May 20, 2018
That’s what a story in the Financial Times tells readers. I don’t think they have much of a case.
The argument is attributed to Hal Varian, Google’s chief economist and a former professor of mine when I was in grad school at Michigan. According to Varian, if we accurately counted the value of software in smartphones it would add $200 billion to US exports, cutting our trade deficit in half.
The first item to point out is that our trade deficit is currently running at an annual rate of $640 billion, not the $400 billion claimed by Varian. A $200 billion reduction is still large, but it would imply cutting the deficit by a bit more than 30 percent, not half.
But the more important issue is the logic of the argument. Varian points out that, while Apple has proprietary software, for which it charges for its use, Google makes its software available for free, but demands ad placement in exchange for its use. This means that Apple’s software should be accounted for in our exports, but Google’s would not. This is indeed a problem, but perhaps not as much of one as Varian implies. (It is worth noting that the value of the software is in principle already counted in our National Accounts, so the Varian critique would imply no change in GDP, but that exports are understated and domestic investment is overstated.)
In effect, he is saying that Google is being compensated for its software by the ads that are subsequently sold on the phone. By contrast, Apple has been fully compensated at the point of sale. If we had proper accounting, then we would also count the value of Google’s software at the point of sale. But look at what happens in subsequent years.
Suppose the Android phone is sold in some third country. Google will be collecting ad revenue from these phones for their full working lives. This ad revenue would then, in principle, (there is an important accounting issue I will address in a moment) be attributed to Google and count as an exported service. By contrast, the Apple phone does not directly generate any further revenue for Apple. This means that we should effectively be picking up the value of Google’s software in Android phones through the ad revenue the phone generates in subsequent years. There is still an issue of timing, and also a definitional one (perhaps the original transfer of software should have been booked as an investment), but we are capturing the value of the software exported in the subsequent income flows from the advertising.
This would not be the case if the Android phone is imported back into the United States since the ad revenue is all domestic income. But there is no problem here because the imported phone costs less than it would have had the software been proprietary like Apple’s. In short, there is not really a major issue here.
Now, there is a very important secondary point. Let’s hypothesize that all of Google’s innovation for its Android phone comes out of its Mountain View campus in California. Suppose to minimize their taxes, Google attributes most of the value and subsequent profits to its subsidiary in low-tax Ireland.
In this case, we would be understating the value of US exports, since the subsequent flows of income would be showing up at Google’s Irish subsidiary, not its Mountain View campus. Clearly, there is much of this sort of gaming taking place, as most of the big tech companies seem to do a surprising share of their innovative work in low-tax countries, although it probably does not get you to $200 billion a year. (And here is my easy fix for this problem, if anyone is interested in a fix.)
Comments