Wednesday 12 August 2015

The Productivity Paradox: Is There a Measurement Problem?

There's been much debate in recent months about the productivity paradox - put simply there's a long standing concern that technology, particularly information technology, does not seem to deliver the productivity gains that might be expected. This concern has resurfaced in the UK, with the Government raising questions about why the UK's productivity has not grown as much as other countries. In fact George Osborne recently called the UK's low productivity growth "the challenge of our time".

This same topic came up in a recent email discussion with colleagues from ISSIP - the International Society for Service Innovation Professionals. This time prompted by an article in the Wall Street Journal entitled "Silicon Valley Doesn't Believe US Productivity is Down". In essence the Wall Street Journal argument was that developments in technology are not captured in the Government's productivity figures - apps that help people find restaurants more quickly or hail cabs from their phones clearly improve the efficiency with which we can do things. Doing more with less is a classic definition of productivity - so these apps must be improving productivity argues the Wall Street Journal (and those it quotes - including Hal Varian, Google's Chief Economist).

While I accept the argument that apps and associated technologies allow us to do more with less, I think there's a need to unpack the relationship between these developments and measures of productivity more carefully. Traditionally governments have measured labour productivity - in terms of GDP per hour worked. As technology replaces labour, GDP stays the same or increases, while labour hours go down - hence productivity increases.

However, there's an interesting new phenomenon which complicates the picture. Take, for example, Uber. I'm a fan of Uber - the app is great. Its convenient. I've never had a bad service from an Uber driver. I love the fact that I can rate drivers and they can rate customers at the end of journeys. I love the fact that the cost of the ride gets charged to my credit card and the receipt automatically emailed to me. But I also love Uber because it is cheaper - I pay less for a Uber car than I do for a black cab in London. Better service, pleasant drivers, lower prices - what's not to like. Other firms have similar business models - think Amazon or Airbnb. Still others provide me a service for free - Google and TripAdvisor - don't charge me for the information they provide, instead making their money through third parties.

When talking about productivity - or the lack of productivity - we need to think about the economic impact of these cheaper and/or free services. Lower prices to consumers must mean lower GDP. The efficiency gains are there, but they are not being captured in productivity gains because the benefits are being passed on to consumers in the form of lower prices, rather than captured in the official GDP statistics. Maybe a more nuanced discussion about productivity is needed - where we look at both sides of the equation - increases in value and hence GDP - and increases in efficiency reflected in lower costs to consumers.


  1. But when prices go down, demand goes up, so P*Q=GDP does not necessarily go down.

    Anyways, although nominal GDP can be affected by price developments, changes in prices over time do not affect real GDP, which is the indicator actually used to compute productivity figures.

    1. Hi - thanks for the comment. Agree that this is the case if you assume the price elasticity curves are perfect descriptors of the world, but of course many would question whether they are. The key point I was trying to make is that we have to rexcognise the different ways in which productivity might be conceived and measured in the digital economy.

  2. Cannot agree more with Andy. In the today's digital economy we cannot be sure what should be integrated into the GDP, can we? Due to digital technological advances many interactions are occurring in creation of value that is not part of GDP. For instance, what does the value of blogging contribute to the GDP in USA? There is plenty of evidence about how perspectives of GDP are supposed to be developed, but we must know that the metrics of measuring are as good as they are made good for measuring something – garbage in garbage out. The GDP lacks integration of today's use of e.g. Internet technology and its value, hence we miss the productivity that is occurring via today’s apps, aren't we?

    In other words, the use of UBER, AIRBNB (think of taxi and hotels value) is good for well-being as Andy’s descriptions reminds us, but on the other hand this is bad for the today’s GDP that is based on money transactions e.g. countries are complaining about missing the collected taxes… I guess we are approaching a change in metrics, since the “fact that GDP may be a poor measure of well-being, or even of market activity, has, of course, long been recognized. But changes in society and the economy may have heightened the problems, at the same time that advances in economics and statistical techniques may have provided opportunities to improve our metrics.” See the GDP Fetishism by Joseph E. Stiglitz at, accessed 24.11.2015 for more.