Global Industries

International

The EV Guide

November 1996

International

Global Industries

The EV Guide
Contents
page 2
3
3
5
7
11

13
13
15
17
18
19
20
21
24

28
28
30
32
33

35
35

Summary
Section 1: The status quo
What's wrong with P/Es
The underlying theory
What about P/E relatives?
What about P/CE multiples?

Section 2: Enterprise value: the price of a business

What do we mean by Enterprise Value?
How do we define non-core assets?
What about minority interests?
What about tax?
The assumptions of EV methodology
So proposition 1 does not always hold. Why?
No-mans land: between tax codes
How are shares actually priced?

Section 3: The real size of a business

How can we best measure profits?
So how can we estimate maintenance capex?
What valuation ratios can we construct?
Book values

Section 4: EV in action
Value vs growth

42

Section 5: Definitions

48

Section 6: a practical example

48
50
51
51
52

54

New BT - New Value

Estimated value of Old BT
Free cash flow of Old BT
Conclusions to be drawn
Extensions

Appendix 1: PER distributions

Authors:
Tom Hill
(+44 171 382 4212)

Tim Orchard
(+44 171 860 0559)

Nainish Bapna
(+44 171 382 4776)

November 1996

Summary

SBC WARBURG

The larger investment institutions are moving steadily in the direction of crossborder, sector-based investment policy. Their rate of progress is limited by,
among other things, a shortage of consistent information. As a result, they are
demanding valuation measures which look through quirks of national or
sector accounting.

We believe that valuation which looks at the total capital structure of

companies offers the best route to developing a consistent, though not
exclusive, methodology which applies across all markets.

This paper sets out the empirical evidence which shows that P/E ratios are
inadequate across borders, develops the concepts underlying EV analysis, and
shows how valuation data fit into the analytical process.

Back-testing of EV methodology already in use shows that it improves the

performance of our recommendations, while P/E ratios do not.

The data also show that stock markets do price in tax effects in countries
where the tax system is not neutral between debt and equity. This paper shows
how, in principle, EV methodology should be modified to allow for this effect.
However, the complexity of tax systems, and the fact that they are steadily
becoming more neutral between debt and equity, make it undesirable to
introduce such a modification at this stage.

Theory is no use unless it is consistently applied; this paper sets out some
standard definitions, to be updated when necessary.

We also show an example of a piece of research which exploits these valuation

techniques to the full.

Section 1: The status quo

Problems with P/Es the influence of accounting standards balance sheet
structures the Gordon model, and why it fails Problems with P/E
relatives P/E distributions and the elusive nature of the market multiple
Problems with P/CE how cash earnings confuse comparisons across sectors

Whats wrong with P/Es?

Accounting varies by
country ...

In contrast with the US or UK stock markets, equity investment in Continental

Europe, emerging markets, and Asia has long been complicated by differences
in accounting standards and practices between countries. Not surprisingly,
investors in these markets routinely adjust reported earnings in order to make
relative investment judgements.

... but investors wait to

compare sectors across
borders ...

There is now a clearly discernible trend among investors towards the

allocation of equity investments by sector as well as by geographical market.
Since those sectors are trans-national, would-be valuers of equities must come
up with yardsticks which allow comparison across borders. Moreover, to be
useful the figures must be comparable between sectors as well as between
countries; the requirement is for a set of universal valuation criteria.

... so need a valuation

toolbox.

At the outset, we must stress that no valuation system, however ingenious, is

perfect; and that calculating multiples is only a part of the problem. We are
not attempting to reduce investment analysis to a formula; merely to create
tools to help tackle the cross-border problem.
The natural inclination is to begin by calculating P/E ratios; earnings per
share, are, after all, supposed to measure the earnings attributable to
shareholders. It turns out, though, that there are (at least) two drawbacks with
EPS as a measure of a companys true profits.

EPS depends on
accounting systems ...

First, net profits are determined by the accounting policies used to prepare a
particular set of accounts; and these vary between companies. Earnings can be
higher or lower depending on how a company accounts for depreciation of
fixed assets, amortisation of intangible assets, provision for deferred tax,
valuation of inventories, treatment of pensions, capitalisation of R&D
expenses the list is endless.

... which may not be

designed for the same
audience.

The extent to which stated earnings reflect a companys economic profits often
depends on the audience the company has in mind. Although shareholders
may stand at the front of the queue in most English-speaking markets, lenders
or employees or the tax authorities often loom larger in the thoughts of
management elsewhere and so dictate different accounting priorities. The
consequence is that an investor who is assessing value within one stock market
may indeed feel comfortable that the accounting policies followed by any two
companies in that market are closely aligned and thus that the EPS of the

SBC WARBURG

two companies are genuinely comparable. But the same investor would be
unwise then to compare the resulting P/Es with those of companies from
another country.
There is no simple
conversion from EPS in
one system to EPS in
another ...

The following chart shows the differences between EPS reported under US and
home accounting conventions for a selection of UK companies. There are wide
differences between the figures reported under each convention, and there is
no simple rule of thumb for converting from one to the other. In many cases,
UK-accounted EPS is significantly higher than in the US, but there is no
pattern to the data.
UK vs US earnings
Mean is 2%

Number of companies

0
-20%

-10%

0%

10%

20%

30%

40%

50%

60%

More

Percentage difference between UK GAAP and US GAAP EPS

...and balance sheet

structures affect
multiples ...

SBC WARBURG

Second, bald EPS and multiples based on them ignore the huge influence
which balance sheet structure can exert on the equity multiple. This is best
demonstrated by an example.

... as an example shows ...

Consider two companies which are identical in every respect and whose
assets are by definition worth the same except for the fact that Company
A has a market capitalisation of $100m and is debt-free while Company B has
a market capitalisation of $50m and has $50m of net debt (costing it, say,
12% p.a.). As the table below illustrates, debt-free Company A stands on a
P/E of 15.4x, whereas Company Bs earnings are capitalised at more than 19x.
Gearing affects P/E ratios
($ million)

Company A

Company B

EBIT

10.0

10.0

Net interest

0.0

10.0

(6.0)

4.0

Net profit

(3.5)

6.5

(1.4)

2.6

Market capitalisation
P/E

100.0
15.4x

50.0
19.2x

Pre-tax profit
Tax (at 35%)

The discrepancy between the two companies P/Es is a function of interest

rates as well as the proportion of assets financed by debt rather than equity.
For instance, if Company Bs debt were to cost it 15%, then its net profits
would be reduced to $1.6m and its P/E inflated to almost 31x. Conversely,
if the cost of its debt fell to, say, 5%, then net profits would rise to $4.9m
bringing the P/E down to a mere 10x.
...because operating
earnings and interest costs
are capitalised at different
rates.

What we are considering here, of course, is the relationship between a

companys earnings yield and its cost of debt: if Company Bs debt costs 10%
(i.e. the same as its pre-interest earnings yield), then the two companies P/E
multiples will be the same, regardless of the relative proportions of debt and
equity in either companys balance sheet. (At this point we are ignoring the risk
of financial distress which would begin to make itself felt if gearing rose to
uncomfortably high levels; more on this subject later. We will also return to the
subject of tax.) But in practice such coincidences are the exception, not the rule.
This example also shows why it is misleading to use P/E multiples to compare
two stocks in different countries with different prevailing interest rates.

The underlying theory

P/Es are based on a
perpetuity model ...

People didnt always use P/E ratios; there was a time when the standard
approach to valuing an equity was to look at the dividend yield and consider
whether it was sufficiently larger than bond yields to compensate the investor
for the risk of holding common stock.
We now know that this overlooks the fact that dividends tend to grow. It was
when this point became generally appreciated that the P/E ratio took over
although it took some time to achieve the enormous popularity it now enjoys.

SBC WARBURG

When investors use a P/E ratio to assess how much they should pay for a
stock, they are implicitly using the Gordon model the model commonly
used to value an investment assumed to grow at a constant rate in perpetuity.
The Gordon model is expressed thus:
P

where:

earnings per share

price per share

expected return

expected growth rate

To relate this to P/E ratios, we can restate the Gordon model:

E/P

where:

(i + r) - g

risk-free rate

equity risk premium

In other words, the earnings yield (i.e. the inverse of the P/E) set by the market
will rise and fall in line with the return which the investor expects from the
share; the lower the return required by the investor, the lower the earnings
yield (and thus the higher the P/E). The model shows why growth stocks
tend to command high P/E multiples because the g in the equation is
relatively high.
... which is fraught with
drawbacks.

SBC WARBURG

Despite its popularity, however, there are several problems in this approach.
Here are just some of them:

Problem

Popular remedies

What popular remedies fail to address

Earnings yield

Are reported earnings the

same as economic earnings?

Adjustments for: extraordinary

items, discontinued businesses,
inventory valuations,
depreciation policies, etc.

Less obvious differences in accounting

policies
Effects of (historic) inflation
Tendency of non-recurring items to recur
Influence of capital structure on earnings

Risk-free rate

What if government bonds

incorporate (default) risk?
How can this be quantified?

P/E relatives

Observation problems (how should market

P/E be calculated?)
Influence of capital structure on earnings
Effects of monetary policy environment

Equity risk premium

Observation problems;
huge discrepancies
within historic data

P/E relatives

Different markets reveal different equity

risk premiums at different times

Expected growth rate

What if growth is not steady?

Use of mid-cycle or
normalised earnings

Estimation errors

But these difficulties are a

profit opportunity.

In a perfect world, all companies would state their earnings in precisely the
same way; but they dont. Indeed, one of the paradoxes of investment is that
the more accessible the information on companies, the less easy it is to spot
bargains. It takes more effort to look beyond the simple P/E ratio, but the
extra work is rewarded with a far better understanding of a share price,
especially in a less well-developed stock market.

What about P/E relatives?

Unfortunately P/E
relatives create as many
problems as they solve ...

As mentioned earlier, it might seem to make some sense for an investor to use
P/E ratios when comparing the valuation of two stocks within the same
market, even if that will probably reveal little about either stocks worth
relative to a similar companies in other countries. Does that imply, then, that
the relationship between a stocks P/E ratio and the P/E ratio of its domestic
stock market (i.e. its P/E relative) is indeed a helpful barometer of value?
The answer to this question is maybe. If EPS is the best indicator of
profitability and if that one stock market represents the investors total equity
investment universe (two pretty big ifs), then it might be argued that a
stocks P/E relative means something.
Applying the Gordon model, the P/E relative for a company should be given
by the formula
P/E (stock)

P/E (market)

I + r(market) - g(market)

I + r(stock) - g(stock)

Under some (rather generous) assumptions, including that the risk premium is
the same for the stock as for the market as a whole, this simplifies to
P/E relative

g(stock) - g(market)
1 +
I

Or that the P/E relative is indirectly proportional to the excess growth rate of
the stock. The problem with this, of course, is that no stock can, by definition,
grow faster than the market in perpetuity; we leave it to the curious reader to
devise a formula which gives the P/E relative assuming a return to market
growth at some point in the future.
... especially across
borders.

Superficially, the P/E relative does away with the problems of risk premium,
because it assumes that any variations affect all stocks simultaneously. But this
is not a good assumption even when investors are confined to one particular
market. When they begin looking across borders, they increasingly distinguish
between country risk and industry risk.
There is another related problem; how should we calculate the P/E of the stock
market (i.e. the P/E ratio against which to compare the P/E ratio of a
particular stock)? There are a number of popular methods; one is to take the

SBC WARBURG

simple average of all the P/E ratios of listed companies; another is to aggregate
the market capitalisation and divide by the net profit (aggregate price divided
by aggregate earnings); yet another is to discard outliers and take the simple
average.
Market multiples are
ill-defined ...

Although this point is at first sight obscure, it is in fact of profound

importance. The following table shows how enormously market P/E ratios
differ according to the definition used.
Market PE ratios on various definitions
Dec 1995
Australia
Canada
France
Germany
Hong Kong
Holland
India
Italy
Japan
Korea
Singapore
Sweden
UK
USA

Median

Mode

Average

12.4
13.6
12.9
17.1
14.7
12.2
10.5
11.2
54.7
15.2
21.9
12.5
17.3
16.1

8.6
9.0
7.8
14.1
9.7
9.7
7.9
11.8
25.3
8.8
19.4
7.5
15.4
12.5

23.3
27.6
38.7
47.0
18.0
19.3
17.2
38.7
99.1
48.0
53.5
18.6
88.8
32.0

Truncated
15.7
17.4
16.0
22.5
14.7
14.2
12.6
14.1
68.6
20.4
27.4
15.4
22.1
20.6

"Market"
15.0
16.0
28.5
25.2
14.1
13.2
12.1
16.9
86.2
11.9
25.5
11.7
14.8
16.7

The median P/E ratio is the one which lies in the middle of the P/Es of all
listed companies; that is, there are as many companies with higher multiples
than the median as there are companies with lower multiples. The mode PE
is the one which occurs most often; the average is the simple average; the
truncated average excludes the top and bottom 5% of companies as
unnecessarily distorting. The market PE is the one used by our own
strategists. In the UK and US, actuarial bodies compute these figures; in Japan,
and most other major markets, by contrast, there is no official definition.
... because of statistical
issues ...

The statistical reason for this finding is that P/E ratios are not normally
distributed across markets, as the following charts for the US and Germany
show. (Appendix 1 shows similar charts for major markets worldwide). The
distribution is usually skewed to the right (there is a long tail of stocks with
high PE ratios by the standards of the market); sometimes, for example during
recessions, the stocks with huge P/E ratios are the countrys industrial or
financial giants which happen to be making temporary losses.

... which affect some

markets more than others.

Typically, the distribution spreads out during periods of weak earnings and
narrows when profits are high. This is hardly a new result the fact that
cyclical stocks have high P/E ratios at the depths of recessions and low
multiples when profits are high is well established yet the implications of
this tend to be overlooked when considering the question of market multiples
as a whole.

SBC WARBURG

USA PER surface

P/Es are tightly
distributed in the US ...

30%

25%

Proportion of sample

20%

15%
10%

5%

90

38
30
23
15

45

53

60

68

75

Year

83

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

0%

PER band

Market multiples - US
... so market multiples do
not vary much with
definition ...
25

20

15

10

0
1980 1981

"Market"
1982 1983

Median

SBC WARBURG

1984 1985

Truncated

Truncated
1986 1987
"Market"

1988 1989

1990 1991

Median
1992 1993
1994 1995

Germany PER surface

... but German multiples
are spread out ...

25%

20%

10%

Proportion of sample

15%

5%

0
15

23

30

38

45

53

60

8
75

83

90

Year

68

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

0%

PER band

Market multiples - Germany

... making the Market
multiple hard to
measure.
35

30

25

20

15

10

"Market"

1980 1981

1982 1983

Median

SBC WARBURG

1984 1985

Truncated

Truncated
1986 1987

"Market"

1988 1989

1990

Median
1991 1992
1993

1994 1995

10

So P/E relatives may work

in some markets
sometimes ...

It is not an accident that P/E relatives have acquired most favour in those stock
markets which offer a broad selection of companies in which to invest and
which represent a reasonable cross-section of the countrys overall economy
the UK and the US, in other words and where accounting earnings are
smoothed as a matter of policy. This is reinforced by the fact that these are
markets where management is under pressure to show a steady progression in
EPS even if some choose to deliver this by manipulation of the accounting
process.

... but generally do not.

Many of the worlds stock markets, however, offer much narrower ranges of
industries; and their composition very often bears only a passing resemblance
to that of the countrys real economy. Moreover, many stock markets include
disproportionately high weightings of particular sectors. Almost one-third of
Switzerlands market, for example, is accounted for by pharmaceutical
companies. So to talk of Nestles P/E relative is tantamount to talking about
the value of a food manufacturing company (and a multinational one at that)
relative to a collection of pharmaceutical companies. Does this make any
sense? An even more extreme example is the Dutch market, half of the
capitalisation of which is accounted for by Royal Dutch. So to value any
Dutch stock relative to the total Dutch market whether using P/Es or any
other benchmark is to value it relative to one of the worlds largest oil
companies. The extreme case at the moment is Japan, where financials make
up one third of the market, and are currently reporting huge losses; the
conventionally calculated market P/E is no guide to the multiple of the typical
stock.

What about P/CE multiples?

Cash earnings may sort
out depreciation policy
differences ...

If EPS is unreliable for cross-border comparisons because of vulnerability to

variations in accounting policy, then it might be thought that cash EPS would
avoid many of the pitfalls provided the investor is given enough
information to be able to spot which P&L items are cash and which are noncash. But cash EPS has three major faults as a measure of profitability.
First is the question of minority interests. Cash EPS is typically calculated by
adding post-minorities net profits to a depreciation figure which is decidedly
pre-minorities. In the case of a group where important subsidiaries are less
than 100%-owned, such a methodology can produce distorted results and
specifically will tend to lower a stocks P/CE multiple, making it appear
cheaper than it really is.

... but are no use across

sectors ...

SBC WARBURG

Second, cash EPS are blind to companies differing capital requirements. To

say that Usinor trades on a P/CE multiple of 2x and that Carrefour trades on
11x is to say nothing about the two companies relative value. Everyone
knows that steel-making is a much more capital-intensive business than food
retailing (viz., ratios of net capital employed to sales of 53% and 6%,
respectively).

11

Finally and most importantly P/CE multiples are even more sensitive to
capital structure effects than P/E ratios, as a few minutes with a calculator will
show.
... and are very distorted
by gearing differences.

SBC WARBURG

All in all, the best one can say about cash EPS is (i) that their calculation forces
investors to consider whether a company has deliberately raised or lowered its
net profits through the use of non-cash items and (ii) that they are possibly of
use when comparing two companies in similar industries with similar capital
structures (a rare enough occurrence in investment analysis).

12

Section 2: Enterprise value: the

price of a business
Definition of EV Miller & Modigliani the distinction between operating
and non-core assets the benefits to investors of thinking about a companys
business first and accounts second summary balance sheets minority
interests Tax economic double taxation evidence that it affects EV
multiples a simple formula to correct for this effect comparison of
theory with facts derivation of an improved EV definition

What do we mean by Enterprise Value?

EV =

Enterprise value is the

price the markets are
putting on a business ...

market capitalisation
plus

market value of net debt (average for the year)

minus

estimated value of non-core assets

It is almost 40 years since Miller and Modigliani first put forward the
proposition (known as their Proposition 1) that the market value of a
companys assets is independent of the mix of liabilities used to finance those
assets. In other words, a company can neither increase nor decrease the value
of its assets by altering the proportions of debt and equity in its balance sheet.
This was later modified by the authors to allow for the influence of tax a
subject dealt with in detail later on. There are still aspects of this proposition,
aside from tax, which remain in dispute; certainly, the imperfections of capital
markets mean that companies are not always able to restructure their balance
sheet at will or at negligible cost. But the basic insight has proved remarkably
resilient, particularly in more efficient markets.

... which is deemed not to

depend on a companys
financial structure ...

SBC WARBURG

Whatever the state of academic debate on the subject, it is clear that

Proposition 1 can only hold true if investors are unwilling to pay a premium
for shares in a company with gearing in its balance sheet (as opposed to a
debt-free company). If the investor himself can borrow in order to buy shares
in a debt-free company, he is in effect able to duplicate the economics of
buying shares in the geared (but otherwise identical) company. Put another
way, if a geared company decides to issue fresh equity to pay off its debt (i.e.
change its capital structure), its shareholders can undo the effects of this
change by borrowing money themselves. This point can be better illustrated
via an example.

13

Let us return to the idea of two companies which are identical except for their
capital structure:
... which an example
shows is a sensible
theory ...

In which company would an investor prefer to invest? A risk-averse investor

would buy shares in Company U (1%, say), which would entitle the investor
to 1% of the gross profits. Such an investment can be expressed thus:
Company U

Investment

Return

Equity

0.01Vu

0.01 profits

An alternative strategy would be to buy the same proportion (1%) of both the
debt and equity of Company G, in which case the investment and the returns
assuming that debt and equity holders are taxed equally would be as
follows:
Company G

Investment

Return

Debt

0.01Dg

0.01 interest

Equity

0.01Eg

0.01 (profits - interest)

Total

0.01(Dg + Eg) = 0.01Vg

0.01 profits

Both strategies offer the same returns, i.e. 1% of the companys profits. In an
efficient market, two investments which offer the same returns must be worth
the same. In other words, 0.01Vu must equal 0.01Vg; and the value of
ungeared Company U must equal that of geared Company G.
Now let us suppose that a more risk-tolerant investor buys 1% of the equity
of geared Company G. His investment and returns would look like this:
Company G

Investment

Return

Equity

0.01Eg

0.01 (profits - interest)

= 0.01(Vg - Dg)

But there is an alternative for the investor: to borrow the equivalent of 0.01Dg
on his own account and then invest the money in 1% of the equity of ungeared
Company U. Borrowing the money generates an immediate cash inflow of
0.01Dg; but the investor must pay interest on the borrowing (equivalent to
1% of the interest paid by geared Company G). The investment and returns
from this strategy would look like this:
Company U and borrowing

Investment

Returns

Borrowing

-0.01Dg

-0.01 interest

Equity

0.01Vu

0.01 profits

Total

0.01(Vu - Dg)

0.01 (profits - interest)

Once again, both strategies offer the same return, i.e. 1% of profits after
interest; and thus both investments must have the same cost. Consequently,
0.01(Vu - Dg) must equal 0.01(Vg - Dg); and so Vu must be equal to Vg.

SBC WARBURG

14

From this we can conclude that an investors appetite for risk has no effect on
the value of either company: both risk-averse and risk-tolerant investors
would agree that Company U and Company G have the same value. As long
as they can borrow on the same terms as the companies (sometimes a big if
in the real world), investors can undo the effects of any change in a companys
capital structure. On this basis, Proposition 1 makes perfect sense: a
companys market value is independent of changes in its capital structure.
... although there is an
untidy tax issue.

Let us accept for now that Proposition 1 does hold true. If the value of a
company is indeed determined by the value of its assets, not its liabilities, then
it follows that investors who we might define as people who want to know
how much corporate assets are worth are missing an important part of the
picture if they restrict their assessment of value to just the equity component
of a companys liabilities.
Further evidence to support the argument that shareholders (i.e. the owners of
companies) ought to be interested in the total capital of companies, not just their
equity, comes from the behaviour of those who buy and sell whole companies at
a time. That is because corporate managements are - or should be - interested first
and foremost in the cash flow which an asset can generate. That cash flow can
then be used to service either debt (through interest payments) or equity (through
dividends). Leaving tax considerations aside for the moment, it is irrelevant
from this cash flow perspective what the mix of debt and equity is.

But the advantages of

thinking about business
values, not eps, are huge.

Another advantage for the investor which comes from looking at the market
price of the whole enterprise (hence the term enterprise value), rather than
just the value of its equity, is that it encourages a distinction between a
companys core and non-core assets. Over the years corporate raiders
and management buyout teams have made handsome returns by identifying
companies which owned assets whose value could easily be realised (through
disposal) without damaging the core business; conversely, failure to dispose of
low yielding assets has been a major drag on the performance of numerous
Japanese shares.

How do we define non-core assets?

The identification of Noncore assets ...

Before including non-core assets as part of a companys EV, an analyst should

answer three questions:

1.

Can I identify assets which are peripheral to the conduct of the groups
business?

2.

If I can identify non-core assets, can I value them with any degree of certainty?

3.

If I can identify non-core assets and value them, is there a reasonable chance
that their value will be realised?

SBC WARBURG

15

It is obvious that only a small proportion of companies will meet all three
criteria. Typically in Continental Europe there are many companies (perhaps
even the majority) with assets that could easily be construed as non-core, but
whose managements have little or no intention of realising their value.
Nevertheless, it is seldom a waste of an analysts time to consider these
questions. At the very least, a discussion with the companys management is
worthwhile: even if the questions are met with blank stares, does that not tell
the analyst something about the managements attitude towards its
shareholders?
... may represent potential
value ...

Classic examples of non-core assets are minority stakes in other companies,

quoted or unquoted (e.g. Saint Louis stake in Danone); property holdings
(e.g. Generale des Eauxs property assets); and tax-loss carryforwards. But
occasionally it may make sense to treat quite large parts of a business empire
as non-core if their earnings profile and capital structure are radically different
from those of the core business. For example, it is helpful to consider the
mobile telephony business of Mannesmann as a non-core asset not because
Mannesmanns management intends to dispose of the operation (far from it),
but because it has little in common with the groups mainstream engineering
businesses from either a P&L or balance sheet point of view. Besides, we can
put reasonable estimate of market value on Mannesmanns mobile telephony
business by comparing it with other quoted cellular operations.

... or associated
companies, often
overlooked by
analysts ...

The identification of non-core assets also provides a way of capturing the

value of associates. Because income from associates is typically taken below
EBIT in the P&L, valuation based on operating earnings risks missing the
(sometimes substantial) value represented by less-than-50% stakes. Although
financial information on an associate tends to be sketchy, there is usually
enough to enable the analyst to value the company sensibly (e.g. as a multiple
of sales and/or earnings). This is certainly much simpler that the other option
of re-consolidating an associate company line-by-line.

... and allow an idealised

balance sheet to be
constructed.

Once we have made the distinction between core and non-core assets, we
are in a position to summarise a companys balance sheet thus:
Assets

Liabilities

Core assets
fixed assets
non-fixed assets (i.e. net working capital)
Non-core assets (including associates)

Net debt

Equity capital

On the asset side of the balance sheet, we have the core assets, i.e. those
assets employed in the companys core business, and non-core assets
(including the value of associates where appropriate). Core assets can be
split in turn into fixed assets (plant & machinery and some types of intangible
assets) and non-fixed assets (i.e. net working capital).

SBC WARBURG

16

On the liabilities side, we have net debt and equity.

Pensions need to be
considered carefully ...

What of balance sheet items such as investments (on the asset side) or
pensions and provisions (on the liabilities side)? By classifying a minority
shareholding as an investment, a company is usually admitting that this is
a non-core asset. Pensions we prefer to treat as debt (being careful to strip out
the net provisions if these are taken above the EBIT line). As for provisions,
these tend to be either of an economic nature (i.e. they refer to future liabilities
such as pensions), in which case they can be treated as debt or as a non-core
liability (e.g. a provision against legal claims), or else they tend to be taxdriven (see below).

... as do intangible
assets ...

There are, of course, grey areas of which the treatment of intangible assets
is perhaps the most obvious. Many companies build up intangible asset value
by spending money on R&D or marketing. From an accounting point of view,
it makes just as much sense to capitalise and amortise such spending as it does
to expense the outlays in the year in which they are incurred (as most
companies do). Nevertheless, because our approach to valuation is based on
cash flow (rather than earnings), the absence of intangible assets from our
definition of fixed capital ought not to affect our valuation conclusions. On
the other hand, it can make cross-sector comparisons of P/BV and RoCE
thoroughly misleading: the apparently huge returns on capital generated by
most European pharmaceutical companies would be materially lower if their
R&D spending were capitalised and amortised over, say, five or ten years.
What about minority interests ?

... and minorities.

This discussion has so far excluded subsidiary companies which are less than
100% - but more than 50% - owned. Group accounts include all the debt and
assets of these companies, and show a balance sheet item for minority interests
as a liability of group shareholders. Unfortunately, the relationship between the
book and economic value of minority interests is tenuous, even by the standards
of historic cost accounting generally. This leaves the analyst with two main
options.
Option 1 is to exclude outside shareholders interests in subsidiary companies
assets and liabilities from each part of the accounts in our formulation, from
core assets and net debt. In practice, though, this can become very difficult
because of shortage of information.
Option 2 which has proved more popular with analysts is to treat the
subsidiaries as permanent members of the group, and include the hypothetical
cost of buying in the minority interest in the EV calculation. This figure can then
be compared with group sales, profits, cash flows, or production capacity,
without having to make further adjustments to the denominator of any EV ratios.

SBC WARBURG

17

In practice, Option 2 is much the easier, although it means that the analyst has
to make an explicit estimate of the value of subsidiaries, and therefore of the
value of outside investors holdings.
The estimate has to be arrived at in precisely the same way as a valuation of
any unquoted company - which is to chose from a portfolio of approaches.
The simplest calculation is to chose an appropriate capitalisation rate of
disclosed profits; but as the entire thrust of this document shows, the precise
rates depend on the capital structure and profits outlook of the subsidiary. We
therefore recommend an EV calculation for the subsidiary if enough
information is available; otherwise, the analyst is forced to rely on a
judgement of an appropriate capitalisation rate. In the case of many groups in
the Far East, the subsidiaries are themselves listed, which gives a precise
market value of the value of outsiders interests in the group as a whole.

What about tax?

If a tax system is not
neutral between debt and
equity ...

There are strong assumptions built into EV methodology, including that tax
systems are neutral between debt and equity. In this section we show how this
compares with reality, and how EV ratios should, in principle, be defined if the
assumption of tax neutrality is relaxed.
In practice, tax-adjusted EV ratios are hard to calculate because they require a
good estimate of the tax wedge - the extra amount of tax levied on returns to
shareholders compared with the tax rate on cash flow distributed to lenders as
interest. The size of the tax wedge depends on the interaction between the
corporation tax code and the income tax code, leading the analysis into the mire
of imputation systems, international tax treaties, and the treatment of pension
funds. We conclude that we cannot estimate the size of the tax wedge with
confidence using tax code data.

... EV methodology would

have to be modified ...

However, if there is a tax effect, it should be visible in the way shares are priced
by stock markets. An empirical study shows that the tax wedge is pronounced in
Japan, but probably small elsewhere in the G7. Unfortunately, we are not
completely confident about the data used. We conclude that while tax effects exist
in some countries, we cannot estimate their size accurately.

... but the size of the tax

wedge is hard to measure.

We therefore intend to continue to produce EV ratios assuming tax neutrality; as

data accumulates, we will home in on better estimates of the size of tax wedges.
At some point we may then modify our EV methodology to include tax effects.

SBC WARBURG

18

The assumptions in EV methodology

The enterprise value of a company is conventionally defined as its equity
market capitalisation, plus its debt, less the value of any peripheral assets. The
total gives the total value - or price - that capital markets are putting on a
particular core business. The concept is that this price can be compared with
various measures of the size of the business - sales, capacity, operating free
cash flow, for example - to determine what expectations are currently built
into the share price.
The arbitrage proof of
proposition one ...

The enterprise value of a company is assumed to depend only on its assets, and
not at all on its liability structure. This assumption is generally justified, as on
page 14, by pointing out that investors, as well as companies, can borrow. If
it were possible for a company to change its enterprise value by altering the
mix of debt and equity, then investors would be able to borrow money to buy
shares in an undergeared company, selling shares in an optimally geared
analogue, and pocket a risk-free profit. Since this violates the precept of
capital market efficiency, it seems a good assumption that EV does not depend
on liability structure.

... assumes tax

neutrality ...

A neat argument, perhaps, but is it right in practice? If EV does not vary with
capital structure, there should be no variation of average EV multiples with
balance sheet gearing. That is, the average EV/operating free cash flow ratio
for a sample of companies with 50% debt and 50% equity should be the same
as the average EV/OpFCF ratio for companies with no debt and 100% equity.
However, in at least one market EV multiples do indeed vary systematically
with balance sheet structure. In the chart below, there is an upward slope in
the trend line; on average, companies with a high proportion of debt in their
capital structure have higher EV multiples.
EV multiples vs capital structure: Japan

... a bad assumption in

Japan.

EV/EBIT
70

60

50

40

30

20

10

0
-120%

SBC WARBURG

-100%

-80%

-60%

-40%

-20%
0%
Net debt / EV

20%

40%

60%

80%

100%

19

So proposition 1 does not always hold. Why?

The arbitrage justification for proposition 1 assumes that investors can
borrow (or lend) at the same cost as companies. This may be a reasonable
assumption as far as interest rates are concerned, but if companies are taxed
differently from individual investors, it breaks down.
The tax wedge lies in the
overlap between
corporation tax and
income tax ...

When a company distributes its operating cash flow in the form of interest
payments to the debt provider, cash flow is only taxed once - as income tax
levied on the lender. However, returns to equity holders are taxed twice; once
in corporation tax, and then either as income tax levied on the dividend
payment to investor, or as capital gains tax on a rise in the share price - which
is related to profits retained in the business. If the rate of tax levied on returns
to lenders is lower than the total tax rate on returns to shareholders,
companies with more debt in their capital structure will have higher enterprise
values, because the total net cash flow to providers of capital will be higher1.

... a fearsomely
complicated zone.

The practical problem in attempting to correct for this is calculating the size
of the tax differential. Our own treatment of the same problem in EV
calculations starts with an abstract discussion of the relationship between
corporation tax and income tax, reviews the tax codes in OECD countries,
and shows that the whole issue is so horribly complicated that attempts to
estimate the size of the tax wedge starting from the tax code can only come up
with a very approximate estimate.

1 In fact, Miller & Modigliani realised this after publication of proposition 1, and in 1963 produced a paper
(Corporate Income Taxes and the Cost of Capital: a Correction) which explains how to allow for taxes in cost of
capital calculations.

SBC WARBURG

20

No-mans land: between tax codes

The diagram shows, in schematic form, how operating profits are taxed. When
operating profits flow to providers of debt capital in the form of interest
payments, a portion is levied by the government as tax. In this diagram, we
assume, for simplicity, that lenders are individuals, and tax is levied at the
marginal income tax rate. More often, money is lent through some sort of
intermediate body, such as a bank or an insurance company, in which case the
effective tax rate on interest payments may be different from the income tax rate.
Schematic of tax
structures ...

Profits which do not flow to lenders are called pre-tax profit. A proportion of
this is extracted by the government in the form of corporation tax; the balance
is either distributed to shareholders as dividends or is retained in the business,
contributing to a capital gain in the shares. Shareholders pay income tax on
the dividends received from the company; they may also pay capital gains tax.
... shows the double tax
effect ...

If the income tax rate on interest is the same as the capital gains tax rate and
income tax rate on dividends, this means that returns to lenders are taxed
once, and returns to shareholders are taxed twice - once at the corporation tax
level and then again as income tax. There could be major differences between
the two rates. Suppose, for example, that the tax rate at each level is 50%;
then the tax rate on returns to lenders would be 50% but the tax rate on
returns to equity would be 1 - (50%*(1-50%)) = 75%

... which may be corrected

by an imputation
system ...

There are a variety of arrangements in place in different countries designed to

correct this effect. In some jurisdictions, dividends are taxed at a
concessionary rate, or not taxed at all; in others, notably the UK, dividends
carry a tax credit; in others, the corporation tax charge depends on a
companys dividend policy.

SBC WARBURG

21

... and the structure of

capital gains tax.

Meanwhile, many countries do not have a capital gains tax; this means that
money retained in the business is taxed only once. Even in countries where
there is capital gains tax, it is usually only payable when the gain is realised,
and is often indexed to allow for inflation. If the corporation tax rate is lower
than marginal income tax rate, such retained profits then continuously
produce a stream of lower tax rate returns. To make the issue more
complicated still, shares are often owned through mutual bodies, such as
pension funds, which may have concessionary tax treatment for both
dividends and for capital gains - and which can distribute tax free to investors.
Finally, if capital providers are located in a different country from the
company they fund, the exact rate of tax on interest, dividends and capital
gains depends on the nature of the tax codes of each country and on a network
of bilateral tax treaties.

An OECD study in 1991

demonstrated the
complexity of the issue.

SBC WARBURG

The following table, taken from an OECD publication, shows how various
countries address the issue of reducing economic double taxation. The fact
that it needs 10 footnotes to explain its terms of reference - even before
turning to the size of the various tax rates - should be enough to convince the
reader that it is very hard to estimate the difference between the tax rates on
returns to equity and returns to debt.

22

Degree of Reduction of Economic Double Taxation (Central Government)

None or very
little reduction
1
Classical system
(1)

Belgium (2)
Luxembourg
Netherlands
Switzerland
United States

Reduction of economic double taxation

Elimination of economic double taxation

Corporate level
2
3
Split rate system
Partial dividend
deduction system

Shareholder level
4
5
Partial imputation
Partial shareholder
system
relief schemes

Corporate level
6
Zero rate system

Lower tax rate

on distributed
income

Partial deduction
of dividends paid

Partial credit for

corporate tax paid

Partial credit for

domestic
shareholders

Zero tax rate on

distributed income

Germany (3)

Iceland (4)
Spain (5)
Sweden (6)

France (7)
Ireland
United Kingdom

Austria (8)
Canada
Denmark
Iceland (4)
Japan
Portugal
Turkey (10)

Greece
Norway

Shareholder level
7
Full imputation
system or full
shareholder relief
system
Full credit for
corporate tax paid
(imputation system)
Australia
Finland
Germany (3) (9)
Italy
New Zealand
Turkey (10)

* Source: OECD
1.

In most of these countries (and in those with reduction at the corporate level) some small degree of reduction is
given to shareholders in the form of a relatively low exemption for dividends received.

2.

Belgium has moved from a shareholder relief system to a classical system but continues to provide relief to
shareholders who invest their dividends in their own professional activity (the use of a so-called mitigation
technique to encourage retentions rather than distribution).

3.

Systems in column nos 2 and 7 are both operative in Germany.

4.

The deductions for dividends paid may in some cases fully offset the corporate income tax and also the personal
income tax, especially for dividends up to 15 per cent of capital value. Dividends exceeding this limit are fully taxed
at both levels. Hence, Iceland is classified both under column 3 and 5.

5.

Spain should strictly speaking be shown under column 5 as well as column 3 but as the credit to the shareholder is
only 10 per cent (and much lower than other countries in column 5), it has been disregarded on de minimus
grounds.

6.

The deduction for dividends paid may in some cases result in elimination of the corporate tax (for dividends on
newly issued shares, maximum 10% per year of the value of the issue with an overall maximum equal to the total
value of the issue).

7.

France is sometimes described as having approached elimination rather than mitigation of the economic double
taxation, as shown in the table, because whilst the rate of corporation tax has been substantially reduced the
amount of credit has not changed, but on the other hand since 1989 retained profits have become subject to a lower
rate than distributed profits (currently 34 and 42 per cent respectively).

8.

As from 1986 dividends paid to residents are taxed at half the normal rate in the hands of the shareholder. The split
rate system (column 2) was abolished as from 1989.

9.

Germany belongs to the extreme right of the table in over-compensating for the economic double taxation of
dividends by giving both full imputation to the shareholder and subjecting distributed profits to a lower corporate
rate than retained profits. On the other hand no credit to the shareholder is given for the payment of the local
business tax.

10 No personal tax is charged on dividends distributed out of profits which have borne corporation tax which means
in practice that the relief is sometimes partial, sometimes complete. This is why Turkey is shown in both columns 5
and 7.

SBC WARBURG

23

How are shares actually priced?

An empirical study of
gearing ratios and EV
multiples ...

Even if it is unfruitful to tackle the tax code from the first principles, there is
an alternative. As the scatter diagram showing EV ratios (p19) in Japan
shows, there may be a systematic variation of average multiple with capital
structure. Assuming that this is to do with the structure of the tax system, this
allows an estimate of the size of the tax wedge, as follows.
Suppose that returns to shareholders are always taxed more highly than
returns to lenders by a proportion T. Each year, the before-income-tax return
to shareholders amounts to (l-T) times pre-tax profits; the extra amount
leaking to the government is T times pre-tax profits.
The present value of all the excess tax payments to government (the double
taxation payments) will therefore be in proportion to the present value of the
returns to shareholders. In algebra
Value of tax wedge
____________________ =
Market value of equity

T
____
(l-T)

The standard enterprise value calculation defines enterprise value as the total
of market capitalisation plus net debt less peripheral assets, the idea being that
this gives the total price being put on the core operating business by the capital
markets. This is customarily compared with the operating free cash flow - or
the current cost profit - before interest payments, returns to shareholders, or
tax.
The double taxation effect means that the company has a peripheral liability
- to pay extra tax - that depends on its capital structure. The operating free
cash flow before tax is split between debt providers, equity holders, and the
extra proportion that flows to the government.
... using simple algebra ...

Therefore, define a taxed enterprise value, assuming no peripheral assets, as

follows;
EV* = Market value of equity + net debt + value of tax wedge
or
Market value of equity
EV* = _________________________ + net debt
(l-T)
If this properly allows for the different tax treatment of return to debt and
return to equity, EV* ratios should not vary with capital structure, whereas
the conventional EV ratio should. Since
EV = Market value of equity + net debt

SBC WARBURG

24

It follows that:
(l-T)
EV = EV* x _________
(l-XT)
Where X is net debt/EV. Therefore the conventional ratio will differ from the
tax adjusted EV ratio by a factor (1-T)/(1-XT). Where there is no excess tax,
the figures will be the same. Where T is a positive number, the higher the debt
in EV, the higher the EV ratio.
So much for the theory; how does it match the data?
The same equation inverted, predicts:
1
1
XT
____ = _________ - _________
EV
EV*(1-T)
EV*(1-T)
The following chart plots EBIT/EV for the Japanese data against X (net
debt/EV). Comparing the slope with the intercept gives a figure for T, the tax
wedge, of 64% (plus or minus 15%).
Tax rate estimate
Shows a significant double
tax effect in Japan ...

EBIT/EV
30%

25%

20%
y = -0.0501x + 0.0789
2
R = 0.2336

15%

10%

5%

0%
-120%

-100%

-80%

-60%

-40%

-20%

0%

20%

40%

60%

80%

100%

Net debt/EV

This is a rather higher rate than would be expected from the tax code. The OECD
reckons that in Japan, on average, interest is taxed at 20% and dividends at 35%.
The average corporation tax rate is 50%. This gives total tax on returns to
shareholders, assuming full distribution, of

plus
equals

SBC WARBURG

Corporation tax
Income tax
Total

50%
50%*35%
67%

25

Compared with an average tax on interest of 20%, theoretically the tax wedge
should be 47%. However, these calculations exclude the effect of tax-exempt
institutions (such as pension funds).
The following pairs of charts show how EV multiples vary with balance sheet
gearing (left hand chart) and the regression which allows a calculation of the
implied tax wedge (right hand chart), for the US and Europe.
US EV multiples vs capital structure

US tax rate estimate

EV / EBIT
30

EBIT / EV
30%

25

25%

20

20%

15

-50%

-40%

-30%

-20%

y = 0.023x + 0.0796
2
R = 0.0117

15%

10

10%

5%

0
-10%
0%
Net debt / EV

10%

20%

30%

40%

50%

-50%

-40%

-30%

-20%

-10%

0%
0%

10%

20%

30%

40%

50%

Net debt / EV

Europe EV multiples vs capital structure

Europe tax rate estimate

EV / EBIT
60

EBIT / EV
40%

35%
50
30%
40
25%

30

y = -0.0124x + 0.0954
2
R = 0.0033

20%

15%
20
10%
10
5%

-100%

-80%

-60%

-40%

0
-20%
0%
Net debt / EV

... but no effect in the US

or Europe.

20%

40%

60%

80%

100%

-100%

-80%

-60%

-40%

0%
-20%
0%
Net debt / EV

20%

40%

60%

80%

100%

In most cases, the data is so scattered that it is hard to decide what the
effective tax wedges are (this shows up in the low r-squared figures on the best
fit lines on the right). For what it is worth, the data implies that the tax wedge
in the US is -29%, and that it is 13% in Europe. However, these estimates are
highly unreliable. A better conclusion is that there is no measurable tax effect.
There are two possible reasons why this should be. The first is that the tax
systems are indeed now largely neutral between debt and equity. This is
regarded by many governments as a desirable feature of the tax system; most
OECD governments - with the notable exception of Japan - have been moving
in the direction of tax neutrality for some time. In addition, cross-border
flows, both via capital markets and via direct investment, will tend to migrate

SBC WARBURG

26

to a more benevolent regime. Note that this is a function of the liberalisation

of cross-border capital flows; where there are legal or de facto controls on
international equity or direct investment, tax systems can remain biased with
less effect on investment patterns.
The other possibility, unfortunately, is that our data may not be accurate
enough. Even though SBC Warburg has been calculating EV ratios for some
time, we cannot be certain that all the components of the calculations are
absolutely standardised.
Therefore, we intend to proceed as follows;
1. To continue to calculate the EVs in the traditional manner - without
attempting to adjust for tax
2. To regularly audit the figures to see if EV ratios vary with balance sheet
structure. If we can demonstrate, to our own satisfaction, that there is a tax
effect and we can estimate it accurately, then we will modify EV
calculations accordingly.

SBC WARBURG

27

Section 3: The real size of a business

Measuring profits how HCA depreciation pollutes accounts why this
matters very much across borders definition of Operating Free Cash Flow
how to measure maintenance capex the importance of competitor
analysis working capital. Valuation ratios - EV/sales, EV/OpFCF,
EV/capacity.

How can we best measure profits?

Many earnings problems
appear after EBIT ...

As we have already discussed, stated earnings are prey to a host of influences

which can either inflate or deflate the published number. A logical response to
this is to move further up the P&L in search of a purer measure of profits.
Now, if we are considering a companys total capitalisation (not just the equity
component), then it requires us to look at profits before interest, tax and
minorities (i.e. EBIT). This ought to be helpful: after all, EBIT is not affected
by differences in capital structure and tax rates.

... but there are still big

accounting problems:

But what exactly does EBIT mean? We might define it like this: EBIT is the
profit generated from a companys assets after making provision for the
reinvestment needed to maintain the value of those assets but before
distributing the money between the lenders, taxman and shareholders. The
difficulty lies in the fact that the reinvestment requirement HCA
depreciation is calculated by rules of thumb which respond slowly, if at all,
to changing circumstances. By definition, then, the EBIT number published by
companies in their (almost exclusively historic cost) accounts is polluted by
the depreciation policies which a company happens to have followed and by
the level of inflation which has prevailed in the countries where the assets are
located.

... inflation ...

When comparing companies in the same country, with uniform depreciation

rules and inflation histories, this problem is of secondary importance. In crossborder research, different accounting and inflation environments catapult the
problem to the first division.
If a company operates in a high inflation environment, it will increase its sales
each year by increasing product prices, but depreciation and some other
costs will only increase later on.
This point is of enormous importance, especially in capital intensive
industries. The following chart shows the stated (HCA) operating profit
margins of automotive assemblers in selected countries. Companies in higher
inflation countries tend to report higher returns on sales than companies in
low inflation countries.

SBC WARBURG

28

Inflation vs margins
... (higher inflation means
higher margins) ...
7.0%
Fiat
BMW

Average EBIT margin, 1982 - 1995

Ford

Honda

6.0%

Peugeot

GM
Chrysler

Toyota
5.0%

4.0%

Volvo

Renault
Nissan

3.0%
Porsche
2.0%

1.0%

0.0%
0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

8.0%

Average CPI inflation, 1982 - 1995

One answer to this problem is simply to move even further up the P&L and
concentrate on EBDIT. This certainly side-steps the issue of whether a
companys depreciation charge is high or low compared with its industry peer
group; but it prevents comparison of the valuations of companies with
different levels of capital-intensity (greater capital-intensity = proportionally
bigger depreciation charge and vice-versa).
... and depreciation ...

... which we replace with

maintenance capex ...

More satisfactory is to try to assess how much a company needs to reinvest in

its core business at current prices in order to maintain both
productive capacity and (where it is an issue) competitive position. In other
words, how much should the company be charging against profits for the use
of its fixed assets on a replacement cost basis? We refer to this number as
maintenance capex; but it is synonymous with normalised, or industry
standard, depreciation. It is only after charging this amount that we are left
with the sustainable operating profit or what can be removed by lenders
and shareholders at years end after enough has been reinvested in the asset
base to allow the company to operate safely for another year. We refer to this
profit as operating free cash flow (or OpFCF).
OpFCF =

EBITDA (i.e. operating profit before depreciation & amortisation)

minus maintenance capex
minus maintenance working capital spending

There are several advantages of basing valuation on operating free cash flow:
OpFCF has the obvious merits of being a real number (in the sense that
it is money that management can reinvest or allocate to owners of the
business rather than an accounting figure). Especially with more stable
companies, quite small variations in free cash flow generation can have a
huge influence on equity valuation.

SBC WARBURG

29

As a universal measure of profit, it is a guide to how efficiently companies are

really converting sales into returns to shareholders.
Calculation of maintenance capex forces an assessment of the real economic
value of a companys assets leading to better return on asset ratios than are
available with conventional combinations of uncorrected profit and balance
sheet totals.
... generating a figure for
free cash flow.

The analysis of free cash flow should give us a reasonable guide to companies
future investment spending and thus to the timing of profit cycles,
especially within more cyclical industries. Companies tend to invest according
to their free cash flow (i.e. internally generated funds): falling free cash flow
is generally followed by falling capex and vice-versa. Thus, if we observe that
free cash flow generation (relative to sales and/or assets) is relatively modest,
it is likely that capex is being squeezed and thus the foundations for a
recovery in profits are being laid. Conversely, relatively high levels of free cash
flow suggest that capex is rising (or will rise), in due course undermining
profitability.
So how can we estimate maintenance capex?

By definition, this is an
analysts estimate ...

Since companies are not obliged to disclose the amount of investment required
to maintain their asset base, any assessment of maintenance capex is bound to
be subjective. But that is no different from most of the processes of equity
investment. And surely an analysts knowledge of a company is missing a vital
component if it does not include an estimate of how much that company must
invest each year to prevent the deterioration of its asset base and competitive
position.
In practice, how can an analyst assess a companys maintenance capex
burden?

... but this is an aid, not a

hindrance, to thought ...

It is logical that the quest for a companys maintenance capex burden should
start with the company itself. Companies in countries with some history of
current cost accounting (e.g. the UK and The Netherlands) are more likely to
have clear ideas on the subject. But in general our experience suggests that
companies are rather more willing to discuss this issue than many others.
Indeed, if inquiries about maintenance capex are met with blank stares from
management, that response in itself should sound an alarm (much as in the
cases of managements who admit to no distinction between core and noncore assets).

... especially in
commoditised industries.

In some industries, maintenance capex can be estimated with some degree of

confidence by referring to the replacement cost of capacity. An extreme example
is the airline industry: the free availability of information on aircraft values (new
and second-hand) allows us to put fairly accurate replacement values on any
airlines fleet. Likewise, the replacement cost of capacity in industries such as
car-making and steel manufacturing is reasonably easily established.

SBC WARBURG

30

It helps distinguish
between profits generated
from lucky asset
purchases ...

In others, however, the figure may bear little relationship to the eventual cost
of replacing fixed assets. The most extreme example is oil production, where
the cost of replacing oil reserves might average $2.50 across the industry, but
where it may vary between a few cents (if a company strikes lucky) and far
more (if it doesnt). But this fact can be turned to the investors advantage; by
calculating profits defined as operating free cash flow using an industry
standard cost of replacing capacity, it is easier to see to what extent reported
profits are flattered by previous lucky strikes or hampered by previous highcost developments; and the rating the capital markets are applying to the
companys OpFCF can be interpreted in terms of the amount of faith or
scepticism currently attributed to the exploration programme.

... and those generated

from careful control of
operating costs.

More generally, an estimate of maintenance capex must be put in the context

of both actual capex and HCA depreciation. Actual capex (over several years),
relative to sales or fixed assets, gives us a guide to how much a company has
thought it necessary to invest in its assets though this must be offset by an
assessment of the proportion of that capex which was earmarked for
expansion rather than maintenance. HCA depreciation in most cases provides
a floor below which it would be hard to justify a figure for maintenance
capex (unless a company is blatantly over- or under-depreciating its assets
relative to their economic lives). In other words, maintenance capex will
typically be higher than stated depreciation; and the extent to which it is
higher will depend (in part) on historic inflation rates (the higher the inflation,
the bigger the discrepancy between HCA depreciation and maintenance
capex).

Step changes in
technology ...

Perhaps the most tricky problem in this area arises from step changes in
technology. If a competitor invents a new production process which
dramatically lowers cash operating costs, a company is obliged to decide
whether to invest in the same technology or face steady loss of competitive
position. Even if it judges that the return on that investment will be inadequate,
once a rivals capacity is in existence it will tend to drive industry prices down
to new, lower, levels relative to input costs; so a decision not to invest is a
decision to begin to leave the industry. Perhaps the most topical example of this
is in memory chip manufacturing, where first the Japanese, and then the
Korean, makers have invested enormous amounts of money in more efficient
production lines eventually destroying their own, and competitors, profits.
Yet the industry is oligopolising; when this process is complete, the remaining
participants may be able to earn monopoly profits. Shares in these companies
move well in advance of published profits trends, reacting to changes in the
competitive environment; operating free cash flow, on our definition, suffers
immediately the competitive environment deteriorates.

... can raise an entire

industrys equipment
costs ...

But a company may elect to give up the race in a capital-intensive industry of

this sort, and run its existing plants for cash. Why, then, should an external
analyst produce a far lower profits figure (stemming from a far higher industry
standard depreciation charge) than the official accounts ?

SBC WARBURG

31

... never obvious from

HCA accounts.

The answer lies, as ever, in the interpretation. If a company is failing to invest in

new technology, its official profits HCA basis will usually look fine for a
while; but when industry prices are forced down, it will typically face extinction
(like much of the British car industry) unless it has been lucky enough to retreat
into a niche (like Rolls-Royce, the car company). The trend in operating free
cash flow will give an early warning of this danger.

Similarly, working capital

inflation absorbs cash ...

Finally, a quick word about maintenance working capital. Just as a companys

fixed assets need to be maintained if the company is to operate successfully for
another year, so too must the non-fixed assets (i.e. its working capital) be
maintained. In essence, maintenance working capital spending can be defined as
the inflation adjustment of the average working capital position during the year.
Working capital is already a current cost figure (unlike fixed asset values). So,
leaving aside managements success or failure in reducing working capital
relative to sales, the required investment in working capital over 12 months is
equal to the annual inflation rate ideally in the products concerned, but more
often in the overall rate of producer prices. In many cases, this investment is
quite small relative to sales: if working capital is equal to 20% of sales and
inflation is running at 3%, then the investment will be just 0.6% of sales. In low
margin businesses, though, this adjustment can be a key insight into true
profitability.

... a vital point in high

inflation economies.

What valuation ratios can we construct?

Valuation ratios are a
measure of market
expectations ...

The EV figure is a measure of the total market value of the liabilities of a

companys core business, and therefore of the worth of its core operating
assets as measured by the capital markets. But capital markets base their
estimate of the value of a company on a fluctuating set of assumptions and
preoccupations; in a sense, the analysts job is to uncover these assumptions
and compare them with the evidence.
An important part of this process is to compare the capital markets estimate
of the worth of a company with its true size. Comparison of these ratios
between companies will give an insight into why the share price stands where
it does, and therefore how it might change in the future.

... EV/OpFCF for profits

growth ...

The obvious ratio to construct is EV/OpFCF which is analogous to a P/E

ratio, but has the advantages that accounting and inflation differences across
borders have been corrected, at least as far as possible, and that the ratio is
not confused by variations in capital structure. Companies whose operating
profits are expected to grow faster, or be less volatile, should have higher
ratios; companies with lower growth rates, or whose operations are more
risky, should have lower multiples.
Within a given industry, the EV/EBITDA multiple is analogous to P/CE;
although the size of the ratio should depend on the degree of capital intensity
of the industry in question, it should be similar all other things being equal
for companies involved in the same activity.

SBC WARBURG

32

... EV/Sales for sales

growth ...

Often, profits are expected to be very volatile typically, at the peaks and
troughs of general economic cycles. Construction of EV/sales ratios gives a
sense of how much the markets are paying for a unit of market share in a given
industry, and allows some commentary on the relative price being paid within
an industry in the absence of confident profits estimates.

... EV/size for new or

deeply cyclical
businesses ...

Similarly, EV per whatever-unit-of-size-is-most-relevant can also be helpful;

one example at the moment is cellular telephony, where companies are valued
on the basis of the scope of their licences (in the absence of any profits, and,
in some cases, sales).

... but must be supplied

with their built-in
assumptions in mind.

The more removed the denominator in these valuation ratios becomes from
the profit line, the more implicit assumptions are being made when they are
constructed. The EV/sales ratio is only useful in comparing the share prices of
companies when it is reasonable to expect the free cash flow produced from
the sales to revert to an industry norm. In many industries this is a fair long
run assumption it is hard to think of many areas where competitive
pressures can be kept indefinitely at bay but it does not always hold in the
shorter term, especially when regulators, and governments generally, involve
themselves in controlling capacity and prices. Nonetheless, markets have a
history of believing in permanently high (or low) margins for favoured (or
ostracised) companies for protracted periods of time; so EV/sales, even more
than most valuation ratios, tends not to help investors time markets.
Book values

In the long run, net asset

value does not depend on
the accounting system ...

A word, finally, on book values. One reason for creating our own profit
measure OpFCF is that entire professions are devoted to the task of
producing accounts which show a company to its best advantage. In AngloSaxon regimes, this means maximising reported EPS; in Germany and Japan,
this has meant minimising reported profits and therefore tax payments.
Yet most accounting bodies have reasonably similar rules about measuring
assets and liabilities; most creative accounting measures depend on exploiting
the rules about when these balance sheet items are recognised. Obvious
examples are depreciation, deferred tax, and restructuring provisions; slightly
less obvious, perhaps, are inventory valuation, the treatment of pension funds,
and bad debt provisions.
Consequently, there are few companies which have consistently managed to
produce reported profits which differ widely from economic profits over very
long periods of time. One paradigm is the UK company which reports a
smooth upward trend in earnings, often as it grows by acquisition, followed
by a sudden profits collapse out of all proportion to the fortunes of its
business units as growth slows and expenses which were deferred from
earlier years catch up with the profit and loss account. The other extreme is
the fast growing Japanese company which produces no profits whatsoever

SBC WARBURG

33

while it is expanding, but which produces blossoming earnings and,

usually, a collapsing multiple when it goes ex growth.
This means that, although reported book values (meaning share capital plus
retained profits) can deviate from net economic worth for considerable
periods of time, they tend to come back to the correct level with a rush.
The following chart shows this point at work in the car industry. Although
General Motors produced substantial gains in retained profits, and therefore
book value, in the late 1980s, the stock market, correctly, did not mark up the
share price in tandem; this was subsequently validated by a collapse in
reported profit following huge restructuring provisions. Again, the share price
did not fall as much as reported book value, reflecting the suspicion which
currently appears justified that the restructuring charge was larger than it
needed to be.
GM: Share price and book value
... take GM, for example.
70
Book value per share

60

$ per share

50

High

40
30
20
Low
10

1995

1994

1993

1992

1991

1990

1989

1988

1987

1986

1985

1984

1982

1983

This makes it difficult to use price/book ratios to value stocks across borders,
because in the short run the book value figure is prone to vigorous
fluctuations driven by accounting alone. However, companies cannot
indefinitely overstate or understate book value per share; so long run trends
in book value per share have proved to be an excellent guide to expansion of
shareholder wealth.

SBC WARBURG

34

Section 4: EV in action
Why bother? how value measurements interact with growth evidence
from continental Europe and Japan how use of EV ratios improve
recommendations weaknesses the need to audit the figures.

Value vs growth
Do EV ratios make
money?

Up to now, the discussion has been largely theoretical; is there any evidence
that all the effort needed to calculate these advanced valuation ratios produces
better investment performance?
Although we did not know this when we began constructing these ratios, the
answer is, resoundingly, yes. The following chart shows the distribution of
EV/EBIT multiples for the companies we cover in Japan; this is closer to being
normally distributed than the P/E distribution, printed alongside for
convenience, which suggests that the market does indeed aim off for variations
on capital structure when pricing shares.

They make sense of

Japan ...

(No. of companies)

(No. of companies)
FY96 EV/OP distribution
25
mode
11.4X
20

FY96 PER distribution (SBCW)

35
30
25
20

mode
26.5X

15

Companies outside
the range

median
35.3X

median
15.0X

15

Companies outside
the range

10

10

0
0

16

24

32

40 48 56
FY96 PER

64

72

80

88

96

16

24

32

40 48 56
FY96 EV/OP

64

72

80

88

96

The acid test, though, of this approach is whether it improves investment

recommendations. It is important to stress, at this juncture, that we are not
attempting to write a formula for beating the stock market. Rather, we are
trying to find a way improving the analytical process in situations where
conventional P/E multiples cannot be easily applied.
In principle, any worthwhile piece of investment analysis starts with a careful
estimate of the outlook for a companys finances, and compares these with
market expectations. The decision matrix is shown below.

Expectations

Prospects

Good

Average

Poor

Good
Average

?
Buy

Sell
Hold

Sell
Sell

Poor

Buy

Buy

Most of these are self-explanatory. The times when analytical input is most
valuable are marked ?; when everything seems most rosy, and shares have
been strong, is when prices are most vulnerable to any bad news; when the
shares are down and sentiment is depressed, any ray of hope can galvanise
share prices.

SBC WARBURG

35

The issue, though, is how to measure expectations; most investors use a

collection of techniques, among them:

Recent share price performance

Consensus forecasts

Valuation ratios

... and, in a test on

Japanese
recommendations ...

Within the investment community, there are, of course, the extremes from
funds which buy value and ignore current financial expectations to those
which buy earnings growth at any price. Our efforts focus on the middle
ground; given a set of financial projections, which valuation ratios prove the
most reliable guide to deciding whether the share price is ahead of, up with,
or behind events ?
We began calculating and recording EV ratios on consistent definitions in
Japan in early 1994, and in Europe in early 1993. The Japanese data has been
produced monthly for an average sample of about 400 stocks; the European
data is less frequent.
In Japan, we first performed the following simple test for seven sectors where
we had enough stocks with a long enough history seven sectors in all.
Each month, sort stocks by analysts recommendation, and compute the
return on the buy recommendations less the return on the sell
recommendations. Over two years, the cumulative difference between the
return on the buy recommendations and the return on the sells is a measure of
the value added by following the analysts advice.

... in six sectors ...

SBC WARBURG

The following table shows the results for six sectors. In each case, the analyst
added some value, in the sense of increasing return relative to the sector;
interestingly, the sell recommendations in sectors which performed well (e.g.
pharmaceuticals) produced higher returns than buy recommendations in
poorly performing sectors (e.g. consumer). This seems reasonable if it is
assumed that sector performance is dictated primarily by macro factors but
that analysts tend to be more influenced by factors specific to individual
companies.

36

Value added by analysts*

Technology
Consumer
Pharmaceuticals
Pulp & paper
Trading companies
Iron & steel
Construction etc

Buy(1)

Sell(2)

5.3
3.9
35.5
36.9
41.4
5.4
14.7

-0.7
-5.4
17.1
22.8
31.0
-2.2
-30.9

Buy Sell(3)

Sector(4)

6.0
9.3
18.4
14.1
10.4
7.6
45.6

4.5
1.5
28.0
30.1
37.0
4.5
-10.0

* Source: SBCW Japan 1994 - 1996

(1) Average annualised return of buy recommendations
(2) Average annual return of sell recommendations
(3) (1) - (2)
(4) Average annual return of sector index

... improved results ...

Although analysts were asked to look at valuation ratios during this period,
the recommendation remained their own responsibility. Most analysts are very
sensitive to earnings momentum; it is, after all, hard to recommend a stock
when the most recent decision has been to revise down the profits estimate.
During this period, enterprise valuation methodology was being introduced,
and, partly because we did not know at the time whether it would work, and
partly because some analysts were more comfortable with the concept than
others, no attempt was made to normalise recommendations according to
valuation criteria.
Had we done this, however, the results of the recommendations would have
improved provided we had used the right criteria. The next test used
followed the same pattern for each sector:

... in a test described in

detail.

Each month, estimate the returns on two portfolios: portfolio 1, buy

recommendations with a below median valuation ratio; portfolio 2, sell
recommendations with above average valuation ratio. Over the same time
period, calculate the annualised returns.
The object of the test is to see whether applying a valuation discipline to
analysts recommendation improves or worsens the performance. The results
are shown in the following chart.

SBC WARBURG

37

Impact of valuation filters

60%

40%
30%
20%
10%

C&H+RE

Iron&St.

Trad. Co.

Pulp&P.

-20%

Consum.

-10%

Pharmac.

0%
Technol.

Annualized Buy-Sell spreads

50%

-30%
Sectors
* AR: analysts' recommendations
Only AR*

PABR + AR

EVOP + AR

APER + AR

PER + AR

The average effect of filtering the recommendations with valuation ratios is

shown in the following table. Filtering by PABR (price to book value, adjusted
for unrealised gains on marketable securities) improved the results the most,
by an average 17.2% annualised; applying EV/OP (enterprise value to
earnings before interest and tax) by 7.3%. By contrast, using a PER filter
knocks 6.6% off the performance of recommendations; using the adjusted PE
ratio which nets out the effect of cross-holdings from the PE ratio) reduced
the performance by 1.4%.
Effect of filtering recommendations*
PABR

EV/OP

APER

PER

Technology
Consumer
Pharmaceuticals
Pulp & paper
Trading companies
Iron & steel
Construction etc

5.7
9.3
39.7
34.0
13.9
10.7
6.8

6.9
3.9
19.4
-2.2
3.6
8.2
11.3

-9.5
2.5
9.6
-27.6
1.1
2.8
11.3

-11.1
3.3
1.3
-36.0
-12.9
1.2
8.3

Average

17.2

7.3

-1.4

-6.6

*Each number shows the improvement in portfolio performance generated by filtering analysts recommendations with
valuation ratios.

Profits multiples worked

best in growth sectors,
balance sheets in cyclicals.

SBC WARBURG

It is also significant that the PE filter produced the worst results in cyclical
sectors paper and steel and where balance sheets are particularly unusual
trading companies. In the case of paper, even the degeared PE ratio the
EV/OP detracted from the performance. Conversely, applying a price/book
38

filter improved the performance by the most in all but two cases technology
and construction. These are generally regarded as growth sectors albeit
currently producing depressed results and it is therefore reasonable to
expect valuation ratios derived from income statements to be particularly
significant.
In the case of continental Europe, efforts to perform a similar back test are
hampered by the way valuation data has been recorded. Although PE and cash
PE ratios are recorded reasonably frequently, the difficulties associated with
compiling data from several different countries has, in the past, made issues of
the company analyser irregular. There are also gaps in the historical data for
individual sectors, which makes it difficult to get a long enough time series to
produce meaningful conclusions.
A similar test in Europe ...

For these reasons, we confined ourselves to two sectors, Media and Autos,
where the record is most complete. The first test, to see whether analysts buy
recommendations did better than their sells, produced the results tabulated
below:
Value added by analysts*
Sector

Buy

Sell

Buy-Sell

Autos
Media

16.8
44.8

-6.6
5.1

23.3
39.8

*Source: SBCW Europe, 1992-1995

Again, the analysts added value within their sector. The second test, filtering
recommendations by valuation ratio, however, produced more ambiguous
results, shown in the following table:
Effect of filtering recommendations*
Autos
Media

PER

PBR

EV/sales

EV/OpFCF

38.8
9.8

24.0
6.6

54.4
8.0

55.3
23.0

*Each number shows the improvement in portfolio performance generated by filtering analysts recommendations with
valuation ratios.

... was inconclusive ...

The results for the PER filter make some sense; since Autos are cyclical,
forward PE ought to be a treacherous indicator of good value, while Media is
a growth sector. However, all the other valuation filters for autos also
detracted from the performance, and while traditional filters for media
improved the performance of recommendations, modern indicators
EV/sales and EV/OpFCF made it worse.
There are a number of reasons why this might be. The most obvious is that
the data drawn from published research used to perform the test is too
sporadic to make any of the statistics valid; certainly, the opinion of clients

SBC WARBURG

39

and salesmen who work with the ratios day to day is that these ratios add
value. The second possibility is that analysts have already taken into account
the valuation ratios when making their recommendations in the first place.
Thus, following only those recommendations which also conform to valuation
screens is to deliberately leave aside the other elements in a piece of investment
analysis, and therefore to detract from the performance of the
recommendations. The third possibility is that PERs and PBRs work better in
European media and vehicles stocks than modern ratios a conclusion that
we are clearly keen to reject but cannot on the basis of this test.
To attempt to shed some light on this issue, we performed another test, which
was simply to check the performance of portfolios constructed using the
valuation data alone that is, to buy cheap and sell expensive stocks
regardless of the analysts views. The results of this test are shown below.
Naive use of valuation ratios*
Autos
Media

PER

PBR

EV/sales

-19.2
-0.5

19.7
-45.7

11.7
-4.9

EV/OpFCF
-22.1
-38.0

* Returns on stocks with cheap valuation ratios compared with returns on stocks with expensive valuation ratios.

This shows that the naive investor, following only the results of PBR and
EV/sales valuation screens, would have done better in the auto sector than if
he had followed the analysts recommendations. In the sense that these are the
ratios which ought, in theory, to be meaningful in a cyclical sector, this is an
encouraging result.
... with two tests on two
sectors ...

In Media, by contrast, the naive test produced a worse performance than

would have been achieved by following the analyst, especially if his views were
filtered by the EV/OpFCF and PBR disciplines. This is disappointing, in the
sense that the modern ratio did not improve on traditional ratios, but pleasing,
in the sense that the media team was clearly of benefit to our clients over the
period.
The conclusions from the back testing on Europe, therefore, are as follows:
First, the lack of a sufficiently rigorous data management programme makes
it difficult to be sure how much the calculation of modern valuation ratios
adds to our product. There are certainly benefits in that the act of calculating
the figures forces analysts to consider all aspects of a companys finances,
rather than simply the published eps figure; but in the absence of good enough
data including a rigorous definition of all the elements of an EV calculation
it is hard to be sure whether the ratios arrived at add value.
Longer experience of applying EV ratios in Europe, rather than Japan, may
mean that analysts are already making full use of the figures in conjunction
with all the other elements in a piece of analysis in arriving at their

SBC WARBURG

40

recommendations. Second-guessing the recommendation therefore detracts

from the performance. This appears to be the case with Autos, but the
opposite holds for Media. However, ignoring analysts recommendations
entirely, and investing by the (right) ratios would have improved the
performance in Autos but detracted from the performance in Media.
Therefore, we will continue to calculate these figures but need to make two
improvements to our systems; first, to produce the data more regularly and
check that the calculations have all been made reliably; and second, to make
sure that analysts are using the figures in the same way. The ideal result would
be that following analysts recommendations beats the sector handsomely, and
also beats a naive investment procedure; but that filtering recommendations
by valuation techniques detracts from the performance.
... demonstrating the
importance of data
management as well as
inspirational analysis.

SBC WARBURG

Taking the studies in Japan and Europe together, therefore, we conclude that
modern valuation techniques appear to improve the performance of
recommendations in some cases on the limited evidence, especially where
corporate structures or cyclical factors make traditional valuation tools
suspect. We therefore intend to calculate the same ratios in all markets but
must be mindful of two important lessons: that the databases thus created
have to be very carefully checked and regularly updated; and that the
calculation of a valuation ratio does not in itself constitute an investment
recommendation.

41

Section 5: Definitions
With the foregoing in mind, we set out the definitions of each item of an EV
calculation. Experience, especially in parts of the world which have not yet
begun to apply EV methods, will lead to modification to these definitions; we
plan to issue an updated EV guide at the beginning of each year.

1. Market capitalisation (MC)

1.1 General
Market capitalisation = Ordinary shares in issue * Share price
Where a company has more than one class of equity, market capitalisation is
the total value of all outstanding equity. Convertible bonds, warrants, and
other equity-linked instruments are debt, not part of market capitalisation.
1.2 Exceptions
None at present

2. Net debt (D)

2.1 General
Net debt = Long and short-term debt cash and cash equivalents + pension provisions

All liquid interest bearing instruments including long-term bonds a

company owns count as cash and cash equivalents. On balance sheet
pension liabilities also count as debt, because retirees have a claim on the
liquid assets of a company.
2.2 Exceptions
2.2.1 Off balance sheet pension funds
If a company has an underfunded off-balance sheet pension fund, this
underfunding should be added to net debt.
2.2.2 Equity securities
If a company owns shares in other companies, these are treated as peripheral
assets, rather than as part of the net debt calculation.

SBC WARBURG

42

2.2.3 Financial subsidiaries

Where a company has large amounts of short term debt used to support
financial subsidiaries (for example, auto makers who lend money to their
customers to help them buy a car) the money receivable from customers
should count as a cash equivalent in the debt calculation. But this introduces
a complication in the profits calculation (see below).

3. Peripheral assets (PA)

3.1. General
The value of any assets which do not contribute to core operating profits.
Obvious examples are investments in marketable equity securities, idle land,
value of affiliated companies.
3.2 Exceptions
3.2.1 What is a peripheral asset ?
It is often a matter of opinion whether assets are core or peripheral for
example, loss-making divisions which may have a value to another company.
Some ways of judging core vs non-core assets are shown on p15. Note in
particular the three-stage test of whether assets are peripheral or not. The key
point is that if an asset is counted as peripheral any contribution it makes to
the sales, operating cash flow, or any other denominator in an EV ratio
calculation is stripped out.
3.2.2 Land
Land is not peripheral unless the corporation is not using it or can reasonably
be expected to stop using it. The value put on it depends on whether the
company is likely to sell it; if this is highly improbable, the value is
correspondingly small.

4. Tax wedge (T)

4.1. General
The difference between the total tax burden on corporate cash flow
distributed to debt providers and the total tax burden on cash flow distributed
to equity holders. See page 18.

SBC WARBURG

43

5. Minority interests (MI)

5.1 General
The value of outsiders interests in a companys subsidiaries. This is best
arrived at by another EV calculation. If the information to do this is not
available, some multiple of profits related to the median market multiple will
have to be used.
5.2 Exceptions
None at present

6. Enterprise value (EV)

6.1 General
EV = MC + D + MI P

6.2 Exceptions
There are no exceptions to the formula. However, analysts have discretion
over which assets are peripheral and what they are deemed to be worth.

7. Sales (S)
7.1 General
A companys revenue excluding sales taxes. Any revenues generated by
peripheral assets but included in reported sales must be excluded.
7.2 Exceptions
7.2.1 Agency Businesses
Agency businesses (for example, trading companies, advertising agencies,
investment banks) may report total turnover as sales. We define sales in these
cases as net commission and trading revenues.

SBC WARBURG

44

8. Earnings before interest, tax, and

depreciation (EBITDA)
8.1 General
Sales less all costs except for interest, tax, depreciation, amortisation,
exceptional items, extraordinary items, less any profits associated with
peripheral assets.
8.2 Exceptions
8.2.1 Taxes
In some countries, companies report local taxes as part of cost of sales. These
should be excluded.
8.2.2 Goodwill amortisation
Treated as part of maintenance capex (q.v.).
8.2.3 Tax-based provisioning
In some countries, companies allowed to make a general provisions (against
unspecified bad debts, for example) as a tax break. The net amount provided
should be added back in the EBITDA calculation.
8.2.4 Pension costs
Where companies run on balance sheet pension funds, the annual service costs
should be excluded from the EBITDA calculation because the capital liability
has been included in the EV calculation as part of net debt.
8.2.5 Accounting provisioning
In some countries, companies are allowed to use provisions made in previous
years (for example as part of restructuring) to offset costs in later years. The
net use of provisions should be deducted in the EBITDA calculation.

SBC WARBURG

45

9. Earnings before interest and tax (EBIT)

9.1 General
EBITDA minus depreciation and amortisation.
9.2 Exceptions
None at present

10. Maintenance capital expenditure (MCAPEX)

10.1 General
Annual reinvestment requirement, as defined by the relevant sector team.
Typically this will be a given percentage of normalised sales, or a given
amount of $ per unit of capacity, or a given amount of $ per unit of output.
For more details, see page 30.
10.2 Exceptions
This figure is, in fact, a collection of exceptions and rules of thumb (see main
text).

11. Net working capital inflation (NWC)

11.1 General
Usually, average working capital during the year multiplied by the local
inflation rate.
11.2 Exceptions
11.2.1 Special product price inflation rates
Where the inflation rate in the companys product price is identifiably
different from the local inflation rate, use this rate instead.

SBC WARBURG

46

12. Operating free cash flow (OPFCF)

12.1 General
OpFCF = EBITDA MCAPEX NWC

12.2 Exceptions
None

13. Adjusted book value (ABV)

13.1 General
Shareholders funds plus unrealised profits on peripheral assets plus (minus)
pension fund surplus (deficit)

14 Standard ratios
14.1 General
The standard ratios for comparing stocks across borders will be
Income statement

Balance sheet

EV/S
EV/EBITDA
EV/OpFCF

Price/ABV

The standard ratio for balance sheet structure will be

MCAP/EV

SBC WARBURG

47

Section 6: a practical example

Even in the UK, EV
analysis helps understand
share prices ...

The EV framework has been developed to give the analyst insights into why
share prices stand at the levels they do, a problem which is particularly acute
comparing companies across borders. But it is not limited to comparative
analysis; frequently, complex companies are misvalued by markets, especially
those with an overzealous concentration on bottom line earnings. The UK
stockmarket, for example, systematically overvalued Hanson shares for some
time, concentrating on group accounts rather than on the economic value of
the operating enterprises it controlled. In this application, EV analysis is a
cousin of sum-of the-parts valuation; but as the following example - an
analysis of BT - demonstrates, it is more sophisticated, and can produce more
profound conclusions.

New BT - New Value

... for example, BT.

One of the first of the UK privatisations, and one of the first national
telecommunications companies to be floated, British Telecom (BT) is evolving
from a state utility, offering traditional wireline telephone services, to a
diversified company with interests in a range of communications services. Its
traditional business is regulated on the now conventional UK model, whereby
prices are forced to decline in real terms. Initially, the main question for the
stock was whether the company could drive its costs - acknowledged to be
excessive, especially personnel - down faster than its prices were falling, and
how the enormous investment in modernising the telephone network was to
be financed.

The company is changing

from a utility to a growth
stock ...

Over time, these issues have become less important, to be replaced with new
questions about how rapidly revenues and profits from new businesses cellular telephony in the UK and overseas, global telephone services for
multinationals, interactive services in the UK, and a range of international
alliances - will grow.

... changing the growth

and risk profiles of the
stock.

Consequently, the profits profile is changing from dependable, but pedestrian,

to less predictable, but higher growth. Moreover, some of the new businesses
will not produce profits in the early stages; but the fact that they do not
contribute to earnings yet does not mean they are worthless. In the context of
a small start-up, this is an uncontroversial point - consider the prices paid for
computer software start-up companies in the US recently - but it is completely
missed if an analyst attempts to value BT on the basis of group earnings per
share growth over the next two or three years.
The following table shows the headline profits projections, which look
distinctly boring. The most that a PER-driven analysis can do is suggest that
fair value would be at a higher multiple on the basis of better long run growth
prospects - although it is very hard to quantify just how much higher the PER
should be.

SBC WARBURG

48

BT: Profits projections and P/E ratios

Year to
end

Pre-exceptional
PBT
EPS
PER

March 95
March 96
March 97e
March 98e
March 99e

3,557m
3,440m
3,645m
3,657m
3,756m

34.6p
36.7p
37.6p
36.7p
36.4p

10.2
9.7
9.4
9.7
9.7

Net
Div

Gross
Yield

17.7p
18.7p
19.7p
20.7p
22.0p

6.2%
6.6%
6.9%
7.3%
7.8%

(data correct on 24th July, share price 355p)

The company is valued

like a utility ...

These PE and yield figures are comparable with listed UK utilities. But BT is
less and less a UK utility; drivers for the share price will increasingly be the
fortunes of the growth businesses within the group, and decreasingly
fluctuations in the domestic regulatory environment or dividend growth
funded by cash squeezed out of an inefficient, recently privatised, business.

... which is less and less

appropriate.

It is more helpful to consider each of the businesses separately. Notionally, BT

can be divided into two parts; Old BT, the wireline operator, a domestic
utility, and New BT, the growth business. The growth businesses are treated
as peripheral assets; the EV of the core telephony business is then defined as:
EV = Market cap + Net debt - Value of growth businesses

Instead, split the company

into growth and core
businesses ...

The notional balance sheet is shown below: the market value of equity and net
debt are reasonably exact figures, the value of the growth businesses is
estimated explicitly, and the value the market is implicitly putting on the utility
business is deduced from these figures.
Assets

Liabilities

Growth businesses
Core businesses

Net Debt
Equity

Estimated value of New BT

... estimate the value of
New BT ...

In summary, the analysis values New BT is worth about 10 bn, as follows:

Value of New BT
60% of Cellnet
20% of MCI
75% of Concert
Interactive service
International alliances
Total

Value (m)

Valuation method

2,833
2,091
2,236
1,993
889
10,043

DCF
Market
DCF
DCF
Per pop

Cellnet is the number 2 cellular telephone operator in the UK. It is

consolidated in BTs group accounts. In this case, we have treated it as a
peripheral asset, valuing BTs 60% stake in the business at 2.8 bn. In the
interests of brevity, we do not show the details of the DCF calculation here;
they are available from our global telecoms team.

SBC WARBURG

49

... deduct their

contribution from group
profits ...

This treatment means that the debt of Cellnet, and its contribution to BTs
group profits, must be stripped out of the accounts to get a picture of the core
business. In this case, the debt of Cellnet is very small; in the year to March
1996, it contributes 110m to group EBIT and free cash flow.
In the case of MCI - which is separately listed - the value of BTs 20% stake,
2.1 bn, is treated as a peripheral asset. Being less than 50% owned, the
results of MCI are not consolidated in BTs accounts, and no adjustment is
needed.
BTs 75% stake in Concert is valued at 2.2 bn. However, we estimate the
company will make an operating loss of 98 mn in the year to March 1997,
reaching break-even in 1998.
The interactive services are valued at 2 bn on the basis, again, of a DCF
calculation. These businesses are wholly owned by BT, so they can be treated
as being debt-free, and the PV of the operating cash flows treated as a
peripheral asset.
Finally, the international alliances are valued at 899 mn; on the basis of the
number of people covered by the licences multiplied by the average value of
the right to service one member of the population. In each case, BT owns less
than 50% of the international venture, and therefore does not consolidate the
results in the group accounts.
Estimated value of Old BT

... producing an implied

multiple for Old BT.

Therefore, the market is valuing the core business at

EV

market cap

net debt

peripheral assets

13.2

22.4

0.8

10.0

Note that this is by no means a definitive calculation, because the value put
on the peripheral assets is a judgement made by the analysts. The virtue of
looking at the company in this way is that the need to spell out the
assumptions and risks buried in the valuation of the peripheral assets makes
it easier for the investor to see what considerations are important for the
shares. This is fundamentally different from an analysis of profits sensitivity;
in the short run, fluctuations in the profitability of the core business will have
a much larger effect on group profits than fluctuations in the profits of the
growth business. But, while the value of the core business is reasonably stable,
the value of the growth businesses depend very much on longer range factors;
competition, sales growth, discount rates, and so on.

SBC WARBURG

50

Free cash flow of Old BT

The implied value of
Old BT ...

... can be compared with

its profits ...

Since the growth businesses are being treated as peripheral assets, their
contributions must be excluded from group profits,to derive an estimate of the
multiple the market is putting on the core business. This calculation is shown
below.
Year to March

1996

1997e

1998e

1999e

Group EBITDA
of which
Cellnet
Concert
Interactive
Core
Depreciation and amortisation
of which
Cellnet
Concert
Interactive
Core
EBIT
of which
Cellnet
Concert
Interactive
Core

5,289

5,412

5,550

5,765

220
-86
0
5,155
2,189

280
28
-14
5,118
2,249

340
167
-70
5,113
2,341

400
303
-51
5,113
2,406

110
76
0
2,003
3,100

130
126
1
1,992
3,163

150
161
29
2,001
3,209

170
182
88
1,966
3,359

110
-162
0
3,152

150
-98
-15
3,126

190
-6
-99
3,114

230
121
-139
3,147

5,118
-2,100
+105
3,123
4.2

5,113
-2,200
+110
3,023
4.4

5,113
-2,300
+120
2,933
4.5

Calculation of core business Operating Free Cash Flow

EBITDA
Maintenance capex in core business
Working capital inflation in core business
Core operating free cash flow
EV/OpFCF

5,155
-2,000
+100
3,255
4.1

Conclusions to be drawn
... showing that the
shares ...

An EV/OpFCF ratio of 4.2x for the core business is one of the lowest figures
for a major business; the arithmetic therefore suggests that expectations are
for the profits performance to be deeply uninspiring. This is somewhat more
negative than our opinion - we expect free cash flow from the core business to
be static - so we conclude that the shares are undervalued. The key element of
this calculation is our valuation of the peripheral assets; and the risks and
rewards for the shareholder depend on the fortunes of the growth businesses.

... are undervalued if our

analysis of the growth
business is correct.

In a recent piece of research, we argued that excessive focus on the rather dull
short run group profits progression meant that the market had overlooked the
value of the growth businesses. This is, of course, contentious - we may be
wrong about the prospects for new technology communications - but the
analysis makes it much clearer where the issues lie.

SBC WARBURG

51

Extensions
The conclusions can be
quantified ...

In this example, we have drawn the soft conclusions from the analysis (what
issues are important for the share price ?). It is also possible to produce a
quantitative opinion of fair value for the shares. The idea will be to take the
full financial projections (profits, dividends, and balance sheet) and combine
them with measures of risk and required return to arrive at an appropriate
target share price.

... on which we will

publish later.

There are various ways of combining these elements, of different degrees of

complexity. At some point, however, any analyst will be forced to justify a
target share price with reference to the way investors at large price shares either explicitly (what is an appropriate EV/OpFCF multiple for a company
with these growth prospects) or implicitly (what is the required return for a
stock like this ?) Before deciding which formulation of the calculation we
prefer, we need sufficient data on the way stocks are valued at present. We will
therefore extend EV analysis to include quantitative measures of risk, required
return, and allowance for changing debt levels once the data has been
assembled.

SBC WARBURG

52

SBC WARBURG

53

SBC WARBURG

Year

15

23

30

38

45

53

60

8
0
10%

15%

10%

Proportion of sample

15

23

30

38

45

53

60

68

75

83

90

Year

68

75

83

90

98

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995

15%
Proportion of sample

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

Appendix 1: PER distributions

UK PER surface

25%

20%

5%

0%

PER band

Canada PER surface

25%

20%

5%

0%

PER band

54

UK market multiples

30

25

20

15

10

0
1980 1981

"Market"
1982 1983

Median

1984 1985

Truncated

1986 1987

Truncated
1988 1989

1990 1991
1992

"Market"

Median
1993 1994
1995

Canada market multiples

40
35
30
25
20
15
10
5
0

"Market"

1980 1981

1982 1983

Median

SBC WARBURG

1984 1985

Truncated

Truncated
1986 1987
"Market"

1988 1989

1990

Median
1991 1992
1993 1994
1995

55

Year

15

23

30

38

45

53

60

15

23

30

38

45

53

60

68

75

83

Year

68

75

83

90

8
0
20%

15%

10%

15%

10%

Proportion of sample

SBC WARBURG

90

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

25%
Proportion of sample

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

France PER surface

35%

30%

5%

0%

PER band

Germany PER surface

25%

20%

5%

0%

PER band

56

France market multiples

40
35
30
25
20
15
10
5
0

"Market"

1980 1981

1982 1983

Median

1984 1985

Truncated

Truncated
1986 1987

1988 1989

1990 1991
1992

"Market"

Median
1993 1994
1995

Germany market multiples

35

30

25

20

15

10

"Market"

1980 1981

1982 1983

Median

SBC WARBURG

1984 1985

Truncated

Truncated
1986 1987

"Market"

1988 1989

1990

Median
1991 1992
1993

1994 1995

57

Year

15

23

30

38

45

53

60

15

23

30

38

45

53

60

68

75

83

Year

68

75

83

90

8
0
30%

20%

25%

20%

15%

10%

Proportion of sample

SBC WARBURG

90

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

40%
Proportion of sample

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

Netherlands PER surface

60%

50%

10%

0%

PER band

Hong Kong PER surface

40%

35%

30%

5%

0%

PER band

58

Netherlands market multiples

18
16
14
12
10
8
6
4
2
0
1980 1981

"Market"
1982 1983

Median

1984 1985

Truncated

Truncated
1986 1987

1988 1989

"Market"

1990 1991
1992 1993
1994 1995

Median

Hong Kong market multiples

30

25

20

15

10

0
80

"Market"
81

Median

SBC WARBURG

82

83

84

Truncated

85

Truncated
86

87

"Market"

88

89

90

91

Median
92

93

94

95

59

SBC WARBURG

Year

15

23

30

38

45

53

60

68

75

83

90

98

40
35
30
25
20
15
10
5
0

45

50

55

60

65

70

75

80

85

90

95

12%

10%

8%

6%

15%

10%

5%
Proportion of sample

Year
1990
1991
1992
1993
1994
1995
100

14%
Proportion of sample

1986
1987
1988
1989
1990
1991
1992
1993
1994
1995

India PER surface

20%

18%

16%

4%

2%

0%

PER band

Italy PER surface

25%

20%

0%

PER band

60

India Market multiples

30

25

20

15

10

5
Truncated
0
1990

Median

1991
1992
1993
Median

1994

"Market"

1995

Italy Market multiples

140

120

100

80

60

40

20

"Market"

0
1986

Truncated
1987

Median

SBC WARBURG

1988

1989

Truncated

1990
"Market"

1991

Median
1992

1993

1994

1995

61

SBC WARBURG

15

23

30

38

45

53

60

68

75

83

90

98

15

23

30

38

45

53

60

68

75

83

90

8
0

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

Year

12%

10%

8%

6%

4%

Proportion of sample

Year

1987
1988
1989
1990
1991
1992
1993
1994
1995

6%

5%

4%

3%

2%

Proportion of sample

Japan PER surface

8%

7%

1%

0%

PER band

Korea PER surface

16%

14%

2%

0%

PER band

62

Japan market multiples

80
70
60
50
40
30
20
10
0

"Market"

1980 1981

1982 1983

Median

1984 1985

Truncated

Truncated
1986 1987

1988 1989

"Market"

1990 1991
1992 1993
1994

Median
1995

Korea market multiples

30

25

20

15

10

"Market"

0
1987

Truncated
1988

Median

SBC WARBURG

1989

Truncated

1990

1991

"Market"

Median
1992

1993

1994

1995

63

year

15

23

30

38

45

53

60

68

75

83

15

23

30

38

45

53

60

68

75

83

90

Year

12%

10%

8%

6%

Proportion of sample

SBC WARBURG

90

8
0

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

25%

20%

15%

10%

Proportion of sample

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

Australia PER surface

35%

30%

5%

0%

PER band

Singapore PER surface

18%

16%

14%

4%

2%

0%

PER band

64

Australia market multiples

25

20

15

10

0
1980 1981

"Market"
1982 1983

Median

1984 1985

Truncated

Truncated
1986 1987

"Market"

1988 1989
Median
1990 1991
1992 1993
1994 1995

Singapore market multiples

50
45
40
35
30
25
20
15
10
5
0
1980 1981

"Market"
1982 1983

Median

SBC WARBURG

1984 1985

Truncated

Truncated
1986 1987
"Market"

1988 1989

1990 1991

Median
1992 1993
1994 1995

65

SBC WARBURG

Year

0%

30
23
15

38

45

53

60

68

75

83

90

1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
98

Year
15

23

30

38

45

53

60

68

75

83

90

98

1986
1987
1988
1989
1990
1991
1992
1993
1994
1995

8
0

20%

15%

10%

Proportionof sample

Proportion of sample

Sweden PER surface

25%

5%

0%

PER band

USA PER surface

30%

25%

20%

15%

10%

5%

PER band

66

Sweden market multiples

Off scale
50
45
40
35
30
25
20
15
10
5
"Market"
0
1982 1983
Truncated
1984 1985
1986 1987
1988 1989
Median
1990 1991
1992 1993
1994 1995
Median
Truncated
"Market"

USA market multiples

25

20

15

10

0
1980 1981

"Market"
1982 1983

Median

SBC WARBURG

1984 1985

Truncated

Truncated
1986 1987
"Market"

1988 1989

1990 1991

Median
1992 1993
1994 1995

67

Other Group Offices

SWISS BANK HOUSE

1 HIGH TIMBER STREET
LONDON EC4V 3SB
TELEPHONE: 0171 329 0329
TELEX: 887434 SBCO G
1 FINSBURY AVENUE
LONDON EC2M 2PP
TELEPHONE: 0171 606 1066
TELEX: 937011 WARSEC

New York

+1

212

224 7000

Tokyo

+81

5402 9111

Amsterdam

+31

20

551 0100

Auckland

+64

307 4800

Bangkok

+66

693 2435

Frankfurt

+49

69

714 010

Hong Kong

+852

Johannesburg

+27

11

322 7000

Kuala Lumpur

+60

201 3011

Luxembourg

+352

Madrid

+34

436 90 50

Milan

+39

725 271

Montreal

+1

514

842 8726

Paris

+33

48 88 30 30

Seoul

+82

399 5566

Singapore

+65

Stockholm

+46

453 7300

Sydney

+61

324 2000

Taipei

+886

717 5606

Toronto

+1

416

364 3293

Zurich

+41

238 1111

2971 8888

421 211

538 8611

ISSUED BY SWISS BANK CORPORATION (SBC), ACTING THROUGH ITS DIVISION SBC WARBURG, REGULATED IN THE
U.K. BY THE SECURITIES AND FUTURES AUTHORITY. A MEMBER OF THE LONDON STOCK EXCHANGE.
This document is not intended to be an offer, or a solicitation of an offer, to buy or sell relevant securities (i.e. securities mentioned herein or of
the same issuer and options, warrants or rights to or interests in any such securities). The information and opinions contained in this document
have been compiled from or arrived at in good faith from sources believed to be reliable. Relevant confidential information, if any, held within
SBC or elsewhere in the SBC Group (i.e. SBC and its subsidiary and associated undertakings), which is not available to our research department
by virtue of Chinese Walls or other internal procedures, is not reflected in this document. No representation or warranty, express or implied, is
made by SBC or any other member of the SBC Group, including any other member of the SBC Group from whom this document may be
received, as to the accuracy or completeness of the information contained herein. All opinions expressed herein are subject to change without
notice. At any time SBC and other companies in the SBC Group (or employees thereof) may have a long or short position, or deal as principal or
agent, in any relevant securities or provide advisory or other services to an issuer of relevant securities or to a company connected with the issuer.
This document may not be reproduced or copies circulated without authority; it is not intended for and must not be distributed to private
customers. Additional note to persons receiving this research in Italy: Should you require further information or wish to effect transactions in
relevant securities please contact either Giubergia Warburg SIM in Milan or SBC Warburg Italia SIM SpA, a member of the SBC Group, in
Milan or its London branch.