Real-Time Economic Indicators

Real-time economic indicators are statistics published at high frequency, usually daily or weekly and with little delay, that give an early read on economic activity. Examples include card and retail payment volumes, restaurant bookings, online job postings, electricity use, road traffic, flight departures, shipping movements and weekly claims for unemployment benefits. Central banks, statistics offices, governments and investors use them to see what is happening now, weeks or months before official figures such as GDP are released.

The term has a second, older meaning in economics: "real-time data" can also refer to the figures as they were known on a given date, before later revisions. Both senses share the same concern, which is that decisions are made with incomplete and preliminary information. Real-time indicators complement, rather than replace, leading and lagging indicators and the official national accounts.

Official economic statistics are like a detailed school report that arrives a few weeks after the end of term. It is careful and complete, but it describes the past. Real-time indicators are more like a parent glancing at a child's homework every evening. Each glance is rough, and one bad night means little, but over a few weeks a clear picture forms long before the report comes home.

When a shock hits suddenly, the evening glance becomes very valuable. If shops close overnight, card spending and footfall show the drop within days, while the official retail sales figure may take more than a month.

• Payments data. Aggregated, anonymised card transactions from banks and card networks show consumer spending by sector and region almost as it happens.
• Labour market data. Weekly initial unemployment insurance claims in the United States and online job adverts collected by recruitment sites track hiring and layoffs.
• Mobility and activity data. Road traffic counts, public transport use, flights, hotel occupancy and restaurant reservations show how much people are moving and going out.
• Production and trade data. Electricity demand, satellite images, port calls and ship-tracking data give early readings on industry and exports.
• Financial market prices. Share prices, bond yields and credit spreads change by the second and reflect market expectations, as discussed under stock market.

Several official bodies publish dashboards built on such data. The Federal Reserve Bank of New York's Weekly Economic Index, launched in 2020, combines ten weekly series into a single measure scaled to annual GDP growth. The UK's Office for National Statistics began a regular release of real-time indicators during the COVID-19 pandemic, covering items such as card spending, traffic and online job adverts.

Real-time data feed nowcasting models, which estimate current-quarter GDP before the official number is out. The Federal Reserve Bank of Atlanta's GDPNow model updates its estimate of US quarterly growth several times a month as new data arrive, and the New York Fed and many central banks run similar models. Nowcasts use the tools of econometrics, such as dynamic factor models and mixed-frequency regressions, to combine many noisy series into one estimate. They are useful for the Federal Reserve, the ECB and other central banks because monetary policy decisions are taken between official data releases.

In the second sense of the term, economists study how much early estimates change. First estimates of GDP, employment and the output gap are often revised substantially, sometimes changing the story of a period. The Federal Reserve Bank of Philadelphia keeps a Real-Time Data Set for Macroeconomists, which stores each historical "vintage" of key US series so that researchers can see what policymakers knew at the time. Studies using such data have shown that estimates of the output gap in the 1970s were badly wrong in real time, which helps explain policy errors during that decade's high inflation. This work underlines the case for caution when setting policy with rules such as the Taylor rule.

The main strength is speed. During the 2020 pandemic, real-time data showed the collapse and partial rebound of spending weeks before official statistics, helping governments design support schemes and central banks judge the need for measures such as quantitative easing. They also allow detail by region, sector or income group that official surveys cannot provide quickly.

The limitations are significant:

• Coverage is partial. Card data miss cash spending and the informal economy, which is large in many emerging markets.
• Samples shift. A data provider's customer base may change over time, creating trends that are not real.
• Seasonality is hard to remove from short daily series, so seasonal adjustment is often rough.
• Privacy and access. Many series are owned by private firms, may be discontinued and must be aggregated to protect personal data.

For these reasons, statistics offices treat real-time indicators as experimental and continue to rely on official surveys for final figures.

• Federal Reserve Bank of Atlanta: GDPNow
• Federal Reserve Bank of New York: Weekly Economic Index
• Federal Reserve Bank of Philadelphia: Real-Time Data Set for Macroeconomists