Functional Principal Component Analysis of Density Families with Categorical and Continuous Data on Canadian Entrant Manufacturing Firms

• Activity Number: 472
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Government Statistics
• Author(s): Kim P. Huynh*+ and David T. Jacho-Chavez and Robert J. Petrunia and Marcel Cristian Voia
• Companies: Indiana University and Indiana University and Lakehead University and Carleton University
• Keywords: Firm Distributions ; Functional Principal Components ; Nonparametric Methods ; Administrative Tax Data ; Balance Sheet ; Entrant Firms

Abstract

This paper investigates the evolution of firm distributions for entrant manufacturing firms in Canada using functional principal components analysis (FPCA). This method is nonparametric, describes the dynamic of marginal densities and illustrates the efficacy of functional principal components to analyze firm distributions. We modify the Kneip and Utikal (JASA, 2001) method to allow for the inclusion of qualitative information in the form of discrete variables such as industry and region. The results indicate that there are substantial differences in the dynamics of firm size, labour productivity, and leverage distributions between two cohorts. Accounting for industry and regional effects matter for inference on the dynamics of the density. These results show that the importance of including qualitative information when applying FPCA to account for potential heterogeneity.