Big Data and New Ways to Measure the Economy

How economists now use satellite images, online prices, card transactions and tax records to measure things traditional statistics miss.

For most of the twentieth century, economists relied on official surveys and statistics released monthly, quarterly or yearly. Today, a flood of new data sources lets them measure the economy in more detail and closer to real time.

Seeing the economy from space

Satellite images of night lights have become a well-known tool. Economists including Vernon Henderson, Adam Storeygard and David Weil showed that the brightness of lights at night tracks economic activity. This is especially useful in countries where official GDP statistics are weak or unreliable, and for measuring activity in small areas like individual towns.

Online prices

The Billion Prices Project, begun by Alberto Cavallo and Roberto Rigobon at MIT, collected prices from online retailers daily in many countries. It provided a fast, independent way to track inflation. In Argentina, where official inflation figures were widely doubted in the early 2010s, online price data showed inflation running well above the official rate.

Administrative and transaction data

Administrative data, records collected by governments for running programmes, such as tax returns and social security records, can cover entire populations rather than small samples. Raj Chetty and colleagues used anonymised U.S. tax records covering millions of people to map how children’s chances of earning more than their parents vary by neighbourhood. Their work showed large differences between places only a few miles apart.

During the COVID-19 pandemic, researchers used anonymised card transaction data and mobile phone location data to track spending and movement week by week, far faster than official statistics.

Measuring a small town's economy

Official GDP figures might only be available for a whole country or large region, once a year. But satellite images can show whether lights in a particular town brightened after a new road opened, and card spending data can show whether local shops saw more customers. Researchers can study effects that official statistics could never reveal.

Cautions

New data brings new problems. Online prices may not represent what people pay in shops. Card data leaves out people who use cash, often poorer households. Privacy must be protected carefully. Big datasets are only as good as the questions and methods applied to them.

Thinking more data automatically means better answers

A huge dataset can still be biased if it leaves out important groups. And more data does not solve the problem of separating cause from correlation. Big data is most powerful when combined with careful research design.

Key takeaways

New data sources let economists measure the economy in more detail and closer to real time.
Night lights from satellites track economic activity, especially where official statistics are weak.
Online prices and administrative records such as tax data have transformed research.
New data can leave out groups, raise privacy issues and still needs careful analysis.