Lecture 15

IIT Kanpur July 201
20 followers
1 Video View·Feb 2, 2024
[Music]
uh welcome back my dear friends and
participants for this course uh
multicriteria decision making under the
mo and P series and as you know my good
name is raghunandan s Gupta from the imy
department at
iano so we are in the phase of
discussing the concepts of multiat
decision making and the broad umbrella
which we'll continue for a few more
lectures is basically concept of utility
Theory concept of Safety First principle
stochastic
dominance and this is the 15th lecture
where the coverages in general we have
considered certainty equivalent and we
have considered two different examples
we'll further on go into the utilization
of certain equivalent why what is the
use and if you remember there were two
points one was basically to understand
how
utility uh functions or utility Concepts
can be used for
ranking and also to find out the
so-called functional form of the utility
function considering that the person can
take a decision but you as an
academician want to find out what type
of util function and which class of
utility function the person has or if
the parameters are to be found out then
we'll go into the concept of geometric
mean methods and how geometric mean
methods can be utilized for for
different sort of ranking and later on
obviously we'll cover Safety First
stochastic dominance H hyperbolic
absolute risk cion properties but these
are as we proceed so hopefully we will
be able to cover the certainty
equivalent and geometric mean models in
this 15th
lecture out of the 61 so if you see this
graph and this is where certainty
equivalent concept and gamble will be
utilized to understand what type of
person he or she is so whether the
person wants to take risk one is
indifferent of taking the risk and is
basically trying to avoid risk so first
let us
specify what we
are drawing and why this diagram is here
I'll use different color
to to denote the first is that we will
denote the values of the wealth which
will denote by a b A1 B1 A2
B2 along the x-axis as noted and we
denote the certainty value which is
given by C if you remember we have used
that uh symbol
C being denoted on the y- AIS so the
values again on the the x-axis as I'm
marking putting a single tick mark and
certainty value along the y- axis where
I'm putting a double tick
mark now what we do and how we
proceed so we consider a 45° line as
given so why this 45° line so this we
should at least mention this angle is
45° and Y you'll understand now consider
on one
table table one which I marked by one we
place wealth of amount A and B which are
different amounts as given here along
the
x-axis and we consider
intrinsically from our side as we are
doing the experiment we considered the
utility function is linear which means
the utility function which we going to
consider for our comparison is
UW is equal to W and the wells we place
A and B the outcomes are based on the
fact when you toss a coin you get
probability half and half it is an
unbiased coin so if I consider the
overall expected value for a and b so I
denote by E only so consider e is case
basically or I'll put
the expected value for a and b will come
out to be the utility for a would be a
because the util function is linear so
it will be a this star sign is
multiplication sign multiplied by half
because half is the probability of
getting a head considering a comes out
for a head and B comes for for a tail so
it will B multiplied by half
so the value would be a + b by
2 now what we do is that we I'm using
the black color so that the consistency
of the notation and there's no confusion
so A + B by half would be a middle point
we draw a straight
line so that so and if you remember I've
mentioned that we place A+ B
on table one on table
two there are no values and the person
comes and we ask him or her that this is
a
gamble with a B coming out with
probability half B coming out to be
probability half and what will M make
you indifferent or see for example we
place C on table two to make things much
more uh explicit and simple C value is
placed so the person comes and we ask
that what will he take or she take the
gamble with AB or the certain value with
the probability one which is
C if the person says the value of C
which has been placed on the table two
if he or she agrees which means the
point based on which we have found out
this would be C
because we have formulated the C value
such that A + B by 2 is equal to C this
is the midpoint so the the point would
definitely be the point which is marked
by the this black dot would definitely
be on on the 45° straight
line now in case the person says no uh
he or she will not be able to take the
gamble but will go for a certainty value
higher so obviously in that case the C
value which will
be told by him or her with with respect
to the C value which is there on the
table the person says the C value is
higher so obviously the x value which is
a plus b by 2 remains same the C value
is higher so consider the C value
according to him or her is
here so we note it down and we note down
this point the second Point here
so let me Mark it
with
the red so this is the
point this
one and in case so I should Mark with
with uh green because the green line is
there just wait 1
minute
so the person says green so this is the
point if the person says no he or she's
would be sure to take that gamble with
respect to a certain value which is
lower than c c play already
placed then the
person's corresponding C based on which
he is proceeding or she is proceeding
would be
below
so we will
denote the higher so consider C is which
is kept on the table again I'm
mentioning and the person one who comes
he or she says that the C value is above
which is on the Green Point and we'll
denote it by C with a green
color this is
C so which means this is C
and actually values which you have
placed on the table was C the black
color and for the second
person the C
value would be here which is red color
so I'm not using any suffix I'm just
differentiating with the color so that
it's easy for us to communicate now what
we do is that we change once we know
whether the person value C green is
above or C red is below or C black
one is on the line we change the values
of A and B so consider A and B are
changed so here is see for
example new value is
A1 new value of B is B1 probability
remains half half the midpoint is now
here
again the values are as I mentioned A1
B1 and the value of black mark is see
for example
C1 where C1 is equal to A1 + B1 by
2 and this C was if you remember we have
mentioned is a + B by
2 so again the person is asked here or
she says no the C1 value which is given
doesn't satisfy him or her the C1 value
for him or her which they say is higher
so this is C1 so this would be
C1 or else it can be
below this is
C1 which
means this is
C1
or this is
C1 and the third case can be where the
person agrees with the C1 and is
indifferent so this would be C1 so C1
but with different colors denoting
different values but C1 is the symbol
which I'm using for the second case
now I keep repeating this question so I
change A1 B1 to A2 B2 same probabilities
I put a C2 C2 is basically A2 + B2 by
2 the
person can give the answer
as the value is exactly
matches so which is indifferent along
the straight 45° line so it will be
C2 color is different again I'm
mentioning
variable concept is C2 but color
different means different values or
else it can be
lower which is red C2 and if it is
higher it will be green
C2 similar
for the case again I change a and b this
is A3
B3 there is an C3 which is half of A3 +
B3 by 2 and
correspondingly the person gives the
answer accordingly which is it can be C3
above above the 45°
line C3 along the 45 digit
line and C3
below so what you see as we keep asking
the question change values of A and B
and also corresponding value of C you
can have three different types of
profile profile one would be along the
45° line which we have
drawn which means the person is based on
A and B and C value a b and changing
with suffixes one then two then three
which means the person is indifferent
based on the Gamble and the certainty
value which is being placed
because the rate of increase is
increasing but at a constant rate if you
remember the three graphs for the case
when the person is risk averse or
risk uh
lover the profile for the person who is
risk AE would be increasing better
decreasing
rate which is the green
curve and for the person who wants to
take the risk is increasing and
increased rate which is the red curve so
these are the curves which we have
considered the marginal
utilities so in this way certainty value
can be utilized considering that we very
simply do the
experiment
now what I said is exactly mentioned
here in this slide so A and B are
wells as we discussed A and B are
wells and we consider the utility
function to be W hence the straight line
45° line you form a lottery that has an
outcome of a with probability half as
already mentioned B with the probability
half then obviously the utility which I
will only denote by U would be a +
b by 2 A +
B by
2
now we place C exactly equal to the half
of a plus b by 2 as we have
mentioned you change keep changing the
values in a b and ask the investor how
much certainty value will make him or
her take that
gamble so again over the 45° line below
the 45° line or along the 45° line so
the expected value of the lottery is
which I have mentioned time and again
is a plus b by 2 which is equal to C the
risk averse risk neutral and risk
seeking person would
prefer C would be based on the the
concept would basically take the
decisions
accordingly if he or she risk averse
person would prefer a higher value of C
would be indifferent between C and A
plus b by 2 the value which you given
and would prefer a value which is lower
depending on that the person profile
will be a straight line which is
increasing at constant rate curved the
green one which is increasing at
decreasing rate and the red one is
increasing at an incre increasing rate
plot the values of c as you V vary in B
and this variation of C for the answer
which he or she is given that means the
red colored C or the green colored C or
the blue color C the value of C which
you're putting is fixed but his answer
or her answer is basically what you're
trying to
plot a risk neutral seeking an averse
person would have as I mentioned the
part if it is neutral straight line
again I'm mentioning the person is R
covers the curve which is green above
and slowly sloping down and for a
risk-seeking person would be below and
sloping up now how can we use the
constant of of C which is certainty
value further on to find out an
expressit form of the utility function
suppose you know that the actual form
form of the utility function form you
know but exact parameters you don't know
the form which you know I'll just
highlight is an exponential part with a
values parameters not known so you'll
ask the
person who has which you think or you're
doing an experiment where the util
function should be exponential and you
ask that person you give a Lotter with a
50/50 chance probably half and half and
unbiased coin being tossed and the
values of winning 50 with 50%
probability is 10 lakhs and another 50%
winning is 1 lakh now you place that and
ask that person what is the amount he or
she is going to pay to buy that lottery
so obviously somebody has to buy and
consider the person says okay I'm
willing to pay 4
LHS so 4 L means that he or she is
trying to balance the expected value of
the lottery with the value of 4 lakhs
because in that case the person would
exactly understand that the amount of 4
lakhs I'm
um paying to buy in the long run if he
wins any two amounts either 10 lakhs or
1 lakh the overall expected return would
be 4 lakhs so in that case for the
certainty
value the overall expected value of the
utility would be this which is because I
put W which is 4 lakhs into the equation
a value is not known and the probability
is one because he or she has already
purchased and the corresponding side for
the lottery would be the utility for 10
lakhs is what I'm underlying with the a
little bit darkish yellow color and
obviously multiply it by half and for
the value which the outcome is
one lakh this is the expected value
where a value is again unknown and this
is the probability so you put this
equation do some iterative methods and
try to solve and find out the values of
a so this concept which I used can be
extrapolated for any other UTD functions
and the parameters can be estimated not
found out
exactly now few concepts for every
gamble there is a certainty equivalent
such that a person would be indifferent
so obviously we we saw from the
experiment that if C is the value or the
certainty value then a gamble whatever
the value is a b x y there would be a
certainty value for which the person
would definitely be indifferent between
the Gamble and the certainty value which
means that if I draw the diagram and I
denote the certainty value with the blue
color which I have drawn but I'll still
mention it again this is the blue color
probability is one I'm marking it here
and if I consider the certainty value of
C then the corresponding utility would
be UC U utility function is given
whatever whatever form is and that would
be
equivalent consider the gamma has two
outcomes it can be three four so on and
so forth but consider very simply a two
outcome and I'll denote the gamble using
the green
color one arm two arm probabilities
being
half half the wealth being
W1
W2 then the corresponding expected value
which should match UC which is in the
blue color I'm using the green color for
the gamble so U of w
would be equal to half into
uw1
plus half into U
W2 so whatever the values of the wealth
is given the util function these two
values would exactly match I'll use the
red color to symbol where this one U see
and UW would exactly match they has to
match and they would be some value of
which will make them
equal now extending that idea to the
capital asset pricing model which very
use different concept of optimization so
few important points investors have
quadratic util function and this is
important which I'll mention and share
prices are log normally distributed here
one important fact is which I'm marking
in red is quad Dy function would mean
that if the utility function is I'm only
using U is W - half of a
w² or to extend it that if you have an
investment and you consider the return
and the return concept uh I'm trying to
utilize the green color so there are two
returns one is Total return one is rate
of return small R so if small R is
normally
distributed with a mean mu
R and variance of Sigma
R then the wealth or the investment
based on which the small r or the rate
of return is being
realized would be quadratic means if R
is normal the util function is quadratic
if the util function is quadratic R is
normal this would be true share prices
are log normally distributed and based
on that if you remember one of the
examples where we
plotted uh and did the optimization for
when we are discussing the introduction
part optimization of the either the
return or minimization of the risk
maximizing the return or minimizing the
risk for a portfolio considering of NC
stocks so there we consider the small r
as as log normal return and we used if
you
remember again a different color being
used in order to make things so R was
calculated by nepan log of P2 by
P1 this P2 and P1 are the prices which
are basically wealth if you can
understand price P2 is the closing price
for day two and P1 is the closing price
for day
one now comparing the point which I just
mentioned normality and quadratic util
function U function being
quadratic and rate of return being
normal so comparison between mean
variance which is considering normal
distribution utility function where we
consider the utility function as given
here which I'm underline is is basically
quadratic so this is the same form W -
half of a w
s so here a by 2 is considered as B so
this is just a constant
and we are trying to utilize that
considering there are three
assets and the assets are a b c the
prices are
given second
column third column price for
B fourth column price for
C and I use the total return so I could
have also used small R there's no only
the con the few formulas would change
the concept remain the same Just for
information I'll write down here if my
wealth or investment the word wealth
investment I'll be for this example I'll
be using
interchangeably consider capital r which
is Total return would be investment at
time period 2 divided by investment at
time period one that means if I invest
100 today and I get back 110 then the
capital value of R would be 110 by 100
if I invest 100 and if I get the value
of 90 then capital r would be 90 by 100
similarly small R would be the
difference of I2 minus i1 in the
numerator divided
by
i1
so if I have 100 investment i1 I get 110
so in the numerator it will be 110 - 100
divided by 100 will give me me small R
so this is the small R return
distribution of we talking about
normality so R let us consider based on
the fact that capital r variation would
also be normal so consider we find out
capital r so capital r would be what the
for a would be 110 by 100 which is 1.1
115 by 110 1.05 120 -
115 and so on and so forth so the values
are found out similarly for
RB which is for uh asset 3 S2 it is 115
by 5 for
B and uh then then it will be for 120 by
115 125 by 12 20 and 130 by 125 so this
is for
B and for C I get I have the values as
90 by 80 95 by 90 105 by 95 130 by 105 I
get the SEC this second last column of
capital
RC and the probabilities I consider as
1/5 for each of these outcomes So based
on that I find out about the average
return so average return for I'm
marking with the orange on this a little
bit darkish yellow color R Bar A R Bar B
R Bar C what is that these returns
capital r which I have calculated for a
add them up divide by the total number
of such instances there are four
instances because prices are five but
when I calculate at capital r it will be
four in number similarly for RB again
adding up the four returns for B divide
by four for C it is adding up four
returns for C divide by 4 so the
corresponding values are given
here 1.06 1.05
1.14 similarly I find out the standard
deviation for a b c from the sample this
is the sample so given that standard
deviation formula for the sample given
the mean value for the population is
unknown mean values are given from the
sample as here 1.06 1.05
1.14 I find out the standard
devision and just for
information
the value of the
variance given by population mean is
known we know that so this was done in
dadm 1 the quotes which I have taught or
in any statistics books you can find if
the mean value expected value of the
population is known which is Mu
the best estimate for the variance is
given as 1 by n summation of x i - mu
whole s and I is equal to 1 to n where n
is the sample
size if I want to find
out the this is a different
estimate and why because the population
mean is unknown I replace the population
mean with
the sample
mean this is for the normal distribution
remember because X Barn is the best
estimate for the population mean X Barn
is the sample
estimate so if you use any of these
formulas obviously the second one will
be used which I'm just marking in a
square box because the population mean
is unknown So based on
that I find out then I find out the
weights the weights are just simply the
wealth my apologies is we wealth I find
out just the sum of the weals for each
investment divide by the number of
instance because there are five so they
these values come out to be 114 119
110 so considering the time limitation
for this lecture I will continue this
discussion further on in the next
lecture sorry for that but because we
have been discussing many things so I'll
continue from the same slide and
consider in the 16th lecture the um
ending of this problem thank you very
much and have a nice
day
[Music]
[Music]
[Music]
