This article constructs seven leading indexes of New Zealand employment and assesses their relative usefulness in terms of forecasting quarterly employment growth. Leading indexes have been widely used since their introduction in the late 1930s. One construction method dominated until academic research interest into alternative techniques re-appeared in the late 1980s. What has been missing so far in the literature is a thorough comparison of old and new techniques in terms of forecasting performance. This article is a step in that direction. The methods covered here reflect varying degrees of technical sophistication, ranging from simple scoring of changes to relying on frequency domain methods to extract dynamic latent factors from a large dataset. The results show that no single index dominates in terms of forecasting employment growth one to four quarters ahead. This suggests that relying on a suite of models may be the optimal forecasting strategy.
|Publication status||Published - 2011|