I get a sense that I am approaching the end of this series of weekly posts, or perhaps I am just hoping that it is true, as the COVID crisis continued to play out in markets in the last two weeks, albeit on a more subdued scale and with a more positive twist. In this post, I will, as in the prior weeks, update the prior weeks’ market action (for two weeks, from April 4 to April 17) in different asset classes, and within equities, across regions, sectors and stock classifications. I will close this post by looking at how pricing tools, including a range of multiples (from PE ratios to price to book to EV to EBITDA multiples) will become shakier and less reliable in the aftermath of the crisis, and suggest ways in which we can compensate for the uncertainties.
I started my crisis clock on February 14, reflecting the fact that I am US-based, and markets outside China did not wake up to the crisis until that week. In the weeks since, we have seen volatility rise and equity markets get whipsawed, with much of the pain being dispensed in February and the first three weeks in March. In the last month, equity indices around the world have seen positive returns, and in some cases, very good positive returns, as can be seen in the table below.
The week of April 10 to April 17 was a benign week, at least in sum, even though individual days still brought big movements, with most indices flat for the week. Moving on to the treasury market, we also saw steadiness, with both short and long term rates staying close to the lows that they hit just two weeks ago.
Looking at commodity prices, the divergence between oil and copper illustrated again the unique travails of oil, where a detente between Russia and China did little to stop oil prices from continuing to drop, while copper prices changed little.
In the two months since February 14, oil prices have dropped more than 65%, providing a contrast to copper, another commodity sensitive to the global economy, which has declined less than 10%. (To top of the craziness, the price of futures on Texas crude dropped below zero on April 20, but that is a story for another day/post.)
To complete the picture, I looked at gold and bitcoin, and while both have settled into holding patterns, the divergence since February 14 is stark.
In sum, gold has held its own, increasing 6.3% since February 14, though investors holding it were undoubtedly expecting a bigger pop, given the economic and market chaos, but bitcoin has disappointed those who believed it would play the role of a crisis asset, down 31% since the start of this crisis.
In my last post, I focused on how the price of risk has changed since February 14, 2020, starting with the corporate bond market, where default spreads have changed significantly over the last few weeks.
Note the surge in default spreads across bond ratings classes from February 14, 2020 to March 3, 2020, though there has been a drop off from highs in the last two weeks. The fear that has played out in the bond market has also affected the price of risk in equity markets. In the graph below, I updated the equity risk premiums, by day, that I computed in my post two weeks ago, through April 17.
The implied equity risk premium which I computed to be 6.01% on April 1, 2020, has declined to 6.27% if I compute the risk premium using the (now stale) earnings and cash flows, and 5.60%, if I assume a 30% drop in S&P 500 earnings this year and a substantial drop in buybacks. I have a feeling that this roller coaster ride is not quite done and I will continue to estimate the numbers daily.
In keeping with my practice in my prior posts, I looked at market capitalizations of all publicly traded companies (36,481 companies with market caps exceeding $5 million on February 14, 2020) and started by computing changes in market capitalization, by region:
Looking at the aggregated returns since February 14, 2020, the worst performing regions in the world are Africa and Latin America, and China remains the standout as the best performing market. Australian and Canadian stocks have been punished, largely because of their natural resource focus, and globally, stocks have lost $15.2 trillion in value, a huge amount but about half as large as the loss was four weeks ago. In the next table, I break market cap changes down by sector:
There are few surprises in this table, with healthcare and consumer staples being the best performers, and energy and financial services the worst. Breaking down the sectors into finer detail, I look at companies classified into 95 industries, and list the ten best and worst performers over the crisis period (2/14 - 4/17):
I know that it's fashionable to talk about how inefficient and volatile markets are but this crisis, in many ways, has been surprisingly orderly and markets have dispensed punishment judiciously, for the most part.
I also looked at other classifications, from pricing levels (PE and PBV) to momentum to dividends/buybacks and found no significant differences across companies. In fact, the evidence seems to more strongly support the notion that the market is punishing low PE, high dividend yield stocks that had little momentum coming into this crisis more than high PE , non-dividend paying stocks. That is disappointing news for value purists who have been waiting a long time to say "I told you so" to momentum and growth investors. In fact, the only variable that seems to offer support for financial moralists is financial leverage, as can be seen in the table below, where I break down global stocks based upon how much debt they had at the start of the crisis:
The most highly levered companies, with leverage measured as debt scaled to EBITDA, have suffered more damage as this crisis has played out.
In my earlier posts, I argued that just because uncertainty has increased, there is no excuse for abandoning valuation first principles or process; you can still value companies, albeit with a much wider range of outcomes. One common counter that I got to this argument was that valuation is pointless when the uncertainty is so great and while most of those marking this argument did not bother presenting alternatives, my guess is that many will fall back on pricing metrics to decide what to buy or sell. Put simply, they will use a PE ratio or an enterprise value multiple of EBITDA or sales to decide what stocks to buy or sell, acting under the delusion that this will allow them to escape having to make assumptions in the future. In this section, I will start by breaking down pricing multiples and then use simple valuation algebra to argue that there are assumptions about cash flows, growth and risk embedded in every pricing multiple. I will close by noting how multiples behave in a crisis, and report on pricing multiples, broken down by region, sector and industry, pre and post crisis.
Anatomy of a Multiple
I think of multiples as standardized prices, allowing investors to get past the challenge of comparing per share values, which are determined by share count. That said, it is easy to be overwhelmed by the number of multiples you see in practice, with some in wide use (PE, Price to Book, EV to EBITDA) and some obscure (EV per subscriber, EV to Invested capital). To put these in perspective, I will start by breaking down the choices that you have in constructing a multiple and using it to make a pricing judgment:
The numerator for any pricing multiple is a market value of equity, firm or operating assets, and the denominator is a scaling variable: revenues, earnings, cash flows or book value. There is no one "best" multiple or timing choice, since that will vary across time and across sectors, but here are two simple consistency rules to keep in mind, when constructing and using multiples:
- Equity/Firm: If the numerator is an equity value, the denominator should be an equity value as well, while if the numerator is a firm or enterprise value, the denominator has to be an operating value. Thus, PE (market cap, an equity value, is divided by earnings per share, an equity value) and EV to EBITDA (EV is a market value of operating assets and EBITDA is a measure of operating cash flow) are consistent, but Price to EBITDA is an inconsistent abomination and Price to Sales is almost as badly constructed.
- Timing: Multiples are constructed for comparisons across companies, not as stand alone measures. It follows therefore that you should be consistent in the timing you use for your scaling variable (revenues, book value, earnings) across companies. Thus, if you choose to use trailing earnings for your company to compute PE, you have to use trailing earnings for all your companies.
Put simply, in pricing, you estimate a value for a business or its equity, based upon how "similar" companies (equities) are being priced in the market place.
Determinants of Multiples
Many analysts who use multiples to find under and over priced stocks do so because they do not want to confront the uncertainty associated with forecasting future growth, margins and cash flows in intrinsic valuation. That is an illusion, since embedded in every multiple are assumptions about growth, risk and investment efficiency. When you pay a hundred times earnings for a stock or ten times book value, you are assuming high growth in earnings for the former and a monstrously high return on equity for the latter. In the picture below, I list out the fundamentals that are embedded in each multiple:
This cheat sheet, designed to find the variables that are embedded in a multiple, brings home a reality about pricing that should make anyone using it uncomfortable. The difference between intrinsic value, where you try (sometimes desperately) to forecast future growth and cash flows, and pricing, where you use a multiple, is that you are explicit about your assumptions in the future, making them both more transparent and easier to critique, and that you are implicit in your assumptions with the latter, making them easier to defend but also more dangerous.
Pricing and Crisis - A Time Line
In the aftermath of every crisis, investors abandon fealty to fundamentals, on the premise that they are in unique times and fall back on pricing. I am sure that will happen with this one as well, but if you decide to go this route, the nature of this crisis will make pricing much more difficult. To see why, take a look at how multiples will move as this crisis unfolds:
- In phase one of this crisis, the market reacts to the crisis by marking down stock prices almost immediately, but the scaling variable (revenues, earnings, book value) does not, partly because it takes time for the crisis to show up in operating numbers and even longer for accountants to record that in the financial statements. Consequently, as the crisis first unfolds, stocks will look cheaper on a trailing basis, as the market price drops and earnings/revenues/cashflows stay stagnant, and analysts/companies are too uncertain to offer guidance about future operating results.
- In phase two, analysts and companies start to provide forward guidance, and you can switch to forward values, if you trust them, but since the crisis can cause more companies to lose money, you will also see a greater dependence on revenue multiples in pricing.
- In phase three, as operating results more completely reflect crisis effects, trailing multiples will reflect the updated operating results, but you should not be surprised to see companies trade at much higher multiples of trailing earnings (if earnings are still positive), or have earnings multiples that are not meaningful. Again, a naive comparison of the trailing PE to historic norms will lead you to conclude that everything is over priced, even when that is not the case.
No matter which phase you are in, you ultimately have to make judgments about whether the company will come out of the crisis, and if it does, what it will generate as earnings, to make sensible investment decisions. Just as there is no room for lazy and mechanistic valuation, in the midst of a crisis, there is no payoff to lazy and mechanistic pricing.
We are still in phase one of this crisis, though we are hopefully approaching its tail end. Not surprisingly, as market prices have dropped and trailing operating numbers reflect what companies did in 2019 (pre-crisis), there has been a drop in trailing pricing multiples across all regions of the world:
The same story unfolds across different sectors:
In some sectors, such as financial services, energy and airlines, where the punishment meted out by the market has been severe, you should not be surprised to see stocks trade at extraordinarily low multiples of trailing earnings. At the same time, companies have been reluctant to offer guidance for the coming year, making it difficult to shift to forward values. You could, of course, get ahead of the curve and try to forecast earnings in a post-virus world, say in 2022 or even 2025, and scale market capitalizations to those values.
As companies start to report their first quarter earnings, you are starting to get a glimpse of the damage created by the crisis and my guess is that you will start to see more analysts and companies start to forecast forward numbers. For those companies where forward earnings are positive, you can switch to forward PE ratios, but expect these numbers to be much, much higher than historical norms. For those companies that expect losses in the next year, you will see revenue multiples or creative variations on future earnings, from earnings before COVID to earnings in 2025 used as the scalar. Later this year, as companies report numbers for the second and third quarters of 2020, the trailing operating numbers will finally catch up with the crisis, but they will come with caveats. Put simply, if you are abandoning or refusing to do intrinsic valuation, because you feel uncomfortable with having to make assumptions, the same uncertainty is going to pervade your pricing as well.
Related: Make America Go Back to Work Again?