13/05/2011

Progress in This Course as a Function of Chapters

CIVL3812 – Project Appraisal
Lecture 11.1 – Probabilistic Risk Analysis Examples involving
Discrete and Continuous Random Variables Foundation Application Integration
Chapter 4: Chapter 5: Chapter 14:
Introduction of EW Evaluating a Single Decision Making
Equivalence Relationships Project Considering
Chapter 7: Chapter 6 & 9: Multiattributes
Depreciation and Tax Comparing Alternatives
Chapter 8: Chapter 11 & 12:
Price Changes and Dealing with uncertainties
Exchange Rates
Chapter 10:
B/C Ratio Analysis
Semester 1 - 2011 Chapter 13:
Capital Budgeting

Last Week This Session

› Public vs. private projects Chapter 12:

› Social discount rate
› Brief recap of random variables from lecture 9.2
› B/C ratios:
EW(B) › Evaluation of projects with discrete random variables
Conventional B/C = ----------------------------------- › Evaluation of projects with continuous random variables
I –EW(MV)+EW(O&M)

EW(B) - EW(O&M) Learning Objectives:

Modified B/C = ------------------------------ › To understand and be able to apply the basic statistical concepts in
I-EW(MV) decision-making situations involving risk and uncertainty.
› To be able to carry out probabilistic analysis for projects involving discrete
› Using B/C ratios to evaluate single projects or continuous random variables
› Using incremental B/C analysis to select MEAs
› Pros and cons of the B-C ratio method
3 4

Random Variables

› Factors having probabilistic outcomes

› Many economic factor values are usually a subjectively estimated
Chapter 12.3 – Distribution of Random Variables likelihood that an event (value) occurs

› Using random variables we can work out expected values and variances
“In reality, killing time is only the name for another
- Particularly helpful in decision making by making the uncertainty associated with
of the multifarious ways by which time kills us.” each alternative more explicit

~Osbert Sitwell

1
 13/05/2011

Notation for Random Variables Discrete Random Variables

› Capital letters such as X, Y and Z are used to represent random variables › A random variable X is discrete if it can take on (at most) a finite number of
values (x1,x2…xL)
› Lower-case letters (x, y, z) denote the particular values that these
variables take on in the sample space (i.e., the set of all possible › The probability that a discrete random variable X takes on the value xi is
outcomes for each variable) given by
Pr{X = xi} = p(xi) for i = 1,2,….,L (i is a sequential index of the discrete values, xi,
that the variable takes on)
› When random variable X follows some discrete probability distribution, its where p(xi) > 0 and  p(xi) = 1
i
probability mass function is usually indicated by p(x) and its cumulative
distribution function by P(x)
› When random variable X follows a continuous probability distribution, its
probability density function is usually indicated by f(x) and its cumulative
distribution function by F(x)

7 8

Continuous Random Variables Mathematical Expectations

› A random variable X is continuous, if a non-negative function f(x) exists, › The expected value of a single random variable X, E(X), is a weighted
such that for any real numbers [c,d], the probability of the event occurring average of the distributed values x that it takes on and is a measure of the
is (i.e. the probability that X is within the set of real numbers [c,d] ): central location of the distribution
d
› Pr{c ≤ X ≤ d} = c f(x) dx › E(X) is the first moment of the random variable about the origin and is


› And f(x)dx = 1 called the mean of the distribution


The probability that the value X is less than or equal to k, i.e. the cumulative E(X) =  xi p(xi ) for x discrete and i = 1,2,…,L
i
distribution function F(x) for a continuous case is:
k 
Pr{X < k} = F(k) =  f(x) dx
 E(X) =  x f(x) dx

for x continuous

› In most applications, continuous random variables represent measured

data, such as time, cost and revenue on a continuous scale

9 10

Variance Multiplication of a Random Variable by a Constant

› For a single random variable, the variance is a measure of dispersion of › When a random variable, X, is multiplied by a constant, c, the expected
the values it takes on around the mean value E(cX), and the variance, V(cX) are:
› V(X) = E(X2) – [E(X)]2 › E(cX) = cE(X) =  cxi p(xi) for discrete
i


› V(X) =  x2p(xi) – [E(X)]2 for x discrete › E(cX) = cE(X) =  cx f(x) dx for continuous
i 


› V(X) =  xi2(x) dx – [E(X)]2

for x continuous › V(cX) = c2V(X)

› The standard deviation of a random variable, SD(X) = [V(X)]1/2

11 12

2
 13/05/2011

Multiplication of Two Independent Random Variables Multiplication of Two Independent Random Variables

› When a random variable, Z, is a product of two independent random • Thus if Z is the random variable then,
variables, X and Y, the expected value, E(Z), and the variance, V(Z) are
› Z= XY
› V(Z) = { V(X) + [E(X)]2 } { V(Y) + [E(Y)]2 } – [E(X)]2 [E(Y)]2
› E(Z) = E(X) E(Y)
› V(Z) =E(X2)E(Y2)– [E(X) E(Y)]2 Or

› But the variance of any random variable, V(RV), is

› V(Z) = V(X) [E(Y)]2 + V(Y) [E(X)]2 + V(X) V(Y)
› V(RV) = E[(RV)2] – [E(RV)]2
› E[(RV)2] = V(RV) + [E(RV)]2

13 14

Evaluation of Projects with Discrete Random Variables Example 12-2

› A discharge channel in a community where flash floods are experienced

has a capacity sufficient to carry 700 cubic feet per second.
› Expected value and variance concepts theoretically apply to “long-run”
conditions where the event will occur repeatedly Water Flow (ft3/sec) Probability of Overflow Capital Investment to
in Any One Year Enlarge Channel to
Carry This Flow
› We can however still apply these concepts even when investments are not 700 0.20 --
made repeatedly in the long-run 1000 0.10 $20,000
1300 0.05 $30,000
› Following examples apply expected value and variance with selected 1600 0.02 $44,000
economic factors modelled as discrete random variables 1900 0.01 $60,000

› Average property damage is $20,000, when storm flow is greater than the
capacity. Reconstruction of the channel will be financed by 40-year bonds
bearing 8% interest per yr. Determine the most economical channel size.

15 16

Example 12-2 Example 12-3

Water Flow Capital Recovery Expected Annual Total Expected

(ft3/sec) Amount Property Damage EUAC › Alternatives (A,B,C) are being evaluated for the protection of electrical
circuits. If a (power) loss does occur, it’ll cost $80,000 with a probability of
700 None $20,000(0.20) $4,000 0.65, and $120,000 with a probability of 0.35. Useful life = 8yrs. S = 0.
1000 $20,000(A/P, 8%, 40) $20,000(0.10) $3,678 MARR=12%. The annual maintenance expenses are expected to be 10%
of capital investment. Determine the best alternative based on expected
1300 $30,000(A/P, 8%, 40) $20,000(0.05) $3,517
total annual cost.
1600 $44,000(A/P, 8%, 40) $20,000(0.02) $4,092
1900 $60,000(A/P, 8%, 40) $20,000(0.01) $5,234
Alternative Capital Investment Probability of Loss in Any Year
› NB: (A/P, 8%, 40) = 8.39% [Capital recovery amount for principal of bond + A $90,000 0.40
interest = 8.39% x capital investment] B $100,000 0.10
› Conclusion: Enlarge the channel to carry 1,300 cubic feet per second, with C $160,000 0.01
the expectation that a greater flood might occur in 1 year out of 20 on
average. (Based on minimum total expected EUAC)

17 18

3
 13/05/2011

Example 12-3 Example 12-5: Analysis Using a Probability Tree

› Improvement project | Analysis period = 2 years, MARR= 12% per year

› The expected value of a power loss if it occurs is: Minimum EUAC, 0.3 $1,000
› $80,000(0.65) + $120,000(0.35) = $94,000 Choose B
$580 0.3 $960
Legend
Alt. CR Annual Maintenance Expected Total 0.4 $800
0.2
Amount Expense Annual Cost Expected EOY0
of Failure EUAC 0.1 $770
A $18,117 $9,000 $37,600 $64,717 0.5 0.8 EOY1
-$1,000 $500 $720
B $20,130 $10,000 $9,400 $39,530
0.1 $680
C $32,208 $16,000 $940 $49,148 EOY2
0.3
0.3 $760
Capital Investment Capital Investment $94,000 0.2
x x x $460 $650
(A/P, 12%, 8) (0.10) (Probability of Loss in Any Year) 0.5 $600
19 20

Example 12-5 Example 12-5

› (a) Find out the E(PW), V(PW), SD(PW) of the project

(1) EOY Net Cash Flow (2) (3) (4) = (2) x (3) (5) = (2)2 (6) = (3) x (5)
› (a) E(PW) = (PW j )p(j) = $39.56
j

j 0 1 2 PW j p(j) E(PW j) (PW j)2 E[(PW j)2]

› V(PW) = E[(PW)2] – [E(PW)]2 = 15,227 – (39.56)2 = 13,662($)2
1 -$1,000 $580 $1,000 $315 0.06 $18.90 99,225$2 5,953$2
2 -$1,000 $580 $960 $283 0.06 $16.99 80,089$2 4,805$2
› SD(PW) = [V(PW)]1/2 = (13,662)1/2 = $116.88
3 -$1,000 $580 $800 $156 0.08 $12.45 24,336$2 1,947$2
4 -$1,000 $500 $770 $60 0.05 $3.04 3,600$2 180$2
(b) What is the probability that PW ≤ 0?
5 -$1,000 $500 $720 $20 0.40 $8.17 400$2 160$2
› Pr{PW ≤ 0} = p(6) + p(8) + p(9) = 0.05 + 0.06 + 0.15 = 0.26
6 -$1,000 $500 $680 -$11 0.05 -$0.57 121$2 6$2
7 -$1,000 $460 $760 $17 0.09 $1.49 289$2 26$2
8 -$1,000 $460 $650 -$71 0.06 -$4.27 5,044$2 302$2 (c) Which analysis results favour approval and which ones are unfavourable?
9 -$1,000 $460 $600 -$111 0.15 -$16.64 12,321$2 1,848$2 › E(PW) is favourable. However, high SD represents high volatility of the PW.
Thus this project has questionable acceptability.
E(PW) = $39.56 E[(PW)2] = 15,227$2

21 22

Evaluation of Projects with Continuous RVs Example 12-6

› Two frequently used assumptions are: › Assume that the annual net cash-flow amounts for this project are
1) Uncertain cash-flow amounts are distributed according to the normal normally distributed with the expected values and standard deviations as
distribution given below and are statistically independent. MARR = 15% per year. Find
the E(PW), V(PW) and SD(PW).
2) Uncertain cash flow amounts are statistically independent (i.e. no
correlation between cash flow amounts is assumed)
EOY, k Expected Value of NCF, E(Fk) SD of NCF, SD(Fk)
 Thus if we have a linear combination of two or more independent cash
flow amounts (e.g. EW = c0F0 + … +cNFN) 0 -$7,000 $0

 …Then using the expressions gained from multiplication of a random 1 $3,500 $600
variable by a constant, we obtain: 2 $3,000 $500
N
 V(PW) =  C V F 
k 0
2
k k
3 $2,800 $400

 And, E(PW) =  C E F 
k 0
k k

23 24

4
 13/05/2011

Example 12-6 Example 12-6

N
› E(PW) =  Ck E Fk 
k 0
3
› SD(PW) = [V(PW)]1/2 = $696
› =  E(Fk)(P/F, 15%, k)
k 0

› = -$7,000 + $3,500(P/F, 15%, 1) + $3,000(P/F, 15%, 2) +

$2,800(P/F, 15%, 3)
› = $153
N

› V(PW) =  C V F 
k 0
2
k k

› =  V(Fk)(P/F, 15%, k)2

k 0

› = 0212 + 6002(P/F, 15%, 1)2 + 5002(P/F, 15%, 2)2 +

› 4002(P/F, 15%, 3)2
› = 484,324$2
25 26

Example 12-7 How to use the Normal Distribution Table

› Based on Example 12-6, find out Pr{IRR < MARR}. Assume that the PW of
the project is a normally distributed random variable.
› From Appendix E in the textbook pg. 648 (Normal Distribution Tables):
( X  )
Z

› Recall that when IRR < MARR, the PW < 0

PW  E ( PW ) 0  153
› Thus, Z    0.22
SD( PW ) 696

› Hence, Pr{PW ≤ 0} = Pr{Z ≤ -0.22}

› From Appendix E, Pr{Z ≤ -0.22} = 0.4129
27 28

Example 12-8 Example 12-8

› Cash flows for a project are shown in the table below with a 5-year study EOY, k Fk E(Fk) = akE(Xk) + bkE(Yk) V(Fk) = ak2V(Xk) + bk2V(Yk)
period. The net cash flow of the project is a linear function of both Xk and 0 F0 $0 - $100,000 = -$100,000 0+(1)2(10,000)2 = 100x106$2
Yk, where both are continuous random variables and are independent of 1 F1 60,000 – 20,000 = 40,000 (4,500)2+(1)2(2,000)2 = 24.25x106
each other. MARR=20%. Find E(PW), V(PW) and SD(PW). What is the
2 F2 65,000 – 2(15,000) = 35,000 (8,000)2+(2)2(1,200)2 = 69.76x106
probability that PW < 0?
3 F3 2(40,000) – 3(9,000) = 53,000 (2)2(3,000)2+(3)2(1,000)2 = 45x106
EOY, k NCF Expected Value SD
4 F4 70,000 – 2(20,000) = 30,000 (4,000)2+(2)2(2,000)2 = 32x106
Fk = akXk - bkYk Xk Yk Xk Yk
5 F5 2(55,000) – 2(18,000) = 74,000 (2)2(4,000)2+(2)2(2,300)2 = 85x106
0 F0 = X0+Y0 $0 -$100,000 $0 $10,000 5


N

1 F1 = X1+Y1 60,000 -20,000 4,500 2,000 › E(PW) =  C E F 

k 0
k k =
k 0
E(fk)(P/F, 20%, k)
2 F2 = X2+Y2 65,000 -15,000 8,000 1,200
3 F3 = 2X3+3Y3 40,000 -9,000 3,000 1,000 › = -$100,000 + 40,000(P/F, 20%, 1) +…+ $74,000(P/F, 20%, 5)
4 F4 = X4+2Y4 70,000 -20,000 4,000 2,000 › = $32,517
5 F5 = 2X5+2Y5 55,000 -18,000 4,000 2,300
29 30

5
 13/05/2011

Example 12-8

End of Sharing
N
› V(PW) =  Ck V Fk  =  V(Fk)(P/F, 20%, k)2
2

k 0 k 0

› = 100 x 106 + (24.25 x 106)(P/F, 20%, 1)2 + ... + (85 x 106)(P/F, 20%, 5)2
› = 186.75 x 106 $2
Thank You
› SD(PW) = [V(PW)]1/2 = $13,666

› Pr{PW < 0}? Assume net cash flow is normally distributed, thus:
PW  E ( PW ) 0  $32,517
Z   2.3794
SD( PW ) $13,666
› Hence, Pr{PW ≤ 0} = Pr{Z ≤ -2.3794} = 0.0087
31