The digital economy and GDP mis-measurement

To paraphrase economist Robert Solow, the digital economy is visible everywhere but in the GDP figures! Mark Cliffe at ING Bank has just published some research (to which I contributed) which argues that GDP is being seriously mis-measured. We estimate that US GDP is under-estimated by at least 0.75% pa, (and inflation is over-stated by 0.4% pa) due to a failure to properly account for the digital economy, together with the exclusion of certain intangibles. We further argue that this mis-measurement has been building up since the early 2000s. The implication is that economic growth in the current upswing has been stronger than officially reported, topping 3% pa in recent years on average (see chart).

The report starts by asking - Is GDP capturing the incredible improvement in quality of the online experience since the early days of dial-up modems – the speed of page-loading, the massive amount of content, including new user-generated content and the availability of streaming? How can GDP measure things which cost nothing, like a Google search or reading a Facebook post? Is GDP really capturing the value of smartphones which bundle together a phone, camera, GPS and maps, web browser, music player, game player and much else all in one product? And what about the effects of digital platforms like eBay, Uber and Airbnb?

To answer, it trawls through an enormous literature ranging from ‘techno-optimists’ like Brynjolfsson who argue that the statistics are getting it completely wrong, to cautious (sometimes slightly defensive) government statisticians who argue that they do take account of it and the under-estimate, if any, is likely to be small. The report sifts the evidence and concludes that the effect is too large to be ignored. And it comes from two main things.

First, the unique characteristic of digital technology is that, once produced, it costs almost nothing to replicate around the world. This makes the cost (often captured in GDP) very small, while the benefit (which should be captured in GDP but often is not) is huge. Secondly, the pace of change in digital technologies is so fast that price indices are failing to keep pace with the rapidity of price decline. To calculate GDP, statisticians add up the total value of goods and services purchased. Then they must work out how much of any increase is just inflation and how much is genuine volume change. The report shows that there is evidence looking at telecoms and the internet that the statisticians record prices declining, but not by nearly enough.

The report calculates the under-measurement of the digital sector to be about 0.5% of GDP annually, which arises because of these rapid price declines, the plethora of new goods and services and the extent of free goods. The improvement in quality is significantly under-measured so that output is understated and inflation overstated, the latter by about 0.4% pa.

A further 0.25% pa should be added to GDP by fully including intangibles. Software and mineral exploration were added in 1999 and scientific R&D spending as well as intellectual property in entertainment, such as movies and books were added in 2013. But US GDP still does not capture investments in brands, non-scientific R&D, training and organisational capital, which is a huge part of spending by big and small organisations alike.

What are the implications of GDP being under-reported and inflation over-reported? They are many and full discussion will have to wait for a later report. But the correction seems to me to make good sense. This has been a more normal recovery than often claimed and the argument that we are living in an era of ‘secular stagnation’ is partially (though not completely) falsified. The lower rate of inflation also seems right. After the devastating slump in 2008-9 it was always surprising that inflation held up as well as it did. Central bankers argued that this reflected the anchoring of inflation expectations. More likely, inflation was over-stated.