Brad Setser

Follow the Money

Cross border flows, with a bit of macroeconomics

Print Print Cite Cite
Style: MLA APA Chicago Close

loading...

Why Doesn’t Apple Export More Services (Wonky)

by Brad Setser
December 15, 2016

Share

A big caveat. Apple is being used to represent a lot of companies with very valuable intellectual property (IP) that have built up similarly outsized piles of cash outside the United States; it is the most high-profile case, but it is hardly unique.

The iPhone, famously, is designed in California and is assembled in China out of parts manufactured (mostly) in Asia.

I do not think the recent coverage about what it would take for Apple to manufacture the iPhone in the United States has added much to Keith Bradsher and Charles Duhigg’s spectacular reporting back in 2012. And, to my mind at least, the big revelation in the Bradsher/Duhigg article was that even the high-end components of an iPhone, and even those high-end components that come from American companies, typically aren’t made in America. Corning now manufactures its gorilla glass mostly in Asia, the chip designers farm out production to Asian fabs, etc.

I suspect not much has changed since then.

I digress.

No one questions that the iPhone is designed in the United States. And a lot of the software that makes an iPhone an iPhone is also created in the United States. And the export of intellectual property rights—the U.S. design and engineering embedded in a “designed in California” iPhone sold in Asia or Europe—should in theory enter into the balance of payments as a services export.

I would think it should show up in the line item for the export of “charges for the use of intellectual property, computer software” though in practice it may enter as a payment for research and development services (see below). Like I said, this is a wonky post.

A while back I thought that maybe U.S. growing exports of software (in the services data) would offset the growing U.S. trade deficit in computer hardware and electronics. But I couldn’t find enough software exports in the balance of payments data. Software exports the past few years have been around $40 billion (see table 3.1 of the international transactions data; R&D exports in 2015 were $35 billion), far less than the U.S. spends on imported hardware.

Here is a chart showing the trade balance on software, R&D, and computer and telecommunications services (“software”) versus the trade balance on computers, cellphones, and telecommunications equipment (“hardware”).*

us-hardware-v-software-trade-balances

Drawing on Nick Confessore’s reporting, and some of the details about Apple’s Irish subsidiary that were in the news a while back—I now think I know why software exports aren’t a bit higher.

Consider the balance of payments impact of the following corporate structure, as reported by Confessore:

“Apple, for example, is an American company, headquartered in Cupertino, California. Most of the research and development that goes into an iPhone happens in California. But according to Apple, if you buy an iPhone in Europe or Asia, the intellectual-property rights contained in your phone actually belong to Apple subsidiaries in Ireland … Apple charges those subsidiaries relatively little for the rights to this intellectual property, yet allows them to collect most of the money Apple makes from selling the phone. In 2011 the Irish subsidiaries…collected two-thirds of Apples worldwide pretax income.”

The actual amounts Apple charges its very profitable subsidiary in Ireland were included in the Senate investigative committee report (the Levin report) on the topic.

“For instance, in 2011, roughly 40 percent of Apple’s worldwide sales occurred in the Americas, with the remaining 60 percent occurring offshore. That same year, Apple’s worldwide research and development costs totaled $2.4 billion. Apple Inc. and ASI [Apple Sales International] contributed to these shared expenses based on each entity’s percentage of worldwide sales. Apple Inc. paid 40 percent or $1.0 billion, while ASI paid the remaining 60 percent or $1.4 billion.”

The U.S. portion of $1.4 billion that Apple charges its Irish subsidiaries for the R&D that generates Apple’s valuable intellectual property rights is likely what enters into the balance of payments as a service export.

The profits earned by Apple’s Irish subsidiaries ($22 billion in 2011, according to the Senate Report) almost certainly enter the balance of payments as investment income. And since Apple Operations International and Apple Sales International do not formally repatriate the income they earn abroad back to the U.S. (to limit Apple’s U.S. corporate tax bill), the net result is also a rise in the reinvested earnings component of the income balance.

Reinvested earnings—income earned abroad and retained in the offshore subsidiary rather than remitted back as dividend payments to U.S. headquarters—now account for the bulk of the foreign direct investment income line in the balance of payments (take a look at Matt Klein’s post on this). Reinvested earnings account for $304 billion of the $413 billion US companies earned abroad in 2015—see table 4.2 of the balance of payments data.

As a result of this transfer pricing structure described in the Senate report, technically speaking, Apple doesn’t export very many “design” or “software” services. The relatively low price it charges to its offshore subsidiary reduces service exports in the balance of payments while inflating the income line in the balance of payments.

Apple here is really a metaphor. It is the most high profile case, but it is—judging from the size of reinvested earnings in the balance of payments and the cash balances various firms have built up abroad—far from unique.

Certainly the profits that U.S. firms report in low-tax jurisdictions—Ireland, the Netherlands, Luxembourg, the Caribbean, Singapore—are now large relative to U.S. exports of software and research and development services.**

us-fdi-v-software-ip-exports

There is a broader point here. Trade theory says that if the winners from globalization compensate the losers from globalization, everyone is better off. But I am not quite sure how that is supposed to happen if the winners are in some significant part able to structure their affairs so that a large share of their income is globally (almost) untaxed.

Appendix

* The Bureau of Economic Analysis only reports  computer software IP royalties back to 2006, so the pre-2006 numbers are estimates based on overall U.S. IP royalties.

** Obviously it isn’t just IT firms that generate large profits in these jurisdictions though they are equally clearly a significant part of the total. Ireland, the Netherlands, Luxembourg, Singapore and the Caribbean account for about half of all offshore profits that appear in the balance of payments data.

Here are graphs that breakout the different components of the services balance in telecommunications and computers, and the goods balance in the same sectors (showing the components that Cole Frank and I used to produce our estimate of the trade balance in hardware and software, we erred on the side of perhaps being too inclusive)

us-software-services-trade-balance-disagg

us-hardware-trade-balance-disagg

14 Comments

  • Posted by michael m

    Brad,
    I don’t know how familiar you are with the Trump/GOP border adjustable tax plan but the basic idea is that it would lower the corporate tax rate to 20% but any “costs” that a company imports would not be deductible from their revenue. Conversely, any revenue that is derived from an export will not be added to taxable income (I give an example at the bottom).
    Before reading your article I thought that Apple would be considered a large net exporter but I guess I was wrong.
    If this plan became law I imagine companies like Apple will be forced to change the structure of how they domicile their IP.

    *For example, under the current law, if Walmart buys a $100 microwave that was produced in China and sells it for $110 in the United States, Walmart is only taxed on their $10 profit and at the 35% corporate tax rate ($3.50 of tax). Under the proposed plan the $100 cost Walmart incurred to procure the Chinese made microwave would not eligible to offset their $110 of revenue. Therefore, Walmart would be taxed at the new 20% corporate tax rate on the entire $110 of revenue (or $22 of tax).

  • Posted by Brad Setser

    Like many I am scrambling to learn about the House Rs border adjustment plan. Apple clearly imports its phones, macs, etc so would pay the border adjustment there. If it located its IPR in the US, it wouldn’t pay corp tax on any revenues derived from exporting that if I understand the proposal correctly. So it wouldn’t need the Irish structure for internationally untaxed income, so long as other countries don’t introduce border adjustments of their own (having income from exports untaxed in US while US taxes imports = target for importers). At least that is my first pass. Am curious to see where the tech companies come down on this … Apple especially b/c of the tax bill it would face of its US sales which could come in part out of its US profit margin given the current markup and b/c it doesn’t currently pay much on its irish income but obviously has some risks there.

  • Posted by Savannah

    I am curious how this Trump plan will work! It will be an explosion on prices!

  • Posted by rn

    It will be easy to beat a system like the one proposed. Chinese supplier will sell the microwave to its US based subsidiary for 10 usd. US subsidiary will sell the microwave to walmart for 100 USD

  • Posted by Blissex

    «built up similarly outsized piles of cash outside the United States;»

    As implied in the latter part of your article, they are only *booked* for tax purposes outside the the USA. Since cash is fungible, that cash is actually invested where it is most profitable and safe, probably mostly in chinese plants and in USA asset speculation.

    «No one questions that the iPhone is designed in the United States.»

    Actually, as written on it, it is designed (by a tiny team) in California… As far as the other 49 states are concerned, Apple is *already* an offshore company. Same as Microsoft and many others.

    «And a lot of the software that makes an iPhone an iPhone is also created in the United States.»

    Only for now, and that involves a small number of programmers concentrated in one location. Plus all big IT companies in the USA have been very, very active opening software development branches in China, India, Philippines, Russia, and moving as much as possible software development to those offshore locations. Prior to this they have been importing large numbers of offshore software developers on temporary visas.

    It used to be that only rich countries could train software developers because the tools of the trade, mainframe computers, required large capital investments, outside the reach of poor countries. But that was decades ago: buying the tools of the trade today is very cheap. This is the result, from 12 years ago already:

    http://money.cnn.com/magazines/business2/business2_archive/2004/08/01/377380/

    The plan of those companies is to move all larger cost centers offshore and keep in the USA just corporate finance, legal, and an USA-only sales and marketing subsidiary; many IT (and non-IT) companies have already done that.

  • Posted by Blissex

    «moving as much as possible software development to those offshore locations.»

    Also venture capitalists in the USA routinely require that the business plans of IT startups that they fund include offshoring all software development. They are not at all interested in wasting their capital on funding the lifestyles of USA programmers.

  • Posted by Blissex

    «Here is a chart showing the trade balance on software, R&D, and computer and telecommunications services (“software”) versus the trade balance on computers, cellphones, and telecommunications equipment (“hardware”).*»

    It would be interesting to know whether the figures for hardware are “hedonically” improved. There have been reports that the “improvements” are very large, both to the GDP and the CPI index:

    host.madison.com/ct/article_71fad514-9caa-11de-9a00-001cc4c03286.html
    «Likewise, quality improvements – better cars rather than just more cars – account for much of the increase in GDP nowadays. But assessing quality improvements is difficult.»

    conversableeconomist.blogspot.co.uk/2014/01/larry-summers-who-always-has-something.html
    «Television sets at five stand out. That is obviously a reflection of a rather energetic hedonic effort by the Bureau of Labor Statistics.»

    globaleconomicanalysis.blogspot.co.uk/2005/05/grossly-distorted-procedures.html
    «The most current figure I have for hedonic adjustment to the GDP is 2.257 TRILLION dollars which is roughly 22% of the GDP.»

    http://www.bloomberg.com/view/articles/2016-10-19/no-u-s-manufacturing-isn-t-really-booming
    «Without adjusting for deflation, value added in computer and electronics manufacturing is up 45 percent since 1997. With the adjustments, it’s up 699 percent! What’s happening here is that the Bureau of Economic Analysis has been trying to account for vast improvements in the processing capacity and thus quality of computers, semiconductors and other electronics equipment.»

    If hardware exports and imports are not “hedonically improved” they are not consistent with GDP data; and if they are their difference (the balance) is also “improved”.

  • Posted by Brad Setser

    the data we used was all in nominal dollars — no adjustments. We didn’t compare with GDP, but it is completely fine to do exports v imports as a share of nominal GDP. We aren’t assessing how much processing power was manufactured in America. Rather we are looking at how much the US paid for imports compared to how much it received in exports.

  • Posted by Alex

    A lot of Apple A-series CPUs are fabbed by Samsung’s Austin, TX plant. IIRC some TSMC ones are fabbed in the US too. The CPU is probably the biggest single contributor to the total BoM although the radio baseband chip will be close. So you really can’t treat it as weightless. Designed in California, made in China was much more true of the first two generations of iPhones, which were almost entirely made of proprietary parts. Since the 3GS, though, the fraction of Apple in-house or at least in-network and US value added has gone up as production scaled up.

    On the other hand, the NAND Flash storage is pretty much a commodity and usually comes from either Toshiba or Samsung export. But Apple sells it dear; the increments from 16GB through 32GB to 64GB sell about $100 each and the additional Flash sells for less than $10 wholesale, so that’s a massive contribution of gross margin and hence US net exports, especially when you think Apple will get a big volume discount.

    Also, Apple owns a lot of tooling in the supply chain and even develops it in-house. They also tend to place very large orders in advance and prepay for them, thus using their (offshore!) cash pile to finance their vendors’ tooling up. Evidently they do this because it suits them; one reason is strategic advantage (eg competitors can’t buy this machine tool because Apple has tied up the whole supply from the one manufacturer in Germany) but another is surely that the vendor offered $435m in upfront cash makes it worth their while on price. I’m not quite sure how you’d model that macroeconomically, but it might look like lower import value into the US with a matching capital export…not necessarily from the US, because offshore cash.

  • Posted by Brad Setser

    Alex. Thanks for the rich detail. Didn’t realize samsung was supplying apple through austin (e.g. generating US exports, not just income for US companies).

    generally speaking, the US semiconductor manufacturing industry — meaning US production of semiconductors, not US design of semiconductors that are made elsewhere — seems to be slipping, at least based on the trade data. As the chart above suggests, the US now is a net importer of semiconductors — which is a shift.

    and other austin based chip companies do not seem to be investing in US production/ fabs: “msung is one of the few Austin semiconductor companies that does any manufacturing here. NXP Semiconductors is another chip company that still does manufacturing in Austin.

    Most of the other Austin semiconductor companies, such as Cirrus Logic, Silicon Labs, and Advanced Micro Devices, have gone “fabless,” outsourcing the manufacturing work and focusing on design.
    Austin’s semiconductor industry accounts for about 10,000 to 15,000 jobs in Central Texas, according to economic development consultant Angelos Angelou. That’s down from about 30,000 jobs two decades ago.

    Some of that workforce reduction comes from closing down fabs and moving manufacturing overseas. From the austinnewspaper
    “Samsung seems to be the only semiconductor company (in Austin) that is actually investing considerable levels of capital to diversify their portfolio of products and adding new equipment and facilities,” Angelou said. The other fabs, he said, “have not necessarily upgraded in a considerable way their manufacturing facilities in Austin.” ” Most of the other Austin semiconductor companies, such as Cirrus Logic, Silicon Labs, and Advanced Micro Devices, have gone “fabless,” outsourcing the manufacturing work and focusing on design.
    Austin’s semiconductor industry accounts for about 10,000 to 15,000 jobs in Central Texas, according to economic development consultant Angelos Angelou. That’s down from about 30,000 jobs two decades ago.

    Some of that workforce reduction comes from closing down fabs and moving manufacturing overseas.

  • Posted by Alex

    This is from 2013, but worth reading – Apple CAPEX into the supply chain was running about $10bn/yr:

    https://www.bloomberg.com/news/articles/2013-11-13/apple-s-10-5b-on-robots-to-lasers-shores-up-supply-chain?cmpid=yhoo

    Currently about half the A9s and A9Xs are Samsung, half TSMC:

    http://www.anandtech.com/show/9708/analyzing-apple-statement-for-tsmc-and-samsung-a9

    Samsung manufactures them in Austin, in Korea, and apparently also at a GlobalFoundries partner site in New York State.

  • Posted by Michael M

    “It will be easy to beat a system like the one proposed. Chinese supplier will sell the microwave to its US based subsidiary for 10 usd. US subsidiary will sell the microwave to walmart for 100 USD”
    -RN

    RN,
    The problem with your example is that you are forgetting that Chinese subsidiary would then have a profit of $90 and so they would be taxed at 20% on their profit. The math below shows how it works out in your example.

    Chinese exporter = $10 of export rev * 20% import tax = $2 of tax
    Chinese subsidiary = $90 profit * 20% corp tax = $18 of tax
    Walmart = $10 profit * 20% corp tax = $2 of tax

    Total tax = $22… The same as my earlier example

  • Posted by Michael M

    Brad,
    I’m hearing a lot of economists claim that the border adj tax plan would not effect our current account balance because the currency would fully adjust and offset this.
    Further, I was just on a conf call with an economist claiming that a border adj tax is not the same as a tariff because the import tax is offset by the export subsidy, which offset each other. That really confused me because, as I see it, the import tax and the export subsidy work in the same direction (from a BoP standpoint) rather than offset.
    Any thoughts on the BoP effects of a border adj tax?

  • Posted by Brad Setser

    Michael — lots of thoughts

    First, i am skeptical of the perfect XR offset. Some countries peg to the dollar. Some manage v a basket that includes the dollar. Others intervene in a discretionary way. some currencies have already adjusted in anticipation of trump (Mexico), some less so (China arguably, as depreciation since election = function of moves in overall dollar). I also tend to think the equal and offsetting impact of the border tax on imports and the XR assumes that the policy is revenue neutral, and my guess is that it will be a part of a broader policy shift that changes the macro fundamentals. Even the micro can have BoP consequences. An import tax on oil for example might not be offset by XR changes, and it raises the domestic equilibrium price as the US is a net importer. That generates higher rents for existing oil producers, but likely squeezes refiners margins.

    Second, you are are right that the import tax and the subsidy for exports (e.g. deduction for wages, which makes this different from a VAT, which is border adjusted) means that it should have a BoP impact in the absence of XR changes.

    that also gets at the WTO challenge. standard VAT adjustments are the border are totally WTO compliant. But the house proposal effectively taxes the consumption of goods made with foreign labor at a different rate than the consumption of goods made with domestic labor. that could be challenged as a prohibited export subsidy, though proponents will say it isn’t specific to exports b/c goods made domestically with domestic labor and sold domestically are treated the same as goods make domestically that are sold abroad (if i understand the proposal correctly) so it isn’t specific to exporting. I would expect some countries to argue it is a countervailable subsidy. Would be lots of uncertainty, b/c it isn’t obviously WTO compliant and thus lots of scope for injured parties to challenge with goal of eliminating or to challenge with the goal of getting right to countervail the “wage” subsidy element that makes it different from a VAT.

    Am sure i am missing something though — hope others chime in

Pingbacks