This statistic rarely provides an accurate picture. During a pandemic that is doubly trueby Peter Kellner / August 10, 2020 / Leave a comment
On Wednesday (12th August), we shall be told how much Britain’s economy contracted in the second quarter of this year. Here are three predictions. First, the figures will show that Britain’s economy has contracted by more than at any time since the Black Death (or some other ancient horror). Second, they will dominate the day’s news. Third, in years to come when we have more data and better ways to measure the economy, the figures will be seen to be nonsense.
To be clear, I’m not arguing that the economy flourished. Plainly, it shrank a lot. My argument is that the particular numbers that will be reported this week will be close to meaningless.
Let’s start with first principles. Gross Domestic Product adds up all the transactions in which money changes hands and that are reported to the Office for National Statistics. This means that it excludes two big features of our society: activities where money does not change hands (such as voluntary work, housework and care for other members of our families) and those activities where money does change hands but the ONS isn’t told (such as sex work and drug dealing). If ways were found to include all these things, GDP would be substantially larger.
However, the main interest in the published figures is not the level of the GDP but the amount it changes in a particular year, quarter or month. As long as the different types of unmeasured activity don’t change too suddenly, this is not a big problem in the short term. In normal times that is a reasonable assumption.
We cannot make that assumption this year. For a start, during the period covered by this week’s figures, April to June, restaurants, theatres and cinemas were closed, cleaners were laid off, nurseries shut down and so on. But we still ate food, cleaned our homes (well, some of us), cared for our toddlers and had the chance to watch films and plays, albeit only at home. The GDP figures will count the lost trade but—with the exception of the rise in supermarket sales and Netflix subscriptions—tell us nothing about how millions of us managed to adjust to these changes. On the other hand, nor does GDP measure the gaps that money can’t fill, such as getting together with friends and family.
The second big problem with the figures concerns public spending—around 20 per cent of measured GDP. We know that schools have shut and hospitals have been operating very differently from normal. But the latest figures will capture almost none of this. This is because the data tell us what these services spend (mostly wages), not what they do. In economists’ jargon, quarterly GDP figures measure inputs not outputs.
To be fair to the ONS, measuring public sector output is a statistical headache around the world and the ONS has one of the best records at tackling it. It has been exploring ways of reporting public sector output and productivity: for example whether more pupils are getting good grades at GCSE. In a paper published last August, the ONS reported progress in measuring the quality of our public services, but says it can provide data only on an annual basis, and with a two-year lag. Quarterly data, such as this week’s release, will tell us, as usual, what schools and hospitals spent during this year’s lockdown, not how much or how well they did for their pupils and patients.
The third big problem did not suddenly appear during the lockdown, but has been intensified by it. For years, rapid advances in technology have messed up the process of measuring GDP and productivity. For example, in a host of areas, we subscribe to services instead of paying for them individually. Instead of paying for each letter, phone call, rented video or record purchase, and hence boosting GDP, we are able to send emails, speak to our friends, watch films and listen to music at zero marginal cost. Many of the extra things we have done during the lockdown have added nothing to GDP. Once again, there is a clear divergence between the transaction-based way we measure GDP and our lived experience. Indeed, the rapid expansion of free services such as Zoom has scarcely been caught by GDP at all.
The challenge of incorporating new technology into our GDP figures has been known for years. Back in 1987 the economist Robert Solow famously said: “You can see the computer age everywhere except in the productivity statistics.” The problem has become far more acute since then. Around the world, productivity figures started to slow down 12-15 years ago—just as smartphones, emails, and services such as Skype and Spotify began to take off. Productivity in the long term is a measure of human ingenuity; the statistics tell an implausible story of sudden global lethargy in the world of innovation.
Even the things we do buy and whose sales do appear in the GDP figures—computers, cars and televisions as well as smartphones—are often very different today from their equivalent five or 10 years ago. Their cash prices are much the same, but their quality is generally higher. But GDP data have never properly captured this. Efforts by the OECD to set international standards for what it calls “hedonic” adjustments to product prices have made little progress.
Two recent attempts have been made to explore the impact of technical change, one in 2018 by Diane Coyle and David Nguyen, both of the National Institute of Economic and Social Research, the other this year by the ONS. Coyle and Nguyen looked at cloud computing, the ONS at telecommunications. Both came to similar conclusions: that on a like-with-like basis, prices have fallen in recent times by around 95 per cent (since 1997 for telecoms, 2009 for cloud storage). Much the same is likely to be true across the technology sector.
This matters. In their paper for the Economic Statistics Centre of Excellence, Coyle and Nguyen conclude that GDP measures product improvements “imperfectly”; worse, “It does not capture process innovations whose outcome is faster delivery of a service or product, greater customisation, or the capability to do something previously out of reach.” New technology is (mainly) a boon to life but a curse to statisticians.
If all these things were measured properly, the outcome would be that GDP and productivity have risen more—cumulatively, much more—in recent times than official figures have shown, inflation would have been lower, and living standards higher. However, inequality would also have risen, because the unmeasured benefits of modern technology have benefitted the better-off most.
Three conclusions flow from all this.
First, any pundit or journalist tomorrow who says that living standards are back to the levels of 1995, 2000, 2005 or whenever is talking through their hat.
Second, our concern should be with not the bogus aggregate but the firm specifics: the job losses, poverty, delayed operations, pupils not being taught, domestic violence, mental ill health and, above all, people dying before their time.
Third, if we end up with a better way to measure GDP, with faster growth and slower inflation, then state benefits and tax allowances linked to inflation will rise by less. The gap between rich and poor will grow. As always, we should be careful what we wish for.