İstanbul ofis kira tahmin modeli geliştirilmesi.

In the last quarter of previous century, economic, social and technological development caused a change in employment structure and an increase in the share of service sector. Demand for modern office space was driven by faster growth in finance, insurance, real estate (FIRE) subsectors. Existing office inventory could not meet modern space requirement, hence new metropolitan areas has started to develop. Office demand has an important role in metropolitan urban development and emergence of new subcentres. Therefore, office rent prediction is a crucial issue in decision making of new office investments and planning strategies. Office rent prediction models have been the major concern of academic research since 1980s. Hedonic office rent prediction models which are the most common method based on multiple regression are well established in the literature. A wide range of variables, categorised as econometric, architectural, spatial and tenure rights, are used in these models for various cities. It is difficult to incorporate large number of variables into a simple mathematical model. As a result, the need has arisen to reduce or group the excessive number of variables to achieve simplier mathematical expressions with greater explanatory power. In the light of previous studies, some difficulties can arise in gathering data and applying the hedonic theory. The major difficulty lies within the hedonic regression models is the multicollinearity problem that may exist between a large number of independent variables. The common solution is exclusion of some variables depending on significance level or using "stepwise" or "backward" procedure in regression models. Another problem with development of rent prediction model is selection of dependent variable. In the literature, asking rent is preferred in some models while contract rent or effective rent are used in others, as the dependent variable. It is reported that the use of contract or effective rent instead of asking rent, can provide more accurate predictions. However, it is difficult to obtain sufficient contract data from real estate firms, due to confidentiality and competition. The aim of this study is to examine the problem with construction of an office rent prediction model and development of a viable prediction model for Istanbul. For this purpose, a proposed regression model is constructed with using asking rent, gross and net contract rents, as dependent variable for 1996 2006 period. First, full model is developed with thirtyfour variables, then a reduced model is constructed by eliminating some variables using "backward" procedure in standard regression model. Finally, the performances of prediction models are compared according to Rsquared and tstatistics. In addition, Akaike Information Criteria and Schwarz Bayesian Criteria are also employed to test the accuracy of proposed models. Based on the findings of most accurate model, the significant variables are defined for Istanbul office market. The similarities and differences from literature findings are discussed. The results confirmed that use of contract rents instead of asking rents can provide robust predictions with higher explanatory power. Besides, the reduce models offer better solution for multicollinearity problem. Building and locational variables are found the most significant office rent determinants for Istanbul metropolitan areas. The findings point out importance of accessibility and locational prestige in site selection for new office investments. Especially, distance from the CBD and important transportation nodes (Bosphorus Bridge, highway connection) have an important role to explain in rental change, in line with global literature. Results reveal that secondary centres gain importance. This strengthens the assertion that the tendency for office investment in Istanbul is away from the traditional centre (CBD) and closer to secondary centres. However, rental values are still higher in the CBD. For further studies, it is aimed to increase number of data to obtain more accurate prediction model for Istanbul. It is expected that office rent prediction models will be helpful to determine new office areas. [ABSTRACT FROM AUTHOR]