The U.S. Bureau of Economic Analysis drew on credit-card transactions, patient visits and other indicators to fill gaps in its first estimate of the second-quarter 2020 downturn. Released July 30, the report put the fall in inflation-adjusted gross domestic product at a seasonally adjusted annual rate of 32.9%. The accompanying technical note shows what sat behind that headline: some series had three months of source data, while others still needed assumptions. [S1][S2]

The annual rate is not a claim that output fell by a third between the first and second quarters. Reversing the annualization of the rounded 32.9% figure gives a quarter-to-quarter decline of approximately 9.5%. That calculation translates the measurement period; it does not forecast four quarters of continued contraction. [S1]

The gaps were not uniform. BEA had three months of source data for consumer spending on goods, capital-equipment shipments, motor-vehicle sales and inventories, among other series. Where only two months were available, or second-quarter data were not yet available, the agency used information including high-frequency private credit-card transactions, news about reopenings and industry reports of activity. A timely national estimate did not mean that every underlying series had reached June. [S2]

For health care, BEA based the reported declines in hospital and outpatient services on volume data for hospital services and outpatient visits, alongside credit-card transactions. In recreation, estimates for clubs, sports centers, parks, theaters and museums relied primarily on card transactions; gambling used state gaming-control data. That mix does not make one card-spending series a measure of all service activity. [S2]

Single-family construction had a different gap. The estimate used April and May Census construction-spending data, with a June assumption based on housing starts. An analyst comparing the release with a construction dashboard needs to keep that June estimate separate from a reported June observation. The GDP release covered the whole quarter; its construction-spending inputs did not yet do so. [S2]

Missing observations were not the only limit. BEA said the full economic effects of COVID-19 could not be separately identified in the GDP estimate because they were embedded in its source data. It made a related point about federal relief: transfers to individuals, business subsidies and grants to state and local governments affected GDP indirectly through spending and investment. The report therefore did not supply a separate total for how much the relief acts changed GDP. [S2]

More complete source data could improve the activity estimate without isolating the pandemic’s effect or the relief programs’ contribution. BEA scheduled its second estimate, incorporating more complete data, for August 27, 2020. [S1][S2]

Source documents

  1. Gross Domestic Product, Second Quarter 2020 (Advance Estimate) and Annual Update — Full Release and Tableshttps://www.bea.gov/sites/default/files/2020-07/gdp2q20_adv_0.pdf
  2. Gross Domestic Product, Second Quarter 2020 (Advance Estimate) and Annual Update — Technical Notehttps://www.bea.gov/sites/default/files/2020-07/tech2q20_adv_0.pdf