•  
  •  
 

Abstract

Integrating time-series and cross-sectional data represents a sophisticated approach in modern quantitative econometric analysis and statistical methodologies used to examine various phenomena, particularly in economics. These two data types are fundamentally complementary, providing a more comprehensive analytical framework than those relying on a single dimension; this approach captures both specific temporal snapshots characteristic of cross-sectional data and the dynamic evolutionary patterns inherent in time-series data. While time-series data is theoretically suitable for estimating economic relationships, its practical application is often hindered by significant challenges, most notably multicollinearity among explanatory variables that tend to co-move over time, leading to imprecise estimations. Conversely, cross-sectional data alone may fail to yield robust estimates for certain variables that remain relatively constant over short durations. Consequently, when both data types are available, researchers focus on optimal integration methods through two primary scenarios: the first involves distinct datasets where multicollinearity prevents the estimation of all parameters in a time-series regression, necessitating the use of cross-sectional information for certain parameters followed by integration using methods such as those developed by Tobin, Koutsoyiannis, Durbin, or Theil and Goldberger. The second scenario occurs when time-series of cross-sections are available, utilizing alternative frameworks such as the traditional pooling model, the aggregation model, dummy variable models, error components models, seemingly unrelated regression (SUR), or random coefficients models; this research specifically focuses on this second scenario, examining the first four of these models in detail.

DOI

10.33095/d2c7pk07

Subject Area

Statistical

First Page

276

Last Page

294

Rights

Copyright (c) 2026 Journal of Economics and Administrative Sciences

Download

Share

COinS