disagree that studies of the political economy can be a "science" as in proper accuracy and falsification, in large part because experiments are difficult and repeatability is pretty rare, and never mind with the measurement problems, because the "Laws of Economics" are not immutable, unless they are kept in “the swamp of vagueness and inconclusiveness”.

Rather than calling it a "social science", which to me seems "scientism", I call studies of the political economy a "discipline", that is a systematic, but somewhat fuzzy and time-dependent, analysis and summary.

Here I suspect that it would be better to write "with precision of two decimal places" rather than “accuracy”:

«Note that income, monetary profit, aggregate demand, GDP can, in principle, be defined and measured with the accuracy of two decimal places»

I reckon that even in the best cases concepts like GDP (or "capital" or "profits") have only imprecise boundaries, in particular but not only because they are aggregates of incommensurables; they make mostly suggestive sense, as long as we realize it.
The national accounts are a classic triumph of hope over experience, even when they are not very cleverly (and often honestly!) biased to deliver the right "optics"...

That unfortunately means that your approach that relies on abstractions like “income, monetary profit, aggregate demand, GDP” seems overambitious to me.

Sandwichman said...

Not being an economist, your rants glide off me with indifference, Egmont. I am actually a literary critic by methodology. One who studies the discourse of economics, as discourse, and who is troubled by the consequences of evident pathology in that discourse.

I would be inclined to agree that economics as practiced is not science but would further agree with blissex view that economics can't be "scientific" in any rigorous way. The proliferation of superfluous mathiness is clear evidence of chronic anxieties about not being a science and an attempt to mask the unscientific premises and assumptions with a language that the layperson associates with science.

Sadly, Egmont, your incessant carping about "science" strikes me as the same sort of hyperventilated disavowal.

Sandwichman said...

"...with all their peer-reviewed articles/textbooks/blogposts have finally to be flushed down the scientific toilet."

O.K. Kakarat, your comments getting flushed. I only tolerate as long as you behave yourself.