ñòð. 76 |

is carried out by parent companies, they may want to focus on cash flows

into their own bank accounts. This might be limited to annual dividend pay-

ments. The different parent companies of a JV might also want to adopt dif-

ferent costs of capital.

There could be real problems if the methodologies are materially different

but small differences such as, for example, a one percentage point difference

in CoC, are usually irrelevant. This is because it is rare for decisions to be so

marginal as to be changed by a single point swing in the CoC.

The point to realise is that the parent companies simply have to learn how

to live with each other or accept that the alternative is to break up the JV.

Issues concerning methodology should be addressed at the formation stage

of the JV so that the company can make good decisions in the future.

The second complication concerns modelling. An additional layer of finan-

cial modelling is usually required in joint venture-economic models. This is

because what might be seen as a cost by the joint venture might well be a

revenue to one of the parent companies. Joint ventures are usually formed

when two companies feel that they can both contribute to the success of the

239 The first pillar: Modelling economic value

venture. This means that material cash flows between parent and JV com-

pany are to be expected. Examples of the sorts of flows that might be involved

would include payment of licence fees for technology, purchase of feedstock,

provision of staff or support services.

The approach to model economic value is first to model the JV as a stand-

alone entity. This analysis will usually be done within the JV. Each parent will

then need to prepare an additional model that adjusts for cash flows between

themselves and the JV.

Any cash flow will need to be checked to see if it creates an additional

contribution at parent company level. For example, are feedstock supplies

charged at cost or market price? If they are at market price, is there an ele-

ment of profit to strip out? It may or may not be necessary to correct for the

profit on a feedstock supply. It would be necessary to correct if the parent

company could not otherwise have made the sale but if a third party sale was

a realistic alternative then there would be no correction to apply. The cor-

rection need not always serve to increase value from the perspective of the

parent. A formula price for feedstock supplies may be below the anticipated

market price.

This analysis should be done within each parent company because it will

require access to confidential information concerning profitability. One par-

ent may well be getting a better deal than another and it would not be appro-

priate for the JV staff to be given access to such information. So in effect a

shareholderâ€™s economic model is required which can sit on top of the basic

cash flow model which the JV company should supply.

Monte Carlo analysis

Monte Carlo analysis is a simulation technique that allows one to investigate

the impact of uncertainty. Basically, one builds a model and then instead of

running it just once with a single set of assumptions, one sets assumptions

as ranges and/or statistical distributions and runs the model, say, a few thou-

sand times in order to investigate the range of possibilities. Each individual

case makes a separate â€˜dipâ€™ into the statistical distribution for the variables

and produces a possible outcome. The range of possible outcomes can then

be tested to find the expected value and also to draw cumulative probability

distributions and probability density functions.

The technique has great intellectual appeal because it does allow one to go

beyond single-point analysis and to take account of the range of possible out-

comes in a statistically justifiable way. The analysis is facilitated via add-ins

240 The three pillars of financial analysis

to the Excel spreadsheet model.24 A typical financial spreadsheet model is an

ideal candidate for Monte Carlo analysis as there are many input assump-

tions and just a few key economic indicators to focus on.

There are, however, significant drawbacks to the approach. It appears to

give scientific precision but is this justified? As usual, the analysis is only

as good as the assumptions but these can tend to be hidden from decision-

makers. It is always important to assess not just the range for each individual

variable but also any correlations that exist between variables. An example

of this could be sales volume and price. Correlations will tend to increase the

range of possible outcomes. Even if the assumptions are good there can be a

feeling of loss of control because one can never say exactly what the answer

corresponds to.

My recommendation is that Monte Carlo analysis should only be used in

special circumstances. These could include two generic situations:

â€¢ where there is an important non-linearity that needs to be investigated;

â€¢ where it is important to study the full range of a statistical distribution as

opposed to focus on just the expected value.

A non-linearity is where an expected value input assumption will not pro-

duce an expected value output. An example might be where the govern-

ment levies a tax charge based on, say, the higher of two methods in order

to ensure that it always receives some tax. I will be considering how to deal

with this kind of situation in more detail later in the book when I deal with

how to value flexibility. In simple terms, however, one needs to use a Monte

Carlo simulation approach that allows for the variability of key variables if

one needs to obtain the most accurate estimate of expected value NPV when

there is a non-linearity.

The second generic situation where we tend to need to use Monte Carlo

simulation concerns when we need to focus on the full range of a statistical

distribution. Suppose, for example, a decision-maker wanted to know not

only what the expected value NPV was but what the probability was that it

would be at least zero or that cash flows would at least reach some specified

threshold. Calculating the probability of achieving at least a zero NPV would

require all variables to be given probability ranges and then a Monte Carlo

simulation would have to be run. A second example of focussing on the full

range of possible outcomes would be where a decision-maker was not happy

The two most popular add-ins are called Crystal Ball and @Risk. Both are very good and are easy to

24

use. They are also quite expensive! There are cheaper add-ins available as well. My suggestion is that

potential users should first test any product before making a purchase.

241 The first pillar: Modelling economic value

simply to characterise the possible outcomes as being just, say, three pos-

sible capital costs each with a probability that between them added to 100%.

This approach would tend to ignore many low-probability but high-impact

outcomes.

My experience is that Monte Carlo analysis is only very occasionally

required and I suggest that readers do not learn the skill unless/until they

need it. The computational side is very easy if one has a good Excel model

and one of the proprietary add-in packages. The difficulty is in establishing

good distributions for all of the variables and any associated correlations.

Our lack of ability to do this well is such that we will always need to view any

Monte Carlo-produced results with considerable caution despite its appeal.

Part 7: Modelling financed cases

Introduction

Although the approach to valuation that is proposed in this book is to exclude

finance effects, there are times when it is necessary to build a model which

does incorporate finance. A brief summary of the additional issues that are

raised is provided in this part. We will also take the opportunity to develop

a financial model of a company that will allow us to review and recap several

of the issues that have been covered in this book.

One thing that we will learn from this part is that we should be very pleased

that the normal approach to finance is to exclude it from investment evalua-

tion. This is because it creates additional difficulties compared with financial

analysis carried out in the conventional manner.

We need to learn how to incorporate finance because valuation is not the

only thing that we will use our financial models for. The real world operates

through debt and equity. Whenever we want to know how our company will

look from the outside we will need to build a financed model that distin-

guishes between debt and equity. The outside is rarely concerned with the

overall value of a company. The outside is concerned either with debt and

how secure it is or with equity and what the share price will be. So a financed

model is going to be necessary when a project is big enough to influence how

a company will be perceived from the outside.25

A financed model of a project is also necessary if the project is taking on debt directly.

ñòð. 76 |