Is US manufacturing turning inward? An 8-chart answer.

Factful Friday by Richard Baldwin @IMD, 26 April 2024

Introduction.

Are American manufacturers turning inward? And what does ‘inward’ even mean? There are two natural ways of thinking about turning inward: the sales side and the sourcing side.

• Turning inward on the sales side means relying increasingly on domestic customers for your sales.

A classic way of measuring this is to look at the share of manufacturing output that is exported. If the US share falls, the US is turning inward.

• On the sourcing side, turning inward means relying less on imported inputs.

A classic way of measuring this is to look at how much foreign value added there is in US exports. If the foreign content share falls, the country is relying more on its own inputs, thus, in this sense, turning inward.

Turning inward: The sales side.

The charts below show that the US manufacturers are turning inward on the sales side, but it was not always so. From 2012 to 2012, manufacturing exports accounted for an increasing share of US manufacturing production (left chart). This was when US manufacturing was opening to the world. But since 2012, the share of US production sold to foreigners has fallen from 15% to 12%. As shares have to add to 100%, we have half of the answer to the question in the title:

• US-based manufacturing firms were increasingly relying on US-based customers since 2012.

A pair of pearls for your pondering: First, the date is important. Jot 2012 down in your notebook. 2012 is WAY before geopolitical tensions were all the buzz. That was the year that Obama started his second term.

Second, the level seems too low, no? Is it really true that 88% of US manufacturing production is sold domestically? What happened to: “The World Is Flat” and the “Death of Distance”? Well, as it turns out, those memes were complete BS. The fact is that mega-economies like the US, the EU, and China are quite closed. Distance is a huge inhibitor of sales, so countries like the US, which have lots of local customers, tend to sell local. I even wrote a whole Factful Friday looking at whether the world was closer to autarky or free trade and concluded that it was much closer to no trade.

The globalisation ratio on a value-added basis.

The right chart above shows the same flows but measured on a value-added basis instead of a gross basis. Remember the difference. Gross exports are the dollar value of exports as they cross the border. In other words, ‘gross exports’ are what most people – including customs officers – call ‘exports.’ We trade economists occasionally add the ‘gross’ when we are about to speak of ‘value added exports.’ And that is the case right now.

The dollar value of exports reflects their cost of production with any profit margin relabelled into a cost (cost of capital). The cost of production comprises the value-add cost – which means the payments made to primary factors of production like labour and capital – and the inputs purchased by the manufacturer. Most of the inputs will be bought within the US and thus ultimately reflect US value added, but some of them are imported. In extreme cases, the imported inputs can seriously distort the picture. Singapore’s gross export to GDP ratio, for instance, is over 300%. Value added exports are just exports with the cost of imported intermediate inputs stripped out.

In short, value-added exports are gross exports minus the cost of imported inputs. In the right panel, the denominator is the manufacturing output on a value-added basis. This is as it should be. Both top and bottom are measured on a value-added basis. Note that another common name for production on a value-added basis is GDP.

So, pulling it all together, the left chart is exports and production both measured on a gross basis. The right chart is the same but with both measured on a value-added basis. I call these the value-added globalisation ratio (VGR) and gross globalisation ratio (GGR).

Having finally got past the definition, consider what the right chart tells us about the question in the title. The time pattern of the VGR is quite similar to the GGR, but its level is higher. The higher level indicates, logically, that US manufacturing is more exposed to exports as a share of value added than as a share of gross production. This, in turn, must mean that intermediate goods are less important in US exports than they are in US production. In other words, US is exporting manufactured goods that have a high value-added content relative to the average US manufactured good. This is an interesting fact per se, but I don’t think it is important for answering the question in the title (but let me know if you can think of something I missed).

Trade to GDP: If you insist.

There are surely some readers who would like to see the mixed ratio since it is so often used, namely the trade to GDP ratio. This mixed ratio has gross exports in the top and net production (value added) in the bottom. The mixed ratio mixes up things and so it is not the right ratio to use. Nevertheless, it provides approximately the same answer. See Annex 2 for a discussion of why we should not use it when we have the possibility of using the GGR or VGR, and for an illustration of the point that the answer is the same for the GGR, the VGR, and the mixed ratio.

Globalisation ratio by manufacturing subsector.

The charts above are all for the US manufacturing sector as a whole and thus may be hiding some important things like a changing composition of the export basket.

The next pair of charts shows the answer is the same for all the subsectors. Both panels display the exports to production ratios (GGRs) for each manufacturing sector. The numerator is the sector’s exports while the denominator is the sector's production. The left panel focuses on the most open sectors, the right panel on the least open sectors. They are separated to reduce clutter. The aggregate GGR is included in both as a reference point. This aggregate line is the same as the one in the left chart above.

What we see is that the peak exposure to foreign markets occurred at about the same time – around 2012 – for all the sectors. The only standout sector is Chemical where the sector did not turn inward, but also did not continue to rely more heavily on foreign sales.

Turning inward: The sourcing side.

?The answer on the supply side is also pretty clear-cut: Yes, US-based manufacturers are turning inward.

In today’s world, no manufactured good is made fully in just one country (Antras 2015, Baldwin, and Lopez-Gonzalez 2013). Companies buy inputs ranging from raw materials like iron ore and phosphate to highly specialized components like semiconductors and ball bearings that are made by very few firms in the world.

For large economies like the US, most of this sourcing is done domestically since the local economy offers most of what is needed. But some industrial inputs are imported and the question at hand is: Are US-based manufacturers buying more or less foreign inputs?

The left panel below shows that the foreign value-added content of US exports has been falling since 2012. This is a sign that US manufacturing is relying less on foreign inputs and this, in turn, is a sign of the sector turning inward.

During the phase of globalisation that I describe as the ‘second unbundling’ in my earlier work (Baldwin 2006, 2016), the reliance on foreign inputs rose from 12% to about 19% in 2012 (with a spike during the 2009 Great Trade Collapse). This was driven by US-based and other G7-based firms offshoring some of their stages of production along with the necessary knowhow. As the parts were made with low wages and high technology, this offshoring was attractive economically and it proceeded at a roaring pace. But for reasons that are not fully clear, the trend reversed in the 2010s. The US reliance is now down to 13.5% just a bit above where it started in 1995.

One reason for the increased domestic value-added content of US exports is the rather spectacular development of US oil and gas production. As the right panel shows, the US’s sector balance of trade in mining and fuels shifted from a deficit of about 1% of gross production to near balance. This replacement of foreign energy with domestic energy clearly boosted the domestic content of most US manufactured goods.

But what about the manufacturing subsectors? Is the story the same for all of them?

As I did for the sell-side, the two charts below illustrate how the basic pattern from the aggregate numbers holds in most of the subsectors. The sectors all became more reliant on foreign suppliers and then less reliant. The only sector that is distinguished by its difference is the Etron sector, which is short for ‘Computer, electronic and electrical equipment’. Its foreign reliance ratio did not rise until the early 2010s and then fell faster than most after 2012. This may have more to do with the selection of which US electronic goods continued to be exported than anything to do with, say, the sectors’ trade policy or technical advances.

Summary and concluding remarks.

I guess I should have titled this week’s Factful Friday “Yes US manufacturing is turning inward since 2012 and on both the sales side and on the supply side.” But I had done that, tell me honestly: Would you have read beyond the headline? So as not to deprive you of the deep insights you gained from having slogged through the 8 charts, I put a click-bait question in the title. Mea culpa.

So, what are the insights beyond the fact that US manufacturing is turning inward? Here are my takeaways:

• Whatever it is that is driving the inward trend, it happened long before Trump started his war on trade.

The war started for real in 2018 – as Chad Bown and Melina Kolb of the @PIIE have demonstrated so well.

Faithful readers of Factful Friday will know that I think a lot of the trends currently ascribed to geopolitical strive are, in fact, part of technologically driven, long-term trends. Of course, President Trump’s war on trade – and President Biden’s refinement and refocusing on China – did have some impact on deglobalisation, but not as much as most think.

Look back at the first pair of charts and remember that Trump started his trade war for real in 2018. The opening salvo was a blast of tariffs on friends and challengers alike. And they all retaliated almost immediately (Bown and Kolb 2023). The impact in the left chart is hardly noticeable between 2017 and 2018, and hardly massive between 2018 and the year the tariff exchange culminated, namely 2019. The US-China Phase I trade deal, which was signed in January 2019, called a ceasefire on the tariff retaliation cycle. But in fact, there was a big drop between 2019 and 2020. That, however, is related to Covid trade disruptions, not geopolitics.

So what?

In closing, and at the risk of redundantly repeating myself:

• The trends documented here started many years before Trump triggered the geopolitical conflicts that we read about every day.

Without denying that the political climate for trade is as bad as it has been in decades, it is important not to conflate long-term trends driven by technology, and the political thrashings that have arisen since the former hegemon has woken up to its declining status.

US manufacturers are indeed turning inward, but not due mostly to geopolitical strife.

References.

Antràs, P. (2015). Grossman-Hart goes global: Incomplete contracts, property rights, and the international organization of production. Journal of International Economics, 95(2), 108-120. https://doi.org/10.1016/j.jinteco.2014.11.005.

Baldwin, R. (2006), “Globalisation: The great unbundling(s)“,?Economic Council of Finland?20(3): 5-47.

Baldwin, R. (2006), “Industrial production processes are defragmenting worldwide: A six-chart portrait,” Factful Friday, 26 January 2024, Linkedin post.

Baldwin, R. (2016). The Great Convergence: Information Technology and the New Globalisation. Harvard University Press.

Baldwin, R. (2024). Is the world closer to the free trade or no trade? Factful Friday, Linkedin post, 12 January 2024. https://www.dhirubhai.net/pulse/world-closer-free-trade-richard-baldwin-rpkxc/

Baldwin, R. and Lopez-Gonzalez, J. (2013). "Supply-Chain Trade: A Portrait of Global Patterns and Several Testable Hypotheses." NBER Working Paper No. 18957, April 2013, DOI: 10.3386/w18957 https://www.nber.org/system/files/working_papers/w18957/w18957.pdf

Baldwin, R. editor (2009). The Great Trade Collapse: Causes, Consequences, and Prospects, a VoxEU.org eBook. https://cepr.org/system/files/publication-files/68568-the_great_trade_collapse_causes_consequences_and_prospects.pdf

Bown, Chad and Melina Kolb (2023). “Trump’s Trade War Timeline: An Up-to-Date Guide,” Peterson Institute for International Economics, https://www.piie.com/blogs/trade-and-investment-policy-watch/2018/trumps-trade-war-timeline-date-guide

Annex 1: Sector categories.

?

Annex 2: Why Trade to GDP is the wrong ratio to look at.

Surely the leading contender in the globalisation-ratio race is the trade-to-GDP ratio. It is so widely used that many articles don’t even bother to define it clearly. They just introduce it as the openness ratio without specifying which types of exports are in the numerator (trade in goods, trade in goods and services, only manufactured exports?)

This mixed ratio, the trade-to-GDP ratio, is flawed in many ways, but the one I focus on today is the mixing flaw. Among the most technically skilled analysts, this is known as the apples-and-oranges issue.

The numerator is gross exports (which includes the double counting that arises from the fact that the value of intermediate goods can cross borders multiple times). Double counting be damned, gross exports are a good starting point since everyone knows what they are.? But the excellent starting point is besmirched by dividing gross exports by net production (i.e. GDP which is a value-added, or net, measure that strips out the production of intermediate goods).

This apples-and-oranges thing is especially problematic since the offshoring boom that started in the late 1980s drove up gross manufactured exports much faster than GDP exactly due to the double counting. That is, the difference between net and gross got more important when the ratio was rising. As a result, the classic trade-to-GDP ratio became increasingly distorted during the exact time frame when it was most interesting to peak-hunters.

The charts below make the point graphically. The left chart shows the three ratios together. The salient takeaways are:

• All three ratios peaked around 2012 and show the reliance on foreign markets has been falling ever since.

This is reassuring in some sense. Regardless of how we measure things, US manufacturers are turning inward on the sales side.

• The level of the apples-and-oranges ratio is much higher than the other two.

This is to be expected since the trade-to-GDP ratio strips out double counting from the bottom but not the top.

• The apples-and-oranges ratio rises much faster from 2000 to 2012 than the GGR and VGR and falls much faster after 2012.

This reflects the changing importance of intermediates in manufacturing. As the right chart shows, the ratio of manufacturing output measured on a value-added basis and on a gross basis is anything but flat. It falls during the 1995-2011 phase. This is when Information and Communications Technology (ICT) was allowing manufacturing processes to be fragmented to save costs. The result was that intermediate goods occupied a growing share of the cost. After 2011, the trend is reversed. This is part of defragmentation – a thing I pointed out in an earlier Factful Friday, titled “Industrial production processes are defragmenting worldwide: A six-chart portrait”.

In case you didn’t pick up on this yet, let me state the implications. Dividing gross exports by net production sounds awkward (and it is), but its real flaw is that it distorts the picture exactly during the time frame we are most interested in. ?