Approximations of the risk value driver

A key finding in this thesis is that industry affiliation seems to capture valuation specific information not captured by the fundamental selection variables. As explored in the previous section, a sensible explanation is that we did not include good enough approximations of the risk value driver. We also saw that valuation errors increase when adding Size and EBIT margin to the SARD equation for EV/EBIT valuations. The immediate rationale behind the result is that selection variables should be tailored to different multiples. However, another explanation could be that our selection variables simply are not good enough proxies of the relevant fundamental value drivers. The following is a discussion of the limitation of one of our main assumptions: the chosen selection variables are good approximations of the underlying drivers of multiples. We have focused this discussion on the risk variable since it is more difficult to quantify compared to profitability.

We included Size and Net debt/EBIT as variables meant to capture aspects of risk relevant to valuations. We learned from the mathematical derivations in Chapter 2

102 6.4 Discussion of assumptions and limitations

that price multiples are driven by the cost of equity (r_e) as a measure of risk, while the enterprise-based multiples are driven by WACC. We ask ourselves whether we included the most optimal proxies to capture the same aspects of risk as captured byr_e and WACC. This is not necessarily the case.

An option would be to include beta as a selection variable. Both re and WACC depend on beta. Beta is used in the Capital Asset Pricing Model (CAPM) where the expected return is expressed as

r_e =r_f +β(r_m−r_f) (6.1)

Whererf is the risk free interest rate, rm is the return on the market portfolio, andβ is the beta for the firm under consideration. Like many other selection variables, beta is based on historical input, which does not necessarily reflect future expectations.

Beta is a measure of the stock’s volatility versus the market. A beta greater than one indicates that the stock is more volatile than the market. A beta less than one indicates that the stock is less volatile than the market. Beta is founded on two important elements: Business risk and gearing. We do not capture business risk with any of our applied SARD selection variables.

Only a few previous studies include beta as a selection variable for comparable firm selection. Alford (1992) briefly mentions that he considered to use market beta but did not report it because it was just as accurate as total assets. Serra and Fávero (2018) use beta as a measure for risk but comment “it is more common to think of size as a proxy to risk or cost of capital”. They find that beta is correlated to almost all multiples in the US (five percent significance) but not in Brazil, which is considered a small and undeveloped market. Earlier studies examine the connection between risk and different valuation multiples. One example is Beaver and Morse (1978) who study the relationship between risk and the P/E multiple and find that risk, measured in terms of market beta, has low explanatory power of the observed persistence in P/E ratios. The argument is that stock earnings generally move together due to economy-wide factors. In years of low earnings, for instance due to financial downturns, the market-wide P/E will likely be high. The hypothesis

6.4 Discussion of assumptions and limitations 103

is that stocks with high betas will tend to have higher P/E ratios in these years because their earnings are more sensitive to economy-wide events. Vice versa, in years of generally high earnings, high beta stocks will have even lower P/E ratios than average. However, Beaver and Morse (1978) cannot confirm this hypothesis.

Another way to capture aspects of risk would be to include a measure of credit risk as selection variable. This would specifically capture risk related to a firm’s ability to pay its financial obligations in a timely manner and the probability of default. This can be done using appropriate financial ratios assessing financing structure, ability to repay debt obligations, liquidity reserves or short-term liquidity risk (Petersen et al., 2017). We already include one such financial ratio, Net debt/EBIT. Net debt/EBIT is a measure of risk that reveals if funds from operations are sufficient to repay debt.

Thus, we already, to some extent, capture credit risk. Nonetheless, it is possible that another financial ratio, or other measures of credit risk, could better capture valuation relevant credit risk and hence increase valuation accuracy. Credit risk can also be measured using statistical models. Two of the most commonly used are Altman’sZ-score or Ohlson’s logit model. These statistical models predict the probability of bankruptcy by returning a value that can be compared with a priori cut-off points, which then determines if a firm is considered to have a low or increased probability of default (Petersen et al., 2017). In our set-up, it would be relevant to use the returned value as selection variable.

All these proposed alternative measures suffer to some extent the same disadvantage compared to industry affiliation. While Size, Net debt/EBIT, and credit risk all say something about the idiosyncratic risk of a firm, they do not capture the full effects of how firms respond to more systematic risks, such as recessions or even pandemics. Different industries are not equally sensitive to cyclicality and black swans, such as the COVID-19 virus. The SARD method provides a snapshot of firms that are similar in terms of the applied selection variables. However, these firms may respond very differently to external shocks and cyclicality in general. Firms within the same industry are generally exposed to the same types of external risks and we expect them to respond similarly to systemic risks. While some industries, such

104 6.4 Discussion of assumptions and limitations

as the hospitality industry, loose nearly all revenue from an external shock such as the COVID-19 virus, other industries, such as the pharmaceutical industry, may be affected, but not to the same extent. Thus, an advantage of using industry affiliation as selection method is that it, to some extent, controls somewhat for differences in market risk and external shocks.

In addition to the variables tested in our study, we found in the literature review that the following variables have been used in similar research: Total assets and beta to approximate risk, ROA to approximate profitability, and revenue growth to capture the growth aspect. From conducted interviews we learned that professionals look at:

Sales growth, EBITDA margins, PPE/sales, intangible assets/sales, R&D/Sales, and capital expenditures/sales. A way to optimise our study would be to test some of these selection variables in different combinations.