Professional Documents
Culture Documents
com
Decision Support
Abstract
This article presents an evaluation study of residential properties carried out together with real estate agents in the city of Volta Red-
onda, Brazil. The study aimed to dene a reference value for the rents of these properties using the TODIM method of Multicriteria
Decision Aiding. By applying this method to the ordering of properties with dierent characteristics, a ranking of all the properties
was obtained and, as a result of this, diverse ranges of rental values for the properties under analysis. The study was complemented
by an analysis of the sensitivity of the numerical results obtained.
2007 Elsevier B.V. All rights reserved.
1. Introduction these methods, the most commonly used are the Compara-
tive Method and the Income Method. In the Comparative
Brazil is the fth largest country in the world. It has a Method, similar properties at locations close to the prop-
population of around 190 million people and 4500 miles erty in question are checked, and based on comparisons
of coastline. Due to the scale of the Brazilian territory, with the data obtained, the value of the rent is established.
the real estate market presents a wide variety of economic This method is the most practiced in the real estate market
conditions as well as diering supply and demand charac- and the one which normally presents the value closest to
teristics. The percentage of housing decit varies according the market value. The Income Method denes the value
to the region of the country. It has historically been highest of the rent as a percentage of the sale value of the property.
in the Northeast and lowest in the Southern region of the It is simple to apply, but depending on the supply and
country. Real estate activity has been intense even in the demand in the region in question, may not generate com-
most developed Brazilian states such as Sao Paulo and petitive and practicable values in the market.
Rio de Janeiro. This article describes the use of a Multicriteria Decision
The rental evaluation of a property is one of the most Aiding method in evaluating residential properties available
important tasks for those who work in the Brazilian real for rent in the city of Volta Redonda, in the State of Rio de
estate market. This evaluation is generally based on quan- Janeiro, Brazil. In the majority of cases, the criteria used to
titative and qualitative criteria employing various simple make the evaluation of the alternatives are conicting. For
methods (Guimaraes Neto, 1992, Pessoa, 1992). Among example, what would be most valued: a small, old house
in an excellent residential neighborhood, close to the center
*
or a large, new property with a swimming pool and leisure
Corresponding author. Tel.: +55 21 45034053; fax: +55 21 25126173.
E-mail addresses: autran@ibmecrj.br (L.F. Autran Monteiro Gomes),
area in a neighborhood far from the downtown area? In this
duncan@metal.eeimvr.u.br (L.A. Duncan Rangel). study, the TODIM method (Gomes and Lima, 1992a,b;
1
Tel.: +55 24 33424896. Trotta et al., 1999) was used to order the alternatives of
0377-2217/$ - see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.ejor.2007.10.046
L.F. Autran Monteiro Gomes, L.A. Duncan Rangel / European Journal of Operational Research 193 (2009) 204211 205
the residential properties for rent. After the ordering, it From the construction of the aforementioned TODIM
becomes easier to dene the rental values, once previously additive dierence function, which functions as a multiat-
evaluated properties, that is, those with previously dened tribute value function and, as such, must also have its use
rental values, have been included in the set of alternatives. validated by the verication of the condition of mutual
Other examples of discrete multicriteria valuation can be preferential independence (Keeney and Raia, 1993;
seen in Stagl (2004, 2007) and Gamper et al. (2006). Clemen and Reilly, 2001), the method leads to a global
ordering of the alternatives. It can be observed that the
2. The TODIM method construction of the multiattribute value function, or addi-
tive dierence function, of the TODIM method is based
The TODIM method (an acronym in Portuguese of on a projection of the dierences between the values of
Interactive and Multicriteria Decision Making), conceived any two alternatives (perceived in relation to each crite-
in its current form at the beginning of the nineties, is a dis- rion) to a referential criterion or reference criterion.
crete multicriteria method based on Prospect Theory The TODIM method makes use of pair comparisons
(Kahneman and Tversky, 1979). This means that underly- between the decision criteria, using technically simple
ing the method is a psychological theory, which was the sub- resources to eliminate occasional inconsistencies arising
ject of the Nobel Prize for Economics awarded in 2002 from these comparisons. It also allows value judgments
(Roux, 2002). Thus, while practically all other multicriteria to be carried out in a verbal scale, using a criteria hierar-
methods start from the premise that the decision maker chy, fuzzy value judgments and making use of interdepen-
always looks for the solution corresponding to the maxi- dence relationships among the alternatives. It is a
mum of some global measure of value for example, the noncompensatory method in the sense that tradeos do
highest possible value of a multiattribute utility function, not occur (Bouyssou, 1986).
in the case of MAUT (Keeney and Raia, 1993) the Roy and Bouyssou (1993), talking about the TODIM
TODIM method makes use of a global measurement of method, state that it is: ...a method based on the French
value calculable by the application of the paradigm of Pros- School and the American School. It combines aspects of
pect Theory. In this way, the method is based on a descrip- the Multiattribute Utility Theory, of the AHP method
tion, proved by empirical evidence, of how people eectively and the ELECTRE methods (p. 638).
make decisions in the face of risk. Although not all multicri- The concept of introducing expressions of losses and
teria problems deal with risk, the shape of the value func- gains in the same multiattribute function, present in the
tion of the TODIM method is the same as the gain/loss formulation of the TODIM method, gives this method
function of Prospect Theory. The use of TODIM relies on some similarity to the PROMETHEE methods, which
a global multiattribute value function. This function is built make use of the notion of net outranking ow. Barba-
in parts, with their mathematical descriptions reproducing Romero and Pomerol (2000) have stated the following in
the gain/loss function of Prospect Theory. The global mul- respect of the TODIM method: it is based on a notion
tiattribute value function of TODIM then aggregates all extremely similar to a net ow, in the PROMETHEE
measures of gains and losses over all criteria. sense (p. 229).
In its calculations the TODIM method must test specic Consider a set of n alternatives to be ordered in the pres-
forms of the losses and gains functions. Each one of the ence of m quantitative or qualitative criteria, and assume
forms depends on the value of one single parameter. The that one of these criteria can be considered as the reference
forms, once validated empirically, serve to construct criterion. After the denition of these elements, experts are
the additive dierence function of the method. This notion asked to estimate, for each one of the qualitative criteria c,
of an additive utility function is taken from Tversky (1969). the contribution of each alternative i to the objective asso-
The additive dierence function is indeed a global multiat- ciated with the criterion. This method requires the values of
tribute value function and reects the dominance measure- the evaluation, of the alternatives in relation to the criteria,
ments of each alternative over each other alternative. In to be numerical and to be normalized; consequently the
this sense, TODIM maintains a similarity with outranking qualitative criteria evaluated in a verbal scale are trans-
methods, such as PROMETHEE ([Brans and Mareschal, formed into a cardinal scale. The evaluations of the quan-
1990), because the global value of each alternative is rela- titative criteria are obtained from the performance of the
tive to its dominance over other alternatives in the set. alternatives in relation to the criteria, such as, for example,
Although it appears complicated to have to test the validity the level of noise measured in decibels, the power of an
of the application of the paradigm to the database, which engine measured in horsepower or a students mark in a
may on occasions oblige the decision analyst to use other subject, etc.
forms of the losses and gains functions, in fact it is not TODIM can therefore be used for qualitative as well as
so. Since the rst practical uses of the TODIM method, quantitative criteria. Verbal scales of qualitative criteria
back in the nineties, the same two mathematical forms have are converted to cardinal ones and both types of scales
been used successfully, and have been validated empirically are normalized. The relative measure of dominance of one
in dierent applications (Gomes and Lima, 1992a,b; Trotta alternative over another is found for each pair of alterna-
et al., 1999). tives. This measure is computed as the sum over all criteria
206 L.F. Autran Monteiro Gomes, L.A. Duncan Rangel / European Journal of Operational Research 193 (2009) 204211
of both relative gain/loss values for these alternatives. The Thus d(Ai, Aj) represents the measurement of dominance of
parts in this sum will be either gains, losses, or zeros, alternative Ai over alternative Aj; m is the number of crite-
depending on the performance of each alternative with ria; c is any criterion, for c = 1, . . . , m; wrc is equal to wc di-
respect to every criterion. vided by wr, where r is the reference criterion; Pic and Pjc
The evaluation of the alternatives in relation to all the cri- are, respectively, the performances of the alternatives Ai
teria produces the matrix of evaluation, where the values are and Aj in relation to c; h is the attenuation factor of the
all numerical. Their normalization is then performed, using, losses; dierent choices of h lead to dierent shapes of
for each criterion, the division of the value of one alternative the prospect theoretical value function in the negative
by the sum of all the alternatives. This normalization is car- quadrant.
ried out for each criterion, thus obtaining a matrix, where all The expression Uc(Ai, Aj) represents the parcel of the
the values are between zero and one. It is called the matrix of contribution of criterion c to function d(Ai, Aj), when com-
normalized alternatives scores against criteria. P = [Pnm], paring alternative i with alternative j. If the value of
with n indicating the number of alternatives and m the num- Pic Pjc is positive, it will represent a gain for the function
ber of criteria, as shown in Table 1. d(Ai, Aj) and, therefore the expression Uc(Ai, Aj) will be
After the attribution of the weights of the criteria and used, corresponding, that is, to Eq. (2). If Pic Pjc is nil,
their normalization, the partial matrices of dominance the value zero will be assigned to Uc(Ai, Aj), by applying
and the nal matrix of dominance must be calculated. Eq. (3). If Pic Pjc is negative, Uc(Ai, Aj) will be repre-
The decision makers must indicate which criterion r is to sented by Eq. (4). The construction of function Uc(Ai, Aj)
be chosen as the reference criterion for the calculations in fact permits an adjustment of the data of the problem
according to the relative importance assigned to each crite- to the value function of Prospect Theory, thus explaining
rion. In this way, the criterion with the highest value the aversion and the propensity to risk. This function has
accorded to its importance will usually be chosen as the ref- the shape of an S, represented in Fig. 1. Above the hor-
erence criterion. The weight of each criterion is determined izontal axis, considered as a reference for this analysis,
by the decision makers on a numerical scale (e.g., from 1 to there is a concave curve representing the gains, and, below
5) and is then normalized. Thus, wrc is the weight of crite- the horizontal axis, there is a convex curve representing the
rion c divided by the weight of the reference criterion r. losses. The concave part reects the aversion to risk in
Using wrc allows all pairs of dierences between perfor- the face of gains and the convex part, in turn, symbolizes
mance measurements to be translated into the same dimen- the propensity to risk when dealing with losses.
sion, i.e. that of the reference criterion. The measurement After the diverse partial matrices of dominance have
of dominance of each alternative Ai over each alternative been calculated, one for each criterion, the nal dominance
Aj, now incorporated to Prospect Theory, is given by the matrix of the general element d(Ai, Aj) is obtained, through
mathematical expression the sum of the elements of the diverse matrices.
Expression (2) is used to determine the overall value of
X
m
alternative i through normalization of the corresponding
dAi ; Aj Uc Ai ; Aj 8i; j: 1
dominance measurements. The rank of every alternative
c1
originates from the ordering of their respective values.
When:
Value
s
wrc P ic P jc
Pm if P ic P jc > 0; 2
wrc
c1
Uc Ai ; Aj v if P ic P jc 0;
0 3
u m
u P
t wrc P jc P ic
1
h c1
wrc
if P ic P jc < 0; 4
Losses Gains
Table 1
Matrix of normalized alternatives scores against criteria
Alternatives Criteria
C1 C2 ... Cj ... Cm
A1 P11 P12 ... P1j ... P1 m
A2 P21 P22 ... P2j ... P2 m
... ... ... ... ... ...
Ai Pi1 Pi2 ... Pij ... Pim
... ... ... ... ... ...
An Pn1 Pn2 ... Pnj ... Pnm
Fig. 1. Value function of the TODIM method (Gomes and Lima, 1992a).
L.F. Autran Monteiro Gomes, L.A. Duncan Rangel / European Journal of Operational Research 193 (2009) 204211 207
Pn Pn
j1 dAi ; Aj min j1 dAi ; Aj
Table 2
ni P n P n : 2 Valuation of criterion C1 location
max j1 dAi ; Aj min j1 dAi ; Aj
Location Associated score
Therefore, the global measures obtained computed by (2) Periphery 1
permit the complete rank ordering of all alternatives. A Periphery/Average location 2
sensitivity analysis should then be applied to verify the sta- Average location 3
bility of the results based on the decision makers prefer- Good location 4
Excellent location 5
ences. The sensitivity analysis should therefore be carried
out on h as well as on the criteria weights, the choice of
the reference criterion, and performance evaluations. dened as presented in Table 3. Considering a property
which has been constructed to a high standard of nish-
3. Case study ing, the property receives a score equal to 3, the maximum
value for this evaluation criterion.
The city of Volta Redonda is situated in the South of the C4 State of conservation: This is a qualitative criterion
State of Rio de Janeiro, Brazil. It has approximately used to evaluate the general state of the property. This cri-
256,000 inhabitants and there are currently a large number terion considers whether any work has been carried out on
of properties, residential and commercial, rented or avail- the property, as well as habitation conditions. Table 4, as
able for rent. follows, presents the associated scores for the state of con-
The study sought to determine a reference value for the servation of the property. For example, an apartment t
rent of residential properties, serving in this way as an for habitation, needing minor repairs, in other words in
important source of information for landlords and real an average state of conservation, receives a score equal
estate agents. A set of 6 alternatives with known rental val- to 2. This criterion is not directly related to criterion C3,
ues was used as a guideline. Computations in this case as the state of conservation is independent of the standard
study were performed using an Excel-based spreadsheet. of nishing of the property. A property might have a high
Commercial software for the calculations of the TODIM standard of nishing, using high quality material, but be in
method is currently under development, though. In order need of repairs because of its age.
to evaluate the properties, it was necessary to identify C5 Number of Garage spaces: This is a quantitative
which criteria should be considered in the study. For this criterion much valued by people who need to rent a prop-
purpose, research was carried out together with several real erty and have one or more cars.
estate companies in the city of Volta Redonda and inter- C6 Number of rooms: This is a quantitative criterion
views conducted with specialists in the eld of property which considers for valuation the number of rooms in the
evaluations. property: living rooms, bedrooms, kitchen/breakfast area
and bathrooms. The dierence between this criterion and
3.1. Denition of the criteria criterion C2, which deals with the area constructed, should
be highlighted. It is important to evaluate the property by
Based on the points of view expressed in the interviews the number of rooms, as two properties of the same total
with realtors, it was concluded that eight criteria should size may be in question, but one of them may have large
be selected for this particular application. All other possi- rooms and the other, a larger number of smaller rooms.
ble criteria would already be reected in these eight criteria, C7 Attractions: This is a qualitative criterion which
given the characteristics of the study area. The eight most considers the existence of leisure areas such as swimming
important evaluation criteria for this analysis were identi-
ed, and are described as follows:
Table 3
C1 Location: This is the most important evaluation Valuation of criterion C3 construction quality
criterion for rental properties. It is a qualitative criterion Construction quality Associated score
which seeks to dene whether the property is in a valued Low standard of nishing 1
location. The scale presented in Table 2 as Associated Average standard of nishing 2
Score was dened to evaluate the performance of the alter- High standard of nishing 3
natives in relation to this criterion. In this way, a property
evaluated as being in a good location, receives a score
equal to 4. Table 4
C2 Constructed area: This is a quantitative criterion Valuation of criterion C4 state of conservation
which describes the size of the property constructed. In this State of conservation Associated score
study, the unit of measurement was square meters (m2). Bad 1
C3 Construction quality: This is a qualitative criterion Average 2
which denes the standard of nishing of the property. Good 3
Very good 4
Three levels of nishing and their respective scores were
208 L.F. Autran Monteiro Gomes, L.A. Duncan Rangel / European Journal of Operational Research 193 (2009) 204211
Table 11 out, the value used for h was 1. In the sensitivity analysis
Properties with a previously dened rental value the value of h was altered from 1 to 5. When this alteration
Property Monthly rental value was carried out, the only change in the ordering of the val-
A3 US$ 214 ues occurred in the values of A13 with A15, indicating that,
A4 US$ 309 in spite of the alteration in the value of the attenuation
A5 US$ 712 index of losses, the ordering obtained was consistent. In
A9 US$ 133
A10 US$ 166
this particular application the decision makers felt that sen-
A11 US$ 414 sitivity analyses should be conducted on the weight of the
reference criterion and the value of h only.
4. Conclusion
method, A8 is in the 8th place, A3 is in the 9th place, with a
rental value of US$ 214, and A4 is in the 7th place, and
The analysis of the alternatives using the TODIM
should be rented for US$ 309. It can then be said that A8
method led to an ordering which showed it to be satisfac-
must be rented for a value between US$ 214 and US$
tory and in agreement with the expectations of the special-
309. In the same way it can be seen that property A14 is
ists. Through its formulation it became easier to resolve
in the 2nd place. A5 is in the 1st place, with a rental value
conicts between criteria, as, sometimes, in order to
of US$ 712, and A11 is in the 3rd place with a rental value
achieve a good performance in a determined criterion of
of US$ 414. Therefore property A14 must be rented for a
the analysis, it is necessary not to be concerned about per-
value of between US$ 712 and US$ 414. If the ranking
formance in another (Belton and Stewart, 2002).
order obtained by TODIM did not coincide with the rank-
In the case of property evaluation, the method is capable
ing based on dollar values this would mean that the prop-
of assisting professionals in the real estate market to eval-
erties dollar values would have to be reconsidered and
uate the alternatives more clearly in relation to the criteria
then changed accordingly.
dened by the specialists.
These value intervals are dened for each alternative
In this study it was also possible to identify that three
under analysis, serving as a reference to obtain values
properties, A1, A13 and A15, fell into the same range estab-
for the rent. In this way, the decision process becomes less
lished by the alternatives A4 and A11. From Table 10, it can
complex. Table 12 suggests the appropriate intervals for
be seen that the order of these three properties is: A13, A1
the properties with non-dened rental values. This shows
and A15. In this way, the rental values of these three prop-
that realtors would considerably benet from using a
erties could be the same or, alternatively, giving a reference
more comprehensive, multicriteria approach through the
for the rental value of the property according to the order-
use of TODIM when they assign dollar values to the
ing supplied by the method, the greatest value attributed to
properties.
A13 and the lowest value to A15.
Thus, the analysis and the solution of the problem pre-
3.5. Sensitivity analysis sented here, by means of the TODIM multicriteria method,
reected in their results the preferences of the decision
After obtaining the results through the implementation agents, experts in the multiple dimensions of the problem
of the TODIM method, a sensitivity analysis was carried analyzed. Consequently, it can be concluded that the
out varying the weighting given to the criterion of greatest method constitutes ecient support for the evaluation of
importance to the decision makers, which is criterion C1. property. As new properties are included in the portfolio
When its weighting was reduced from 5 to 3 there was no of a realtor, the TODIM method must be run again taking
variation in the ranking of the 15 alternatives. Another into account the characteristics of these new properties.
analysis was carried out varying the value of h, the attenu- After the rank ordering is obtained for these new properties
ation factor of losses. In the rst implementation carried their suggested rental values will be determined with the
help of Table 11.
Realtors concluded that applying the TODIM method
Table 12 to the rental evaluation of residential properties can pro-
Suggested rental values of the alternatives analyzed
vide a considerable help to them, given the extreme dicul-
Property Ordering Monthly rental value ties in assigning dollar values to all evaluation criteria.
A1 5 Between US$ 414 and US$ 309 Given new evaluation scenarios, with a new set of evalua-
A2 10 Between US$ 214 and US$ 166 tion criteria, however, new applications of the multicriteria
A6 11 Between US$ 214 and US$ 166
analysis would have to be performed. Changes in the sce-
A7 15 Less than US$ 133
A8 8 Between US$ 309 and US$ 214 narios can lead to changes in estimated rental values even
A12 13 Between US$ 166 and US$ 133 for properties whose rental values have previously been
A13 4 Between US$ 414 and US$ 309 dened. For future studies eorts should also be focused
A14 2 Between US$ 712 and US$ 414 on trying to quantify the monetary consequences associ-
A15 6 Between US$ 414 and US$ 309
ated to every criterion.
L.F. Autran Monteiro Gomes, L.A. Duncan Rangel / European Journal of Operational Research 193 (2009) 204211 211