The Discipline of Value Investing

Value has a long history as an investing style, backed up by empirical evidence that portfolios of the cheapest stocks outperform the broad market. The strategy behind value investing is simple: buy stocks with a low price relative to their current financial metrics like earnings, EBITDA or cash flow. The value of the company is sum of the current and future economic surplus. Estimating future surplus starts with current metrics like earnings or cash flow, so using the most recent financial information against the market valuation is a good indicator of the relative cheapness of a stock. Stocks with very strong valuations on current earnings may come with lowered expectations on future earnings, which is why value investors are often called contrarian. Growth investors emphasize strong future earnings as the investment opportunity, creating the two investing styles of “Value” and “Growth”, instead of “Cheap” and “Expensive”.

value2016-image-1

The historical long-term effectiveness of value investing is compelling. Table 1 shows the excess returns for a number of valuation metrics within the U.S. Large Stocks universe, stocks trading in the U.S. with a market capitalization greater than average from 1964 to 2015. Each valuation factor separates winners from losers, where the more expensive a stock is the worse it is for the investor. This works whether it is expensive on trailing operating metrics like Sales, Earnings, EBITDA or Free Cash Flow, or on the return of capital through dividends and repurchases measured by Shareholder Yield.

Individually, each of these metrics do a good job of identifying companies that are attractive valuations, but they are not perfect predictors of outperformance. One example is the difference between measuring companies based on sales versus earnings. Using Sales as a financial metric is beneficial because it is very consistent across different companies and tough to manipulate. The downside is revenue does not capture all the costs of doing business: cost of goods sold, sales, general and administrative costs, depreciation of capital equipment, obligations to debtholders or the government. Earnings are a more comprehensive financial measure of the economic output, but earnings are calculated using company-specific accounting assumptions, incorporating inconsistencies when measuring the cheapness of one company against another.

One way to improve on valuation is to combine the multiples into a single measure, ensuring broad cheapness across a number of metrics. This avoids investing in a company that may look attractive on one metric, but expensive on another. In that same example of using sales and earnings, sometimes a company may look very cheap on the top line, but very expensive on the bottom, or vice-versa. Table 2 shows the historical returns of stocks based on how they rank across those two metrics: Sales-to-Price and Earnings-to-Price. The right panel shows the percentage of companies within each bivariate grouping, and most companies show alignment where the company looks expensive or cheap on either metric. It’s less common, but sometimes a company will fall into the cheapest quintile on one metric, and the most expensive on the other. A simple static combination of metrics like Sales-to-Price, Earnings-to-Price, EBITDA-to-EV, Free Cash Flow-to-EV and Shareholder Yield can create a Value Composite that does a better job of separating winners from losers than any single metric.

Table 1: Historical Excess Returns by Value Factor, Large Stocks 1964-2016
Table 1: Historical Excess Returns by Value Factor, Large Stocks 1964-2Q2016
Table 2: Bivariate Returns and Observations, Large Stocks 1964-2016
Table 2: Bivariate Returns and Observations, Large Stocks 1964-2Q2016

To show this in practice, we can look at six stocks from the GICS Retailing Industry Group within the Consumer Discretionary sector, as of March 31st of this year. Focusing on a basic valuation comparison of Amazon.com versus Macy’s, Amazon had almost four times the sales that Macy’s did, at $107bn versus $27bn, but Macy’s actually made almost double in earnings, $1.07 billion versus $596 million for amazon. But Amazon’s market cap is $256 billion, versus $13billion for Macy’s, over 19 times more expensive. For the ratios, Amazon is over five times as expensive on Sales and EBITDA, and thirty-seven times as expensive on earnings. Amazon is going to have to grow a lot to justify the higher valuation. To set scale of how much growth the company needs, Amazon already has a good deal of market share with almost 15% of all retailing revenue. If Amazon were to capture the entire $726 billion all retailing revenue with the same profit margin, it would still have a P/E ratio of 63. Perhaps instead of growing revenue, it can grow earnings by raising prices and improve margins. Keeping revenue the same, and moving from 0.5% margin to the retailing median of about 6.2%, it would still have a P/E of 38. To get to a market-level P/E in the low 20’s, it would have to both revenue and margins.

This is a very basic analysis, and an informed bull perspective on Amazon could argue several points to why Amazon is a good investment; Amazon.com is growing its core retailing business as well as businesses like web services and media. The point of this post is not to debate whether an investment in Amazon is a good idea. We are exploring the point of view of a Value investor, and to a Value investor the stock looks expensive. The question is whether Amazon is a good investment, and sometimes great companies are bad investments. Investing is about taking chances, and maximizing your odds of success, and Value investing is about avoiding stocks that priced for a perfection they might never meet.

Macy’s is at the opposite end of the expectations spectrum. Recent news highlights significant hurdles in their business with pressure to maintain revenue and earnings. Again, the exercise is not to defend Macy’s, but the type of investment it represents. Macy’s is priced for extremely low expectations and all the company has to justify its valuation is maintain existing earnings and the market should bring the stock back to a reasonable valuation. Add in that Amazon is diluting shareholders by one percent in the last twelve months, versus Macy’s which is returning capital through dividends and share repurchases at a rate of twelve percent, and you get a complete picture of why Macy’s looks attractive to a value investor.

Table 3: Valuation Ratios as of March 31st, 2016
Table 3: Valuation Ratios as of March 31st, 2016

It is a fair question to ask whether you can compare Amazon.com and Macy’s, given their different business models. They are both retailers, but Amazon is an internet retailer. GICS classification has them in the same Industry Group, but separates them at the Industry level into Internet Retail and Multiline Retail. The differences in cost structure and client base raises the question of whether it makes more sense to rank on a relative valuation within the GICS groupings.

As with many things in portfolio management, going relative is a trade: you are comparing stocks of like cost structures against one another, but you are potentially moving yourself into a more expensive area of the market. In Table 4 below, you can see how this trade works: moving relative gives up return and lowers tracking error. In practice, there are constraints working with a smaller universe of stocks, a limit to how far you can make that trade. The Large Stocks Universe today has only has about 750 names for investment, is that when you slice the universe into smaller and smaller groupings, there are fewer and fewer stocks to compare. There are 62 distinct industries, 42 of which have fewer than ten names. In the Large Stocks universe, the Industries of Internet Retail and Multiline Retail, which contain Amazon.com and Macy’s, have only six names each. Pretty hard to split into deciles at that point.

Table 4: excess returns Value Composite, Large Stocks 1964-2016
Table 4: excess returns Value Composite, Large Stocks 1964-2Q2016

 

Table 5: number of companies in Large Stocks June 2016
Table 5: number of companies in Large Stocks June 2016

It’s one thing to go through the academic exercise of researching value, where the analysis is done over very long periods of time, and a completely different thing to use Valuation to invest in stocks every day. The returns we’ve been looking at so far are from 1964-2015, over fifty years. Most investors don’t have a fifty year investment horizon.  One way to look at the effectiveness over shorter investment horizons is through Base Rates: how often the strategy is outperforming the average stock in the universe on a rolling investment horizon.  Table 6 shows the base rates for the best decile of Value factors within the Large Stocks Universe.  Even though investing in the best decile of a composite of value factors averages out to have excess returns of almost four percent annualized, when looking at shorter investment periods it only works a little better than two out of three years on a one-year basis. On a quarterly basis, it is only a little better than average.

Table 6: Base Rates of Best Decile within Large Stocks, 1964-2016
Table 6: Base Rates of Best Decile within Large Stocks, 1964-2Q2016

The second quarter of 2016 is a good example of a period where Value managers struggled. According to the eVestment database, only 14% of Large Cap Value managers beat the Russell 1000 Value on a gross basis, and very few beat it by a large margin. This is following up a tough fourth quarter of 2015. Combined, the market has created the worst 1-year investing environment for Large Value managers since the dot-com crisis ended in 1999. Chart 2 shows a universe of active Large Cap Value mutual funds built from the CRSP Mutual Fund database, and it shows that the average gross underperformance of the portfolios over the last four quarters has been -4.4%, the worst twelve months that active Large Cap Value managers have had since the dot-com crisis.

Chart 1: eVestment Universe Large Value Managers, 2Q2016
Chart 1: eVestment Universe Large Value Managers, 2Q2016
Chart 2: Rolling 12-Month Gross Excess Returns of Large Value Managers, CRSP Mutual Fund Database
Chart 2: Rolling 12-Month Gross Excess Returns of Large Value Managers, CRSP Mutual Fund Database

Having a strong track record didn’t help:  this environment punished those managers with the most success over the previous three years. Looking at both the eVestment and CRSP Mutual Fund databases, the managers that did the worst for the trailing twelve months were those with the best 3-year track record as of June 30th, 2015.

Table 7: 1-Year Performance of Managers based on trailing 3-year Quartile Ranking
Table 7: 1-Year Performance of Managers based on trailing 3-year Quartile Ranking

When an active manager underperforms, there are questions about whether something is broken in the investment process. Was there a change in how you select stocks? Has there been a change in portfolio managers? Was there been style drift? When decomposing what happened in the second quarter, the answer is that Value stocks were penalized heavily over the last twelve months. The following chart shows that the best decile of stocks based on the value composite have underperformed the average investment by over -11% over the last twelve months. In historical context, this underperformance is abnormally large: over the last fifty years, only 2% of the time the Value Composite has had underperformance of -10% or more.

Chart 3: Rolling 12-Month Excess Performance Value Composite, Large Stocks, 1964-2016
Chart 3: Rolling 12-Month Excess Performance Value Composite, Large Stocks, 1964-2Q2016

The statement “Value stocks underperformed” is a bit detached. What does that really mean? Retail is a great example, where multi-line retail stocks struggled in the second quarter. Macy’s was the bellweather, posting a -7% decline in sales for the quarter versus the quarter from the year before. This contrasted with Amazon.com, which did post sales growth of 28%. The narrative became that big-box retailers are going to be put out of business by the internet, and in the Russell 1000 Multi-line Retailers went down -8% while Internet Retailers were up 10.3%.

The Energy Sector is another great example for the quarter. Every Large Cap Value manager was facing a choice of which Energy stocks to pick.  How to allocate among subindustries of a) Integrated Oil companies, like Exxon, and b) Exploration & Production companies that dig the oil out of the earth or c) refiners that buy oil, process it and resell it. Oil had declined about 75% from the middle of 2014. When diving into the valuation ratios based on trailing earnings and free cash flow, the energy sector offered a choice was between E&P and Integrated oil companies that had sustained large drops in their earnings, and Refiners who had an earnings yield close to 12%, and had seen an uptick in earnings.

Table 8: Financial Metrics for Subindustry, March 31st, 2016
Table 8: Financial Metrics for Subindustry, March 31st, 2016

Hindsight bias is a strong phenomenon, because it seems obvious now that oil prices could go on a rally and improve 100%, moving from a February low of $26, to $48 at the end of the quarter. It’s hard to remember that back in February news articles were speculating oil could drop to $15 as we running out of places to store it. The future price of oil was unknown, and in that uncertainty Value managers looking the companies with strong earnings and free cash flow chose Refiners over E&P and Integrated oil.

In the second quarter, the Russell 1000 Value was up 4.6%. Energy stocks did very well and was up 10.8%,but the split within the subindustries was large. Integrated companies were up 12.3%, and companies focused just on E&P were up 17%, benefitting from the higher oil prices. Refiners received the opposite effect from the upswing in oil, and were down almost -11% for the quarter.  Looking across the holdings from the CRSP Mutual Fund database, we can see that the mutual fund managers that underperformed the most had the largest underweights to energy, with most of the underweight coming from the Integrated and E&P subindustries. The managers that underperformed the most had an overweight to refiners.

Table 9: 2Q2016 returns by GICS grouping
Table 9: 2Q2016 returns by GICS grouping

 

Table 10: Mutual Funds by Performance, Allocation Effect from Energy 2Q2016
Table 10: Mutual Funds by Performance, Allocation Effect from Energy 2Q2016

The reason Value managers underperformed was not because they were stupid, they underperformed because they kept to their investing style. And those with the strongest conviction on value investing were punished the most. Using the CRSP Mutual Fund database again, the mutual funds with the cheapest holdings had the worst performance. This holds across metrics based on Earnings, EBITDA or Free Cash Flow: the cheapest 20% of the universe by any of these metrics were the ones that underperformed the most.

Table 11: trailing 12-month performance of Mutual Funds by Value Metrics, June 2016, source CRSP Mutual Fund Database
Table 11: trailing 12-month performance of Mutual Funds by Value Metrics, June 2016, source CRSP Mutual Fund Database

The strategy of Value investing is to buy an asset for less than it is worth and benefit when the market corrects the pricing mistake. The deep historical evidence suggests that indeed there are often large gaps between the market’s perception of a company (the price it sets) and the reality (the current and future state of that company’s fundamentals like sales and earnings growth). Find stocks where expectations are wrong, avoid the stocks where those expectations are too high, and buy those where they are too low. The key to implementing the strategy successfully is separating your investment process from short-term results: to have discipline. Discpline ties good research to good long-term investment outcomes. Research looks broadly at the market, theorizing investment strategies and looking at large historical datasets to find trends that work more often than not. Investing is putting money into individual stocks and living with the daily outcomes of your choices. Investing is buying Macy’s and Oil Refiners in the second quarter, and making sure the underperformance does not affect your investment philosophy.  One bad quarter does not mean that the investment process is broken, that it was a mistake to invest in Macy’s or Oil Refiners or Value stocks in general. The way to make a mistake is to abandon your philosophy when challenged.

When is a “Value” Company not a Value?

Value has broadly been accepted as an investing style, and historically portfolios formed on cheap valuations outperformed expensive portfolios.  But value comes in many flavors, and the factors(s) you choose to measure cheapness can determine your long-term success.  In particular, several operating metrics of value, like Earnings and EBITDA, have outperformed the more traditional Price-to-Book ratio.  A possible reason for the limited effectiveness of P/B is because of the increase in shareholder transactions, primarily through the increase in share repurchases.

Valuation ratios have the benefit of being simple, but can also have flaws.  Sales-to-Price has the benefit of measuring against revenue which is tough to manipulate, but doesn’t take margins into account.  Price-to-Earnings measures against the estimated economic output of the company, but also contains estimated expenses which can be manipulated by managers.  EBITDA-to-Enterprise-Value has the benefit of including operating cost structures, but misses out on payments to bondholders and the government.  Even with these flaws, the ratios are effective in practice.  Historically, portfolios formed on cheap valuations outperformed expensive portfolios.  The following charts show the quintile spreads two ratios within a universe of Large U.S. Stocks, stocks with a market cap greater than average, from 1964-2015 [1].  Earnings/Price, or Earnings Yield, generates a spread of 5.1% between the best and worst quintile, and EBITDA/EV generates a 6.0% spread.

Quintile Spreads for Earnings-to-Price and EBITDA-EV
Quintile Spreads for Earnings-to-Price and EBITDA-EV, 1964-2015

Book-to-Price is perhaps the most widely used valuation metric in the investing industry.  Russell, the top provider of style indices for the U.S. market, uses the metric as its primary metric to separate stocks into Value and Growth categories.  They use B/P in combination with forecasted 2-year growth and historical 5-year sales per share growth, but Book-to-Price is the chief determinant at 50% of the methodology.  Their choice of Price-to-Book most likely comes from its long history in academic research.  The seminal work on Book-to-Price was the 1992 Fama-French paper “The Cross-Section of Expected Stock Returns”, which established the 3-Factor model of Market, Size and Book-to-Price.

But when you start looking at the metric of Book-to-Price, a few issues start to become apparent.  First, the overall spread on the factor isn’t as strong as operating metrics like Earnings and EBITDA:  the spread between the best and worst quintile is only 2.8%, versus 5.1% for E/P and 6% for EBITDA/EV.

Quintile Spreads on Book-to-Price

Second, when breaking down the effectiveness of the factor based on market capitalization, Book-to-Price is least effective with the largest cap stocks.  The following chart shows the same quintile spreads of B/P in the Large U.S. Stocks universe, but separates out the smallest and biggest largest third based on market cap.  Book-to-Price degrades in effectiveness as you move up the market cap range, with the quintile spread within the largest third of stocks only at 1.2%.  This is especially noteworthy because Russell market-cap weights their benchmark, and about two-thirds of the benchmark is in that top-third by market capitalization.

B/P Quintile Spreads by Market Cap Tertile
B/P Quintile Spreads by Market Cap Tertile

Last, the effectiveness of Book-to-Price has been waning, especially since the turn of the century.  The following chart shows the rolling 20-year quintile spread, the difference between the two portfolios of the cheapest 20% and most expensive 20%.  For Book-to-Price against EBITDA/EV and Earnings-to-Price, you can see how all three metrics behaved very similarly before 2000.  They had generated consistent outperformance until being inverted in the dot-com bubble of the late 1990’s, where the most expensive stocks outperformed.  But coming out of the internet bubble, Book-to-Price has started behaving differently than other valuation ratios, degrading to the point where for the last twenty years it has had almost no discernible benefit on stock selection.

Rolling 20-Year Quintile Spread in Large Stocks
Rolling 20-Year Quintile Spread in Large Stocks, 1984-2015

On the surface, using book value in relation to price makes intuitive sense.  The book value of equity is the total amount the common equity shareholders would receive in liquidation, the accounting value of the total assets minus total liabilities and preferred equity.  The P/B ratio is meant as a quick measure to see how cheaply you could acquire the company.  The ratio will move around based on changes in either the market value or book value of equity.  But the ratio comes with assumptions.  “Clean surplus accounting” is based on the assumption that equity only increases (or decreases) from the earnings (or losses) in excess of dividends.  In practice, there is another influence on equity:  transactions with shareholders.

When a company repurchases shares, the market effect is straightforward.  The number of shares outstanding are reduced while the price remains the same, so the market capitalization goes down.  When accounting for the share buybacks for financial reporting, the repurchase of shares does not create an asset as if the company had repurchased equity in another company.  Instead, the equity value is decreased by the amount spent in purchasing the shares.

As a hypothetical example, take a company with a $200m market cap, $100m in book value of equity, and $10m in earnings.  The company has a P/E ratio of 20, and a P/B ratio of 2.

If that company becomes an aggressive repurchaser, and decides to acquire $50m worth of its own equity, it will alter the ratios significantly.  The earnings remain the same, but the market cap goes down, and the P/E will adjust down to 15.  But the P/B ratio will be reduced on both the top and bottom of the ratio, and it will actually increase to 3.

As a practical example, Viacom has been aggressively repurchasing its own shares since separating from CBS in 2006, spending almost $20bn over the last ten years.  In 2015 alone, it repurchased about $1.4bn in shares.  So even though the company has been seeing retained earnings of about $1.5bn per year, its common equity has reduced from $8bn to $4bn over that same time frame. [2]

Historical Financial Metrics for Viacom
Historical Financial Metrics for Viacom

You can see how this distorts valuation ratios:  Viacom trades at a significant discount on earnings versus the median P/E for other Large Stocks, while looking like it trades at significant premium on the book value of equity.

Historical Valuation Ratios for Viacom
Historical Valuation Ratios for Viacom

A company issuing shares will have the reverse effect:  the company will actually increase its book value, even though the earnings and cash flows are diluted across more investors.  Any transaction for a company through the issuance or reduction of equity, flows through the book value of the equity.

The following table compares median valuation ratios for companies with a market capitalization greater than average.  Two groups are compared with the median Large Stock:  those companies that have repurchased the most shares over the last 5 years, and those that have issued the most shares.  The top 25 companies repurchasing their shares have better operating valuation metrics (i.e. Sales, Earnings, EBITDA, FCF) than the median, and the top 25 diluters have worse ratios, with the standout exception of Price/Book.   Repurchasers have an average Price/Book of 4.5, almost 20% higher than the median 3.8, while Diluters look cheap with a P/B of only 2.7, an apparent discount of almost -30%. [3]

Valuation Ratios for Large Stocks by Share Activity
Valuation Ratios for Large Stocks by Share Activity

This distortion means using Price-to-Book could lead to misclassifications of stocks as a Value investment.  Stocks that are cheap on operating metrics like Sales, EBITDA or Earnings could wind up classified as Growth.  On the flip side, that universe could include a company that has issued a lot of stock and has inflated its book value of equity.  This is something to keep in mind, as a number of quantitative managers start with the benchmark as their universe, and starting with the Russell 1000 Value could bias you towards a number of companies that look cheap on Price-to-Book, but are not cheap on other metrics.

Over the last fifty years, there has been a gradual increase in the amount of company equity transactions.  In particular, larger companies have been increasing their share repurchase activity.  In classifying companies based on a trailing 5-year change in shares outstanding, we can see which companies have consolidated shares by more than 5%, issued shares more than 5%, or have been relatively inactive.  In 1982, the United States loosened regulation around the company’s restrictions for repurchasing shares, and there has been a marked increase in activity. This has led to a change in the overall market, where the percentage of companies inactive has been reduced from almost 60% in the 1960’s, down to around 28%, with the activity mainly being driven from companies consolidating shares. [4]

Large Stocks by Share Activity

Large Stocks Share Activity by Decade
Large Stocks Share Activity by Decade

This begs the question whether the gradual increase in shareholder transactions has resulted in the gradual ineffectiveness of Book/Price as a valuation factor.  The first rule in analysis is not to confuse correlation with causation, but the rolling 20-years when Price-to-Book has been less effective coincides pretty well with the increase in shareholder transaction activity.  Price-to-Book is also the least effective in the largest cap stocks, which have the largest volume of dollars affecting book value of equity.  Perhaps the most interesting analysis is looking at the effectiveness of Price/Book within those Large Stocks that have been relatively inactive with shareholders over a trailing 5-year period, versus those that have been active, either on issuance or repurchase.  Looking since the 1982 the legislation change, there is a different effectiveness of valuation metrics between companies active or inactive with shareholders.  If your investments are focused on companies with share issuance or repurchase activity, there has been no relative benefit to buying companies that look cheap on P/B, and there’s almost no difference between high and low valuations.  But if limiting to companies that are relatively inactive, you can get a spread of 6.4% between the best and worst 20% based on the B/P ratio.  Using another valuation metric, like EBITDA/EV, works well independently of independent of a company’s activity in issuing or repurchasing shares.

Even with the long-term degradation of returns from Book-to-Price, it is possible that Book-to-Price will revert to an effective investment factor.  Book-to-Price has been off to a strong start in 2016 and is outperforming other valuation factors, particularly in small cap stocks.  But there are structural challenges to the factor, and before you use it you should be aware of the embedded noise from repurchases that could mislead you.

Book-to-Price Quintiles by Share Activity, 1983-2015
Book-to-Price Quintiles by Share Activity, 1983-2015
EBITDA/EV Quintiles by Share Activity, 1983-2015
EBITDA/EV Quintiles by Share Activity, 1983-2015


Footnotes

[1] Quintile portfolios are formed on the “Large Stocks” universe, stocks in Compustat with a market capitalization greater than average, rebalanced every month with a holding period of one-year.
[2] Compustat used as source for the Viacom Data
[3] Compustat source used for Russell 1000V constituents, as of May 31st, 2016
[4] Large Stocks universe, with Compustat as source for Share Repurchases