Relative Valuation - Part 2

When the subject of relative valuation analysis is broached, one of the first questions that is usually put forward is what ratio will give the most 'correct' estimate of value. Even the most novice investor has heard of at least half a dozen financial valuation ratios, and sophisticated investors will throw around hundreds at a time. The answer is not a simple one and depends on what the analysis is being used for and what data you have access to. Usually, your analysis will triangulate between several different ratios in order to overcome some of the idiosyncratic issues of one measure. 

I have spoken in previous posts about incentives - and the incentives to sell a stock to a client will normally deliver you a set of ratios that may not be useful if you were on the other side. Ratios are not always complementary - they can sometimes even tell you a completely different story. This is why the definition step is so important - if we know what goes into the ratio, then we can ensure that we are asking the right question the next time someone proffers us an investment based on some good looking ratios. The second point about data availability is a less important one, but necessary nonetheless. If a firm is not yet mature, burning through cash but generating good growth, then you may not be able to compute some of the better known ratios that are applied to mature firms. Is P/E for example, the best measure of a firm that is yet to generate positive earnings? Sure you can forecast earnings ahead and then apply a forward looking P/E to future EPS but is the forecast error worth the effort?

In my last post I spoke about some of the critical steps of relative (multiple) valuation analysis. Although they may appear somewhat intuitive, it is worth spending a little more time to ensure that we get the basics right! 

Just like in intrinsic valuation analysis, multiples can be used to calculate enterprise value or the market value of equity. If you are a trader, chances are your focus is on the market value of equity, and if you are interested in purchasing the business, then your interest may be geared to the former. The definition step of relative valuation analysis is about ensuring that your multiple is defined in such a way what that the output reflects either either equity or enterprise value. 

What's in a definition?

If we take the most popular ratio we know, Price / Earnings, it is clear that this is a measure of the firm's outstanding equity. Price, as we all know, is a function of the market value of equity, derived from the expected cash flows to equity holders. In the denominator - earnings (Net Income / Shares Outstanding or EPS) are measured after interest payments (to debtholders) and taxes. It is therefore, an accounting proxy of the cash flow available to equity holders. At a very basic level, the price / earnings is consistently defined. 

If I chose P / EBITDA in my valuation analysis, is this measure consistently defined? The numerator, unchanged, reflects the market value of equity. In this ratio, however, EBITDA includes interest payments to be distributed to debtholders. If we have two identical firms, in all respects other than the second firm has a substantially higher proportion of debt, then what your analysis will show is that the second firm is comparatively undervalued! Why? Think about how increasing your proportion of debt affects EBITDA. Relative to price, a more geared firm will generate a comparatively undervalued (but incorrect) signal.

There are other similarly critical aspects of the definition test that we need to think about. We know that EPS (when looking at P/E) is not always defined in the same way. We have forward and backward looking EPS, normal and diluted EPS, and EPS before and after abnormals. If we are comparing P/E's across different countries, then we also need to think about the fact that different accounting standards will not always produce the same result. There is a lot to think about here, which is why one should never rely on P/E's provided by data providers. These are what I like to call indicative P/E's - your analysis needs to get knee deep and ensure that if you chose diluted EPS after abnormals, that the comparative firms in your sample are treated in the same way.

The Analysis

I mentioned in the last post that the interpretation of the output is often misunderstood even by the people who perform this valuation analysis day in and out. This failure is largely the result of our neglect for basic statistical lessons. For example, statistics 101 says that if we have a distribution of values that is skewed in any one direction, then standard mean variance properties cannot be used to describe the 'average' result. If I have the value 4, 5, 6, 5, 100, then is the average closer to 24 or 5. It obviously depends on the what these values represent, but most people in this situation would select 5. In multiple analysis, we often deal with two sorts of problems of inference. 

Firstly, if you take a ratio like P/E, you know that the lowest value possible is 0 (because price can't be less than 0) and the highest value is infinity. Naturally, your values are going to be skewed, especially if your sample mixes growth and mature stocks. Therefore, we need to consider other more reliable outcomes - the median / mode. While these are not perfect they are better than focusing on the average. An alternative is that if your analysis is not normally distributed, then to make it so. This means that you will have to identify the outliers in your sample and drop them from your analysis (some people use log-distributions to overcome this problem but I don't really see the point). If a firm has $0.01 worth of earnings and its stock price is 100, then is this ratio really informative. If you have good reason to drop these stocks from your sample, and the distribution becomes normal as a result, then going back to the average is fine. 

The Drivers

Every multiple contains a driver (or drivers). These drivers can be related to growth, risk, productivity, the firm's capital structure, capital intensity etc... There are normally many drivers at work that will affect whether a firm appears cheap or expensive and unless you can relate the former to these drivers then making any inferences will produce erroneous results. Implicit in every valuation analysis are certain assumptions. Unlike DCF or intrinsic valuation methodologies these assumptions are not explicit, rather they are determined by the market and the sample of firms in your analysis. We can therefore learn plenty by stepping up our analysis and deconstructing these multiples to learn what makes a firm cheap or expensive. 

Going back to the P/E multiple - we should know that P / E = Price / EPS. We can restate this equation using the DDM model because we know that Price = DPS(1) / r - g. If we take this equation and divide both sides by EPS then we end up with the following equation 

                                                              P/E = Payout Ratio*(1+g) / r-g. 

Why did we do this?

Let's look at the result. The last stated equation says that P/E is driven by the payout ratio, growth, and the discount rate. The payout ratio is a function of a firm's cash flows. If theory is to be believed then there should be a positive relationship between growth, the payout ratio and the P/E and a negative relationship between risk and the P/E ratio (you'll have to take my word for the moment that it is but i'll see if i can produce some pictures in the near term to prove my point). What this tells us is that any conclusion about whether a firm is cheap or expensive needs to go hand is hand with an assessment of the firm's risk, payout, and growth ratio.

We need however, to be careful that we don't focus on the drivers individually but consider them as a whole. Why? Generally, a high growth firm will require higher re-investment needs, so you are unlikely to see the growth and payout ratio (in this case) moving in the same direction. If that is the case how do we interpret our P/E result? There is no simple formula to solve this problem. We need to apply judgement to the problem and assess the firm within the context of its operating environment.The same kind of analysis can be applied to any number of ratios from EV/EBITDA to the PEG ratio. We should always have an idea what makes our ratios look cheap or expensive.

Sector of Market Multiples

The final issue that I'd like to elaborate on is whether one should choose sector or market multiples. You may not think so, but you know the answer to this question already. If you think that markets make mistakes on individual firms but that the valuations on the sector are on average correct then you will use sector averages. If you think that sectors are mispriced, but on average that the market is priced correctly then you will use market-wide multiples. We know that not all sectors move together. Take the Australian agriculture sector of late - which is surging on the back of growing worldwide demand. Stocks like GrainCorp and Rural Co have increased over 25% YTD. In any relative valuation, you should triangulate between sector and market multiples but don't sit on the fence in terms of the weightings you assign. Weigh up the issues, the benefits, and costs and run with it!