Value Investing Revisited

This interview originally appeared in Morningstar Magazine.

Any way you cut it, value stocks have been out of favor for quite some time. From the end of 2006 through May 2020, the Russell 3000 Value Index lagged its growth counterpart by a staggering 5.9 percentage points annualized.

Is systematic value investing broken? Andrea Frazzini, principal and head of the global stock selection team at AQR Capital Management, and Andrew Ang, head of factor investing strategies at BlackRock, agreed to tackle the question and discuss their approaches to value.

We spoke on June 8, and valuation and performance data is as of that date. Our conversation has been edited for length and clarity.

Alex Bryan: Value investing is an intuitive strategy—buying stocks that are trading at low valuations makes sense. Why has value underperformed for so long?

Andrew Ang: I will start by emphasizing just how bad it has been. If we look at Fama/French data, which starts in 1925, this value drawdown we've had since January 2017 is the worst one, and it now has the longest duration.

Part of this is that value is inherently cyclical and tends to perform the worst during economic slowdowns and contraction—which is where we are today, and we’ve been in a late-stage environment for the past couple of years. All the other bad episodes for value—the 1930s, the early 1980s, the late 1990s—have also corresponded to very late-stage economic cycle contractions. Value tends to perform best coming out from the bottom of economic troughs. Value companies have operating leverage, which makes them very efficient, and they tend to do very well in the early parts of the cycle.

Andrea Frazzini: You are right, Alex, that value investing is fairly intuitive. When you have two companies with very similar cash flows or earnings or revenue into the future, and one of them is trading at a lower price relative to the other, you expect to have a higher return if you buy the cheaper company.

What can go wrong? Two things: Either your forecasts of future long-term cash flows are wrong or prices take a long time to converge—they can even diverge in the short term. Over the past few years, there’s not a lot of evidence that the fundamentals have been moving in a way that is extremely abnormal relative to history. But we have seen systematic evidence of prices really deviating from those fundamentals. That has led us to the recent situation where the spread in valuation ratios between cheap and expensive companies has reached historically high levels, levels that we haven’t seen since the peak of the tech bubble at the end of the 1990s.

There are lots of different ways of measuring valuation spreads. You can look at different types of value portfolios or different measures. You can adjust for industry. But the overall evidence shows that value stocks are very cheap. Over the past couple weeks, the spread has come down a bit, but even with that slight rebound for value, I would say it’s still in the ballpark of what we saw during the tech bubble.

Ang: Since the bottom of the stock market on March 23 to the close on June 5, we have seen quite a remarkable turnaround in value. The S&P 500 from that nadir of stock markets is up around 40%. Value has outpaced the market and is up more than 6% above momentum.

Is This Time Different?

Bryan: That gets back to the cyclical issue: A lot of value companies seem to do better in early parts of a recovery, such as after the recent coronavirus sell-off. But the past 13 years or so have been a challenging time for value. What are the structural issues playing into this?

Ang: Some of this is absolutely cyclical. We moved firmly into contraction in April. Value companies only manufacture one good or one type of service. They have inflexible business models and tend to underperform during a late economic cycle or entering into a contraction. But those economies of scale give them a leg up at the bottom, which accounts for some of the outperformance more recently.

To look at the structural issues, it’s informative to think about the other episodes where value performed badly. The late 1990s was the age of the Internet, which connected us all and completely changed our lives. The 1980s saw the rise of free trade and globalization, and it was the era of the personal computer. As for the 1930s, we remember the massive company failures and record unemployment, but it was a time of new growth, too. We had the rise of modern entertainment, of movies and radio. We use the term soap opera because Procter & Gamble PG started those in the 1930s as radio serials to sell consumer products, such as soap. Disney DIS and HP HPQ started during these downtimes as well.

What’s common to these periods is an acceleration in the decline of industries. In the Great Depression, it was things like textiles and leatherworking. Today, it’s old-fashioned retail. But we also see massive shifts in technological adoption or types of things that have increasing returns to scale. Today, we’re undergoing something similar.

The rise of intangible assets is economywide. In the 1990s, papers by Janice Eberly and other macroeconomists estimated intangibles at around 10% of total firm assets. In the late 2010s, this reached close to 15%–20%. Today, the fraction of intangible assets is over 40% in sectors like technology. It’s hard to think of a great company today that doesn’t have large amounts of intangible assets. But these are harder to estimate with traditional balance sheet and earnings statement information, so we have to start exploring different value metrics.

Bryan: The world has certainly changed over the past few decades in ways that critics of value investing would argue reduce its efficacy. There is the rise of intangibles. There are structural headwinds facing value-oriented sectors like energy. In financial services, you hear about the impact of low interest rates. The market is becoming more competitive as access to information becomes more widely disseminated. So, is this time really different?

Frazzini: Whether this time is different is very hard to test empirically. It's always possible that the world changed yesterday in a way that is completely unpredictable and not related to past history.

The world is, of course, different and constantly changing, which makes our job harder, but we haven’t seen any evidence that makes us reject that basic premise of value investing. Fundamentals still drive stock returns. What might have changed is the way you capture and measure those fundamentals. If intangibles are a larger part of the earnings-generating power of a firm today, it means, as Andrew rightly pointed out, that valuation metrics have to account for that.

At the same time, if you look at the value performance over the past few years, the underperformances have been widespread across a variety of ways of measuring value, from basic valuation ratios all the way to metrics that account for intangibles or metrics that are unaffected by intangibles. You can look at valuation metrics, price-based reversals, or empire-building type of metrics, they all underperform by a large amount.

The other component is industry adjustments. At AQR, we don’t make industry bets based on valuation metrics. When we think about whether a company is cheap or expensive, we always do that relative to industry peers. That helps control for a variety of structural exposures. Intangibles might be important, but they’re less important when looking at two companies that are in the same line of business.

Mispricing and Risk

Bryan: Prices seem to be more disconnected from fundamentals today than in the past. Is that primarily due to the difficulty of measuring those fundamentals, or is it the case that prices are not anchoring as much to what the true fundamentals are?

Frazzini: Both are possible. I can only test hypotheses with the data we have. What I don't have available is what the actual realized earnings five years from now will be. It's possible that the very expensive companies will realize such high earnings, revenue, and cash flows to justify prices today. I can't refute that possibility. But in history, on average, that has not happened. Valuation ratios tend to revert to the mean.

Ang: This goes back to the economic fundamentals of why buying value should result in a long-term—and the emphasis is on long-term—return. There are two economic rationales for value. The first is the reward for bearing risk. We may have short-term cyclical losses, but over the long run, we're compensated with a value risk premium.

There’s also a behavioral aspect of value. The behavioral story is that we get taken in by the high-flying growth stocks. We drive up those prices and ignore the staid, boring value companies. I think that’s happened for several years, up until now. Since we ignore them, those prices tend to be low, but the corresponding forward-looking expected returns tend to be high.

But during periods of extended value drawdowns—including today, the late 1990s, the 1980s, and the 1930s—there is a window of time where it is possible to generate increasing returns to scale that benefit some companies tremendously. Today, it’s streaming technology, machine learning, and massive data sets. These certainly lead to increasing returns to scale, as did using the Internet to reach consumers in the 1990s, global supply chains and computers in the 1980s, and mass entertainment in the 1930s. During these periods, we got both cyclical reactions and also sentiment changes, then behavioral reactions. Both contribute to the existence of a long-run value premium.

Bryan: What about the argument that the behavioral element is less pronounced today because more people are trading on value?

Frazzini: Value is an active strategy. For every dollar every value investor is overweight a cheap stock, there is someone out there who must be underweight that stock in the same dollar amount. The universe of active investors has to hold the market, by definition. So, valuation changes in the premium have to do with not just the amount of people tilting toward a particular factor, but also how large the other side is.

Having said that, active investing is a competitive business. I compete with Andrew, and we compete with everybody else. Therefore, you might see returns deteriorating, but that’s not special to value investing. That’s true for every active strategy. It’s normal that returns might not be as good now as they were 40 or 50 years ago.

You can have a risk-based explanation for value, meaning that cheap stocks are riskier in some sense, and therefore, you have to be compensated for holding them. Or a mispricing behavioral explanation that relies on investors who either have some biases or tune their portfolios in a particular systematic way. But both the risk-based explanation and the behavioral-based explanation always require somebody to be taking the other side. So, when we think about whether a value premium has deteriorated, we’re really thinking whether or not the number of growth investors is going up or down. And there’s not a lot of evidence that it’s shrinking toward zero.

Ang: This is a really important question on flows or crowding, and it leads to short- term cyclicality. We've recently moved to a slight overweight position in value, and we moved to an underweight position in momentum at the beginning of May. For most of 2019 and all of 2018, we were significantly underweight value. We never want to disinvest from a factor, but during those times, we were at the minimum weight.

We look at valuation and relative strength, which is a form of momentum that’s played against value over the last couple of years. And we also look at where we are in the business cycle. Our business cycle indicator by itself is neutral to value. Those indicators account for some of the larger positions we’ve put on to value over the past couple of weeks. We’re not significantly overweight, but there is a large risk of being underweight value. We’ve seen some of the snapback in value since the bottom of the COVID drawdown, and sometimes those snapbacks can offer great opportunities.

Does Timing Matter?

Bryan: Andrea, you seem a little bit more skeptical about trying to time exposure to value. What is the rationale behind always having some exposure to value?

Frazzini: We are fairly conservative about timing exposure to factors in our portfolios. We do a small amount of timing, and we do that in extreme circumstances. Like Andrew, we look at valuation ratios and relative strength, what we call factor momentum. We look at how cheap inexpensive companies are relative to expensive companies, as well as movement in those valuation ratios—whether those spreads are widening or tightening. It's a balance between the two.

We are currently overweight value. It's not a secret; [AQR co-founder] Cliff Asness recently put up a blog piece about it. As valuation ratios are extreme, we do have a value tilt. We don't have a very large tilt. It's a positive tilt that reflects how confident we are in our ability to time, which is not that much. We believe factor-timing is deceptively difficult and that investors are better off holding a well-diversified portfolio of factors.

Bryan: Are you using any type of cyclical economic indicators to time your exposure to value?

Frazzini: This is probably where Andrew and I disagree. This is not for lack of trying. We would like to be able to time those factors, but we haven't found data on macroeconomic exposures that lead us to include that into a timing model.

Bryan: Andrew, it seems intuitive that value stocks should do better during the early stages of an economic cycle, especially if they have more operating leverage, as you suggest. But can you actually tell in real time where we are in the business cycle?

Ang: A paper that we published on factor tilting in the Journal of Portfolio Management in 2017 argued that the economic regime was the stand-alone signal that had the highest predictability. But you're right, sometimes it's really hard to call business cycles. The National Bureau of Economic Research just announced today that it dates our current recession back to February. Ex post, even, it's a hard exercise. And ex ante, it's very difficult. That's why we like to use several indicators—economic regime, valuations, relative strength, dispersion, some measures of crowding—to inform our view, so we get some diversification across different indicators.

I'd like to go back to crowding and flows. There is an area of the market that I think does have significant crowding; our forthcoming research in the Journal of Investment Management looks at areas of crowding and what happens when there are more funds than stocks. In an analysis of crowding and style factors in individual equities, it turns out that one area that really does have significant crowding is large-cap growth. That's the complete opposite of where value is today.

Bryan: How has your view of value investing evolved over the past decade?

Frazzini: We constantly innovate our investment process—the measures that we use for value, refinements in portfolio construction, the way we think about the risk exposure, all these are areas where we've made changes in order to make our process as effective as possible. However, we have not changed our thinking on value. Buying companies that are trading at a cheap price relative to their fundamentals is associated with higher-than-average returns over the long run. Our recent contrarian tilt into it shows our confidence in that long-term evidence rather than the short-term performance.

Ang: I completely agree with Andrea. We are value investors at heart. We believe in the underlying economic concepts, and that gives us confidence that value will be around for decades to come. But to be better investors, we do want our research to evolve.

I'll give you three examples of my personal evolution. The first one is that growth is not the opposite of value. There are reasons why we might pay up for an expensive company. It might have high-quality earnings, or positive momentum, or stable returns. These aspects of being expensive aren't necessarily bad. Value versus growth doesn't necessarily mean cheap versus expensive.

Conditioning, as we touched on before, is another way we can better implement value. With intangible information, we might do this within and across sectors or industries, or around the world. Finally, we have dynamic notions of when we might want to tilt into value, such as today, or tilt out of value, such as the last two to three years.

Apples With Apples

Bryan: How can you implement value in a more thoughtful way than traditional value indexes, many of which use low growth as value signals? Andrea, can you talk more about sector neutrality?

Frazzini: As I mentioned, we don't take industry bets based on valuation metrics. We place all our bets within an industry. That means buying technology stocks that are cheap relative to technology stocks that are expensive, or utilities stocks that are cheap relative to utilities stocks that are expensive.

There’s a variety of reasons why that is reasonable. One is the empirical data. Historically, buying cheap securities within an industry has been a better bet than buying cheap industries and selling short expensive industries. You get a better Sharpe ratio and a better return by controlling for industry components.

There are also ex ante reasons. It’s extremely hard to compare a technology company to a utilities company or to a retail company. They have different business models. They might have different levels of cash flow, or leverage, or macroeconomic exposure. You would ideally compare companies that are selling an identical product. Adjusting for industry is a way to approximate that.

Another reason is macroeconomic exposure. If you make industry bets based on which companies are cheap versus expensive, you might end up having large macro exposures. An example would be interest-rate exposure, which has been talked about a lot recently in the financial press. Controlling for industry helps mute, if not eliminate, that exposure. From a risk perspective, it allows you to focus on what value investing is really about: companies trading at different prices relative to future fundamentals.

Bryan: Andrew, do you take a similar industry-relative approach to value?

Ang: We also like to compare apples with apples. We control for sectors, industries, and regions.

I’ll give some examples of the way we’ve evolved the value signals in our active portfolios. We talked about the importance of intangible assets. A lot of accounting rules tend to expense investments rather than capitalize them. To get a better read on intangible assets, we can undo some of those accounting transactions and try to capitalize some things that generally have been expensed.

One direct measure of intangibles is patents. They directly measure intellectual property and the ability of a company to innovate and offer new technologies and services. This is another way that we can capture intangible assets.

We’re actively researching peer groupings. Some of the most famous companies and the largest ones, like Amazon.com AMZN, have changed sectors just because the definitions of the sectors themselves have changed. We can find better peer groups for these companies. One way, potentially an extreme case, is to compare a company’s value metrics with its own historical value metrics—using a peer comparison of itself as a comparable.

Intangible value and traditional value signals sometimes go in the same direction, as for, say, Merck MRK. In some cases, like Ford F, the intangibles constitute a very small part. But in other companies, like Uber UBER or some of the technology names, we might see traditional value signals score negatively, but intangible signals or time series score positively. They can attenuate the impact of traditional value measures.

Man Versus Machine

Bryan: Fundamental stock-pickers who use a more qualitative approach have been trying for many decades to buy stocks for less than they're worth. Most of them have failed to meet their benchmarks. Why would a systematic approach to value investing work better?

Frazzini: It's a good question that gets at the heart of what we do. One thing to keep in mind is that value is a factor that we think you should be exposed to, but it's not the only factor. Even in periods where value is attractive, your portfolio should not be 100% value. Value should be a part of what you do, and maybe today a little bit bigger part.

Why would a systematic approach work better than a fundamental approach on individual companies? The approach is similar. Deep-value investors who hold a concentrated portfolio might know a lot more about one individual company. We replicate that process across thousands of companies, and that diversification helps. We may not be sure about a particular name, but we do know that on average, buying cheap securities relative to fundamentals can generate positive returns. We take a bet that prices will revert toward the fundamentals. We might do it across thousands of names. Fundamental investors might do it across a few names. But the process is similar in spirit.

Ang: There is a role for both systematic or factor investments as well as more discretionary or fundamental investments in a portfolio. They hit at different things. We believe in indexes plus factors plus alpha. Market-cap indexes are standard benchmarks for good reason: They're transparent and they can be implemented at extremely low cost. Factors, where Andrea and I live, are broad and persistently rewarded sources of returns in the long run, but there can be some short-term cyclicality. The tremendous value losses over the past two to three years is testament that these effects have been very broad in the wrong direction!

For the last component, alpha, what we really mean is returns in excess of what you can get from indexing and factors. A fundamental investor can concentrate on just a handful of names in a given sector or country and know everything about them—the analyst might walk a factory floor or be able to interpret emotion on an earnings call. That’s very specialized knowledge. Or you could generate alpha by sophisticated machine-learning techniques that require massive amounts of data or infrastructure and specialized skills.

There’s room for indexing and factors and alpha, and we need all three more than ever in today’s environment.

Bryan: Thank you both. Value investing is front and center for a lot of investors we work with, and your perspective is a helpful one.

Disclosure: Morningstar, Inc. licenses indexes to financial institutions as the tracking indexes for investable products, such as exchange-traded funds, sponsored by the financial institution. The license fee for such use is paid by the sponsoring financial institution based mainly on the total assets of the investable product. Please click here for a list of investable products that track or have tracked a Morningstar index. Neither Morningstar, Inc. nor its investment management division markets, sells, or makes any representations regarding the advisability of investing in any investable product that tracks a Morningstar index.