GDP all over again


11 March 2014

Q: If Gross Domestic Product (GDP) estimates are quarterly, why are there one or more releases each month?

A: The U.S. Commerce Department’s Bureau of Economic Analysis (BEA) calculates estimates of GDP. Estimates get better over time with improved completeness or refinement of source data.

  • “Advance” estimate, near the end of the first month after the end of the quarter
  • “Second” estimate, near the end of the second month
  • “Third” estimate, near the end of the third month
  • “Latest” estimate, in following quarter, annual and 5 year comprehensive or “benchmark” revisions. Some comprehensive revisions reflect a changing understanding of the economy such as the 2013 change to intellectual property or international investment position.

For 2014, the schedule is

Q: How big are average revisions of estimates?

A: For Real (also known as “constant currency” or “inflation adjusted”) GDP from 1983 to 2010, the BEA calculates:

Estimated change from:    Average (without regard to sign)    Standard deviation of revision (with regard to sign)
Advance to Second          0.5                                                   0.4
Advance to Third             0.6                                                    0.4
Second to Third               0.2                                                    0.2
Advance to Latest            1.3                                                    1.0

The BEA explains change is greatest to the “Latest” estimate because of those comprehensive revisions.

Quarterly estimates correctly indicate:
Direction of change of real GDP 97% of the time
Whether GDP is accelerating or decelerating 72%of the time
Whether real GDP growth is above, near, or below trend growth more than 80% of the time

Q: With all the revisions, why are the “Latest” data still “estimates” rather than “the” number?

A: Because the GDP is compiled from multiple estimates based on a range of sources. Estimate quality depends on completeness of data, accuracy of data in original form, sampling approach, field procedures, office processing quality and broader considerations such as “are we really measuring what we think we are measuring?” For example, the revision in international transactions added hedge funds that have substantially grown over time. Other numbers are squishy such as cash transactions in households.

This is why:

  • Policy makers have long been concerned about the quality and timeliness of data
  • Investors (in contrast to traders) need not react to each data point – trends are more meaningful.

Much of this post was adapted from the BEA’s comment


Leave A Reply