The “Uberisation” of GDP?

The latest US economic data raises a serious question of coherence. On the one hand, a fast-paced rise in employment. On the other, the volume of economic activity as measured by GDP is very sluggish, resulting in the apparent virtual stagnation of productivity over the past few quarters. This is even more surprising given the talk around the growth of the digital economy.

One possible explanation for this is that US GDP forecasts are notoriously volatile and subject to frequent revision. So not too much importance should be given to these diverging signals in the short-term. It is a problem, however, when considering longer periods: in the United States, the trend growth of labour productivity has been just over 0.5% per year since 2010, compared to 2.5% growth over the previous fifteen years. Hence the return to the forefront of the debate on the impact of New Technologies of Information and Communication (NTIC) on productivity.

It's an old debate. In 1987, the winner of the Nobel Prize in Economic Sciences, Robert Solow, quipped: “You can see the computer age everywhere but in the productivity statistics”. In fact, one difficulty that statisticians have been coping with for years is the volume-price split because it relies on the measurement of constant-quality price indexes. In a sector such as NTIC, where the power of products (computers, smartphones, etc.) grows relentlessly, it is very difficult to measure constant-quality price change: hence the tendency to overestimate price growth and to underestimate volume growth. To overcome this problem, several methods are used, including the “hedonic” method, which consists in determining the quality of a product based on its intrinsic characteristics (speed, memory, etc.). This does not completely eliminate the problem of overestimating price growth, but the real question is knowing whether or not the magnitude of this residual bias has increased in recent years.

The newest element in the debate does not reside in NTIC products in and of themselves but rather in the ongoing explosion of services with high NTIC content. The traditional problems encountered when measuring value added and the price of service activities are multiplied in the case of the digital economy. Initially, these new services were primarily within B2B (Business-to-Business), which is recognised as inputs to production in a nation's accounts: therefore, any statistical biases had no impact on whole-economy total value added (which is shown in GDP); instead, they influence its division into areas of activity. Today, however, we are seeing spectacular growth in both B2C (Business-to-Consumer) activities and C2C (Consumer-to-Consumer) activities, which is likely to have a direct impact on final expenditure and GDP. Firstly, a substantial proportion of new services rely on unprecedented and disruptive models that are replacing – at least in part – traditional B2C activities. It is therefore extremely difficult to ascertain whether or not the price reductions they are causing are taking place at constant-quality service levels. This is all the more so with the development of the so-called Sharing Economy, where C2C is replacing B2C. Secondly, disintermediation of services often occurs without giving rise to transactions between households. For instance if, instead of using a travel agency, households made their own reservations over the Internet, there would be -all other things being equal- a negative impact on measuring GDP because the services that a household provides to itself are not recognised. Thirdly, free services are very widespread on the Internet. As such, goods purchased at a price of “zero” are not recognised in the national accounts nor are they used for calculating volumes. Time spent on Facebook or Google has no impact on GDP unless, of course, it prompts Internet users to make purchases they would otherwise not have considered because of advertising. Lastly, C2C activities, which are more informal, are less likely to be reported and are therefore harder to include in official statistics.

Overall, productivity gains (and lower prices) linked to the expansion of the digital economy are presumably underestimated. That said, the magnitude of this underestimation should not be exaggerated. On the one hand, the activities at issue, although they are booming, still only account for a modest percentage of the productive economy. On the other, we need to question the measurement of employment and hours worked. The Internet and the “Uberisation” of the economy are increasingly blurring the lines between work, domestic activity and leisure.

Olivier Garnier, Chief Economist for Societe Generale Group

Column for Agefi-Hebdo weekly, 25/06/2015