Key Findings

Now that we have established a premise for discussing the findings from our case study, we will present them, with a special focus on the use of our model in an applied setting. Our findings are based upon direct feedback from the case participants and our own observations. As the findings are ultimately a result of our own interpretation of the case study, we emphasize that they do not directly represent Orkla. However, we note that we have worked with Orkla to ensure that the presented content is in accordance with the experience that the case participants had during the case.

CHAPTER 7. CASE STUDY WITH ORKLA

7.4.1 Sequential Thinking

A somewhat limiting assumption of our model is the assumption of systematically time inde-pendent cash flows. Note that cash flows can be deinde-pendent in time, as long as this dependence is not conditional upon the market states, i.e. the cash flows can exhibit a non-systematic dependence. This requires the user of our model to think about the evolution of cash flow streams in a slightly unintuitive way. After all, to state that future cash flows are unaffected by previous systematic outcomes is questionable, at least for many firms. Our opinion is that in general, the assumption of systematic time independence in cash flows is slightly unrealistic.

For example, the revenue in a booming economy for a given year is likely to be dependent on the state of the economy in the previous year. Therefore, it is certainly intuitive to think sequentially. In fact, it could be strictly necessary to achieve a realistic valuation. However, the problem with sequential thinking is that the number of cash flow estimates are increasing in time, as the different paths leading up to a future point in time grows to an increasingly large number. Sequential thinking gives rise to both a strength and a weakness of our proposed model. The weakness lies in the difficulty of estimating cash flows that are independent of the past, as pointed out above. The strength however, is how the model can allow for such cash flows by extending the number of scenarios. This is not something that would be possible with the CAPM, as the CAPM does not incorporate market states. For concrete examples of such extensions, the reader is referred to chapter 8. Nevertheless, the user of the model should be made explicitly aware that cash flows are systematically independent in time when using our proposed model in its most basic form. We conclude that without this awareness, the user of our proposed model might make choices which are in contradiction with the theoretical framework of the model.

7.4.2 Ease of Communication

Our proposed model builds on topics that are not necessarily clear for many practitioners.

Hence, we are aware that it is probably a bit ambitious to assume that practitioners have a complete understanding of what is going on ’under the hood’ of the model. Theoretically, the user only needs to be able to estimate the state-contingent cash flow of the asset subject to valuation. After all, the estimation procedure of the state prices could certainly be conducted through a software. However, we deem that the user would most likely not accept such an approach to valuation without further clarification, as this would be a complete black box.

The reason is that to understand the output of the valuation, i.e. the value, the user needs to understand the intuition behind the model. Furthermore, in a real life decision making process, the output of a valuation is just one of many factors that contribute to the investment decision.

Hence, if the drivers behind the value are not fully understood, the valuation is likely to receive less weight in the decision-making process, and might even be ignored. As such, it seems evident that the user must understand the intuition behind the model. Nevertheless, we do believe that the intuition can be communicated relatively efficiently. In addition, the case participants also emphasized the fact that the users of the model are not the only ones that have to understand the intuition behind the model. This results from the fact that the estimated value of an asset has to be communicated to many other stakeholders in the decision making process, which do not have a finance background. Orkla emphasized that business rationale and practical judg-ments are just as important aspects in investment decisions. Consequently, a highly complex and precise model could in fact be value destroying for the investment/divestment decision as

CHAPTER 7. CASE STUDY WITH ORKLA

a whole, if it suppresses the practical aspects of the decision making too much.

We believe the discussion on the proposed model’s ease of communication can be further sep-arated into two problems. The first is with regards to the theoretical framework of our model being relatively unknown to the vast majority of practitioners. This includes concepts such as Markov chains, BSM pricing and linear algebra. However, we argue that it is not a require-ment to understand the mathematical concepts of the model to use it. This is analogous to the many different devices practitioners use every day, without necessarily understanding the underlying mathematical concepts. For example, we believe that most practitioners would fail to conduct an OLS regression analytically to estimate the beta in the CAPM, even though they can calculate it successfully with the aid of statistical software. In fact, we believe that for all practical purposes it is perhaps more important to understand the underlying intuition of a beta estimate, not necessarily the intricate details of the machinery which is being used to arrive at the estimate.

To understand the second problem we wish to address, consider the fact that most, if not all business students, are taught the CAPM as part of their degree. Moreover, the concepts of the CAPM take time to understand (or just accept). After all, business students spend at least a year of their degree on the subject. Hence, we believe that there is a challenge of introducing something new, especially since the burden of proof lies with us. Luckily, some of the logical building blocks needed to understand state-contingent pricing is already partly internalized by many practitioners. For example, the concept of market risk should be fairly intuitive for most. Thus, we believe that if practitioners can understand the concept of beta, they can also understand the concept of state prices. This belief was strengthened through our case study with Orkla.

7.4.3 Forecasting State-Contingent Cash Flows

As we mentioned in section 5.1.3, for every additional state defined in our model, the user has to estimate an additional cash flow in each time period. Furthermore, we also addressed the fact that additional states should be justified by increased precision. For instance, zero-beta assets have the same cash flows across all states, and could therefore easily be valued simply by discounting with the risk-free rate. Hence, our proposed model is more value adding when cash flows have high systematic risk (both negative and positive systematic co-movements), and especially when the systematic risk is asymmetric between states. Furthermore, any rational and pragmatic individual should choose the simpler model when faced with the choice of two models yielding the same end result. As we mentioned in our brief introduction of Orkla, it is a company with relatively low systematic risk. Accordingly, its M&A department is mostly concerned with the estimation of cash flows that have low systematic variation. Consequently, the added complexity might not justify the use of our model for many of Orkla’s projects.

However, the case study revealed certain areas where our model can prove to be quite useful, as will be further discussed in sections 7.4.5 and 7.4.6.

CHAPTER 7. CASE STUDY WITH ORKLA

7.4.4 Forecasting in Real Terms and Nominal Terms

If done with consistent assumptions, valuation in nominal- and real terms yields identical present values. However, a drawback to our model is that forecasting in nominal terms might be slightly difficult. This results from the fact that it requires the user to estimate inflation expectations conditional on the states of the economy, as expected inflation is a key-component in the nominal risk-free rate. The problem of state-contingent inflation is effectively eliminated by forecasting in real terms and using an unconditional real risk-free rate, which is what we do in our proposed model. However, in many applied scenarios, working with nominal cash flows is perhaps most natural. As an example, consider a company with much of its future cash flow determined by contracts. The expected cash flows will then likely correspond to the payments agreed upon in the contracts, and these will under normal circumstances be nominal.

This implies that to value these cash flows, the user of our proposed model will have to make assumptions about inflation anyway, i.e. to convert expected cash flows to real terms. On the other hand, if cash flows are not determined through contracts, it is perhaps easier to forecast in real terms. In such cases, focus should be on the real growth of the business, which means that considering inflation only complicates the forecasting activity. In our opinion, this is espe-cially true for long-term forecasts, where deciding on appropriate inflation expectations is most certainly difficult, if not impossible.

Moreover, our impression is that many practitioners generally prefer to work with nominal cash flows. This was also confirmed during the case study. There can be many reasons for why, one being that cash flow expectations are in nominal terms, as illustrated with the con-tract example in the previous paragraph. Another reason, which we notably did not encounter in the case study, might be the fact that when using nominal cash flows in the CAPM, the expected inflation rate does not enter the equation for expected return directly, since it is im-plicit in the observed risk-free rate. Note however, that this subtle inclusion of inflation in the discount rate can lead to a mismatch between the risk-free rate and the assumed cash flow growth, i.e. if practitioners think that this frees them from making assumptions about infla-tion. To illustrate, consider the fact that the long-term growth rate of nominal cash flows, i.e.

the real growth rate and inflation, has to be consistent with the risk-free rate used to calcu-late the appropriate discount rate. For example, the long-term growth rate of nominal cash flows, cannot be higher than the long-term nominal risk-free rate, since the long-term nominal risk-free rate is capped by the long-term nominal growth in the economy. This results from the same logic applied when discussing the bounds for the real risk-free rate in section 3.3.2.

Hence, there is no way around the fact that assumptions about inflation have to be made when forecasting nominal cash flows, even if they are made implicitly in the growth of the cash flows¹. Lastly, we mention that an imperfect solution to using nominal cash flows in our model is to let inflation be independent of states. Nevertheless, the exact choice of conducting a valuation in nominal or real terms should ultimately be dependent on the situation, and we leave this choice up to practitioners and users of our proposed model.

1This issue is similar to the issue addressed by Aswath Damodaran in his blogpost: As g-> r...To Infinity and Beyond!, which can be read at http://aswathdamodaran.blogspot.dk/2016/11/myth-52-as-g-rto-infinity-and-beyond.html

CHAPTER 7. CASE STUDY WITH ORKLA

7.4.5 Combining the Model with the CAPM

Throughout the case, it became evident that it can be value adding to combine the model with the CAPM. For example, imagine that we have three possible non-recombining paths that a project can take a year from now, which are contingent on the realized states. In other words, the state one year from now determines how the unconditional cash flows of the project will look in the future. We could then use the CAPM framework to arrive at three respective present values one year from now, i.e. an unconditional forecast and valuation for each state a year from now. Since the present values one year from now are state-contingent, we could use our model to find the present values today. Another way to combine our proposed model with the CAPM framework could arise if there was a significant asymmetric development in cash flows for the next few years, but cash flows were expected to become symmetric after this period. In similarity to the previous example, we could then use the CAPM to find the present value in the latter period and use our model in the former and asymmetric period. In a sense, such an approach is similar to a sum of the parts valuation, but where the different parts are segments of the future cash flow stream.

Before we proceed with our discussion, the reader should note that these type of approaches are certainly theoretically inconsistent. The fact that we advocate for such approaches might also seem hypocritical, especially since we have argued for the consistency of other theoretical aspects. However, we want to stress that we are not trying to camouflage that what we are proposing is not perfectly consistent. On the other hand, we also want to emphasize that we believe that creating models which are supposed to provide a holistic description of reality, is a certain road to failure in the real world. The reason being that economics and finance are social sciences, where plenty of relations may never be possible to quantify and/or measure exactly.

Thus, they stand in contrast to other sciences, such as physics and mathematics. Instead, we are proponents of creating models that do an adequate job at describing specific parts of reality.

We believe that combining models, even though somewhat theoretically inconsistent, can add more value than adhering to one single theoretical framework, which might be a bad description of reality as a whole.

Returning to our discussion, we will address a shortcoming of our model. As we will describe in section 8.2, our model faces some challenges in an applied setting, which result from the fact that there is an underlying assumption of cash flows being completely non-systematic in their variation within states and/or with time. As we will show, this creates a bias towards over-valuation of cyclical assets, which is increasing with time. This bias could possibly be minimized by approaches like the ones above. Furthermore, combining models may be particularly useful for reducing complexity, as it can have the ability of simplifying forecasting activity and allows users of our proposed model to do much of their work in well-known environments. Note that the user can still benefit from the strengths of our model. This includes elements such as real options, which in many ways are very similar to the examples we gave in the first paragraph of this section. We will get back to this topic in 8.1, but we note that real options can be important components in dynamic capital budgeting, as they can have substantial implications on the value of a project.

CHAPTER 7. CASE STUDY WITH ORKLA

7.4.6 Using the Model for More Specific Purposes

We have already touched upon the subject in the previous paragraph, but one of the most important findings of our case study, was the fact that our proposed model perhaps excels the most when pricing relatively specific parts of the cash flow of a project. This is directly related to the complexity of conducting a valuation using our proposed model, which can increase quite considerably with the number of years in the forecast, and the complexity of the project being valued (given that we have multiple states). Furthermore, the amount of work required when estimating conditional cash flows, might as mentioned previously not be justified if the systematic variation is low. Therefore, our proposed model might be most appropriate to use for specific and limited purposes, when the user can benefit from the ability to model asymmetric systematic variations in cash flows, and/or the implementation of real options. We find that focusing the scope of the valuation can make the model less overwhelming, as the conditional cash flow forecasts become more tractable. For instance, the model could be used to value a potential synergy in an acquisition, which is contingent on the evolution of the market in the near future. This could for example be the possibility of deciding not to extend a company’s product portfolio if the market enters aLow state in the near future. This is an example of a real option, which as already mentioned will be discussed more in-depth in section 8.1. Nevertheless, the real option gives an example of how the model can be used for specific valuation purposes.

This accentuates a key message in our thesis, which is that we do not strive to replace the conventional approaches to capital budgeting, merely to provide a complement and alternative.