Lecture 11

IIT Kanpur July 201
20 followers
Feb 2, 2024
[Music]
a very good morning good afternoon good
evening to all the participants and the
students for this Mo course titled
multicriteria decision making and my
good name is raghunandan s Gupta from
the IM department at tanur in India
so as you know this course is for 60
lectures which is for 12
weeks and we will be
starting uh the next lecture uh which is
the 11th one that means we are starting
the third week and the broad outline for
this set of lectures which are happening
is about definition Concepts utility
Theory Safety First principle stochastic
dominance so under that we are
proceeding slowly
the coverage for this 11th lecture which
is the first lecture in the third week
will be utility Theory examples some
properties so obviously what are
mentions we'll try to cover whatever is
not covered obviously that will be
mentioned in the slides when they are
uploaded which I've already uh did me
did uh point it out in one of the
classes earlier and there would be
pointwise if you go to the the details
risk aversion neutrality seeking
properties
risk-seeking properties marginal utility
absolute dis oftion relative disc oftion
and these examples for certainty
equivalent geometric mean methods Safety
First principle stochastic dominance
hyperbolic hyperbolic absolute dision
functions and others so let us start
with an example so
consider
venture capitalist so he wants he or she
wants to uh invest and there are two
possibilities in front of him or
her so the first
alternative is to buy government bonds
treasury bills which cost 6 lakh that's
the cost I have not come into the
concept of utility yet the second
alternative has three possibilities as
noted in point number three and the cost
for these three alternatives are
respectively 10
lakhs 5 lakhs and 1 lakhs and with their
respective probability of
2.4.4 and as you do note the sum of the
probabilities which have already
mentioned should be double check of
2.4 +4 comes out to be 1 as it should
be now consider for the vental
capitalist the utility function is W ^
half that's the utility functions value
how you define now if I consider the
first alternative which is a sure event
in the sense there's only Government
Bond so obviously there's no risk which
means probability is one so if I want to
find out the probability uh expected
value for that first option first part
which is 6 lakh to the^ half is the
utility for investing 6 lakhs multiplied
by 1 which is the probability the value
comes out to be
775 now if I consider the second uh
alternative I'm going to use a different
color so the utility would be there are
three Alternatives under that so the
first one for investment is 10 lakhs 10
lakhs to the^ half gives you the utility
and what is the probability if you
remember it's half for the
second um outcome for the second
alternative it is 5 lakhs 5 lakhs to
the^ half is the utility multiplied by
its probability and the last one is 1
lakh to the^ half which is the utility
function multiplied by4 which is the
probability that value comes out to be
69 now if I consider and if I compare
the value of 609 to
775 obviously the expected value for the
government investment is much more than
the second alternative so obviously the
decision maker based on the expected
value of the utility will choose the
first alternative which is the
government one and not the second one
where there were three possible outcomes
so I just found out the probabilities
multiplied them with the um utility
function and found out the expected
value now
just another important fact which which
I will mention which you would already
would have understood but I'm still
mentioning it so I'm going to erase all
the colors and not make
it too
cluttered so consider now due to some
reason the utility function was not W
to^ half but uh it was only
W only W so that can be U function also
so in that case
the expected value of the utility for
the first alternative which is the
government
one would now be
6 into 10^ 5 which is lacks and
obviously it is W so there would not be
any power 6 lakhs is the utility
multiplied by 1 that would be
given as the expected value for the
first one I'm denoting E1 as for the
government
one and for the second case when is the
second alternative again the values are
same 10 lakhs 5 lakhs 1 lakhs and the
corresponding probabilities are also
same4
2.4.4 the expected value for the second
alternative would
be 10
lakhs multiplied by
0.2 the second arm would be 5
lakhs into 4 which is the probability
and the third one um is 1
lakh multiplied by
0.4 so find out the expected value for
the first case which was given by E1
the second case
alternative which was for
three such decisions consider it is
E2 find out the expected value for
this and at the end of the day you
basically compare these values E1 with
respect to E2 which one is more you take
take the decision accordingly so
obviously U function changing may give
you definitely different values of
expected value and you can find out the
DS accordingly tank means
come simple question is would the above
problem give a different answer if we
use in util function of the form w^ half
+ C so this is a
constant so adding that constant would
do what only this constant is outside so
obviously if you multiply by the
probabilities on the left hand side and
right hand side they would not affect
the overall decision because in that
case what you would have is for the
government
one it will be W to the^ half +
C into
1 so this will be
E1 and for the second case which is the
venture capital is invest
it will be W1 that is the values which
are given 10 lakhs 5 lakhs 1 lakhs so W1
is the 10
LH +
C into
0.2 + W2 which is 5 lakh to the^ half +
C into
0.4 and the last value is W3 which is 1
lakh again to the^ half half means
because the U function is not changing
plus
C multiplied by
0.4 and if you
notice the red portion which is for the
case when you
have the government one so this is you
take out a c into 1 which is C and if I
consider the values of c into
to2 C into point4 C into point4 so the
value comes out to be c as usual so c
will cancel each other from the left
hand side and the right hand side now
consider another example in utility
Theory consider in a span of 6
days the prices of security fluctuates
and a person makes his her transaction
only at the folling prices and we assume
the utility to be logarithm MC so
there's a util function logarithmic of
that
form the day numbers are given on the
First Column the prices are given in
rupees dollars Euros yens whatever it is
and when you convert using this
logarithmic util
function so log of
1000 log of which is 6.9 91 log of 975
6.88 so on and so forth so the third
column gives the utility based on the
logarithmic utility function and the
number of outcomes are given what does
the number of outcomes mean actually the
day
one let us not be confused so what it
actually means is
this a price of
1,000 occurs 35 number of times in the
total number of samples or the
observation which is done so if I count
the total number of of days so obviously
it is 100 so out of this 100 35 number
of instances are there where the price
was th000 20 number of instances were
there out of the 100 where it was 975
and so on and so forth so based on the
number of outcomes I find out the
probabilities if you remember
probability was NW number of times for
that
price summation of n n w so NW summation
is basically
100 so for the first case 35 by 100
gives you 35 20 by 100 gives you 0.1 and
similarly for the last one 15 by 100
gives you
.15 so given the utility which is there
in the third column and given the
probability which is in the last column
I need to find out or we need to find
out the expected value so here it is
calculated 6.91
the first value multiplied by
probability 35 6.88 into
2 then the third one is 6.86 into .1 and
I keep multiplying the third column
respective values with the last column
probability the value comes out to be
6.91 which is the expected value of the
utility based on the prices and where
the utility function is logarithmic in
nature
now let us change the utility function
but the senar unity Remains the
Same consider now the utility function
is p to the^
.25 so if I find out
those utility those values would be
given like this I need to erase the
colors in order to
make things visible to all of you so
so utral function is
p^
25 and the utilities are accordingly so
for the first price
1,000 I find out the utility which is
1,000 to the^ 1/4 for the second one
which is
975 is 975 to the^ 1/4 third one is 950
to the^
1/4 fourth one is is
10550 to^
1/4 then the LA second last one is 925 ^
1 14th and the last one is 1025 to the^
1/4 so this should come
here and interestingly the pro outcomes
are same so we already know the outcomes
are
given the corresponding probabilities
are already known multiply them so if
you see 35 for the first probability 2
second case .1 third case .15 fourth
case
05 the fifth case and 0.15 for the last
case when you multiply them the
corresponding expected value comes out
will be 5
62 so using that you will see that based
on the fact that there are two utility
fun different utility functions for the
same scenario you'll get two different
expected
values now we will consider some of the
important properties of utility function
so one there would be four properties so
we'll go slowly and discuss them one by
one the first properity is known as the
property of non satiation that
means people are not satisfied the first
restriction play based on the util
function is that is consistent with more
being always being preferred than less
so more I give you more you want and it
is always
increasing so which may not be true in
many of the Practical examples but it is
considered as one of the important
properties for util function this means
that between two certain Investments
which are there we will take the one
which has the largest outcome among them
so if I consider that obviously W + 1
which is one unit more of wealth is more
than W so we'll consider the utility
function of w + 1 is
greater than
UW which when
translates in simple concept of
mathematics means the first derivative
for the util function is always
positive and this will be utilized in
many of the cases later on so first
derivative of U function is positive the
second property is that we'll consider
Investments investors or decision makers
perception of risk and based on this
concept of risk which would come out
from Property 2 any decision maker any
investor can be classified any one of
these three
categories which is the first one mean
people are risk ofs I'm putting one tick
mark there the second characteristics
under Property 2 would be people are
risk neutral that means they are neither
here nor there increasing risk
decreasing risk doesn't matter for the
first one is that averse means I want to
avoid risk and the third one where I'm
putting three tick
marks is the case where people want risk
want to face risk and we'll see that how
Property 2 using using simple
mathematics can be formulated
accordingly so for that let us consider
a very simple
example the example is like
this in case one consider on the left
hand side of yours there is a table
which is marked one on the right hand
side for yours there's a table which is
marked
two and in table one there is a coin
which is an unbiased coin which with a
head and a tail and probability is
obviously for a head is half tail is
half
and the overall investment which you
do is if a head comes technically
whatever investment means what is the
outcome you are going to get whatever is
the input you are going to give is
immaterial now for the time being so
consider that with probability half you
get two and for probability half for the
tail you get zero
none and for the table
two it's a coin which is both head or
both tail that means sure so for that
with probability one you will get one so
the left one which I'm circling now on
table one is known as a fair gamble in
the sense his values is exactly equal to
the decision making of not investing
Fair Gamble also consider the
probabilities are half and
half if I consider the expected value of
the fair gamble it will what two is the
value I'm getting or somebody's getting
multiply the probability half plus 0o
when a tail comes multiped by
probability half the value comes out of
one and for the sure event which is on
table two the value which person gets is
one multiplied by the probability one
and is one so both the values are
same so now let us change the values for
the same example setup remains same but
the values are
changed consider four instances of the
problem first case in the first case for
which is there for the
gamble it it is now not 2 it is
2,000 and for the one which you are not
getting it's still
zero so the corresponding probabilities
would be what it will be
2,000 multiplied by half plus 0 into
half gives you a value
1,000 and for the sh one so this was on
table
one this is in table two so I mark it as
T2 T1 table 1 table two and for table
two the investment is now the value
which you get is 1,000 probability
remains one so when you multiply 1,000
into 1 the value is 1,000 again again if
I follow want to follow the concept of
expected value they are
same consider the second instance of the
problem again the same same setup but
now the investment value have increased
by 100 100 times so from instance one so
it is now 2
lakhs for table one which is change Zero
remains same probabilities are half and
half as usual U unbiased
coin and for the table two the
investment of the values now increases
to 1 lakh so if I want to find out the
expected value it is 2 2 lakh * half + 0
into half gives you a value of 1 lakh
and similarly 1 lakh for the table two
so instant two again
same expected values are
same so if I want to
denote so this is expected value for
table
one the thing which is there on table
one and this this is for expected value
for table
two so here table 2's expected value
which was
1,000 exactly matches table one's
investment equal and for this second
instant problem
also expected value for table two
is 1 lak
and for table one the value is also one
lakh now go to the third instance so if
I find if I
check the value is
now to 20 lakh again value is zero for
the tail probability is a half and half
and the sh event is one 10 lakh the
probability is one the expected value as
usual comes out to be one lakh for table
1 one lakh for table two table one table
two are
the the games the fear Gamble and the
certainty one you think think so table
one expected
value table
two expected
[Music]
value it's same
similarly when I increase it
to 2,000 lakhs or 200 200
million for instant
4 for the Gamble and another arm is zero
the probabilities are also half and half
and for the Sher event which is table
two it is 100 million or 1,000
lakhs and the probability is one as as
I've been mentioning so if you find out
the expected value which is given
here which
is 200 million into half + 0 into half
gives you the value
of 100 million and for table two which
is the sure one 100 million into 1 which
is the probability becomes also the same
value of 100
million here also expected value for
instance
4 is exactly equal to expected value of
table two for the
same instant now why did I do instant 1
2 3 4 if I ask you the question that for
instant
one which one will you take the sh one
or the
gamble maximum of our of us will take
the gamble very few would go for the sh
event and some may go either for the
gamble or the sh event now as the stakes
increase 2,000 becomes 2
lakh then it is almost certain that the
number of people who were willing to
take the gamble would now start
thinking and they may switch to the sure
event
that means the number of people who are
also undecided May slowly shift to the
sure event so people are shifting that
means people are shifting from the case
who were risk seeking slowly to risk
neutral and slowly to
the risk averse that means I'm trying to
avoid the gamble which is there on table
one
third instance the Val Stakes have
really increased now there are 20 lakh
and zero for the Gamble and um 10 lakh
for the sheer event so now a whole lot
of people will be shifting for the sure
event and for the case when it inst
stands four there would be almost none
who are willing to take the gamble which
means depending on the output human
beings will slowly keep changing their
risk Properties or risk taking
properties so depending on the instance
or what is the outcome people may move
from risk seeking to risk neutrality to
risk hating or averse and there can be
instance where people can move the other
way around so this EX four instances of
the same problem gives the idea that
depending on the outcome people may
change maybe if I consider a person
salid person
who is working and who has basically a
salary of about 15 lakhs
per year he or she may go to the case of
about instance two but it will be a a a
rarity or a rare example if a person of
that salary 15 lakhs per year would go
for instance three or instance four so
depending on your age depending
depending on an income depending on
risk- taking
abilities people may take different
decisions for the same problem but the
values being different so what I mean by
that is as
follows if I
consider the case as
U1 being the case for the gamble one arm
so I'll consider the gamble I'll use the
red color so this is one arm this is
second n the probability is P1 here
probability is P2 here what is P1 P2
I'll
denote the values which are being
invested is
i1 gives you a utility of
ui1 is amount this basically a certain
value of for w
I2 Gives You
U2 and on the other
side there is a sh
event that means there is a investment
of two consider
I2 probability is one so if somebody
wants to
balance
equivalence so if I'm willing to take
the gamble which is red color that means
I am a risk-seeking
person as you see the greater than sign
here if I am
neutral equality sign so I'm risk
neutral person and for the case
when my overall value for the expected
case wherefore the sure event is more
which is greater here because the right
hand side is the sheer event then I'm a
risk averse person so depending on the
examples and the three bullet points you
can classify the human beings
accordingly this example which we
considered about the risk Awards risk
neutral and risk-seeking
persons can be judged considering the
sure event which is marked in blue and
the red one which is the Gamble and
based on that human beings can be
classified as I'm told in Risk discovers
risk neutral and risk seeking person and
we'll continue discussing this use
utilizing Property 2 to analyze how a
human being or a decision maker will
behave thank you very
[Music]
much
[Music]
[Music]
