Explanation of Structural Complexities in Tehran’s Population Projections

Projecting the spatial distribution of population and employment is the basis of urban planning, particularly in metropolitan areas. Errors in these projections may lead to significant economic, social, political and environmental costs for local authorities, residents and businesses as well as other stakeholders. In this paper, we first evaluate the accuracy of population projections of Tehran’s master plan (2007), Tehran’s transportation and traffic master plan (2004) and Tehran’s transcendence plan also known as Tehran’s second master plan (1992) against the results of 1996, 2006 and 2016 national population and housing censuses. Measures used include coefficient of determination (R squared) for evaluating goodness of fit, Mean Absolute Percentage Error (MAPE) and Root Mean Squared Error (RMSE) for measuring average relative errors, number of outliers (extreme errors) and bias (prevalence of over- or underestimates). Then, potential factors leading to projection errors as well as variables potentially affecting spatial distribution of population in urban regions which have been neglected in previous projection models are investigated. For this purpose, a number of inferential statistical analysis tools including means comparison, linear regression coefficient and goodness-of-fit hypothesis tests have been applied. The results of the aforementioned analyses are as follows: 1) performance of all projections as measured by relative error, number of extreme errors (outliers) and bias in projections are not satisfactory, 2) amongst potential factors affecting the accuracy of these projections, length of projection period and mutual effects of sectional projections on the performance of each other are found to be significant, while other factors including the effects of inaccurate inputs and magnitude and direction of changes in population are insignificant, 3) amongst the potential factors affecting increases / decreases in population, area of the region’s urban decay is found to have significant negative effects on the tendency of people moving to the region (i.e. leading to lower population growth rates and in some cases decreases in the region’s population). Additionally, distance from the Central Business District (CBD) is found to have significant positive effects on population growth. Other potential factors including area of the region, population density, transit accessibility, perceived conditions of environment, urban services and entertainment facilities, average property prices and changes in property prices are not significant. Accordingly, it seems that most factors regarded to affect spatial distribution of population are not addressed in previous projections and simple Land Use Transportation Interaction (LUTI) models are not capable of accurately projecting the spatial distribution of the population. In order to improve the performance of the population projections, implementing register-based up-to-date urban databases including economic, social and demographic data of the residents and businesses as well as shortening the projection and consequently planning horizons are suggested. Furthermore, it is recommended that population’s tendency to avoid settling in urban decay areas as well as area adjacent to activity centers be regarded in future population projections as well as urban planning practices. Finally, it is highly recommended that instead of sectional projections, local authorities collaborate in projecting future population and employment and unanimously adopt the results.