ASPECTOS Y ACCIONES INICIALES PARA LLEGAR A LA

The general company valuation model, depicted in equation 5.1, consists of three components. From now on, we will deepen into the operationalisation of the second component, strategic Real-Options value. The general model serves as guidance to properly include a company’s strategic Real-Options value in a company valuation, as defined by this research.

A theoretical guidance, based on this research, to use option-pricing methods to value strategic Real- Options is stated in section 5.1.2. We will try to apply option-pricing methods in combination with the MADD approach on strategic Real-Options retrieved from a PhiDelphi case, in order to demonstrate our concept and to find out and show to what level of detail we are able to operationalise our (generalised) model.

Section 5.4 will introduce the case, which is concerned with the valuation of a strategic opportunities to expand manufacturing and sales to other countries. But first, we will operationalise the MADD approach in order to apply this on our case.

5.3.1

Operationalising the MADD approach

Copeland and Antikarov (2001) have established the MAD assumption, which is according to them and to many others (e.g. (Trigeorgis, 1999), (Brealey et al., 2011)) the best estimate for the underlying asset’s value. In creating the binomial lattice, subsequently, Copeland and Tufano (2004) do not propose a general method to determine the project’s volatility. Instead, they recommend practitioners to look at the value drivers. In an example, they use the volatility of the spread between the price of the factory’s output and input, which are both traded in the financial markets. Brand˜ao et al. (2005) rightly remark that this method is too complex and not intuitive for practitioners.

Copeland and Antikarov (2001) suggest however another method to estimate the volatility of a project that has multiple uncertainties. Besides Monte Carlo methods, they suggest to use subjective estimates provided by management. According to Copeland and Antikarov (2001), most managers and industry experts have subjective, non-formal, non-statistical estimates of volatility in their heads. If you let management answer what are the highest and lowest values that a variable (e.g. FCF) could take in some year, withα% confidence, and you assume the uncertainty follows a Geometric Brownian motion, you can establish an estimate of the variable’s volatility. Copeland and Antikarov (2001) show that when

αis 95%, the volatility of the variable could be determined according equation 5.2. Where,V_TU pper and

V_TLower are the highest and the lowest estimate respectively for the variable’s value afterT periods,ri is the growth of the variable in eachith period andV0 is the value of the variable after 0 periods, which is

the start value of the variable.

σ= ln( V_TU pper V0 )− Pn i=1ri 2√T , σ= Pn i=1ri−ln( VLower T V0 ) 2√T (5.2)

In a good free cash flow forecast, all risks attributable to that project are incorporated. The forecasted free cash flows are expected values. If management can provide us with good estimates of the highest free cash flow in some future period T, with α% confidence, we can determine the project’s volatility. Fortunately, management often makes multiple cash flow forecast for projects they plan to undertake, so practitioners will be familiar with such a method. In our model, we choose αto be as well 95%, as we want to be as close to 100% as possible, but don’t think management’s estimates can be any better. Besides, we will always want to know the value ofV_TU pper, as we are interested in the potential growth. Based on symmetry of the standard normal distribution, we can then accordingly model the cash flows forVLower

T .

Unfortunately, the free cash flow series of a project are probably not homoscedastic, but heteroskedastic, which means that the variance of the free cash flows differ through time. Especially for projects in the earlier phases, free cash flows will probably have a high variance (hence high volatility). In later phases, the amount of free cash flows will probably stabilise and the variance (hence volatility) will drop. We will therefore ask management as well to identify periods with about the same level of volatility, for each period they need to estimate again with 95% confidence to what level (or with which ratio) free cash flow will grow. The disadvantage of this method is that by allowing the volatility to change, the binomial lattice becomes non-recombining. A risk-adjusted decision tree can however include multiple binomial lattices.

The main advantage of using binomial trees is that Real-Options could easily be incorporated and valued. Incorporation of flexibility alters the fair discount rates, but by using risk neutral valuation, the discount rate is for each branch in every phase equal to the risk free rate of interest.

The main advantage of the MADD approach is that it integrates the value drivers of an underlying asset, just like Monte Carlo simulation does. In contradiction with many other methods, this method does not force you to make a value driver – which is traded in the financial markets (e.g. an asset, spread, commodity, etc.) – responsible for all the project’s uncertainties. In many cases, it will not be true that the volatility of the underlying asset is exactly replicated by some asset traded in the financial markets. However, when an asset’s uncertainty is replicated by some traded asset, as for example with Real-Options in the oil industry, volatility derived from the financial markets should always be used instead of management’s estimates. Furthermore, the MADD approach can also be used to estimate the volatility of variables that influence the project’s free cash flow.

A Matlab function, with the input parametersF CF,W ACCandV T, can be found in appendix B. F CF

is a (1x m) matrix with the forecasted amounts of free cash flows of the project till timem, W ACC

is the company’s weighted average cost of capital and V T is the highest possible amount ofF CF for the project after m periods with 95% confidence. This function could be called as follows: [sigma, S] = waarde(FCF, VT, WACC). Matlab will then return the asset’s value S (based on the provided free cash flows), the asset’s volatility sigmaand a multiplot that includes the range of free cash flows and the expected free cash flows based on the asset’s volatility and based on the actual growth and average growth of free cash flows.