Diane Coyle explores data and its value – could we be paid for our individual data? What can it reveal about consumption? And why data could be much more than the sum of its parts.
How valuable is data? It’s a question that needs answering if you’re wondering whether digital companies should pay individuals for data, or an investor being pitched by a start-up boasting about its clever algorithms, or the Office for National Statistics thinking about whether data is an asset that should be included in the official economic statistics.
The Bennett Institute and Open Data Institute hosted a workshop on 18th July to discuss the issues raised by the question of the value of data, and focus on what we’d need to know to answer it.
This is more than just an academic question: as a report commissioned from the ODI points out, there is a lot of potential benefit to be gained from untapped data if it can be made widely available, but data owners are unwilling to open up their data sets without knowing the value of what they might both give and gain through sharing. The contribution to the economy and society of untapped data assets could be large. Artificial intelligence is making this seem an urgent question, given the appetite algorithms have for data, and the competitive advantage data-using companies seem likely to gain.
The workshop participants came from academia, official organisations, tech businesses and the voluntary sector. Three speakers kicked off the discussion: AI expert Azeem Azhar, curator of Exponential View; intangibles guru Professor Jonathan Haskel from Imperial College Business School; and Will Page, Director of Economics at Spotify.
The discussion (held under the Chatham House Rule so participants could speak freely) underlined the salience of some key issues from my perspective as an economist:
The value of aggregated data is far greater than individual data. Paying each of us for our personal data wouldn’t transform anyone’s fortunes. Yet aggregate data is highly informative. A good example is the information in the heat maps published by fitness app Strava earlier this year: inadvertently, in publishing these, the company published maps of military bases because so many troops have the app on their watch as they run round the perimeter.
There are going to be many unlocked benefits in joining up data sets, if ways can be found to build the technical protocols and trust to enable some pooling.
Estimating the economic value of data will require some standardisation of the types of data, and a taxonomy of how they fit together (something the OECD is working on).
Economists will also need to understand how quickly or slowly different types of data depreciate (very fast if it’s targeting ads for restaurants as you walk down the street, but very slowly if at all in some other cases); what the marginal value of additional data is when adding to an existing data set; how to conceptualise the rate of return on data as an asset.
A final intriguing insight was that GDP measures purchases in the economy (the price of a purchased CD), but data from services like Spotify enables us to measure actual consumption (which songs purchasers listen to and for how long). There has been discussion of the ‘attention economy’; perhaps data to measure it are in sight.
One thing everybody agreed on was that data is not the ‘new oil’. It may become essential fuel but has completely different economic characteristics from oil. If it is going to be as essential as fossil fuels have been, though, getting to grips with how to value it is key. Jeni Tennison (@jenit) of the ODI and I (@DianeCoyle1859) will be taking some work on this forward.
The views and opinions expressed in this post are those of the author(s) and not necessarily those of the Bennett Institute for Public Policy.
