ECONOMICS FOR MANAGEMENT

LECTURE 21 - DECISION MAKING UNDER RISK

Prof. THIAGU RANGANATHAN

 FRAMEWORK

Consumers
maximize
their levels
of
satisfactio
Opportunity n
• Time, Money, and Costs Markets allow
other resources are interaction of
scarce • While we do so, we
• We make choices in account for the two and
presence of this opportunity costs we obtain an
scarcity equilibrium

Scarcity
Producers
maximize
their profits
PERFECT COMPETITION

 Large Number of Buyers and Sellers

 Identical Products

 Free Entry and Exit

 Full Information – What about Imperfect Information and Asymmetric

Information?

 Negligible Transaction Costs

Risk and uncertainty

 What is uncertainty?

 Is Risk different from Uncertainty?

 Risk and Uncertainty are different – How?

 In Risk, you know the potential outcomes and the probability of outcomes

 With Uncertainty, you don’t know the probability of outcomes, at times we do not
know the outcomes also
Risky prospects

 Risky prospects are characterized by two aspects – the outcomes and the
probability of outcomes

 The theories of decision making under risk try to explain how people perceive these
two aspects
HOW ARE DECISIONS MADE UNDER
RISK?
Expected value theory

 People evaluate risky prospects by treating both the probability and outcomes in a
linear manner

 If an action A will lead me to a profit of Rs. 10 lakhs with a 30% probability and a
profit of 2 lakhs with a 70% probability, what is the expected value of the action?

 If I use expected value to make decisions, will I take that action if someone says if
you do not take action A, your profits will be Rs. 4 lakhs with 100% probability?
St Petersburg's Paradox

 Consider this game. I will toss a coin. If

it is a head, you win Rs.2 and the game
ends. If it is not, I will toss again, If it is a
head then, I will pay you Rs.4 22  and
the game ends. Else, I toss again and
you could get Rs.2  8
3
or the game
continues and so on. How much will you
pay me to play the game?
St Petersburg's Paradox

 What is the Expected Value of the game?

 What were you willing to pay?

 Why do people value risky prospects at less than the expected value of the
prospect?
Risk aversion

 Risk Aversion is the tendency of people to evaluate gambles lesser than the
expected value

 If that is the case, do we also see people value risky prospects more than expected
value? Examples….

 What do we call that tendency as?  Risk Seeking

Non-linear valuation of outcomes

 In Expected Value, both the probabilities Probability of

Winning at (0.5 ^ i )
2^i i
(lnWeighed
2 ) Non-linearly
and outcome are treated linearly Toss (i )
That Particular
Toss
Outcomes weighed linearly Outcomes

1 0.5 2 0.693147181
2 0.25 4 1.386294361
3 0.125 8 2.079441542
4 0.0625 16 2.772588722
 Risk Aversion or Risk Seeking could be 5 0.03125 32 3.465735903

explained by non-linear valuation of 6 0.015625 64 4.158883083

outcomes. 7
8
0.0078125
0.00390625
128
256
4.852030264
5.545177444
9 0.001953125 512 6.238324625
10 0.000976563 1024 6.931471806
11 0.000488281 2048 7.624618986

 In the St. Petersburg Example, let us 12 0.000244141 4096 8.317766167

13 0.00012207 8192 9.010913347
consider a case where we replace the 14 6.10352E-05 16384 9.704060528
outcome function by Ln (x) 15 3.05176E-05 32768 10.39720771
16 1.52588E-05 65536 11.09035489
17 7.62939E-06 131072 11.78350207
18 3.8147E-06 262144 12.47664925
19 1.90735E-06 524288 13.16979643
20 9.53674E-07 1048576 13.86294361
1.386279818
Total Value 20
Ln(4)=1.39
Expected utility theory

 People evaluate risky prospects by treating the probability linearly, but outcomes
either in a linear or non-linear manner

 The monetary outcomes are transformed into utility and then the expected utility is
calculated to arrive at a decision

 For a person with utility functionu ( x)  x , which of these options is attractive:

 Option A that offers 4 with a probability of 0.3 and 16 otherwise
 Option B that offers 4 with a probability of 0.9 and 121 otherwise
Expected utility (EU) theory

 In EU theory, the shape of the curve

determines the risk attitude of the
person

 A Person with a linear utility function will

be risk-neutral (will evaluate risky
choices by its expected value), a person
with concave utility function will be risk-
averse and a person with convex utility
function will be risk-seeking
RISK AVERSION
RISK LOVING
ALTERNATING UTILITY – INSURANCE AND LOTTERY
INSURANCE

 Property worth 40,000. A chance of fire of 1% which could cause a loss of Rs. 30000. What
is the risk premium for a person with a utility function of square root of the value of property
 Actuarial value of loss = 0.01*30000=300
 Uno-ins=0.01*sqrt(10000)+0.99*sqrt(40000)=199
 Uins=0.01*sqrt(40000-30000+30000-P)+0.99*sqrt(40000-P) = sqrt(40000-P)
 For me to have insurance, the utility with insurance should be atleast equal to that without
insurance
 Sqrt(40000-P)=199; P=399
 The Insurance company can charge a premium of 399 for a loss with a actuarial value of
300
 Risk premium = 399-300 = 99
Reflection effect

 Decision makers might prefer a sure v ( x)  x ^ 0.7 if x>=0

prospect less than expected value of the  -(-x) ^ 0.7 if x<0

gamble in the gain domain, but might For simplicity, we assume reference point = 0

prefer the gamble to the expected value 2500

in the loss domain

1500

500

 ‘Risk-averse’ in the gain domain and -50000 -40000 -30000 -20000 -10000 0 10000 20000 30000 40000 50000

‘risk-seeking’ in the loss domain -500

-1500

-2500

-3500

-4500
Weighing losses more than gains
v ( x )  x ^ 0.7 if x>=0
 Reflection effect - ‘risk-seeking’ in the  2.25*( x ^ 0.7) if x<0
loss domain and ‘risk-averse’ in the gain For simplicity, we assume reference point = 0
domain 2500

2000

1500

1000
 Add to that - pain of losing a certain 500

thing/amount is much more than the -50000 -40000 -30000 -20000 -10000
0
0 10000 20000 30000 40000 50000
pleasure of gaining the same {Loss -500

Aversion} -1000

-1500

-2000

-2500

-3000

-3500

-4000

-4500
LOSS AVERSION IMPLICATIONS – STATUS QUO
BIAS/omission bias

 Loss Aversion in action leads to status  Person A owns shares in company A.

quo bias or omission bias During the past year he considered
switching to stock in company B, but he
decided against it. He now finds that he
would have been better off by Rs.12000
if he had switched to stock of company
B

 People prefer ‘not to act and cause no

harm’  Person B owned shares in B. During the
past year he switched to stock in
company A. He now finds that he would
have been better off by Rs.12000 if he
had kept his stock in company B
But….

 Do we treat probabilities linearly?

 Is an increase of probability by 0.1 seen in a same manner irrespective of whether

we start at p=0, p=0.5, or p=0.9?
Probability weighting
Probability weighting

 A reduction of 16.7% probability is not treated the same in both the cases

 More sensitive to increase to 10% from 0% and to 100% from 90% than from 40%
to 50% or 50% to 60%?

 Pessimistic or optimistic where we might treat 40% as more than 40% or less than it
Probability weighting
Probability weighing

0.9
w( p)  1/ exp(ln(1/ p)) α
0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

alpha=0.25 alpha=0.5 alpha=1 alpha=1.5 alpha=2

Probability weighting

 People typically display an inverse shaped probability weighting with pessimism. So

they overweigh probabilities close to 0% (possibility effect) and 100% (certainty
effect), while underestimating moderate probabilities

 People can be risk-seeking for small probability gains (lottery) and be risk-averse for
small probability losses (insurance and warranty)
Probability weighting implications

 Consider warranties. $400 media player with an optional two year warranty for $
50. Why would anybody buy that?

 Say the probability of repair is 0.05 (a very high estimate), the expected value of loss
is $20. Why would anyone pay for that?

 But with probability overweighting if w(p)=p-squared (pessimist), the valuation for a

person with a linear value function is 0.125*400 = 50
Probability weighting implications

 Lawsuit on Nissan Car Dealers revealed the following:

 Cost of typical warrant - $ 795

 Cost of covering Repairs - $ 131

 Administrative Costs - $ 109

 Dealer Profit - $ 555 !

Prospect theory

 We treat both the probability and outcomes non-linearly

 Gains and Losses in the outcomes are treated differently (concave in gains and
convex in losses)

 People also feel the pain of loss of an amount much more than the pleasure of
gaining the same amount (loss aversion)
Prospect theory implications

Situation Lower Probability Events Moderate Probability Events

Risk Seeking
Gains Risk Averse
(Gambling)

Risk Averse
Losses Risk Seeking
(Insurance)

Mixed Risk Averse Risk Averse

The turkey problem – where you think uncertainty is risk