You are on page 1of 40

3- 2

Overview
DCF and the relevant project cash flows
Mutually exclusive investments Projects
with unequal economic life
A formula for the infinite replications NPV
Optimal investment timing with growth
The simple problem (without rotations)
Timing with replications (rotations)

3- 3

## Overview further topics

Capital rationing and constrained optimization
One-period rationing
Multi-period rationing
Probabilistic constraints (Budget-at-Risk)

3- 4

## In the past, other methods beyond NPV, like the Payback

period and the Internal Rate of Return (IRR) have also been
used. These methods are synopsized below.
1. Payback method: the number of years to recover the cash
outlay. This method ignores a) time-value-of-money
b) cash flows.
2. Internal Rate of Return (IRR): the rate that equates
present value of cash outflows and inflows (accept if IRR is
greater than opportunity cost of capital).
This method does not maximize wealth. It assumes
reinvestment at the same rate, ignores different project
scales, and may lead to multiple IRR solutions.

3- 5

## The NPV criterion

Net Present Value (NPV): accept project if Net Present Value
(of discounted cash flows) is greater than zero.
The only method that allows correct decisions and is
consistent with wealth maximization, considers all cash flows
and preserves the value additivity principle (of present
values).
To calculate NPV cash flows are discounted at the weighted
average cost of capital (WACC).

3- 6

NPV

## For capital budgeting (NPV calculations),

a) we ignore noncash expenses (depreciation),
b) we ignore interest payments, debt principal repayment, and
interests tax effects, but
c) we include the tax credit from depreciation.
Then we discount cash flows at the appropriate discount rate
(the WACC).

3- 7

Discounting Review

Ct

t
t 1 (1 K )
The present value is a sum of present
values.

3- 8

constant growth perpetuity
growth rate g:

i.e. Ct= (1+g) Ct-1

PV C /( K g )

3- 9

Annuity -
() (annuity)

(PV) =

1
C
t
K K 1 K

3- 10

NPV
NPV and a simple example
If the economic life of a project is 10 years, the investment
capital required is equal to 5000, the company is at a tax
bracket of 50%, is using a 10% discount rate for the cash
flows of the project, and uses straight line depreciation, what
is the NPV of the project?
We will use the pro forma income statement below to
estimate the cash flows.

3- 11

NPV

## Pro forma income statement

-------------------------------------------------------------------------Rev revenue
2500
-VC variable costs
-700
-FCCfixed cash costs
-300
- dep noncash (depreciation)
-500
------------EBITearnings before interest and taxes
1000
-kdD interest payments (kd = interest, D = principal)
-100
------------EBT earnings before taxes
900
-T taxes (at c = 50%)
-450
------------NI net income
450

3- 12

NPV
The true cash flows to use are

## (2500 700 300)(1 .50) + .50(500) = 750 + 250 = 1000

Thus NPV =
PV(1000 for 10 years discounted at 10% p.a.) 5000 =
6144.57 5000 = 1144.57.
The project has a positive NPV and is thus accepted

3- 13

NPV
Another example
An investment to improve production efficiency requires
capital outlay 10000, it has an economic life of 5 years and
saves annually 3000 in salaries. Investment is financed with a
loan of 15% per year, the principal to be repaid (amortized)
by 10000/5 = 2000 per year. Taxes are at 40% and after tax
cost of capital is 20%.
What is the NPV? (the 5 year annuity factor is 2.991)

3- 14

NPV
Cash flows are: 3000(1-0.40)+0.40(2000)
=2600 per year.
Present value of cash flows: 2600(2.991)
= 7776.60
NPV = 7776.60 10000 = -2223.40
NOT ACCEPTABLE

3- 15

## Mutually exclusive investments Projects with

unequal economic life
We will consider projects with unequal economic lives, under
the assumption that they can be replicated. The following
methods have been used:

## 1. Simple method: replicate till they reach the same

economic total life, and then compare NPVs.
2. Find the equivalent annuity value for all, and then
compare.
3. Compare using the NPV with infinite replications
(assuming replication at constant scale).

3- 16

## Mutually exclusive investments Projects with

unequal economic life

From the three methods above, only the third (the infinite
replications NPV) is always correct.
If the projects under comparison have the same risk (thus
they have the same relevant discount rate), then the first two
methods are correct too.

3- 17

## Equivalent Annuity Value

Say, projects A, B,, have NPV(A), NPV(B), and
economic life T(A), T(B),, and discount rates K(A) and
K(B).
Each has a PVAF(K,T) = present value annuity factor(K,T).
By definition, for all projects
NPV = PVAF(K,T)(equivalent annuity value)
=> (equivalent annuity value) = NPV / PVAF(K,T).

3- 18

## Infinite replications NPV(N,)

Also, the infinite replications (rotations) NPV for a project
with an economic life N (and a relevant discount rate K),
denoted as NPV(N,) equals

## NPV(N,) = (equivalent annuity value)/K = a perpetuity,

and of course
NPV(N,)K = (equivalent annuity value)
The last is another formula to get the equivalent annuity
value assuming that we have calculated the NPV(N,).

3- 19

## Infinite replications NPV(N,)

A formula for the infinite replications NPV.
We often need a formula for the infinite replications
(rotations) NPV(N,). We can easily show from
NPV(N,) = NPV(N) + NPV(N)/(1 + K)N
+ NPV(N)/(1 + K)2N +
that
NPV(N,) = NPV(N) (1 + K)N/[(1 + K)N 1]

3- 20

## Infinite replications NPV(N,)

To show that the above formula holds, simply remember that
the value of every project rotation at the end of its economic
life equals NPV(N) (1 + K)N, since the effective
compounding per period of N years equals (1 + K)N. This is
repeated an infinite amount of times, with a discount rate per
period (of N years) equal to [(1 + K)N 1]. Using the
perpetuity formula it is obvious that the above holds. In case
the project can be replicated n times,
NPV(N,n) = NPV(N) + NPV(N)/(1 + K)N
+ NPV(N)/(1 + K)2N + + NPV(N)/(1 + K)Nn
which with the use of an annuity formula gives
NPV(N,n) = NPV(N) (1 + K)N[1-(1+K)(-nN)]/[(1+K)N 1].

3- 21

Example
We have two mutually exclusive projects with different lives.
Their cash flows are:
Year
Project A
Project B
0
-10.00
-10.00
1
6.00
6.55
2
6.00
6.55
3
6.55
The opportunity cost of capital is 10% for A, but 40% for the
riskier project B. For each project we must calculate the
simple NPV, the NPV(N,), and the annual equivalent value.
Which project should be accepted and why?

3- 22

Example
The simple NPV first:
Year CF(A) PVIF(10%) PV
CF(B) PVIF(40%) PV
0
-10.00
1.000 -10.00 -10.00
1.000
-10.00
1
6.00
.909
5.45 6.55
.714
4.68
2
6.00
.826
4.96 6.55
.510
3.34
3
6.55
.364
2.39
------------.41
.41

## Using the simple NPV both projects are equally appealing.

3- 23

Example
The NPV(N,) for each project is:
NPV(N,) for A =

.4 1

1 .1 2
1 .1 2 1

. 4 1(1 . 2 1 / . 2 1) . 4 1( 5 . 7 6 1 9 ) 2 . 3 6

NPV(N,) for B =

. 4 1 1 1. 4. 43 1 . 4 1( 2 . 7 4 4 / 1 . 7 4 4 ) . 4 1(1 . 5 7 3 4 ) . 6 5
3

## The annual equivalent value, given NPV(N,), is:

For A: k NPV(N,) = .10(2.36) = .236
For B: k NPV(N,) = .4(.65) = .260
Answer: We accept project A (contributes the greater wealth
if replicated forever at constant scale)

3- 24

## OPTIMAL INVESTMENT TIMING WITH

GROWTH
The simple problem (without rotations)
Assume that the project cannot be replicated. We seek to
find the optimal timing of a project that maximizes value,
given growth on the underlying revenues (and/or the costs).
This problem is often called the optimal harvesting or optimal
tree-cutting problem in economics. Note that the revenues
Rev(t) are a function of time. Often and especially in new
sectors, new products, etc., the more we wait, the higher the
(future) value of investing. But there is an opportunity cost in
the sense that we defer the arrival of revenues (heavier
discounting). Thus, we have a trade-off.

3- 25

## OPTIMAL INVESTMENT TIMING WITH

GROWTH
The more general problem would involve two capital costs.
Cost I is paid at time zero in order to buy the lot, plant, etc.,
and cost X is paid at harvest time (say the cost of harvesting).
The cost X(t) could in general also be a function of time, but
capital I is paid up front and is treated as a sunk cost. We
assume a continuous discount rate equal to K for both the
revenues and the costs of harvesting. So we maximize the net
revenues after having incurred cost I:
Max(t) [Rev(t) e-Kt - X(t) e-Kt - I] = Max(t) [(Rev(t) - X(t)) e-Kt]

3- 26

## OPTIMAL INVESTMENT TIMING WITH

GROWTH
In order to find the optimal timing that would maximize
investment value, we set the derivative in respect to time
equal to zero:
0 = - K e-Kt (Rev(t) - X(t)) + e-Kt [dRev(t)/dt - dX(t)/dt]
and after removing e-Kt and solving for K we get the important
result
K = [dRev(t)/dt]/Rev(t) when X(t) = 0,
or in the more general case
K = [dRev(t)/dt - dX(t)/dt]/(Rev(t) - X(t)).
Note: you must remember how to calculate derivatives

3- 27

## OPTIMAL INVESTMENT TIMING WITH

GROWTH
This result says that the maximum value is when the marginal
rate of return by holding the tree lot equals the opportunity
cost of capital K. Obviously in any realistic example this rate
of return is eventually decreasing, so beyond this point it is
not beneficial to hold the tree lot and we should harvest. By
taking the derivative of the revenue function in respect to
time and solving for time, we find the optimal investment
timing.
Then, we still need to find the NPV of such an investment.
Having found the optimal timing does not guarantee that
NPV would necessarily be positive.

3- 28

Example
The concepts will be further demonstrated with an example.
At time t = 0 we invest capital cost I = 15000 in a tree lot. When
we harvest we get revenues Rev(t) = 10000sqrt(1+t). There is
an opportunity cost of capital K = 5% with continuous
discounting. In this example the expenses in order to harvest
are zero (constant in time and X = 0).
K = [dRev(t)/dt - dX(t)/dt]/(Rev(t) - X(t)), so
K = [dRev(t)/dt]/Rev(t).

3- 29

Example
K = [dRev(t)/dt]/Rev(t).
= [(0.5)10000/sqrt(1+t)]/[10000sqrt(1+t)]
= 1/[2(1+t)]
and since K = 0.05

## 0.05 = 1/[2(1+t)] or (1+t) = 1/0.10 = 10

and optimal t = 9 years.

3- 30

Example

## Note that at t = 0 we pay I = 15000, and we must still calculate

the NPV. Finding the optimal timing is not enough.
For t = 9, net present value:
(10000sqrt(1+9))exp(-0.05(9)) 15000
= 31622.78(0.637628) 15000 = 20163.57 15000
= 5163.57
It is positive and we accept the project.

3- 31

Example - discussion

## Another way to look at this problem is that at time zero we

pay I = 15000 (an option price). This gives us the option to
the optimal timing of the investment on an underlying asset
valued at time zero at Rev(t=0) = 10000sqrt(1+t) = 10000.
Rev(t) is the underlying asset of this investment option (net
of any exercise costs).
This simple (investment) option involves no uncertainty, it is
simply a timing option.

3- 32

## A Different Visual Example

(discrete discounting)
We will see another example, that we will solve iteratively
(no equation solution). This will demonstrate that optimal
solution is when g = K. Note that here we have two growth
rates, one for the revenues and one for the capital cost, so that
the aggregate growth rate differs (is a function of time).
There is an investment with two years life, capital cost I =
10000 and net (after tax) revenues equal to 6000 per year (for
two years). The discrete cost of capital is K = 10%. If the
investment decision is deferred, the revenues grow at a
discrete growth rate of 5% per year, and the capital cost to
invest I grows at a discrete growth rate of 4.75%.

3- 33

Visual Example
If for example we invest now, NPV equals 413.2231. If we
wait for a year, the capital cost I = 10475, and the yearly
revenues equal 6300 per year. The NPV one year later would
be equal to 458.8843, exhibiting a growth rate of 11.05%.
The present value of that investment today equals 417.1675.
Looking at the numerical results for timing of the investment
up to 10 years later, it is obvious that we maximize NPV if
we wait for about 4 years. At that time, the growth rate of
waiting equals 10.1234 that is close to the 10% cost of
capital.
The analytical calculations follow:

3- 34

Visual Example

Value at T

NPV

## values growth rate:

413.2231

413.2231

---------

458.8843

417.1675

0.1105

508.016

419.8479

0.107068

560.8482

421.3736

0.103997

617.625

421.8462

0.101234

678.6056

421.3607

0.098734

744.0648

420.0052

0.096461

814.2947

417.8619

0.094387

889.6048

415.0072

0.092485

970.3237

411.512

0.090736

10

1056.8

407.4421

0.089121

3- 35

Visual Example
Note that if there was a single growth rate (the same growth
rate for net revenues and the capital cost), if g > K we should
wait for ever (!!!), if g < K we must make the investment
K we are indifferent about the timing.
These are of course not realistic examples, aggregate g for the
project value in general should not be constant. The visual
example instead is very realistic since .

3- 36

## OPTIMAL INVESTMENT TIMING WITH

GROWTH & ROTATIONS
Assume a project can be replicated many times (rotations).
We need an initial capital I in order to start the process (to
own the rights to this investment). Every time we harvest,
we get net revenues equal to (Rev(t) - X(t) ). This allows us
to get to the next harvest, and so on. We face the
optimization problem

Max(t)[NPV(t) =
I+(Rev(t) X(t))e-Kt +(Rev(t) X(t))e-2Kt +(Rev(t) X(t))e-3Kt
= I+PV(t)]
where PV(t) equals the present value of the net revenues for all
rotations.

3- 37

## OPTIMAL INVESTMENT TIMING WITH

GROWTH & ROTATIONS
For N rotations we have (easy to prove)
PV(t) = [Rev(t) X(t)](1 e-KNt)/[(1 e-Kt) eKt]
and for infinite rotations (N ) we get
PV(t) = [Rev(t) X(t)]/(eKt 1).
So finally for infinite rotations we get the NPV(t) equal to
NPV(t) = I + (Rev(t) X(t))/(eKt 1).
Note: not optimal t yet.

3- 38

## OPTIMAL INVESTMENT TIMING WITH

GROWTH & ROTATIONS
To optimize this function we set the first derivative equal to
zero (note that I is not a function of time):
0
= dNPV(t)/dt = 0 + dPV(t)/dt = d[(Rev(t) X(t))/(eKt 1)]/dt
= [d(Rev(t) X(t))/dt]/(eKt 1)] + (Rev(t) X(t))d(eKt 1)-1/dt
= [d(Rev(t) X(t))/dt]/(eKt 1)]
+ (Rev(t) X(t))(1)KeKt /(eKt 1)2
= [d(Rev(t) X(t))/dt] + (Rev(t) X(t))(1)KeKt /(eKt 1)
=>
d[Rev(t) X(t)]/dt = [Rev(t) X(t)]K/(1 e-Kt)
or
K = (1 e-Kt) {d[Rev(t) X(t)]/dt}/[Rev(t) X(t)]

3- 39

## Rotations Problem Proof

(not optimal t yet)
Call the discount factor e-Kt = U and rewrite
NPV(t) + I = (Rev(t) X(t))U + (Rev(t) X(t))U2 +
+ (Rev(t) X(t))UN
and multiply both sides by U
(NPV(t) + I)U = (Rev(t) X(t))U2 + (Rev(t) X(t))U3 +
+ (Rev(t) X(t))UN+1
and subtract the first from the second to get
(NPV(t) + I)(U 1) = (Rev(t) X(t))UN+1 (Rev(t) X(t))U
and for N->: (NPV(t) + I)(U 1) = 0 (Rev(t) X(t))U
so NPV(t) = I (Rev(t) X(t))U/(U 1)
= I + (Rev(t) X(t))/(eKt 1)

3- 40

Numerical Example
We use X(t) = X = 15000, K = 0.05 and
Rev(t) = 10000sqrt(1+t),
we use d[Rev(t) X(t)]/dt = [Rev(t) X(t)]K/(1 e-Kt)
we get
d[Rev(t) X(t)]/dt = d[Rev(t)]/dt = 10000d[sqrt(1+t)]/dt
= 10000(0.5)/[sqrt(1+t)]
and [Rev(t) X(t)]K/(1 e-Kt)
= [10000sqrt(1+t)-15000](0.05)/(1 e-0.05t)
which due to the non-linearity is solved by approximation
(numerically-with trial and error) to give (approximately)
t = 4.6. The NPV must also be calculated depending on I to
decide if the project is finally accepted or not.