Nikolai Siniak
Associate professor of Belarusian State Technological University,
Member of gif German Society of Property Researchers
E-mail:
siniakn@mail.ru
siniakn@tut.by
“Paper presented at the seminar ”Advances in mass appraisal methods”, Delft, The Netherlands, 30-31, October, 2006. 2
Mass appraisal in Belarus used mostly for
taxes
Mass appraisal
Land
(official mass appraisal)
Flats
(scientific work)
Purpose of the research is with a help of existing examples (Belarusian
experience of mass valuation) to find a disadvantages of traditional approaches
in mass appraisal and consider a possibilities to use fuzzy system in mass
appraisal. We come to the conclusion on the existing stage of development
fuzzy system that we should combine traditional statistic approaches (for
example multyregression techniques) with fuzzy logic and neural system. We
can use fuzzy in individual valuation or in mass appraisal if the valuation of
factors is making at time, when their future estimation is labored (has not a
sufficient probabilistic bases). We also should develop fuzzy approaches and
methods and first of all computers fuzzy technologies. It is necessary to
develop big the program "of fuzzy researches directed on achievement by
valuation and real estate economy sphere of a qualitatively new level of
self-consciousness”. These questions can be solved, if it is created and within 2-3
years minimally financed working group of the researchers in area Fuzzy Sets
and Control.
208 cities and
towns
24 000 villages
It was completed cadastral
valuation of the lands of
5 000 garden
companies
4
• The Main methdological principle
of the methods cadastral valuation
of the town lands - an
Briefly technology cadastral
estimations of the lands of the
populated points possible to describe
the following operation (the stage):
• The Stage 1. Shaping the thematic layers to source
information in format ArcView GIS.
• The Stage 2. Merit zoning.
• The Stage 3. The Determination of market value test land
area.
• The Stage 4. The Determination of the base cost of the lands.
• The Stage 5. The Determination cadastr cost of the merit
6
Conclusions of 1
st
part
• Technology of cadastral valuation of
the town lands in Republic Belarus
provides
broad
using
Research purpose of flat mass apprisal
Jointly with Sergio Coppiello (Italy)
8
Analysis model
Mass appraisal
Mass appraisal models
•
The models for mass appraisal are the two following:
•
P =
+
j
• X
j
+
[4]
•
P’ =
+
j
• X
j
+
[5]
•
where P represent total house prices, P’ house prices per
10
Mass appraisal models
•
In statistical terms [4] and [5] are linear models, which can be
handled with standard regression techniques, namely least
quares estimators and test of significance based on t-statistics.
•
Models are tested both on original variables and on normalized
variables [6], in order to highlight their relative importance:
•
X
ij
= 2 • (X
ij
– X’
j
) / (X
maxj
– X
minj
)
[6]
Empirical evidences:
total price model
•
The model results for total price of house are:
P = –13.583 +665 X
3+1.056 X
5+1.764 X
8+31 X
9+3.239 X
11+
(16,06) (79,87) (13,83) (6,96) (2,89) (11,30)
+2.228 X
14–1.272 X
16+10.365 X
17–983 X
23+4.522 X
24
(4,42) (3,40) (8,12) (3,02) (12,66)
12
Variables of analysis
Variables
P
supply price ($)
P’
supply price ($ per square meter)
X
1number of rooms
X
2Neighborhood
X
3total square meters of flat
X
4square meters of the living part
X
5square meters of kitchen
X
6number of floor of flat
X
7number of floors of building
X
8flat not at first or last floor
X
9age of the building in year
X
10prefabricated building
X
11brick building
X
12cement building
Variables
X
13monolith building
X
14presence of phone
X
15one separate toilet and bathroom
X
16one common toilet and bathroom
X
17two separate toilet and bathroom
X
18two common toilet and bathroom
X
19presence of balcony
X
20presence of loggia
X
21presence of veranda
X
22a closed one of X
19, X
20, X
21X
23wood as floor surface
• the model can not be ideal, since it is based not
on 100% market information: incorrect deals,
bad database brings about that that in 9 events
model gives close to ideal results, but in tenth -
surge. Particularly this often occurs on unique
object.
14
Reasons of inaccuracy of the valuation
• 1. Analysis market and collection of data is
conducted on the first stage.
• Database is formed on the second stage.
• At this stage, inaccuracy appears at a rate of
formalizations. For example, the factor wall
building is assigned as panel, block, brick,
wood and multifunction wall.
16
• The Third stage - a choice of the type of
• In the step of calibrations appears the last type of
inaccuracy connected with choice of the vector of
initial importances for iterative methods, with number
by cast-off filter data, finally, with number of
iterations.
• In the course of mass valuation of flats it was found
out inaccuracy about 4 %. From the words of mass
valuation specialists inaccuracy is available up 5-10
%.
18
Cadastral value can be defined:
normative method
expert method
• VL = PV + (I - E) : R (*),
CALCULATION of the CAPITALIZED VALUE
Ретроспектива по годам
Parameters of calculation per year
5 year
4 year
3 year
2 year
1 year
rate of the discounts-Kd
17,0%
Real rate of growth of the income on years- G
39,7%
16,6%
10,1%
7,1%
Difference between rate of the discounts and
rate of growth on year
20
On the graph you can see that change the rate of the
discounts or rate of growth on 1% changes importance of the
cost of the valued object on 8-20%. Thereby, mistake in value
of the rate of the discounts in amount more than two percents
brings about appearance of the mistake in the price of the
estimation in 16-40%.
Dependency net capitalized income (Ck) from rate of incom growth on
years (G) and discounting rates (Kd)
230,9 184,7 153,9 131,9 115,4 102,6 92,4 84,0 77,0 71,0 131,9 115,4 102,6 92,4 84,0 77,0 71,0 66,0 61,6 57,7 0,0 50,0 100,0 150,0 200,0 250,0 | 1% | | 2% | | 3% | | 4% | | 5% | | 6% | | 7% | | 8% | | 9% | | 10% |
Rate of ne t income growth on ye ars (G)
Not good statistical explanation
Average price of 1 sq.м. dweling of Мinsk ($)
y = -0,0113x
6+ 0,5029x
5- 8,5284x
4+ 70,938x
3-
314,44x
2+ 749,15x - 401,32
R
2= 0,9923
200
400
600
800
1000
1200
Real estate prices
22
Tendency of growth
0
200
400
600
800
1000
1200
19
92
19
94
19
96
19
98
20
00
20
02
20
04
U
SD
p
e
r
s
q
u
a
re
m
e
te
rs
1000
2000
3000
4000
5000
6000
7000
8000
9000
Eu
ro
p
e
r
s
q
u
a
re
m
e
te
rs
24
Literature research. What kind of fuzzy
is available?
• From "Fuzzy Sets" in 1965 to Perception-Based Theory... "
(Zadeh, 2000).
• "Linear Systems Theory-The State Space Approach"(1963)
"Frequency Analysis"(1950), "Wiener's theory of
prediction"(1950), "Sample-Data Systems"(1952), "Probability
Measures of Fuzzy Events" (1968), "Outline of a new
Approach to the Analysis of Complex Systems and Decision
Processes"(1973), "Fuzzy Sets as a Basis for a Theory of
Possibility"(1978), "A Theory of Approximate
Reasoning"(1979), "The Role of Fuzzy Logic in the
Management of Uncertainty in Expert Systems" (1983),
"Fuzzy sets" (1985) etc. (Zadeh)
• Fuzzy mathematics, cognitive and decision process
were being to be developed by Kaufmann(1975),
Zadeh, Fu, Tanaka, Shimura(1975), Neogita and
Ralescu, (1975).
• D.Dubois and H.Prade, 1980
26
Very simple to use
valuation cost =
(9,2,10,11)/(0,14;0,15;0,16).
D=[9,2+0,8a,11-1a]/[0,14+0,01a,0,16-0,01a]=
=[(9,2+0,8a)/(0,16-0,01a),(11-1a)/(0,14+0,01a)].
It is evident that the property cost is not lower $ 57500 and not above $ 78571.
With 100% certainty, we can state that the cost of the estimated property is $66 666.
28
Disadvantages of fuzzy system:
• 1. Absence of understanding in mass appraisal
specialists.
• 2. Absence of special computer programs
available like Excel. The best way out is to
integrate fuzzy system in Microsoft Office and
Excel.
For mass appraisal
• On the existing stage of development fuzzy system that we
should combine traditional statistic approaches (for
example multyregression techniques) with fuzzy logic and
neural system.
30
Expert’s opinions
Experts
Income (V)
Costs (C)
Net income (I)
We a help of average fuzzy numbers
• Valuation of I expert
• With a help of
-cuts
32
)
636
,
619
,
596
(
)
6360
,
6190
,
5960
(
10
1
1
n
i
C
)
96
,
40
,
5
(
)
960
,
400
,
50
(
10
1
1
n
i
I
The most probable income is 40. Than we put the figure in formula
* and get exact result. Than we use traditional approaches.
Conclusions
Professor Zadeh
In humanistic systems, human reasoning and decision making is not just
"measurement" based, as we are taught through out our academic education,
rather "perception" based. "Fuzzy Sets" in 1965 and came to surface toward the
beginning of this Millennium in "Toward a Perception-Based Theory... " (Zadeh,
2000).
Conclusion: from the above analysis we can state that the
application of fuzzy numbers in the process of property evaluation
enables to determine property value with much higher probability
(100%) in comparison with the traditional approaches of evaluation
34
We need to work on further developments of fuzzy theory in
particular on fuzzy knowledge representation and reasoning
in real estate field. This is more acutely needed in the
development of humanistic decision making domains which
Professor Zadeh have been urging us to direct our attention
over the last thirty five years or so.
We need to create of an fuzzy economy, valuation and management of real estate
organization. It is aimed to establish an institute, which accepts fuzzy valuation
and economy as a profession, to control and manage the applications in respect
of education, rules and standards. It is also aimed to standardize fuzzy
valuation, approaches, rules and factors which must be taken care during the
fuzzy valuation and management. We also need to provide by methodology and
standards valuers and mass appraisal issues and define the role fuzzy valuation
in investment decision-making process.
36
Welcome to Belarus
You will have a chance to see some from Belarusian
architecture and hospitality of Belarusian people.
Conferences:
1. International real estate conference
15-17 November 2006, Minsk
For more detail: www.expozona.lt
2. The 2-d International Conference
ECONOMY, VALUATION AND MANAGEMENT OF THE REAL ESTATE AND
NATURAL RESOURCES
Minsk, May, 3-5, 2007.
38
• 1.Bellman R., Zadeh L. “Decision-Making in a Fuzzy
Environment”, Management Science (17), pp. 141- 164,
(1970)
• 2. Zadeh L. “Fuzzy Sets”, Information and Control (8), pp.
338-353, (1965)
• 3. Zadeh L.A., The concept of a linguistic variable and
its application to approximate reasoning I, II, III, Inf.
Sci., 8(1975),199-257, 301-357; 9(1975), 43-80.
• 4. Novak V., Fuzzy logic as a basis of approximate reasoning.
In: Zadeh L.A., Kacprzyk J. Fuzzy Logic for the Management
of Uncertainty. Wiley & Sons, New York 1992.
• 6.
I.BURHAN TÜRKŞEN. Operations Research and
Management Science Applications of Fuzzy Theory.
University of Toronto, 2000.
• 7. L.A. Zadeh, "Toward a Perception-Based
Theory of Probabilistic Reasoning", Key note
address; Fourth International Conference on
Applications of Fuzzy Systems and Soft
Computing, June 27-29, Siegen, Germany,
40