Caveat WACC: Pitfalls in the Use of the Weighted

Average Cost of Capital

A comment on Miller R.A., The weighted average cost of capital is not quite right, The

Quarterly Review of Economics and Finance (2007), doi:10.1016/j.qref.2006.11.001

Sebastian Lobe*

University of Regensburg

Abstract

In Discounted Cash Flow valuations, the WACC approach is very popular. Therefore,

knowing which limitations the concept inherits is essential. The objective of this paper

is thus twofold: First, it is clarified that a constant WACC rate must fail if the implied

leverage ratio is time-varying. This seems to be the rationale for defining a nonlinear

WACC (NLWACC). However, the NLWACC appears to be rather artificial when

allowing for time-varying WACCs. Second, although the NLWACC approach is

further amplified in this paper, it must be emphasized that this approach is, even then,

applicable only under specific conditions while a time-varying WACC is still able to

provide reliable results. In conclusion, the WACC approach is a valid workhorse whose

results can be economically interpreted.

J.E.L. classification: G11, G3

Keywords: WACC; Cost of Capital; Discount rate; Financial structure; Tax shield

* Center of Finance, Chair of Financial Services, University of Regensburg. Universitätsstraße 31, 93053
Regensburg, Germany. E-mail [email protected], phone +49 941 943 2727, fax
+49 941 943 4979.

Electronic copy available at: http://ssrn.com/abstract=1561188

1. Introduction

Methodological issues on valuation have experienced a remarkable renaissance in the

financial literature over the last years. Fernandez (2004) initiated a provocative

discussion claiming that the value of tax shields is not equal to the present value of tax

shields. This claim is indeed provocative as it implies inter alia that the principle of

value additivity is not working and that the seminal propositions of Modigliani and

Miller are flawed. The subsequent discussion led in several journals revealed that the

claim was not well substantiated. For example, Arzac and Glosten (2005) reconsider

tax shield valuation looking at a value-based debt policy in the spirit of Miles and

Ezzell (1980), and prove the validity of the respective valuation formula. Comments to

Fernandez (2004) are given in a comparable vein also by Fieten et al. (2005), Cooper

and Nyborg (2006), (2007), and Massari et al. (2008).

The advanced textbook by Kruschwitz and Löffler (2006) offers a rigorous and

authoritative perspective on Discounted Cash Flow valuation leading to many new

insights. Cooper and Nyborg (2008) analyze the case of tax-adjusted discount rates with

investor taxes and risky debt.

Ruback (2002) advocates the Capital Cash Flows methodology as a simple approach to

incorporate the value of the debt tax shield in valuation formulae. His methodology is

founded on a value-based debt policy. Booth (2007) makes the case that the Capital

Cash Flows and the Adjusted Present Value methodology is not easy to handle in

specific valuation scenarios while the weighted average cost of capital (WACC) is a

more flexible methodology.

For valuing companies or projects, WACC is the dominant Discounted Cash Flow

approach in practice. The idea of this long-lived concept is that the total market value

(debt and equity) is calculated by discounting the unlevered cash flows with the

Electronic copy available at: http://ssrn.com/abstract=1561188

weighted returns for shareholders re and bondholders rd.[1] The tax rate t adjusts the

return to bondholders downward to reflect the interest tax shield.

WACC = w d rd (1 − t) + w e re (1)

Recent survey evidence from the US, UK, and Germany supports the WACC’s

dominance in practice.[2] Three reasons could support this success: First, the input

parameters are rather easily estimated from market data. Second, applying WACC does

not require a commitment to judge how risky debt tax shields are. In other words, the

debt policy does not have to be specified. This is not an appealing constellation as it is

unknown how much the debt tax shield contributes to the company value. However,

given today’s knowledge, only a vague idea exists which debt policies companies

actually apply. Therefore, putting a valid value on the debt tax shield given these

estimation problems is not such an easy task. Third, the WACC is computationally

elegant. An iterative or recursive procedure is generally not necessary. However, the

WACC also has its known shortcomings. It implies by definition that a periodic

rebalancing of debt takes place to maintain the capital structure set forth in the WACC

formula. If this inherent mechanism is not acknowledged, WACC is prone to errors.

(Miller, 2007) challenges WACC with a nonlinear WACC (NLWACC). My concern is

to evaluate the validity of his assertion that WACC is not quite right, and to examine

the potential contribution of NLWACC to the literature.

The remainder of the paper is organized as follows. In section 2, the newly introduced

NLWACC is motivated and applied. Also, the NLWACC is generalized to allow for

annuities with growth rates g ≠ 0%. In section 3, the concept of the NLWACC will be

revisited from the perspective of rebalancing the capital structure using WACC before

taxes. The merits of WACC and NLWACC are discussed. Taxes being crucial in this

context as shown in the seminal work by Modigliani and Miller (1958), section 4

2
analyzes the after tax-case. Finally, section 5 summarizes shortly the findings and

offers an outlook.

2. WACC and modified WACC: an example

The notion of the modified WACC shall be covered briefly here:

1) Looking at a levered project with given outlays IC0, cost of capital WACC, and

duration N, which break-even unlevered cash flows CF has the project to deliver to be

acceptable? Financial acceptance is measured with the net present value NPV. To

derive a unique solution for this question, an annuity structure (allowing for geometric

growth) is imposed:[3]

⎡ (1 + g) N ⎤
⎢
CF 1 − ⎥
⎢⎣ (1 + WACC) N ⎦⎥ IC ⋅ (WACC − g)
NPV = 0 = −IC0 + ⇔ CF = 0 (2)
WACC − g (1 + g) N
1−
(1 + WACC) N

Adopting the numerical example of Miller (2007) which assumes g = 0%, IC0 =

$200,000, t = 0%, WACC = w d rd (1 − t) + w e re = 0.25 ⋅ 0.06 + 0.75 ⋅ 0.12 = 0.105 , and N

= 8, leads to

$200, 000 ⋅ 0.105

CF = ≈ $38,173.86 .
−8
1 − (1.105)

This threshold operating cash flow CF belongs to the shareholders and bondholders of

the company.

2) Having performed this exercise Miller (2007) further asks what the equivalent

annuity for shareholders CFe and bondholders CFd is? Equivalence here again is

3
achieved by equating the NPV with zero at time T = 0 respectively. This leads in

analogy to Eq. (2) to (assuming g = 0%):

w e ⋅ IC0 ⋅ re 0.75 ⋅ $200, 000 ⋅ 0.12

CFe = = ≈ $30,195.43, and
1 − (1 + re ) − N 1 − (1 + 0.12) −8

w d ⋅ IC0 ⋅ rd 0.25 ⋅ $200, 000 ⋅ 0.06

CFd = = ≈ $8,051.80 .
1 − (1 + rd ) − N 1 − (1 + 0.06)−8

Obviously, the sum of both annuities differs from the annuity of the sum of both flows:

CFe ($30,195.43) + CFd ($8,051.80) ≠ CF ($38,173.86)

  

CF( $38,247.23)

3) Discounting the sum of both annuities ($38,247.23) with the textbook WACC of

10.5% will overestimate the project value in this numerical example.[4] To overcome

this misvaluation, Miller (2007) suggests discounting this cash flow with a modified

version of the WACC dubbing it NLWACC (nonlinear WACC). This modified WACC

is an internal rate of return r which can be expressed as an implicit function when

incorporating also the possibility of geometrically growing annuities.[5]

r −g w e (re − g) w d (rd − g)
= + (3)
(1 + g) N (1 + g) N (1 + g) N
1− 1− 1−
(1 + r) N (1 + re ) N (1 + rd ) N

In the numerical example with g = 0%, interpolating for r yields according to Eq. (3):

r 0.75 ⋅ 0.12 0.25 ⋅ 0.06

= + ⇔ r ≈ 0.105553
1 − (1 + r) −8 1 − 1.12−

8

−8
1 − 1.06

=0.191236

Discounting with r now leads to a zero-NPV according to Eq. (2):

⎡ (1 + g) N ⎤
CF ⎢1 − ⎥
⎢⎣ (1 + r) N ⎥⎦ $38,247.23 ⎡1 − (1 + 0.105553) −8 ⎤
NPV = −IC0 + = −$200, 000 + ⎣⎢ ⎦⎥ ≈ $0
r−g 0.105553

4
The final deduction of Miller (2007) is that WACC is more or less flawed and that

NLWACC is needed.

In the following section 3, I put the NLWACC in perspective using the initial example.

The analysis reveals why a traditional WACC can not work in this scenario. Also, I

show how the challenger, the NLWACC, can be interpreted, and I demonstrate that the

NLWACC is a rather narrow concept which is not superior to the WACC.

3. WACC and modified WACC revisited – before taxes

First, it is helpful to remember that in this before tax-setting and under the assumptions

set forth by Modigliani and Miller (1958) the capital structure irrelevance theorems

hold. (To be absolutely clear, the just mentioned Nobel Prize laureate Merton H. Miller

is not to be confused with Richard A. Miller who proposed NLWACC.) This implies

that no additional value can be created while dividing the financing funds in a different

manner. In other words, the weighted average cost of capital equal the unlevered cost of

equity. Secondly, it is necessary to introduce a general return definition. A

straightforward and commonly applied definition incurs the total return RT in period

T[6]

DT + VT − VT −1
RT = , (4)
VT −1

where DT are inflows/outflows in period T (dividends, etc.), and VT is the market value

at the end of period T. The total return R can be defined as a return for shareholders of

levered (unlevered) projects re (ru), bondholders rd and for both claimholders combined

as WACC. WACC with time-varying input parameters can be written down more

generally than in Eq. (1) as follows:[7]

WACCT = w d,T −1 ⋅ rd,T ⋅ (1 − t) + w e,T −1 ⋅ re,T (5)

5
One of the constituent characteristics of the WACC concept is that returns to

shareholders and bondholders are weighted with their respective market value weights

Vd,T-1
of the prior period: w d,T-1 = , and w e,T-1 = 1 − w d,T-1 . It is important to
Ve,T-1 + Vd,T-1

emphasize that the definition of WACC is based on market, and not on book value

weights. Thus, discounting multi-period cash flows with a time-constant WACC

implies ceteris paribus a constant relative capital structure over time. This is

demonstrated for the initial example in Panel A of Table 1 showing the expected levels

of equity and debt over time given the information at T = 0.

Based on the calculations in Panel A of Table 1 it is natural to compute the implied

cash flows to shareholders and bondholders over the life of the project in Panel B. The

results reveal two remarkable points: First, the cash flow when divided between

bondholders and shareholders is sufficient to pay each group its necessary cash flow.

Discounting these cash flows with their respective returns leads to their implied market

values. Second, both flows do not conform to an annuity. In fact, the equity cash flows

decrease while the debt cash flows increase over time. Therefore, forcing the cash flows

to shareholders and bondholders to be annuities portrays a rather different scenario (see

section 2, part 2)). This scenario II is investigated in Table 2.

6
Table 1: Scenario I (before taxes)
Panel A: Total market value, equity market and debt market value over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
CFT (in $) 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86
VT (in $) 200,000.00 182,826.17 163,849.06 142,879.35 119,707.82 94,103.28 65,810.26 34,546.48 0.00
Ve,T (in $) 150,000.00 137,119.63 122,886.79 107,159.51 89,780.86 70,577.46 49,357.70 25,909.86 0.00
Vd,T (in $) 50,000.00 45,706.54 40,962.26 35,719.84 29,926.95 23,525.82 16,452.57 8,636.62 0.00
we 0.750 0.750 0.750 0.750 0.750 0.750 0.750 0.750 -
wd 0.250 0.250 0.250 0.250 0.250 0.250 0.250 0.250 -
CFT is the aggregated cash flow to shareholders and bondholders during the period T. VT is the total market value (debt and equity, Vd,T + Ve,T) at date T. Ve,T is the equity
market value (we⋅VT) at date T, and Vd,T is the debt market value (wd⋅VT) at date T. No taxes t = 0, WACC = ru = wd⋅rd + we⋅re = 0.25⋅0.06 + 0.75⋅0.12 = 0.105, and N = 8.
VT is the present value of CFT given WACC. we is the equity weight (equity/total value), and wd is the debt weight (debt/total value). The weights are constant over time.

Panel B: Implied cash flows to shareholders and bondholders over the life cycle of the project
T 1 2 3 4 5 6 7 8
CFe,T (in $) 30,880.40 30,687.19 30,473.70 30,237.79 29,977.11 29,689.06 29,370.76 29,019.04
CFd,T (in $) 7,293.46 7,486.67 7,700.16 7,936.07 8,196.75 8,484.80 8,803.10 9,154.82
CFT (in $) 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86 38,173.86
CFe,T is the implied cash flow to shareholders during the period T: CFe,T = CFT – CFd,T. CFd,T is the implied cash flow to bondholders during the period T based on Panel A:
CFd,T = Vd,T-1⋅rd + (Vd,T-1 – Vd,T). The aggregated cash flow to shareholders and bondholders during the period T is CFT = CFe,T + CFd,T. Discounting these flows at re = 0.12
for CFe,T, and rd = 0.06 for CFd,T leads to the same values of Ve,T and Vd,T as in Panel A.

7
Table 2: Scenario II (before taxes)
Panel A: Annuity cash flows, and implied market values over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
CFe,T (in $) 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43
CFd,T (in $) 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80
Ve,T (in $) 150,000.00 137,804.59 124,145.71 108,847.77 91,714.07 72,524.33 51,031.82 26,960.21 0.00
Vd,T (in $) 50,000.00 44,948.22 39,593.31 33,917.11 27,900.34 21,522.56 14,762.11 7,596.04 0.00
VT (in $) 200,000.00 182,752.81 163,739.02 142,764.88 119,614.41 94,046.89 65,793.93 34,556.24 0.00
CFe,T and CFd,T is the annuity cash flow to shareholders and bondholders, respectively. The implied equity market value Ve,T at date T is the present value given CFe,T and
re = 0.12. The implied debt market value Vd,T at date T is the present value given CFd,T and rd = 0.06. VT is the total market value of debt and equity (Vd,T + Ve,T) at date T.

Panel B: Implied capital structure ratios and returns over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
we,T 0.750 0.754 0.758 0.762 0.767 0.771 0.776 0.780 -
wd,T 0.250 0.246 0.242 0.238 0.233 0.229 0.224 0.220 -
WACCT = ru,T 0.10500 0.10524 0.10549 0.10575 0.10600 0.10627 0.10654 0.10681
we,T is the time-varying equity weight (equity/total value), and wd,T is the time-varying debt weight (debt/total value) based on Panel A. The time-varying WACCT is
computed via Eq. (5) as WACCT = wd,T-1⋅rd + we,T-1⋅re = wd,T-1⋅0.06 + we,T-1⋅0.12, and is confirmed by Eq. (6) calculating ru,T. Applying Eq. (7), the total market values of
Panel A are confirmed by discounting CFT (= CFe,T + CFd,T) with WACCT

8
Panel A of Table 2 now shows that market values (i.e., Ve,T, Vd,T, and VT) are identical

with market values given in Panel A of Table 1 only at T = 0. At other valuation dates

T, the values differ, hence exhibiting a different scenario than scenario I considered in

Table 1. Scenario I (rebalancing at a constant ratio of debt to market value), and

scenario II (rebalancing at an implicit time-varying ratio of debt to market value) are

obviously not directly comparable.

Panel B of Table 2 analyzes the capital structure ratios over time in scenario II and

highlights several implications for the returns. First, the debt ratio is expected to shrink

over time. Appendix B shows that under plausible conditions this is generally true. It is

only the same as in Panel A of Table 1 at T = 0. Second, this observation has an impact

on the WACC returns according to Eq. (5). Under plausible conditions, WACC

increases over time. Third, because the return to equity is supposed to be constant in

this valuation exercise a very specific behaviour of the operating returns over time is

implied when the capital structure is changing. Building on the (Modigliani and Miller,

1958) proposition 2, Eq. (6) postulates that the operating return ru,T is expected to

behave over time as follows.

w d,T −1
re + rd
w d,T −1 w e,T −1
re = ru,T + (ru,T − rd ) ⇔ ru,T = (6)
w e,T −1 w
1 + d,T −1
w e,T −1

This built-in feature seems not very appealing. Confirming the irrelevance theorem

again, WACCT = ru,T proves to be true.

Panel B of Table 2 also explains why discounting with a time-constant WACC has to

fail. The WACC is simply time-varying in scenario II. Textbooks usually do not

emphasize that the WACC can also be time-varying.[8] However, the use of a time-

varying WACC has the advantage of an easy economic interpretation. A changing

WACC points at a changing capital structure over time. The NLWACC does not

9
provide this information. Thus, discounting debt and equity annuity cash flows

($38,247.23) with time-varying WACCs leads to consistent results shown in Panel A of

Table 2.

CFT
Vτ−1 = ∑ T =τ
N
(7)
∏ (1 + WACC )
N
j=τ j

How can the modified WACC which was employed in section 2 be interpreted now?

Instead of using time-varying WACCs, the NLWACC allows to calculate the present

value at T = 0

• with a single (amalgamated) discount rate

• if the cash flows to shareholders and bondholders exhibit an annuity structure,

and

• if the operating cost of capital incidentally follow a specific structure to ensure a

constant equity return over time (as in the no tax-case shown by Eq. (6)).

The NLWACC certainly is not a return as defined by Eq. (4). It is an amalgamation of

time-varying returns into one discount rate, as typically is the case with internal rate of

return procedures. This limits its economic interpretation. If one wants to use

NLWACC also for prospective valuation at dates T > 0, it has to be updated, that is, N

in Eq. (3) has to be updated. The conditions set forth in scenario II are rather limiting

(even when allowing for g ≠ 0%) and only then the use of NLWACC is admissible.

Clearly, the NLWACC does not help in situations where cash flows are not annuities. A

time-varying WACC does not share these limitations. Its use leads to consistent results.

However, it is not clear why under scenario II the flow to equity approach is not used

instead. This would be the natural choice. The NLWACC (and WACC) seem like a

detour.

Under scenario I, WACC is definitely the right choice, and even under scenario II a

time-varying WACC (keeping the linear structure), solves the problem as does the

10
NLWACC. WACC obviously is a technique better able to handle more general

situations than NLWACC.

4. Analysis after taxes

Modigliani and Miller (1958) have shown in their seminal work that without taxes the

WACC concept does not render any additional insights in comparison to the unlevered

cost of capital. The after tax-WACC is covered lengthy in Miller's (2007) paper. Thus,

to reconcile the argument taxes have to be considered.

Again, in scenario I, an after tax-annuity is discounted with a constant after tax-WACC

to arrive at V0 = $200,000, which implies ceteris paribus, a constant relative capital

structure over time. The after tax-annuity now is $37,488.80. This is demonstrated for

the initial example in Panel A of Table 3 showing the expected levels of equity and

debt over time given the information at T = 0.

Based on the calculations in Panel A of Table 3 it is obvious to compute the implied

cash flows to shareholders and bondholders over the life of the project in Panel B. The

results again reveal that actually both flows do not conform to an annuity. In fact, the

equity cash flows decrease while the debt cash flows increase over time. Therefore,

forcing the cash flows to shareholders and bondholders to be annuities portrays a rather

different scenario which is investigated further in Table 4. To strengthen the argument,

an alternative calculation based on the Adjusted Present Value technique (APV) is

provided in Panel C. As is well known, the operating value (unlevered value) Vu and

the value of the debt tax shield Vt are valued separately. Unlike in the no tax-case, it is

now important to specify which financing policy is pursued. Different debt tax policies

will require a case-by-case discussion. For the purpose of illustration, I consider a

value-based debt policy only.[9]

11
Table 3: Scenario I (after taxes)
Panel A: Total market value, equity market and debt market value over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
CFT (in $) 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80
VT (in $) 200,000.00 182,511.18 163,273.50 142,112.05 118,834.45 93,229.10 65,063.21 34,080.73 0.00
Ve,T (in $) 150,000.00 136,883.38 122,455.12 106,584.04 89,125.84 69,921.82 48,797.40 25,560.55 0.00
Vd,T (in $) 50,000.00 45,627.79 40,818.37 35,528.01 29,708.61 23,307.27 16,265.80 8,520.18 0.00
we 0.750 0.750 0.750 0.750 0.750 0.750 0.750 0.750 -
wd 0.250 0.250 0.250 0.250 0.250 0.250 0.250 0.250 -
CFT is the unlevered cash flow during period T. VT is the total market value (debt and equity, Vd,T + Ve,T) at date T. Ve,T is the equity market value (we⋅VT) at date T. Vd,T is the debt
market value (wd⋅VT) at date T. Taxes are considered with t = 0.3333, WACC = wd⋅rd⋅(1-t) + we⋅re = 0.25⋅0.06⋅0.6667 + 0.75⋅0.12 = 0.10, and N = 8. VT is the present value of CFT
given WACC at date T. we is the equity weight (equity/total value), and wd is the debt weight (debt/total value). The weights are constant over time.

Panel B: Implied cash flows to shareholders and bondholders over the life cycle of the project
T 1 2 3 4 5 6 7 8
CFd,T (in $) 7,372.20 7,547.09 7,739.47 7,951.08 8,183.86 8,439.91 8,721.57 9,031.39
Vd,T-1⋅rd⋅(1-t) + (Vd,T-1 – Vd,T) (in $) 6,372.20 6,634.53 6,923.10 7,240.52 7,589.68 7,973.76 8,396.25 8,860.99
CFe,T (in $) 31,116.60 30,854.27 30,565.70 30,248.28 29,899.12 29,515.04 29,092.55 28,627.81
CFT (in $) 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80 37,488.80
CFd,T is the implied cash flow to bondholders during the period T based on Panel A: CFd,T = Vd,T-1 rd + (Vd,T-1 – Vd,T). To compute the unlevered cash flow CFT Vd,T-1⋅rd⋅(1-t) + (Vd,T-1 –
Vd,T) and CFe,T is needed. CFe,T is the implied cash flow to shareholders during the period T: CFe,T = CFT – CFd,T. Discounting CFe,T with re = 0.12 leads to Ve,T, and discounting CFd,T
with rd = 0.06 leads to Vd,T.

Panel C: Adjusted Present Value and market values over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
Vu,T (in $) 196,260.03 179,420.39 160,809.00 140,239.44 117,505.69 92,380.04 64,610.85 33,919.97 0.00
Vd,T-1⋅rd⋅t (in $) 1,000.00 912.56 816.37 710.56 594.17 466.15 325.32 170.40
Vt,T (in $) 3,739.95 3,090.79 2,464.50 1,872.61 1,328.77 849.05 452.36 160.76 0.00
VT = Vu,T + Vt,T (in $) 200,000.00 182,511.18 163,273.50 142,112.05 118,834.45 93,229.10 65,063.21 34,080.73 0.00
For illustration purposes a value-based debt policy is assumed. The unlevered firm value Vu,T is computed by discounting CFT based on Eq. (9) with a time-constant ru = 0.10521
according to Eq. (8) . The debt tax shield Vd,T-1⋅rd⋅t is discounted with ru and rd to arrive at the value of the debt tax shield Vt,T according to Eq. (10). VT is the total market value (Vu,T +
Vt,T) at date T.

12
Table 4: Scenario II (after taxes)
Panel A: Annuity cash flows, and implied market values over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
CFe,T (in $) 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43 30,195.43
CFd,T (in $) 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80 8,051.80
Ve,T (in $) 150,000.00 137,804.59 124,145.71 108,847.77 91,714.07 72,524.33 51,031.82 26,960.21 0.00
Vd,T (in $) 50,000.00 44,948.20 39,593.30 33,917.10 27,900.33 21,522.55 14,762.11 7,596.04 0.00
VT (in $) 200,000.00 182,752.79 163,739.01 142,764.87 119,614.40 94,046.88 65,793.92 34,556.24 0.00
CFe,T and CFd,T is the annuity cash flow to shareholders and bondholders, respectively. The implied equity market value Ve,T at date T is the present value given CFe,T and re = 0.12. The
implied debt market value Vd,T at date T is the present value given CFd,T and rd = 0.06. VT is the total market value of debt and equity (Vd,T + Ve,T) at date T.

Panel B: Implied capital structure ratios and returns over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
we,T 0.750 0.754 0.758 0.762 0.767 0.771 0.776 0.780 -
wd,T 0.250 0.246 0.242 0.238 0.233 0.229 0.224 0.220 -
WACCT 0.10000 0.10032 0.10066 0.10099 0.10134 0.10169 0.10205 0.10242
ru,T 0.10521 0.10545 0.10570 0.10595 0.10621 0.10647 0.10674 0.10701
we,T is the time-varying equity weight (equity/total value), and wd,T is the time-varying debt weight (debt/total value) based on the results of Panel A. The time-varying WACCT is
computed with Eq. (5) as WACCT = wd,T-1⋅rd⋅(1-t) + we,T-1⋅re = wd,T-1⋅0.06⋅0.6667 + we,T-1⋅0.12. The implied unlevered cost of capital ru,T is calculated for illustration purposes in line
with a value-based debt policy according to Eq. (8).

Panel C: Adjusted Present Value and market values over the life cycle of the project
T 0 1 2 3 4 5 6 7 8
CFT (in $) 37,247.33 37,348.35 37,455.44 37,568.95 37,689.28 37,816.82 37,952.01 38,095.32
Vu,T (in $) 196,383.29 179,798.09 161,410.14 141,015.75 118,387.64 93,272.10 65,385.93 34,412.93 0.00
Vd,T-1⋅rd⋅t (in $) 999.90 898.87 791.79 678.27 557.95 430.41 295.21 151.91
Vt,T (in $) 3,616.73 2,954.71 2,328.87 1,749.11 1,226.76 774.78 407.99 143.31 0.00
VT = Vu,T + Vt,T (in $) 200,000.00 182,752.79 163,739.01 142,764.87 119,614.40 94,046.88 65,793.92 34,556.24 0.00
CFT is the unlevered cash flow during period T computed as CFT = CFe,T + Vd,T-1⋅rd⋅(1-t) + (Vd,T-1 – Vd,T); see Panel A, rd = 0.06, t = 0.3333. The unlevered value Vu,T is computed
according to Eq. (9) with ru,T. The debt tax shields Vd,T-1⋅rd⋅t are discounted according to Eq. (10) with ru,T and rd to arrive at the value of the debt tax shield Vt,T. VT is the total market
value (Vu,T + Vt,T) at date T.

13
Miles and Ezzell (1980) and Kruschwitz and Löffler (2006) show for this policy:

⎛ rt ⎞ w
re + rd ⋅ ⎜1 − d ⎟ ⋅ d,T-1
⎛
re = ru,T + ( ru,T − rd ) ⋅ ⎜1 − d
rt ⎞ w d,T-1 ⎝ 1 + rd ⎠ w e,T-1
⎟⋅ ⇔ ru,T = (8)
⎝ 1 + rd ⎠ w e,T-1 ⎛ rt ⎞ w
1 + ⎜1 − d ⎟ ⋅ d,T-1
⎝ 1 + rd ⎠ w e,T-1

The unlevered value is computed with the unlevered cost of capital as follows:

CFT
Vu,τ−1 = ∑ T =τ
N
(9)
∏ j=τ (1 + ru, j )
N

In the special case of a time-constant ru, the formula is easier:

CFT
Vu,τ−1 = ∑ T =τ
N
.
(1 + ru )
T −τ+1

The value of the debt tax shield is risky given this financing policy, and is therefore

discounted with the unlevered cost of equity apart from the cash flow of the previous

period which is certain.

Vd,T-1 ⋅ rd ⋅ t
Vt,τ−1 = ∑ T =τ
N
(10)
∏ (1 + r ) (1 + r )
N −1
j=τ u, j d

Vd,T-1 ⋅ rd ⋅ t
Simplifying with a time-constant ru leads to Vt,τ−1 = ∑ T =τ
N
.
(1 + ru ) (1 + rd )
T −τ

The APV approach confirms the consistency of the calculation in Panel C of Table 3.

To arrive at the equity value, the value of debt has to be subtracted from the total

market value (Vu + Vt).

Under scenario II (forced annuity structure) Panel A of Table 4 exhibits that market

values are identical with market values given in Panel A of Table 3 only at T = 0.[10]

For other valuation dates T, values are different, hence exhibiting a different scenario

than scenario I considered in Table 3. Scenario I (rebalancing at a constant ratio of debt

to market value) and scenario II (rebalancing at an implicit time-varying ratio of debt to

market value) are, again, not directly comparable.

14
Panel B of Table 4 analyzes the capital structure ratios over time in scenario II and

highlights several implications for the returns. First, the debt ratio is expected to shrink

over time again. It is only the same as in Panel A of Table 3 at T = 0. Second, this

observation has an impact on the WACC returns according to Eq. (5). WACC increases

over time. Third, because in this valuation exercise the return to equity is supposed to

be constant a very specific behaviour of the operating returns over time is implied when

the capital structure is changing. The unlevered cost of equity increases over time.

Again, this built-in feature seems not very appealing. Panel B of Table 4 also explains

why discounting with a time-constant WACC has to fail. WACC is simply time-

varying in scenario II. The NLWACC does not provide any information about the

changing capital structure. Thus, discounting unlevered cash flows with time-varying

WACCs leads to results consistent with Panel A of Table 4. The arguments already

established in section 3 can be analogously repeated. Additionally, Miller’s (2007)

claim that in the tax-case the integration of the interest tax shield in the WACC formula

seems misplaced does not have to be followed. The time-varying WACC approach

proves this fact. The APV calculation in Panel C of Table 4 confirms that operating

returns are time-varying, indeed, and also confirms the results of Panel A and B. For

example, the value of the debt tax shield at the beginning of period 6 is calculated based

on Eq. (10) as:

295.21 151.91
Vt,6 = + = 407.99 .
1.06 1.10674 ⋅1.06

The analysis underlines that in the tax-case the situation is not getting any better for the

NLWACC and the aforementioned results hold in principle.

15
5. Summary and outlook

The claim that the NLWACC is conceptually superior to a constant WACC seems for

two reasons rather artificial:

1) A slightly modified WACC which is time-varying is conceptually superior to the

NLWACC. It does not seclude itself from an economic interpretation.

2) Given the special valuation scenario for which the NLWACC is motivated, the flow

to equity approach is the direct solution while the (NL)WACC is a detour.

Therefore, the foundations of WACC are sound. However, valuation is still a field

which has many promising research questions to offer. Just to name a few: Which debt

policies can be empirically supported? Given its autarkic nature, this is a question the

WACC does not necessarily have to approach. How do more realistic tax regimes with

personal taxes influence tax shield valuation? Which terminal value calculations are

plausible? These and others seem to be more pressing questions which deserve further

attention in future research.

16
Appendix A: Derivation of time-varying WACC

The following derivation draws on the proof of a time-constant WACC by Brealey et

al. (2008), p. 533. Starting with the value in the next to last period: VN-1 = Ve,N-1 + Vd,N-1 .

The total cash flow to debt and equity investors is the cash flow CF plus the interest tax

shield: CFN + t ⋅ rd,N ⋅ Vd,N −1 . This total cash flow can also be written based on returns:

⎛ Vd,N-1 Ve,N-1 ⎞
VN −1 ⎜1 + rd,N
⎜
+ re,N ⎟ = VN −1 (1 + rd,N w d,N-1 + re,N w e,N-1 )
⎝ Ve,N-1 + Vd,N-1 Ve,N-1 + Vd,N-1 ⎟⎠

Equate both definitions and solve for VN-1:

CFN CFN
VN −1 = =
1 + rd,N (1 − t)w d,N-1 + re,N w e,N-1 1 + WACC N

The WACC-definition of Eq. (5) shows up. This can be repeated for VN-2. As the

return-definition is based on Eq. (4) the next period’s payoff includes VN-1:

CFN −1 + t ⋅ rd,N −1 ⋅ Vd,N − 2 + VN −1 = VN − 2 (1 + rd,N −1w d,N-2 + re,N −1w e,N-2 )

Solving for VN-2 yields:

CFN −1 + VN −1 CFN −1 + VN −1
VN − 2 = =
1 + rd,N −1 (1 − t)w d,N-2 + re,N −1w e,N-2 1 + WACC N −1
CFN −1 CFN
= +
1 + WACC N −1 (1 + WACC N −1 )(1 + WACC N )

By recursion, one obtains for any arbitrary valuation date τ–1:

CFT
Vτ−1 = ∑ T =τ
N
(7)
∏ (1 + WACC )
N
j=τ j

17
Appendix B: Is the debt to total market value ratio falling over time?

To show the conditions for this relationship, I compare growth rates of debt and equity

Ve,T − Ve,T −1
market value, e.g. the equity growth rate is: g T (Ve ) = . If debt is not
Ve,T −1

growing as strong as equity, the debt ratio has to shrink. Inserting Eq. (2) yields the

following growth rate:

⎡ (1 + g) N − T ⎤ ⎡ (1 + g) N − T + 1 ⎤
CFe ⎢1 − ⎥ CFe ⎢1 − ⎥ N −T

⎢⎣ (1 + re ) N − T ⎥⎦ ⎢⎣ (1 + re ) N − T + 1 ⎥⎦ ⎡1+ g ⎤
− 1− ⎢ ⎥
g T (Ve ) =
re − g re − g
= ⎣1 + re ⎦
N −T+1 − 1
⎡ (1 + g) N − T + 1 ⎤ ⎡1+ g ⎤
CFe ⎢1 − ⎥ 1− ⎢ ⎥
⎢⎣ (1 + re ) N − T + 1 ⎥⎦ ⎣1 + re ⎦
re − g

For the debt growth rate equivalently:

N −T

⎡1+ g ⎤
1− ⎢ ⎥
g T (Vd ) = ⎣1 + rd ⎦
N −T +1 − 1

⎡1+ g ⎤
1− ⎢ ⎥
⎣1 + rd ⎦

Under realistic conditions ( rd < re ) equity growth rates are higher than debt growth

rates. Therefore, wd,T decreases over time.[11]

18
References

Arnold, G.C. and Hatzopoulos, P.D. (2000), “The Theory-Practice Gap in Capital

Budgeting: Evidence from the United Kingdom”, Journal of Business Finance

and Accounting, 27, pp. 603-626.

Arzac, E.R. and Glosten, L.R. (2005), “A Reconsideration of Tax Shield Valuation”,

European Financial Management, 11, pp. 453-461.

Booth, L. (2007), “Capital Cash Flows, APV and Valuation”, European Financial

Management, 14, pp. 29-48.

Brealey, R.A., Myers, S.C. and Allen, F. (2008), Principles of Corporate Finance (9th

ed.), Mc Graw-Hill Irwin, New York.

Bruner, R.F., Eades, K.M., Harris, R.S. and Higgins, R.C. (1998), “Best Practices in

Estimating the Cost of Capital: Survey and Synthesis”, Financial Practice and

Education, Spring/Summer, pp. 13-28.

Campbell, J.Y., Lo, A.W. and MacKinlay, A.C. (1997), The Econometrics of Financial

Markets, Princeton University Press, Princeton, NJ.

Cooper, I.A. and Nyborg, K.G. (2006), “The value of tax shields is equal to the present

value of tax shields”, Journal of Financial Economics, 81, pp. 215-225.

Cooper, I.A. and Nyborg, K.G. (2007), “Valuing the debt tax shield“, Journal of

Applied Corporate Finance, 19.2, pp. 50-59.

Cooper, I.A. and Nyborg, K.G. (2008), “Tax-Adjusted Discount Rates with Investor

Taxes and Risky Debt”, Financial Management, pp. 365–379.

Copeland, T.E., Weston, J.F. and Shastri, K. (2005), Financial Theory and Corporate

Policy (4th ed.), Pearson Addison Wesley.

Daves, P.R., Ehrhardt, M.C. and Shrieves, R.E. (2004), Corporate Valuation: A Guide

for Managers and Investors, Thomson South-Western.

19
Fernandez, P. (2004), “The value of tax shields is NOT equal to the present value of tax

shields”, Journal of Financial Economics, pp. 145-165.

Fieten, P., Kruschwitz, L., Laitenberger, J., Löffler, A., Tham, J., Vélez-Pareja, I. and

Wonder, N. (2005), “Comment on „The value of tax shields is NOT equal to the

present value of tax shields“, Quarterly Review of Economics and Finance, pp.

184-187.

Graham, J.R. and Harvey, C.R. (2001), “The Theory and Practice of Corporate Finance:

Evidence from the Field”, Journal of Financial Economics, 60, May, pp. 187-243.

Koller, T., Goedhart, M. and Wessels, D. (2005), Valuation: Measuring and managing

the value of companies (4th ed.), John Wiley & Sons, Hoboken, NJ.

Kruschwitz, L. and Löffler, A. (2006), Discounted Cash Flow: A Theory of the

Valuation of Firms, John Wiley & Sons, Chichester, West Sussex.

Lobe, S., Niermeier, T., Essler, W. & Röder, K. (2008). Do Managers follow the

Shareholder Value Principle when applying Capital Budgeting Methods? A

Comparison of Theory and Practice based on German Survey Results and Return

Data. Center of Finance UR Working Paper.

Lundholm, R.J. and Sloan, R.G. (2004), Equity Valuation and Analysis with eVal, Mc

Graw-Hill Irwin, New York.

Massari, M., Roncaglio, F. and Zanetti, L. (2008), “On the Equivalence between the

APV and the wacc Approach in a Growing Leveraged Firm”, European Financial

Management, 14, pp. 152–162.

Miles, J.A. and Ezzell, J.R. (1980), “The Weighted Average Cost of Capital, Perfect

Capital Markets, and Project Life: A clarification”, Journal of Financial and

Quantitative Analysis, 15, pp. 719-730.

Miller R.A. (2007), “The weighted average cost of capital is not quite right”,

forthcoming in The Quarterly Review of Economics and Finance,

20
 doi:10.1016/j.qref.2006.11.001

Modigliani, F. and Miller, M.H. (1958), “The Cost of Capital, Corporation Finance and

the Theory of Investment”, The American Economic Review, 48, pp. 261-297.

Ruback, R.S. (2002), “Capital Cash Flows: A Simple Approach to Valuing Risky Cash

Flows”, Financial Management, pp. 85-103.

Stowe, J.D., Robinson, T.R., Pinto, J.E. and McLeavey, D.W. (2007), Equity Asset

Valuation, John Wiley & Sons, Hoboken, NJ.

Titman, S. and Martin, J.D. (2008), Valuation. The Art & Science of Corporate

Investment Decisions, Pearson Addison Wesley.

21
Endnotes

[1] For ease of exposition the notation of Miller (2007) is adopted. Expectation operators are therefore also dropped.

[2]For the US, see Bruner, Eades, Harris, & Higgins (1998), and Graham, & Harvey (2001), for the UK, see Arnold,

& Hatzopoulos (2000), and for Germany, see Lobe, Niermeier, Essler, & Röder (2008).

[3] The following geometric series is evaluated: CF = CF1, CF (1+g) = CF2, … , CF (1+g)N-1 = CFN.

[4] The present value then is $200,384.42 which is marginally higher than the true present value of $200,000.

[5] See (Miller, 2007), p. 8, Eq. (23) for g = 0%. The derivation incorporating g is straightforward, and thus needs

not be shown here.

[6] See e.g. (Campbell et al., 1997), p. 12.

[7] A derivation of Eq. (5) can be found in Appendix A.

[8] See, for example, (Brealey et al., 2008), (Copeland et al., 2005), (Daves et al., 2004), (Lundholm and Sloan,

2004), (Stowe et al., 2007), (Titman and Martin, 2008), (Koller et al., 2005).

[9] For other policies see (Kruschwitz and Löffler, 2006).

[10] Panel A of Table 4 is identical with Panel A of Table 2. Therefore, also the weights in Panel B of Tables 2 and

4 are the same. The returns are different, of course, as different tax situations are considered in Table 2 and 4.

[11] Except for N → ∞, and T = N. Then, gT(Vd) = gT(Ve).

22