The GDP-based recession indicator index comes from an analysis by Marcelle Chauvet and James Hamilton. The motivation behind developing the index was to determine whether we are in a recession as it would be defined by the NBER in real time rather than waiting for before the official NBER announcement, which may occur many months after the recession has ended.
The central idea behind the index is fairly straightforward: the probability of an NBER-declared recession increases when real GDP growth rates are low. The index uses Bayes’ theorem to derive the probability, based on available GDP data, that the economy is currently in a recession.
Chauvet and Hamilton declare the economy to have been in a recession as of the previous quarter when the index rises about 67%. Once the recession warning triggered, it remains in place until the index falls below 33%. The predicted start and end dates are then assigned based on a mathematical formula.
As is obvious from the chart, the GDP-based recession indicator has been remarkably accurate at predicting NBER recession dates in real time.