Examining estimator bias and efficiency for pseudo panel data: a Monte Carlo simulation approach.

Pseudo panel data have been increasingly applied in empirical research as an alternative approach to a longitudinal analysis when genuine panel data are unavailable. However, conventional techniques are typically used to estimate pseudo panel data models without careful consideration to some unique properties of pseudo panel data. Ignoring properties such as time-varying cohort effects, a small number of constructed cohorts, large between-group variance, and trade-offs between cohort sizes and number of cohorts potentially lead to estimation bias or inefficiency not observed in genuine panel data. This paper presents a Monte Carlo experiment with scenarios that are designed to generate, under conditions of limited observations, various data possessing pseudo panel data characteristics, and evaluates the performances of various estimators using the simulation results. The main research findings are that the large between-group variation of the exogenous variable and the variance of fixed group effects in pseudo panel data are the primary causes of estimation bias and inefficiency. Other factors including the cohort size and potential non-spherical errors have a smaller impact on the estimators’ performances. An empirical application using Sydney Household Travel Survey data is also presented to illustrate the simulation findings.

• Record URL:
  http://www.itls.usyd.edu.au/research/publications/working_papers
• Corporate Authors:

  University of Sydney. Institute of Transport and Logistics Studies

  University of Sydney, 144 Burren Street, Newtown, New South Wales, 2042, Australia
  Sydney, New South Wales   
• Authors:
  • Tsai, C -
  • Leong, W
  • Mulley, C
  • Clifton, G
• Publication Date: 2012-4

Language

• English

Media Info

• Pagination: 16p
• Serial:
  • Issue Number: ITLS-WP-12-07

Subject/Index Terms

• TRT Terms: Behavior; Data analysis; Economic efficiency; Human subject testing; Reliability; Research; Research projects; Travel; Travel behavior
• Uncontrolled Terms: Efficiency
• Geographic Terms: Sydney (Australia)
• ATRI Terms: Cohort study; Data analysis; Efficiency; Reliability; Research method; Travel behaviour
• ITRD Terms: 9102: Method
• Subject Areas: Research; I72: Traffic and Transport Planning;

Filing Info

• Accession Number: 01374120
• Record Type: Publication
• Source Agency: ARRB
• Files: ITRD, ATRI
• Created Date: Jun 27 2012 9:56AM