Dynamic Factor Analysis for Panel Data: A Generalized Model∗

We develop a generalized dynamic factor model for panel data with the goal of estimating an unobserved performance index. While similar models have been developed in the literature of dynamic factor analysis, our contribution is threefold. First, contrary to simple dynamic factor analysis where multiple attributes of the same cross sectional unit are measured at each time period, our model also accounts for multiple cross sectional units. It is therefore applicable to a panel data framework (i.e. multiple attributes for multiple cross sectional units observed over time). Second, our model estimates an unobserved index for every cross sectional unit for every time period, as opposed to previous work where a ∗The authors would like to thank Alexander Danilenko for providing data from the International Benchmarking Network. We are grateful to John Buonaccorsi, Klaus Moeltner, Joe Moffitt and John Stranlund for their constructive feedback.Comments and suggestions from seminar participants at the University of Ottawa and the University of Massachusetts Amherst are greatly appreciated. This research was made possible with funding by the NIH grant AG13846. Any remaining errors are ours. †School of Public and Environmental Affairs, Indiana University, [email protected]. ‡Department of Biostatistics, Boston University, [email protected].