Impacts of Traffic Congestion on Regional Production Efficiency: Cases of U.S. Urban Area




Yuan, Junyang

Journal Title

Journal ISSN

Volume Title



This dissertation aims to provide another way to evaluate traffic congestion's impacts on regional economy besides of traditional method of calculating congestion costs in terms of traffic delay and wasted fuel which has encountered considerable critiques on debatable definition and measurement of both benchmark speed and value of time. To additionally measure traffic congestion's indirect and long-term influences, an econometric approach is applied in this study. Since traffic congestion should be considered as a factor that may affect the efficiency of production procedure rather than a direct input in production function, a two-step approach is implemented here. Initially, a Translog production function model with three inputs is applied to calculate the Total Factor Productivity (TFP) growth. Sequentially, regression analysis is conducted to detect traffic congestion's impacts on TFP growth. To comprehensively investigate congestion's influences, a stochastic frontier analysis is further introduced to decompose the TFP growth into technical change, scale efficiency change and technical efficiency change. The relationship between traffic congestion and each component of TFP growth is then probed into. To verify the results from the parametric analysis, a non-parametric analysis is also applied in this dissertation. The Malmquist productivity index as well as its components is calculated using the Data Envelopment Analysis (DEA), and then each of them is regressed based on the same information applied in the parametric analysis. Results derived from both methodologies are compared at the end. The data covers 31 large and very large American urban areas, boundaries of which are defined by the author using ArcGIS software, during the period from 1990 to 2009. Influenced by the conversion from SIC to NAICS in later 1990s, in order to keep data's consistency, the research periods is divided into two segments: 1990-2000 and 2001-2009, which also provides an opportunity to make comparisons between these two periods. After considering various econometric and statistical issues, such as stationarity, spatiality, multi-collinearity, heteroscedasticity, auto-correlation, cross-sectional dependence, and endogeneity, a SYS-GMM (System General Moment Methodology) is implemented and corresponding results show that traffic congestion has significantly negative impacts on TFP growth during 1990 and 2000, while this impact becomes positive and still significant in the successive period. Moreover, congestion didn't affect technical change significantly in both periods. Its impact on technical efficiency change became trivial after 2000, though it negatively influenced this component in the previous period. For scale efficiency change, traffic congestion seems like a positive contribution continuously. Both parametric and non-parametric analysis provide similar results, though still tiny differences exist. Three possible explanations are provided correspondingly: (1) adaptation and adjustment to congestion in a long run; (2) "Hidden-behind" factors of traffic congestion; and (3) Redistribution effects in urban areas. This dissertation also shed light on various ways in mitigating urban traffic congestion presently, and emphasized the pros and cons of implementing traffic congestion pricing which is economically welcomed, but politically objected (in general) in policy implication.



Regional studies, Economics, Statistics, Malmquist productivity index, Regional economic efficiency, Road pricing, Total factor productivity, Traffic congestion