This In Practice piece gives a practitioner’s perspective on the article “Stick to the Fundamentals and Discover Your Peers,” by Jens Overgaard Knudsen, Simon Kold, and Thomas Plenborg, published in the Third Quarter 2017 issue of the Financial Analysts Journal.
What’s the Investment Issue?
In assessing whether a company is over- or undervalued, analysts typically compare it with other companies in the same industry sector, on the assumption that industry peers share similar economic characteristics in terms of profitability, growth, and risk. But diversity among companies in the same industry groups may mean that apparently comparable companies are actually not similar at all and can trade at permanently different valuations.This study examines whether eschewing industry classifications and instead finding peers with similar fundamental value drivers is a more accurate way of assessing value. It also investigates whether creating peer groups with similar fundamental drivers and similar industry classifications could produce even greater value.
How Do the Authors Tackle the Issue?
Comparing companies on the basis of fundamentals rather than industry classification is not new. Other approaches measure profitability, growth, and risk, but those methodologies allow for the selection of peers on the basis of only two variables, which limits the usefulness of these measures. Adding more variables would serve only to cut the shortlist of comparable companies down to a very small number.
A more useful approach is to apply the “sum of absolute rank differences” (SARD) in order to better calibrate comparable companies. SARD allows for an infinite number of variables—that is, proxies for profitability, growth, and risk—which increases the chances of finding close-matching peers. Each company is ranked against the other companies in the sample according to a set of variables.
The authors identified five selection variables. Return on equity (ROE) was used as a substitute for profitability, net debt/EBIT (earnings before interest and taxes) was a surrogate for risk, and analyst forecasts of earnings represented future growth. They also included size—measured by market capitalisation to better characterise risk. A fifth variable, EBIT margin, was used because it has proved to be a significant determinant of the enterprise value (EV)/sales multiple.
Comparable companies were ranked according to the “least sum rank difference” across these variables—that is, they were chosen according to the similarity of their fundamentals to those of the target company. If a potential peer has a low SARD value, the potential peer and the target company can be viewed as similar for purposes of analyst valuations.
The companies with the lowest SARD scores were further ranked against key valuation multiples: EV/EBIT, EV/sales, price/book, and price/earnings (P/E). Numerous combinations of these variables were assessed, leading to the identification of the six most comparable companies for each target company—six being the minimum number the authors believed necessary to minimise error.Finally, the authors tested the SARD approach against the traditional industry classification method and also tested it within industry sectors to see if the accuracy of sector-specific analysis could be improved.
What Are the Findings?
The SARD approach produces substantially more accurate valuation estimates than the traditional industry classification approach, and adding more variables further improves the valuation accuracy.
For example, ROE appears to be the worst-performing variable for the EV/EBIT multiple, with a median valuation error of 0.292. But accuracy improves as more variables are added, and the valuation error falls to 0.222 when all five variables are applied. The results are similar for the three other multiples—the progressive inclusion of extra variables increases the accuracy of the valuation.
Combining the SARD and industry classification approaches results in greater valuation accuracy still—that is, the SARD approach produces an incremental increase in accuracy when used within industries. In contrast, identifying potential peers by industry classification alone produces the least accurate valuation estimates.
For example, the median valuation error of the EV/EBIT multiple is 0.255 when peers are selected on the basis of industry classification alone. But selecting peers in the same industry on the basis of ROE reduces the valuation error to 0.244. By adding net debt/EBIT, size, growth, and EBIT margin as selection variables, the valuation error decreases to just 0.203.The findings were found to be consistent across time (1995–2014), company size, and most industries and with varying numbers of comparable companies.
What Are the Implications for Investors and Investment Managers?
The SARD method challenges investment practitioners to throw out industry classifications as a tool for selecting peer groups of stocks for comparative analysis.
Two advantages of the SARD approach are that it is not restricted in the number of variables that can be used and that it is independent of industry classifications. So, although this study focuses on US markets, the SARD approach should also be applicable to non-US markets, including smaller markets where the testable universe may be limited.
Another advantage is that the SARD method offers investment professionals flexibility: It can be used in combination with other approaches, including the industry classification approach.
In addition, the SARD variables can be tailored to any multiple to fit different strategies and investment processes. Finally, the SARD approach is intuitive in that it is based on a set of financial value drivers that are well recognized and traditionally used in financial analysis worldwide.A challenge that fund managers could face in implementing the SARD approach is that classifying peers according to industry allows for longer time series, because industry composition changes only slowly. That is not the case with the SARD approach. Because financials change every quarter, peer status could be short-lived, making historical comparisons less meaningful.