The purpose of this study is to evaluate the scale of property tax losses due to the tax base within the framework of the results of the field study conducted in Ankara, Gölbaşı, and to demonstrate whether it is possible to determine the real tax value objectively with the mass appraisal system.

Property tax bases in Turkey are re-set every four years by the Real Estate Appraisal Commission, and these values are indexed by the Ministry of Treasury and Finance the following year, with half of the revaluation ratio determined according to the producer price index.

The tax base value of real estate in Turkey is an important indicator for not only real estate property tax revenues but also some other taxes’ revenues. The following examples can be given in this regard: Determination of the value of the real estate transferred by inheritance, the minimum amount of the title deed fee in real estate purchases, and the determination of contribution amount of land/house owners to the road, sewage, and water infrastructure expenses made by municipalities. According to Article 73 of the Income Tax Code, the total amount of annual rental income can’t be less than 5% of the property tax value. As of January 1, 2021, a new additional housing tax for residential real estate with a property tax value over TRY 5 million will be applicable in Turkey.

In recent years, some legislative changes have been made regarding valuation criteria. One of the most important of these is determining the real estate value by mass appraisal methods, which can be defined as the valuation process performed through models established and tested using statistical methods over a selected sample set when more than one property needs to be valued simultaneously. Its process consists of the following stages preparing the data set (data collection, control, coding), choosing the variables (factors affecting the value), establishing the appropriate model, testing the model (model performance criteria), controlling and reporting the predicted values. 

In the field study, the land sales made by the public authorities in 2017 in Gölbaşı, where housing has increased intensively in recent years, were examined to show the scale of the difference between tax and sales values of real estate. According to the result of Multiple Regression Analysis and Artificial Neural Networks models, which are the mass appraisal methods, the sale price or value of a real estate varies depending on its characteristics. It’s also understood from the data set that the property tax value varies according to the property’s location.

According to the multiple regression model, the factors affecting the unit value are: the area, the total construction rights, plan function, proximity to infrastructure services and social reinforcement areas, road front, being within the boundaries of a special environmental protection zone, and sales legislation. 

In the artificial neural network analysis, there were independent variables affecting the unit price in the input layer and unit price in the output layer. The number of neurons from 1 to 15 in the hidden layer was defined and each model was run ten times, a total of 150 trials were made, and the best model was achieved with a 6-neuron model that gives the smallest Mean Square Error value (.0173). According to the normalized importance order of the independent variables in the model, it was seen that the most important variable affecting the unit price was the region, followed by total construction right, sales legislation, parcel area and being within the boundaries of the special environmental protection zone.

While it has been revealed by mass appraisal models that many factors affect the land unit values as well as the spatial characteristics, it is clearly understood from some examples in the data set that an assessment is made only according to the location of the property in determining the property tax values. On the other hand, according to the results of the examination, it is seen that the realized purchase and sale values in Gölbaşı district correspond to an average of 12.34 times the tax value. The fact that the real estate tax value is very low compared to the real property sales value is particularly important in terms of showing the scale of the loss in tax and fee income received by the central and local governments from real estate.

It is clear that the government will benefit from the mass appraisal practices for the valuation of the real estate tax bases in Turkey. This method, based on international standards, would determine their values close to market values, with less cost and different from based on subjective judgments of a real estate appraiser. This will also contribute to the increase in tax revenues of local governments and central budget revenues by establishing a fair tax system