Abstract
Many urban problems are attributed to inappropriate comprehensive planning which fails to account for the dynamic and inter-related nature of urban activities in its forecasts. While many studies have attempted to integrate urban activities using structural models, the time series approach has received little attention. Traditional forecasting approaches, with their highly restricted structural frameworks, reveal an inefficiency in integrating forecasts for urban activities. The purpose of this study was to develop an integrated forecasting model which efficiently captures the dynamic effects and trends of urban development. Recent development of the Vector Autoregressive Representation (VAR) model in the field of economics makes it possible to analyze multi-variate time series data using theoretical intuition. The model is appropriate for an integrated forecasting analysis because of its efficient mode and policy-endogeneity. Although there are countless urban activities, eight was the maximum number of variables to identify the dynamic urban structure because of small degrees of freedom. The chosen variables are: population, factory employment, fiscal deficit, land price index, length of paved road, permitted area of new building construction, area of farmland and forestry, and number of school classrooms. After comparing performances among three alternative VAR models, the Bayesian approach was selected as the most appropriate model for this study. After the forecasting model was obtained, comparison of unconditional forecasts with forecasts conditional upon a policy change, provided the basis to conduct policy analysis. An impulse response function technique was applied to analyze the dynamic response of each endogenous variable to a shock to the system. According to the forecasts in this study, the future of Incheon, Korea is characterized as continuing its explosive pace towards metropolis status. On the other hand, the factory employment and the agricultural land use is decreased during the forecasting period. The results indicate some trade-off relationships between activities. The results also reveal the significant gap between the ideal planned values and the forecasts.
Yoo, Jae Yoon (1992). Integrated forecasting analysis for comprehensive urban planning using vector autoregressive model. Texas A&M University. Texas A&M University. Libraries. Available electronically from
https : / /hdl .handle .net /1969 .1 /DISSERTATIONS -1348982.