Also - when did accounting for inflation become a ‘trick’? No way you can look at nominal gdp on these time frames.

> Real gross domestic product is GDP adjusted for inflation, which allows us to better measure and compare growth over time. If you're measuring and comparing different industries' shares of GDP over time, it's not immediately obvious why you would need to do such an inflation adjustment. The very act of dividing manufacturing's contribution to GDP in 1960 by overall GDP in 1960 does the inflation adjustment for you, given that both amounts are in 1960 dollars.

> The BEA's remedy to the problem is to put up warnings against doing share-of-real-GDP calculations all over its website.

It's worth reading the whole thing.

The manufacturing/services divide itself is a bit silly without careful attention, and I think is incredibly overplayed because "manufacturing" aligns with a bunch of other political/economic geography stuff.

This is true everywhere, not just in computing where it's most obvious. Even agriculture in a lot of ways.

By the time you apply all that careful attention, any sort of meaningful macro work becomes impossible. You're just doing a whole ton of little sector-by-sector analyses.

And it gets worse if you start looking at the labor economics instead of just measuring output. It used to be that manufacturing-related services and the physical work of manufacturing were relatively coupled. The people working in the manufacturing plants more-or-less knew how all the equipment worked. That's not really at all the case anymore.

I believe that was the point of the "share-of-real-GDP" adjustments. Because "prices of manufactured goods have gone up more slowly than those of other goods and services (think health care), Baily and Bosworth reasoned that manufacturing's shrinking share of nominal GDP understates its continuing impact on the economy."

Apparently there is a decent way, at least in principle, to compare mainframes to iPhones; that is, to compare their contributions to improved productivity independent of nominal prices. But it breaks down over longer time scales unless you can plugin real numbers for changes in relative value-add across entire industries, which in this case isn't possible because the trick was used to adjust industries in the aggregate by assuming a constant ratio of value-add.

Jobs, the economy... there is more to life, and more to maintaining a hegemony.

NB: there's also a strategic geopolitical advantage to out-sourcing manufacturing. The US would be a weaker nation if we had to commit a bunch of human capital to textiles.

But to answer your actual question:

> Jobs, the economy... there is more to life, and more to maintaining a hegemony.

You could say the same about freedom. Most people would prefer a good-paying job over global hegemony, and care about the latter only in service to the former.

note: I agree there's a strategic national security/economic/sov. interest in domestic chip manufacturing, hegemony or not. But the general analysis that we should make ourselves poorer in order to be world hegemon is, to most Americans, I suspect a bit backward.

It's not hegemony for the sake of hegemony, but hegemony for the benefits of hegemony, which in our recent history have included technological and economic supremacy, both tightly correlated to one another.

I think globalism can be a good thing, and I hardly oppose the notion of outsourcing some economic activities. However, the original point was regarding ceding our production capabilities for a short term and potentially insignificant improvement in the performance of a single company.

> The US would be a weaker nation if we had to commit a bunch of human capital to textiles.

There is something to be said for balance, taste, and sensibility. Like, it would be cool if we were able to manufacture essential medical PPE during a pandemic. Them be textiles, my friend.

If we are unprepared for a crisis that can gut our economy because we were prioritizing the economy over preparedness... I'm sorry, but that is stupid and short sighted.

Jobs are important. The economy is important. But it's as if all nuance is lost on the GDP-measuring mouth breathers. You want a strong economy? Maintain the hegemony. Be a global leader in many diverse categories.

The post you're relying to literally says the same thing...

> medical PPE be textiles, my friend.

1. No. "Textiles" is a term of art, and it doesn't include N95 masks or 3 ply surgical masks. Full stop. Might include some basic cloth masks that are equivalent to a ripped-off tshirt sleeve with some elastic bands. But we never had anything approaching a shortage of t-shirts to rip up for cloth face coverings... what we're missing is the stuff with melt-blown layers and/or filters.

2. We do make N95 masks and could have surged production... shortages of the most needed types of masks in the US were due to an utter failure of political leadership, not a lack of manufacturing capacity. E.g., https://www.washingtonpost.com/investigations/in-the-early-d...

3. The same is true for 3-ply surgical masks. Repurposing a huge number of t-shirt factories would've taken ages and $$$. The difficult part is the melt-blown layer, which is more like N95 or surgical gowns than t-shirts. Not even clear having an existing t-shirt factory would even speed things up much, since, obviously, the barrier isn't "stitching a couple pieces of cloth together". But even if it did, what are the odds on that happening if we couldn't even give a green light on flipping the switch on existing N95 lines in May? Again, leadership problem, not manufacturing capacity problem.

4. All of that aside, this is the sort of thing you solve with strategic stockpiles and planning, not "make everything ourselves and hope to god some of the other lines can be repurposed ASAP" and especially not "constantly maintain manufacturing capacity and raw resource supply chains for every conceivable long-tail event". No. You figure out for each long tail event how long it takes to ramp up production, and then you stockpile for that amount of time.

The rest of the post is sort of _exactly_ demonstrative of the point I was making in my original post about macro analyses being limited and needing to go sector by sector. So I'm really not quite sure what we're arguing about here.

While I probably agree with the broader idea, I can't help but think that at root, sentiments such as these are the root cause of our industrialization and significant but hand-waved human costs here that have been incurred.

Scenario A: You assemble everything locally with 50% of your components from east asian component manufacturers, the rest from the domestic market. Your costs are $800 per unit. Then you spend $200 on in-house design and sell your product for $1500, for a profit of $500 per unit and a domestic value add of $700 per unit. At that price point, you sell 2/3 of a million units, adding roughly half a billion to domestic manufacturing.

Scenario B: You outsource all of your assembly to China and parts from East Asian component manufacturers and only do design/QA in-house. If you pay $300 to your suppliers, spend $200 on in-house design but sell your product for $1000, you have $700 in domestic value add. Say at $1000 you sell a million units, so you've contributed a billion dollars to the manufacturing share of GDP.

Now if a company switches from Scenario A to Scenario B, most people would (rightly, I believe) assume that some manufacturing has left the economy. But actually the manufacturing share of GDP has increased! However it's a very fragile increase, because a change in the terms of trade and suddenly you pay more for your foreign inputs and at that point it will appear as if manufacturing's share of GDP has fallen. There is also the issue that foreign rivals are going to be in a better position to compete against you, so long term this does not bode well for manufacturing's share of GDP.

Now this is just one company, and buying from domestic suppliers also contributes to GDP. But they have the same issue -- they can outsource production leaving design/QA in house as well. So you can have a situation with high manufacturing share of GDP with nothing produced in the nation other than design/QA. So all these GDP by industry measures, at the end of the day, are measuring industry margins more than actual manufacturing per se, and really weren't designed with outsourcing in mind, and aren't intended to be used in a time series analysis during which a lot of outsourcing is happening.

In most situations you end up using price as an aggregator, even if the price is itself normalized by a scaling factor. That means if you shut down a factory today and have your goods produced cheaply overseas except for a few small finishing steps here, it will be the relative prices of the inputs that determine how much of the good was produced here versus how much was produced overseas, and this wont generally reflect the "real" amount produced here versus overseas as viewed by a layman. In fact, if a producer couldn't outsource cheaply, they probably wouldn't do it, so there will be a consistent gap in terms of what the laymen sees and what the BEA data will show. And note this also creates the appearance of a productivity boost. Imagine Apple firing all their manufacturing employees and employing only a few designers. Now, they can produce a lot of value with only a few cost inputs, hence productivity just went up. Of course, the designers may not be more productive in terms of creating more designs or better designs, but as Apple is able to capture a large share of the total value, then it will be attributed to the super productivity of the designers, which may have more to do with politics in East Asia choosing an export led growth model than anything to do with the iPhone.

And it gets more complicated because the NIPA system doesn't take into account the possibility of economic rents. E.g. going back to the Apple example, if Apple can charge $1000 for the phone, with only $300 of inputs, then this $700 worth of "value added" must be attributable to some portion of labor and capital. If the designers are paid $200 then capital is paid $500. In NIPA parlance, that means $500 worth of capital investment (and or depreciation) must have occurred somewhere in the economy, which again to the layman would require unrealistic measurements of investment. There is no NIPA notion of someone overpaying. And then don't get me started on how corporate shell companies headquartered in tax shelters like the Bahamas or Ireland screw with GDP numbers, especially of the smaller nations that serve as tax shelters, but also with US productivity and investment data.

> a lot of the manufacturing output growth since the mid-1980s, when the BEA started factoring quality improvements in computers into its measures of real output, has been driven by the rising quality of computers and electronic products, not by increases in actual, you know, output.

i.e. one Intel chip produced today is considered to be thousands of times more output than one Intel chip produced in the 80s, because it's thousands of times faster.

This point is a bit of a rorschach... is that an over-estimation or an under-estimation? Sounds like a massive under-estimation to me, in the same sort of qualitative way that it's silly to try to quantify the economic effect of the internal combustion engine in terms of horses displaced... it's measuring something kinda relevant but totally missing a qualitative shift that unlocked impossible-to-quantify economic activity (for better or worse is up to us).

Yes, exactly. My point is that I wouldn't call measuring actual value instead of a pretty meaningless "output" metric a "mistake".

The original point of the pre-existing "output" metric was to approximate actual value creation, after all...

If we treat each computer made in the US today as equivalent to a thousand 1980s computers, then I'd hope to see the weighted data show that US manufacturing has massively increased as a result of making an exponentially larger number of exponentially faster computers. Rather than seeing it stagnate despite weighting each computer as worth so many old computers.

Why would you hope for that?

After that story, they’ve lost a good deal of credibility.

A favorite story on HN, this listening device was not detected for seven years, despite experts extensively looking for it [0]

[0] - https://en.wikipedia.org/wiki/The_Thing_(listening_device)

As a side note, we tend to judge an article if it goes against our belief, but are willing to give it a pass otherwise. Case in point, plenty of Bloomberg's articles about Boeing or climate change being discussed but don't think I've seen anyone questioning their premise or evidence. I'm guilty of this as well.

They haven't; they still stand by it. Since they consider the rest of their output just as trustworthy as that article, I'll take them at their word.

In the end they turned out to be right but it took more than 5 years since the first article. I believe we should consider that Bloomberg could fall into a similar category.