Note: Formulae used on this website are converted into MathML by javascript. To view the formulae correctly you will need to have javascript turned on and MathPlayer (for Internet Explorer users) or MathML fonts (for Firefox, Mozilla, Netscape users).
See our Help Page for more information.

The forecasting model was estimated using ordinary least squares. Following Auerbach (1982), the model includes lags of the index and past employment growth. The approach of broad to specific in the selection of lags of the index and past employment growth was used. Statistical significance and residual based tests led to the selection of the following specification:

`Deltaemp_t = alpha_0 + sum_(i=1)^4 alpha_i Deltalienz1_(t-1) + alpha_5 Deltaemp_(t-3) + epsi_t`

Where `Deltaemp_(t-i)`, `Deltalienz1_(t-i)` indicate the growth rate of employment and index at lag `i * alpha_0` and `alpha_j`, `j=1,...5` are coefficients and `epsi_t` is white noise error.

The composite index forecasting model includes one to four lags of the index and employment growth lagged three quarters.