- Seasonally adjust and smoothen using X11ARIMA to obtain the cycle-trend component for each of the eleven leading indicators and the reference series (non-agriculture component of the GDP).
Remove the trend component from the seasonally adjusted and smoothened series to obtain the cycle component of each of the indicators by using the Hodrick-Prescott (HP) Filter Method. Starting with the Q1 2004 LEI estimate released on 12 February 2004, the computation of the LEI adopts a new detrending procedure called the Hodrick-Prescott (HP) Filter Method, which has effectively addressed observed limitations in the method used for the estimation of the LEI prior to 2004.
- Correlate the cycle of each indicator with the cycle of the non-agriculture GDP (reference series) to obtain the lead period. The lead period determines the number of quarters the cycle series for each indicator is moved forward.
- The index is computed as the linear combination of the indicators using the correlation coefficients of the indicators with the non-agriculture GDP as weights. In determining the relationship between the Non-agriculture GVA and the composite indicator, the following simple linear regression model is used: