Affiliations: [a] University of Massachusetts, Boston, MA, USA | [b] John F. Kennedy School of Government, Harvard
University, Cambridge, MA, USA
Correspondence:
[*]
Alan Clayton-Matthews, Public Policy Program,
UMass Boston, 100 Morrissey Boulevard, Boston, MA 02125-3393, USA. Tel.: +1 617
287 6945; Fax: +1 617 287 6949; E-mail:
[email protected]
Abstract: The Stock/Watson index methodology is applied to the Massachusetts economy to
estimate coincident and leading indexes for the state. A coincident index,
calibrated to trend with gross state product, is estimated as a dynamic single
factor, multiple indicator model, using the Kalman filter and smoother on a set
of coincident indicators. The leading index is a six-month ahead forecast of the
coincident index, based on a regression on recent growth in the coincident index
and a set of leading indicators. Filtering of noisy data and model selection in
the context of a short historical span of data are two issues common to index
construction at the state and regional levels that the authors address.
Keywords: coincident index, leading index, Kalman filter, dynamic single factor model, predictive least squares, Stock/Watson model
