James D. Hamilton*
(Updated October 31, 2011)
The GDP-based recession indicator index is a pattern-recognition algorithm that assigns dates to when recessions begin and end. It is based on the observed dynamics of U.S. real GDP growth. To make a reliable inference, it is necessary to wait one quarter for data to be revised. With the 2011:Q3 advance GDP numbers released by the Bureau of Economic Analysis on October 27, 2011, we can calculate a value of the recession indicator index describing economic conditions for the second quarter of 2011. To maximize usefulness as a real-time indicator, the index is not subsequently revised. The index ranges from 0 to 100, with a value above 50 indicating the data are more consistent with a recession than expansion.
The value for the index for 2011:Q2 is 10.3 percent, down slightly from the 14.4 percent recorded for the first quarter and still significantly below levels that would signal that a new recession has begun. The GDP figures just released indicate that the level of real GDP has finally reached a value above that seen in 2007:Q4, when the last recession began. However, the economy has continued to grow from the trough in 2009:Q2, and our algorithm follows the usual economic convention of characterizing a recession in terms of growth rates rather than the overall level.
Based on the procedure described in a paper by Marcelle Chauvet and James Hamilton (from Nonlinear Time Series Analysis of Business Cycles, 2006, edited by Costas Milas, Philip Rothman, and Dick van Dijk), the most recent recession was determined to have begun in 2007:Q4 and ended in 2009:Q2. These start and end dates for the recession are the same as were announced separately by the Business Cycle Dating Committee of the National Bureau of Research, though NBER did not issue the end-date declaration until September 2010.
If there is a subsequent downturn in GDP growth, this approach would characterize it as the beginning of a new recession. The algorithm would declare that a new recession has begun if the index goes back above 67 percent.
The plotted value for each date is based solely on information as it would have been publicly available and reported as of one quarter after the indicated date, with 2011:Q2 the last date shown on the graph. Shaded regions represent dates of NBER recessions, which were not used in any way in constructing the index, and which were sometimes not reported until two years after the date.
For more details about the method, see Chauvet and Hamilton's paper or the less technical description by Hamilton. You can also download a spreadsheet containing historical values of the index.
*James Hamilton is a professor of economics at the University of California, San Diego.
