MAIN POSSIBILITIES TO USE THE STATISTIC METHODS IN REAL ESTATE VALUATION Assoc. Prof. Ciprian Şipoş, Ph. D. Prof. Alexandru Buglea, Ph. D. Eng. Adrian Crivii, Ph. D. , FRICS West University of Timisoara Darian Rom Suisse Romania e-mail: ciprian. sipos@feaa. uvt. ro alexandru. buglea@feaa. uvt. ro acrivii@darian. ro ERES Conference, Stockholm 24 -27 June 2009
Introduction Using statistical methods in valuating real estate properties is stipulated and clearly regulated in the international valuation standards. The Guidance Notes dedicated to statistical valuation say that the “Mass Appraisal Process may be used as a methodology for Ad Valorem Property Taxation, or statistical and economic studies under government administrative programs”. Assessments through statistical methods are based on market value as defined in international valuation standards. The databases used must consider the categories of properties valuated according to geographical position, property type, technical and economic issues of the property or local taxation laws. Statistical methods selected in this paper with best chance of being successfully used in assessing real estate are the type of confidence intervals based on statistical tests, linear or non-linear multiple regression models and analytical models based on time series. ERES Conference, Stockholm 24 -27 June 2009 2 / 16
Confidence intervals for the property market value Determining the price of a property with statistical confidence intervals is based on the estimation theory. Estimating the confidence interval of the price of a property consists in determining, with a certain probability, a range of values within the unknown property price is. Estimating the confidence interval consist, in fact, in determining the price limits for a given confidence coefficient. The limits of the confidence interval are defined as lower limit (LL) and upper limit (UL). In these conditions, the probability that the unknown price of a property (P) is located inside of the confidence interval is defined as: p (LL P UL) = 1 – (1) If the sample volume is less than or equal to 30 units and the variance of the population statistics is unknown (most often case in the practice), then the confidence interval for mean is determined using Student distribution with ν degrees of freedom. ERES Conference, Stockholm 24 -27 June 2009 3 / 16
Confidence intervals for the property market value Starting from the relation 1, were determined confidence intervals for the prices of apartments in Timisoara, at 95% confidence level, grouped according to two fundamental criteria: number of rooms and the area they are located in the city as can be seen in Table 1: ERES Conference, Stockholm 24 -27 June 2009 4 / 16
Confidence intervals for the property market value Table 1 ERES Conference, Stockholm 24 -27 June 2009 5 / 16
Confidence intervals for the property market value The upper and lower margins prices are statistically reliable. This means that 95% of the market values recorded are included in these intervals. Upper and lower values of the confidence intervals must not be confused with the maximum and minimum recorded on the market. The minimum and maximum prices recorded on the market are the extreme prices and are included in that 5% values which are considered statistic insignificant. This is exactly the role of confidence intervals: the exclusion of the accident occurring on the market, which represents exceptions from rules. As can be seen from Table 1, the fewer transactions were recorded in the case of the more expensive apartments. In this category enter the apartments located in city centre and the apartments with four rooms from almost all city areas. This is largely due to cash crisis on the real estate market and very strict conditions for granting property loans. It can be observed that are also low traded apartments with one room, which are used largely for the purpose of renting. Timisoara is one of the largest university cities in Romania, which makes the rent market of apartments, especially with one room, to be very developed. ERES Conference, Stockholm 24 -27 June 2009 6 / 16
The multiple regression model of valuation The multiple regression models start from the idea that several explanatory variables can be used to predict the value of a dependent variable. The theoretical bases is the linear multiple regression models. The dependent variable in the case of valuation process is the value of a property. The valuation model intends to predict the value of the apartments in the city of Timisoara. The panel data is structured considering as explanatory variables the city areas, the number of rooms, the floor space, the improvements level, the floor level of the apartments and the comfort of the building. It must be underlined that the coefficient’s estimated values and their explanations are not generally valid. They are corresponding only to studied sample at Timisoara city’s level. To apply the proposed methodology to other cities is necessary to create a consistent database referring to the place and the period analysed. The database has to contain relevant information about all the variables of the model. ERES Conference, Stockholm 24 -27 June 2009 7 / 16
The multiple regression model of valuation The city areas are delimited as it is given in the official documents of the Timisoara’s mayoralty and adjusted considering the valuator’s experience. There were selected a number of eight major city areas, descending sorted by medium price. The number of rooms varies between one and four rooms that represents the most common apartments (flats) in Timisoara and, generally, in Romania. The floor space is measured in square meters and means the useful surface of the apartment. The improvements level is diversified, but for the statistical purposes are delimited three major categories: the superior class, the medium class and the low improvement level class. The apartment floors are classified in two categories: first category is the ground floor and the last floor; the second category is the intermediate floors. The building comfort is delimited in three categories: first class, second class and the third class. ERES Conference, Stockholm 24 -27 June 2009 8 / 16
The multiple regression model of valuation The valuation model is given by the equation: Pi = b 1 Ai + b 2 ln NRi + b 3 USi + b 4 Fi + b 5 Ii + b 6 BCi (2) where: Pi is the price of the apartment; Ai is the area of the city; NRi is the number of rooms; USi is the useful surface measured in square meters; Fi is the floor of the apartment; Ii is the improvements level; BCi is the building comfort; Starting from this equation will be estimated the coefficients of the model with very well known Least Square Method. The statistics of the model will obtain based upon a Regression Statistics and Analysis of variances methods. ERES Conference, Stockholm 24 -27 June 2009 9 / 16
The multiple regression model of valuation Table 2. ERES Conference, Stockholm 24 -27 June 2009 10 / 16
The multiple regression model of valuation In the Regression Statistics part of the Table 2, the values close to 1 of Multiple R and R Square mean that the prices of the apartments (Pi) was strongly influenced by the city areas (Ai), the number of rooms (NRi), the useful surface (USi), the floor level of the apartments (Fi), the improvements level (Ii) and the comfort of the building (BCi). . In the Analysis of variance part from the Table 2, the calculated value F is determined as report between value of apartments as a result of those six explanatory variables (Regression MS) and value of apartments as a result of actions of other factors, random or insignificant (Residual MS). It can be noticed that F is much higher than the value of the statistical test Fisher F critical, which shows a very significant econometric model. . The t Stat values for all estimated coefficients are higher than the value of the Student statistical test, t critical. That means that the estimated values of the 1, 2, …, 6 coefficients are fair from statistical point of view. . The greatest value of t Stat is for city area variable, that means that city area is the most important in the establishing the apartment’s value. The second variable is the number of rooms, followed in order by the improvements level, by the building comfort, by the floor level and, in the last position, by the floor surface. ERES Conference, Stockholm 24 -27 June 2009 11 / 16
Analytic valuation models based on time series Analytic methods that use statistical models to study the evolution in time of real estate prices are known as time series. They have the main characteristic expression of a trend time series based on a function as: Pt = f(t) + t (3) where Pt is the range of data studied, considered the outcome or dependent variable (price of property), t is the time factor, regarded as an independent variable, f is the mathematical function that shape the evolution in time of the phenomenon studied and t is the random variable, which shows the influence of random factors at the moment t. Using an analytical function of determining the property price trend sets with greater or lower accuracy the long-term development of the economic process or phenomenon studied, depending on the spreading of values around the central tendency. ERES Conference, Stockholm 24 -27 June 2009 12 / 16
Analytic valuation models based on time series Were eliminated from the analysis city areas 1, 2 and 8 because there were not enough registered transactions. Based on data from Timisoara were developed analytical functions of time for the city areas taken into account and were obtained the following results: Table 3. It can be seen in Table 3 that all values of parameters of time factor t are negative which shows that prices have had a descending trend in all areas analyzed in the city. Of course this trend is followed by the average price of apartments with two rooms calculated on the entire city. The R Square values are very close to 1 which means a strong linear trend. ERES Conference, Stockholm 24 -27 June 2009 13 / 16
Analytic valuation models based on time series Based on the linear trend it can be made a forecast of future developments. Forecast and the price developments in the areas analyzed are shown in following Figure: ERES Conference, Stockholm 24 -27 June 2009 14 / 16
Final issues Assessments of real estate brokers from Timisoara look like the real estate market is in deadlock. In the last three months compared to same period of last year the sales volume has dramatically decreased. If is considered the period since the beginning of the year compared to same period in 2008, the number of property transactions has fallen by at least 50%. According to agents of the companies, compared to last year have increased very much the number of owners who wish to sell their buildings. According to agency representatives, real estate developers have dropped prices by 15 to 20%. Sales have decreased, there is not a demand as large as last years and the real estate market in Timisoara is unstable. That was natural to have a saturation level, because prices were inflamed, unreal, were sold apartments with two rooms to 60. 000 Euro, the banks valuating them, in fact, at 80. 000 Euro, only to give more loans. In other words, has been an overvaluation. According to analysts, the market recorded a tendency to involution of prices and values are, on average, to the next level: a bachelor's rooms at 3035. 000 Euros, an apartment with a room at 40. 000 Euro, one with two rooms at 50. 000 Euro and with three rooms at 70. 000 Euro. ERES Conference, Stockholm 24 -27 June 2009 15 / 16
Final issues The prices of houses are also in involution - 100. 000 Euro without finishing and up to 300. 000 Euro if are completely finished. It must be stressed that last year in the same period a similar house worth 500. 000 Euro. In this situation appear also anomalies. For example, in a certain area of the city, some owners are expected to receive for a square meter of land even 1. 000 Euro. Who is intending to buy may choose from a generous offer and if there is a waiting owner, will buy at a reasonable price. Now in Timisoara are for sale at least 1. 000 apartments with three rooms, all tendered by the owners. Some people have come to give their car to go with flat. In recent months, about ten large estate agents in Timis County have closed their doors. All brokers from the real estate market believe that the market will unlock when the banks will again go to credits. ERES Conference, Stockholm 24 -27 June 2009 16 / 16
Thank you! ERES Conference, Stockholm 24 -27 June 2009
Скачать презентацию MAIN POSSIBILITIES TO USE THE STATISTIC METHODS IN
75b86ef7211440776919fe92c9e627f1.ppt