The cool tools that are reshaping economics
If Hollywood is to be believed, every mad scientist who ever lived has a laboratory full of bubbling flasks, flashing consoles and glowing orbs. Science writer Philip Ball — who has visited countless research labs — tells me that reality is not so very different: the gear may be more subdued, but the gear is always there.
Science depends on tools, often instruments to detect or measure that which was previously undetectable — think of Galileo’s telescope or Newton’s prisms. Nobel Prizes have often been awarded to the physicists who developed such tools: the cloud chamber (1927); the electron microscope (1986); and LIGO, the laser interferometer gravitational-wave observatory (2017).
What, then, of economics? Economics has its own quasi-Nobel Prize, but it is a stretch to find a single example of a prize being awarded for the development of new tools or instruments. Simon Kuznets (laureate in 1971) probably has the best such claim, for developing the ideas behind the gross domestic product measure. Alas, GDP is a broad aggregate with limitations that Kuznets himself understood all too well.
The great Alfred Marshall described economics as being the study of humanity “in the ordinary business of life”. Unfortunately, in Marshall’s day — he died in 1924 — there was no way to observe the ordinary business of life, except perhaps as an anthropologist. Economists spent a lot of time in armchairs, thinking hard about theory rather than measurement.
Some economists now make progress using old tools from other fields. MIT’s Esther Duflo, winner of the prestigious John Bates Clark medal, answers economic questions using randomised controlled trials. RCTs are typically dated back to Austin Bradford Hill’s 1948 trial of streptomycin for tuberculosis. (Hill was trained as an economist, so perhaps we can score that one for the profession.)
But the holy grail is to be able to observe the ordinary business of life in detail, in real time, and at scale — ideally all three at once. That was once an impossible goal, but three new developments put that goal within reach.
The first is the availability of high-resolution satellite images. In the mid-1990s, an economist named Alex Pfaff realised that these images could be used to answer questions about the connection between development projects and deforestation in the Amazon.
Hundreds of others have followed suit. Satellites can easily measure illumination at night, a simple way to track economic activity and patterns of urban development. It is also possible to measure various kinds of air pollution, and to observe the growth of crops. Algorithms are starting to extract subtle information at scale: how many Ethiopian homes have tin roofs? Which roads in Kenya are in good condition? And ever-cheaper small satellites are taking detailed photographs of everywhere, every day.
An even bigger change is that economists are using administrative data. I realise that “economists are using administrative data” is a contender for the most boring sentence uttered in 2018. But over the past two decades or so, this has been a quietly revolutionary move.
Administrative data are the numbers generated by governments or private companies for the purposes of getting things done. Schools keep track of attendance and grades. Tax authorities know your (declared) income — but also where you live, your age, and perhaps who your children are.
As such records have become digitised, they can be used to answer serious questions in research. For example, tax data can tell us the extent to which the children of rich or poor parents grow up to be rich or poor themselves. These detailed data are now at the forefront of empirical economic research.
According to Dave Donaldson, who like Prof Duflo is a John Bates Clark medallist at MIT: “In my field, international trade, I rarely see a paper that doesn’t use customs-level data. Every shipment generates a record which will specify what it is, where it came from, where it’s going, and the tax paid.”
A third measurement tool is the mobile phone. Every time a call is placed, the phone company generates a record of who called whom, when, for how long, and where the phones were, sometimes to within less than a hundred metres. With that kind of “metadata”, economists and other researchers can ask questions such as: how rapidly are people moving around, and to what extent is that correlated with the spread of an epidemic? Is a city’s transport infrastructure working well? How quickly are refugees integrating into a new society?
This is both an opportunity and a challenge for economists. Data scientist and economist Josh Blumenstock told me that “anyone who graduated with an economics PhD more than five years ago has no idea how to handle this data, and is frantically scrambling.”
Surely the scramble will produce results. At last, it is possible not just to theorise about Marshall’s “ordinary business of life”, but to observe it. Our tools are letting us see something new — and what we can see determines what we can think.
Written for and first published in the Financial Times on 13 July 2018.
My book “Fifty Things That Made the Modern Economy” (UK) / “Fifty Inventions That Shaped The Modern Economy” (US) is out now (or in the US, very very soon) in paperback – buy online or through your local bookshop.