Title: Dominant U.S. manufacturing sectors: A factor model analysis
Authors: Soroosh Soofi Siavash - Bank of Lithuania (Lithuania) [presenting]
Abstract: The focus is on the issue of identifying observed time series variables which serve as proxies for the factors underlying sectoral comovement. In the studies using factor model analysis, an approximate dynamic factor model fit sectoral data well. In multisector models with the input-output linkages, the macroeconomic fluctuations are viewed being primarily a result of shocks specific to the sectors which have an important role in supplying products to other sectors, or are of a great size. We use a factor method to investigate whether any observed time series variable of an individual sector serves as a factor proxy in large sectoral panels. We show that the method can identify the factors with a probability approaching unity when $N,T\rightarrow\infty$ even if the factors are relatively weak. In an application of the method to sectoral industrial production growth rates, we find that; (a) growth rates of a few heavy machinery and electrical equipment sectors serve as proxy for a factor, and (b) the sectors identified appear to have a key role in supplying capital products in U.S. economy, but have a moderate role in supplying intermediate products to others, and are of a moderate size.