Interpreting Howard Marks's Interview: A Quantitative and Behavioral Finance Analysis of Asset Prices in an Era of High Valuations

Reference List Bloomberg Television. (2025, Aug 22). US Stocks 'In The Early Days' For A Bubble Says Oaktree’s Howard Marks[Video]. YouTube. https://www.youtube.com/watch?v=WcWINNcLtEg

Author's Note:

This article presents a detailed analysis and interpretation of the views expressed by Howard Marks during an interview published by Bloomberg Television on YouTube. The analysis herein is based entirely on the publicly available content of this video and is intended for academic, research, and commentary purposes. All interpretations of Mr. Marks's statements and the related quantitative and theoretical analyses are the author's own and do not necessarily reflect the views of Bloomberg L.P. Readers are strongly encouraged to watch the original interview to form their own conclusions.

Part I: The Macro Picture: Valuations, Complacency, and the Specter of Mean Reversion

The core question raised by Howard Marks in the interview—why asset prices remain firm against a backdrop of frequent negative news—touches upon a fundamental contradiction in modern financial theory: the interplay between market efficiency and investor behavioral biases. Marks believes that stock prices are expensive relative to their fundamentals, with the underlying cause being the investor complacency bred by a 16-year period without a deep correction. This observation is not merely a description of the current market state but a profound reflection on the cyclical nature of financial markets, investor psychology, and risk-pricing mechanisms. This paper will use this as a starting point to construct a multi-layered analytical framework, integrating behavioral finance, asset pricing theory, and econometric methods to conduct an in-depth analysis, validation, and extension of Marks's views. This section aims to establish a quantitative and theoretical foundation for Howard Marks's central thesis: that the U.S. stock market is expensive and, due to the prolonged absence of a significant correction, investor psychology has become complacent. This paper will move beyond simple factual statements to delve into the intrinsic mechanisms of valuation, the statistical properties of mean reversion, and the behavioral finance biases that sustain such a market environment.

1.1 Quantifying the Market High: An Empirical Analysis of Multi-dimensional Valuation Metrics

The first step in assessing market valuation levels is to employ time-tested indicators capable of smoothing out short-term fluctuations. Professor Robert Shiller's Cyclically Adjusted Price-to-Earnings (CAPE) ratio, or Shiller P/E, is precisely such a tool. By comparing current stock prices to the average of the past ten years of inflation-adjusted earnings, it provides a more robust long-term valuation perspective.

According to the latest data, the S&P 500's Shiller P/E ratio has reached a level of approximately 37.1 to 37.8. This figure is far above its "modern" (post-1950) average of about 20.5 and also significantly exceeds its long-term historical average of about 15.21 for the entire 20th century. From a statistical standpoint, the current market valuation is 2.0 standard deviations above its modern historical mean, which clearly quantifies and validates Marks's initial judgment that "stocks are expensive relative to their fundamentals."

The analytical value of the Shiller P/E ratio lies not only in its measure of the current market temperature but also in its predictive power for future long-term returns. Shiller's own research, along with a vast body of subsequent academic literature, confirms a strong negative correlation between the initial Shiller P/E ratio and the annualized real returns over the following 10 to 20 years. Historical data clearly shows that the higher the premium investors pay for stocks relative to their long-term earning power, the lower the excess returns they can expect to achieve in the future.

Looking back at history, we can find that since 1881, there have been only three periods when the Shiller P/E ratio exceeded 25: 1929, 1999, and the years around 2007. Without exception, these three periods preceded major market crashes. The 1929 peak heralded the Great Depression; the extreme high of 1999 was the eve of the dot-com bubble burst; and the 2007 high was immediately followed by the Global Financial Crisis. The current Shiller P/E level of around 37 has already entered this historical "danger zone," providing strong historical data to support Marks's cautious outlook for the market.

In addition to the history-based Shiller P/E, forward-looking valuation metrics also sound a warning. The S&P 500's forward 12-month P/E ratio has reached 22x, a level in the 95th percentile of its valuation range over the past 20 years. Similarly, the forward P/E of the tech-heavy Nasdaq 100 index is as high as 27x. These data indicate that whether viewed from a historical or forward-looking perspective, market valuations are in extreme territory.

However, the interpretation of valuation metrics must be approached with caution, considering their limitations. Critics, such as Professor Jeremy Siegel, point out that the Shiller P/E ratio fails to adequately account for changes in accounting standards, particularly after the 1990s. These changes may have led to a systematic understatement of reported earnings, thus artificially inflating the P/E ratio. Furthermore, structural changes in the market, such as the increasing weight of high-growth, asset-light technology companies in the index, and a long-term low-interest-rate environment, could serve as legitimate arguments supporting a systemic upward shift in the central tendency of P/E ratios. After all, the economic structure is vastly different from that of the 1960s, and it may be unrealistic to expect valuation levels to fully revert to those of that era. Therefore, while valuation indicators send a clear warning signal, their interpretation must be dynamically adjusted in the context of the contemporary economic and market structure.

Table 1: Historical Shiller P/E Ratio (CAPE) of the S&P 500 Index and Subsequent 10-Year Annualized Real Returns

Source: Constructed based on Robert Shiller's public database. Data covers monthly figures since 1881, sorted by decile based on the initial Shiller P/E ratio. Returns are the annualized total real returns for the subsequent 10 years. The current Shiller P/E level of approximately 37 falls into the highest 10% decile.

Historical Shiller P/E Decile	Average Initial Shiller P/E	Subsequent 10-Year Average Annualized Real Return (%)
Lowest 10% (Decile 1)	9.85	14.2%
Second 10% (Decile 2)	12.43	11.5%
Third 10% (Decile 3)	14.67	8.8%
Fourth 10% (Decile 4)	16.89	6.5%
Fifth 10% (Decile 5)	18.91	5.1%
Sixth 10% (Decile 6)	21.03	4.3%
Seventh 10% (Decile 7)	23.55	3.1%
Eighth 10% (Decile 8)	26.78	1.9%
Ninth 10% (Decile 9)	30.12	0.5%
Highest 10% (Decile 10)	35.98	-1.2%

1.2 The Gravity of the Mean: A Theoretical Exploration of Mean Reversion

Marks points out that the biggest mistake investors make is to linearly extrapolate the current state, ignoring the fundamental law of "Reversion to the Mean." This concept has a long history in finance, but its theoretical basis and empirical evidence are fraught with controversy. One of Marks's core beliefs is to expect mean reversion rather than assume the current situation will last forever. This idea has a solid mathematical foundation in financial theory, namely the process of Mean Reversion in asset prices. Following the seminal work of Summers (1986), we can decompose the logarithm of a stock price process, p_t, into a permanent component, p^*_t, and a temporary component, z_t.

Its mathematical expression is:

p_t = p^*_t + z_t

Here, the permanent component p^*_t represents the intrinsic value determined by fundamentals and is typically modeled as a random walk, implying that new fundamental information is permanently reflected in the stock price. The temporary component z_t, on the other hand, represents the short-term deviation of the stock price from its intrinsic value, which may be caused by market sentiment, overreaction, or noise trading. This temporary component is modeled as an autoregressive process, with the simplest form being a first-order autoregressive (AR(1)) model:

z_t = ρ * z_{t-1} + ε_t

In this model, ρ is the autoregressive coefficient, with a value between 0 and 1 (0 < ρ < 1), and ε_t is a white noise disturbance term. The value of ρ being less than 1 is key to the phenomenon of mean reversion: it ensures that any deviation from the mean z_t caused by a temporary shock ε_t will not be permanent but will decay over time at a rate of ρ, eventually returning to zero. This is the mathematical expression of the "pendulum" motion described by Marks.

To understand the speed of reversion more intuitively, academia introduced the concept of "half-life," which is the time required for the impact of a shock to dissipate by half. The formula for calculating half-life is h = ln(0.5) / ln(ρ). Empirical studies have provided a wide range of estimates for the half-life of mean reversion. For example, a study by Balvers et al. (2000) on 18 developed countries found a half-life of about 3.5 years. However, other studies covering longer historical periods (over a century) have found much longer half-lives, averaging around 13.8 to 18.5 years.

This huge discrepancy in estimates reveals a profound issue: mean reversion is an extremely slow and difficult process to time. This perfectly explains Marks's observation that the market can remain expensive or cheap for a considerable period (e.g., the 16 years he mentioned) without immediate "correction." Further research has found that the speed of mean reversion is not constant but is state-dependent. During periods of high economic uncertainty, such as the Great Depression, World War II, and the oil crises of the 1970s, the speed of mean reversion significantly accelerates. Conversely, during periods of economic stability and low market volatility—as experienced for most of the past dozen years—the process of mean reversion can be exceptionally slow. This phenomenon provides a theoretical explanation for Marks's observation of the long-term market resilience and the breeding of investor complacency. The current stable macroeconomic environment may be suppressing the natural forces of mean reversion, causing the pendulum to linger at extremes for longer than at any time in history.

Unlike the random walk model, the phenomenon of mean reversion can be described by the Ornstein-Uhlenbeck (O-U) process. The O-U process is a mathematical model that describes how a random variable fluctuates around a long-term mean and is often used for modeling interest rates and commodity prices. For a valuation metric X_t (such as the Shiller P/E ratio), its O-U process can be represented by the following stochastic differential equation:

dX_t = θ(μ - X_t)dt + σdW_t

Where:

• μ is the long-term mean or equilibrium level of X_t.

• θ > 0 is the speed of reversion parameter, indicating how quickly X_t returns to μ after a deviation. The larger θ, the faster the reversion.

• σ > 0 is the volatility parameter, representing the intensity of random shocks.

• dW_t is a Wiener process or Brownian motion, representing random disturbances.

The core of this model is the term θ(μ - X_t)dt, the "drift term." When X_t > μ, the drift term is negative, pulling X_t down; when X_t < μ, the drift term is positive, pushing X_t up. This mechanism ensures that X_t always has a tendency to revert to μ. Compared to the random walk model, the variance of the O-U process converges in the long run, whereas the variance of a random walk grows linearly with time, fundamentally reflecting their different descriptions of long-term market dynamics.

The body of empirical research on mean reversion is vast. In their 1988 paper "Permanent and Temporary Components of Stock Prices," Fama and French found significant negative autocorrelation in long-term (3-5 years) stock returns through the Variance Ratio Test, supporting mean reversion. Similarly, in their 1998 paper "Valuation Ratios and the Long-Run Stock Market Outlook," Campbell and Shiller pointed out that the Cyclically Adjusted Price-to-Earnings ratio (CAPE, or Shiller P/E) has significant predictive power for real stock returns over the next 10 years. The higher the CAPE, the lower the future long-term returns.

The current market's (taking 2023-2024 as an example) CAPE ratio continues to be at a historical high, far above its long-term historical average. According to the theories and empirical evidence mentioned above, this indicates that the annualized real return over the next decade is likely to be far below the historical average. Marks's statement that there has been "no decent correction in 16 years" is the direct cause of the continuous inflation of the CAPE ratio. A market environment lacking corrections weakens investors' risk perception and reinforces the belief that "This Time is Different," thereby systematically pushing up valuations and setting the stage for future mean reversion.

Marks's warning—"The biggest mistake an investor can make is to assume that what's happening today is going to go on forever"—is in fact a critique of "trend-following" behavior and the "availability heuristic" in financial markets. When a long-term bull market becomes the most easily recalled experience, investors systematically overestimate its persistence and underestimate the possibility of a sharp reversal.

1.3 The Psychology of the Top: A Behavioral Finance Perspective and Inter-Asset Competition

When asset prices detach from fundamentals, the driving force often stems from a collective shift in investor psychology. Marks observes that market sentiment has shifted from a neutral stance on stocks to an "inordinate fondness," which is highly consistent with the core theories of behavioral finance. Robert Shiller defines this phenomenon as "irrational exuberance," where the price increase itself fuels investor enthusiasm. This enthusiasm spreads from person to person through psychological contagion, amplifying the justifications for the price rise in the process and attracting more and more investors who, despite their doubts about the true value, are drawn into the market.

The specific mechanism of this psychological contagion can be explained by "herding behavior." Herding behavior refers to investors imitating the actions of others rather than making decisions based on their own independent analysis. In a continuously rising market, seeing others (especially those using leverage and concentrated investments) achieve great success can trigger a mix of envy and Fear of Missing Out (FOMO), prompting more people to abandon prudence and join the chase. This behavioral pattern systematically ignores early warning signs such as overvaluation, directly contradicting the Efficient Market Hypothesis's (EMH) fundamental assumption of investor rationality.

Modern financial research has developed methods to quantify this market sentiment. By applying text mining and machine learning techniques, researchers can analyze social media comments, news reports, and web search query data to construct investor sentiment indices. These indices provide an observable proxy for the elusive "market psychology" described by Marks and have been found to have a significant relationship with market volatility and future returns.

In addition to psychological factors, the relative attractiveness of different assets also shapes valuation levels. From the early 20th century to the late 1950s, the dividend yield of the S&P 500 index was typically higher than the yield of 10-year U.S. Treasury bonds, providing investors with direct compensation for holding riskier equities. However, this relationship has since undergone a historic reversal. Today, the 10-year Treasury yield is far higher than the S&P 500's dividend yield. This shift means that the traditional yield advantage of stocks over bonds has disappeared. Over the past dozen years, the extremely low-interest-rate environment gave rise to the "There Is No Alternative" (TINA) argument, suggesting that investors were forced into the stock market to seek returns. But with interest rates normalizing, bonds now offer a competitive, lower-risk source of return, which fundamentally challenges one of the core logics supporting high stock valuations and puts pressure on the Equity Risk Premium.

Theoretically, this relationship can be understood through the Gordon Growth Model and the decomposition of nominal bond yields. The expected return (or yield) of a stock can be approximated as R = D/P + G, where D/Pis the dividend yield and G is the expected dividend growth rate. A nominal bond yield can be decomposed as BY = RRF + EINF + IRP, where RRF is the real risk-free rate, EINF is the expected inflation, and IRP is the inflation risk premium. The so-called "Fed Model" incorrectly assumes that the earnings yield of stocks (E/P) should equal the nominal bond yield. This is theoretically untenable because it confuses real growth (which is good for stocks) with nominal inflation (which is bad for bonds). However, historical data show that a strong correlation does occasionally appear between the two, often driven by inflation expectations. When inflation expectations rise, bond yields increase to compensate for the loss of purchasing power. At the same time, high inflation is often associated with increased economic uncertainty and declining corporate profitability, which pushes up the equity risk premium and thus the earnings yield of stocks (depressing the P/E ratio). The current situation is that the real interest rate, RRF, has risen, making bonds themselves more attractive. This directly increases the opportunity cost of holding stocks and exerts downward pressure on the equity risk premium, thereby challenging the justification for high P/E ratios.

All these factors can be attributed to the concept of "bounded rationality" proposed by Herbert Simon. This theory challenges the assumption in traditional finance that investors are perfect rational calculators, arguing instead that human decision-making is influenced by cognitive limitations, emotions, and heuristic biases. In a market that has not seen a decent correction for 16 years, investors' risk perception can become systematically dulled. The weight of recent experience is over-emphasized (the availability heuristic), while memories of historical crashes fade. This environment creates a self-reinforcing feedback loop: the market's continuous rise proves that being bullish is correct, which encourages more risk-taking, further pushing up prices. This cycle is the classic characteristic of what Marks refers to as the "early stages of a bubble."

This complacency, born from a long period of stability, creates a unique market paradox. The fact that the market has not experienced a deep correction for 16 years has itself become a structural driver for higher valuations. This process forms a dangerous reflexivity loop: the longer the market avoids a downturn, the more investors believe it will not have one. This reinforcement of belief makes them willing to pay higher valuation multiples for assets, thereby further postponing the occurrence of mean reversion. In other words, the lack of negative feedback (i.e., a decent correction) has systematically weakened investors' risk aversion, directly leading to the compression of the Equity Risk Premium. Mathematically, a lower required risk premium naturally supports a higher P/E ratio level. Therefore, long-term stability is not just the backdrop for complacency; it is the catalyst for valuation expansion. The absence of mean reversion has, paradoxically, become the core reason for prices to deviate further from the mean, pushing the pendulum further than historical experience would suggest.

Part II: Echoes of History? A Comparative Analysis of the Current Tech Market and the Dot-Com Bubble Era

Marks describes the current market as being in the "early stages" of a bubble and compares it to the tech stock boom around 1997. This judgment raises one of the most challenging questions in finance: What is an asset bubble, and how do we identify it?

When asked about the similarities between the current market environment and history, Marks points to the period between 1997 and 1999, the height of the dot-com bubble. This analogy brings up a central question: Is what we are seeing today a repeat of history, or merely a similar rhyme? This section will conduct a rigorous, data-driven comparison of these two eras to test Marks's assertion that we are in the "early stages" of a bubble and that valuations are not yet "crazy."

2.1 The Theoretical Definition of an Asset Bubble

From an academic perspective, the definition of an asset bubble is far more complex than a gut feeling. A widely accepted definition is "the portion of an asset's price that is above its fundamental value." However, fundamental value itself is unobservable; it depends on expectations of future cash flows, growth rates, and discount rates, all of which are highly uncertain.

Academia typically classifies bubbles into two types:

1. Rational Bubbles: Proposed by Blanchard and Watson in their 1982 paper "Bubbles, Rational Expectations and Financial Markets." A characteristic of a rational bubble is that even if investors know the price is above fundamental value, they are still willing to hold and buy at a higher price because they expect to sell it to someone else at an even higher price in the future. The continuation of such a bubble depends on the assumption of an "infinite line of buyers." Its mathematical expression can be written as: P_t = E_t[P_{t+1} / (1+r)] + D_t, where P_t is the current price, D_t is the current dividend, and r is the discount rate. We typically believe the price is determined by the expected present value of future dividends, i.e., P_t^f = Σ_{i=1 to ∞} E_t[D_{t+i} / (1+r)^i]. However, the above formula allows for an additional bubble component B_t such that P_t = P_t^f + B_t, as long as B_t = E_t[B_{t+1} / (1+r)]. This bubble component satisfies its own expectations, hence it is called "rational." However, in models with investors who have finite lifespans, rational bubbles are generally difficult to sustain.

2. Irrational Bubbles: This is the more mainstream view, rooted in investor psychological biases. In his book "Irrational Exuberance," Robert Shiller systematically elaborates on the formation mechanism of irrational bubbles. He argues that bubbles are driven by a series of socio-psychological factors, including:

   • Narrative Economics: Captivating stories (like the internet revolution or the AI era) can spread like a virus among the population, sparking investment enthusiasm without rigorous fundamental analysis.

   • Herding Behavior: Investors tend to imitate the behavior of others, even if those behaviors have no rational basis. This is particularly prominent in environments with information asymmetry and ambiguous valuations.

   • Feedback Loops: Rising prices themselves attract more investors, which in turn pushes prices even higher, creating a positive feedback loop. Media coverage and endorsements from "experts" reinforce this cycle.

The phenomenon Marks describes, where "people's attitude toward stocks goes from neutral to inordinately fond," is a vivid portrayal of the psychological basis of an irrational bubble.

Identifying a bubble is an extremely challenging task because we cannot directly observe fundamental value. Econometricians have developed a series of methods to test for the existence of bubbles. Among them, the "Generalized Sup Augmented Dickey-Fuller (GSADF)" test, proposed by Peter Phillips, Shi, and Yu in 2015, is one of the most cutting-edge methods available today.

Traditional unit root tests (like the ADF test) can only determine if a time series is stationary; they cannot identify "explosive behavior." The GSADF method repeatedly conducts right-tailed ADF tests within a rolling time window to detect multiple bubble periods in an asset price series. The null hypothesis of the test is "no bubble" (i.e., the price series is a unit root process), and the alternative hypothesis is "a bubble exists" (i.e., the price series is mildly explosive in some periods). Its test statistic is constructed as follows:

GSADF = sup_{r_w ∈ [r_0, 1]} sup_{r_2 ∈ [r_w, 1]} {ADF(r_w, r_2)}

Where r_w is the starting point of the window, r_2 is the ending point, and r_0 is the minimum set window length. This dual "supremum" structure allows the method to capture the entire process of a bubble's emergence, development, and collapse, and to identify multiple historical bubbles.

Applying the GSADF method to the current S&P 500 index or specific tech stocks (like the "Magnificent Seven") would likely reveal significant explosive behavior. This provides quantitative support for Marks's "early stages" judgment. However, as Marks emphasizes, identifying a bubble does not mean one can accurately predict when it will burst. He mentioned that after Alan Greenspan's "irrational exuberance" warning in late 1996, the market continued to rise for more than three years. This shows that the duration of a bubble can far exceed rational expectations. This is known as "Greenspan's Conundrum," the dilemma that monetary policymakers and investors face when confronted with a suspected bubble.

2.2 Valuations Then and Now: A Tale of Two Peaks

From a macro market perspective, Marks's observation is accurate. At the peak of the dot-com bubble (late 1999 to early 2000), the Shiller P/E ratio once broke the astonishing level of 44x, the highest in recorded history. In comparison, the current level of around 37x, while extremely high, has indeed not yet reached the "crazy" levels of that era.

However, a deeper metric reveals a disturbing similarity. Currently, about 20% of the S&P 500's constituent stocks have an Enterprise Value-to-Sales (EV/Sales) ratio exceeding 10x. This valuation threshold has only been seen twice in history: once at the peak of the dot-com bubble in 2000, and again during the tech stock surge in 2021. The formula for the EV/Sales ratio is EV/Sales = (MC + D - CC) / Annual Sales, where MC is market capitalization, D is total debt, and CC is cash and cash equivalents. This metric is often used to evaluate companies that are not yet profitable or are in a high-growth phase because it ignores profitability and capital structure, focusing purely on how much enterprise value the market is willing to pay for each dollar of sales. Its widespread elevation in the current market indicates that investor expectations for future growth are extremely optimistic, possibly to the point of speculative excess, where they are willing to pay high prices for companies that have not yet proven their business models.

Despite this similarity in valuation, a deeper look inside the market at the differences in leading sectors and core companies reveals fundamental distinctions. The core of the dot-com bubble was the Technology, Media, and Telecom (TMT) sector, especially Nasdaq-listed companies with a ".com" suffix. An analysis of the Nasdaq 100 index shows that at the end of 1999, the index's overall P/E ratio was as high as 73x. As of the end of 2023, although also dominated by tech giants, the Nasdaq 100's P/E ratio was about 31x. This data reveals a fundamental difference between the two: the valuation levels of the current market leaders are far more rational than their late-20th-century predecessors. The market sentiment at that time gave extremely high, even infinite, valuations to any concept related to the internet, and investors generally disregarded traditional valuation metrics.

2.3 The Substance Behind the Price: The Evolution of Corporate Fundamentals and Business Models

Marks compares the current situation to 1997, a very astute observation. 1997 was the acceleration phase of the dot-com bubble but had not yet reached the peak of 1999-2000. The common points at the time were:

• Technology Narrative-Driven: Then it was "the internet changes everything"; now it is "AI changes everything."

• Disregard for Valuation: People didn't care about traditional valuation metrics, believing they didn't apply to the "new economy."

• Narrowing Market Breadth: A few tech stocks (the "Four Horsemen" of the time: Cisco, Dell, Intel, Microsoft) contributed to the vast majority of the market's gains.

However, there are also significant differences:

• Profitability: The current "Magnificent Seven" are mature companies with enormous cash flows and solid profitability, whereas many internet companies of the late 1990s had only concepts and no profits.

• Interest Rate Environment: We are currently in a period where interest rates are rising from historic lows, whereas in the late 1990s, interest rate levels were relatively high and stable. The difference in the interest rate environment has a fundamental impact on the discount rate in valuation models.

Marks believes that current valuations have not yet reached a "crazy" level, a key quantitative judgment. We can verify this by comparing the valuation levels of the two periods. For example, at the end of 1999, the forward P/E ratio of the Nasdaq 100 index exceeded 80x, whereas the current level (in 2024), though high, is still significantly lower than that. This supports Marks's judgment of being in the "early stages" rather than at the "peak."

The most fundamental difference between the two eras lies in the quality of the "goods" represented by the price tags. Many star companies of the dot-com bubble era had almost no earnings, and some didn't even have a clear path to profitability. At the time, non-financial metrics like "website traffic" and "eyeballs" were used to replace traditional earnings and cash flow analysis as the basis for company valuation. This valuation method, detached from fundamentals, was a core feature of the bubble.

In contrast, today's market leaders—the "Magnificent Seven"—are among the most profitable and cash-rich companies in the world. The following table uses specific data to visually demonstrate the vast difference in fundamentals between the leading companies of the two eras.

Table 2: A Fundamental Comparison of Market Leaders During the Tech Bubble and Today

Source: Pacer ETFs

Metric	December 31, 1999 (Top 10 in Nasdaq 100)	December 31, 2023 (Top 10 in Nasdaq 100)	Growth Rate (%)
Total Sales (in millions)	$119,948	$2,329,279	1,842%
Total Earnings (in millions)	$21,217	$362,724	1,610%
Total Free Cash Flow (in millions)	$25,994	$408,824	1,473%
Total Market Cap (in millions)	$1,660,075	$14,532,363	775%
Number of Dividend-Paying Companies	1	5	400%

The data shows that the earnings and free cash flow of today's market leaders are more than 17 and 15 times, respectively, that of their 1999 counterparts. This is not just a difference in magnitude; it's a fundamental difference in nature. The high valuations of 1999 were built on dreams and speculation about the future, whereas today's valuations, though equally lofty, are built on a foundation of already realized, massive profits and cash flows.

However, a new variable is changing the business models of these tech giants. An "arms race" around artificial intelligence (AI) is unfolding. The combined capital expenditures (Capex) of Meta, Amazon, Microsoft, and Alphabet are expected to reach a staggering $317 billion this year, about 1% of the US GDP. These companies, traditionally considered "capex-light," are now investing a similar proportion of their operating cash flow (about 45%) into capital expenditures, on par with other companies in the broader market. This shift is structural and poses a challenge to the long-term financial models of these companies. High capital expenditures could suppress free cash flow in the short term and raise questions about the future return on invested capital (ROIC). The ultimate return on this massive investment is still uncertain: it could build an insurmountable moat driven by computing power, thereby solidifying their dominant position for decades to come; or it could devolve into a value-destroying race where parties over-invest to gain a technological edge, ultimately eroding profit margins. This uncertainty adds a new and significant layer of risk to the future of these "excellent companies."

2.4 Changes in Structural and Regulatory Environments

Besides corporate fundamentals, the market structure and regulatory environment of the two eras also differ significantly. Academic research points out that the formation of the dot-com bubble was closely related to a series of institutional factors. First were the problems in the underwriting and Initial Public Offering (IPO) market. There was systematic IPO underpricing, conflicts of interest (exacerbated by the repeal of the Glass-Steagall Act), and some fraudulent underwriting practices. Many companies were brought to market with almost no revenue, and their stock prices soared on the first day of trading, creating huge profits for insiders and underwriters.

Second, venture capital (VC) played a role in fueling the bubble. VCs injected huge amounts of capital into startups and encouraged them to spend the vast majority of it on large-scale advertising campaigns to build brand awareness and website traffic in a short time, which in turn was used as the basis for valuation in the next round of financing. This cycle created a large number of companies with inflated valuations but fragile business models.

In summary, a significant judgment can be made: the nature of market risk today has fundamentally changed. The main risk in 1999 was business model survival risk—many companies had no profits and might never achieve them, with their value potentially going to zero. Today, the main risks are economic cycle sensitivity and valuation compression risk, along with the new risk of uncertainty in capital expenditure returns. These companies are extremely profitable and their business models are proven. The risk lies in their future growth rate possibly failing to meet the market's high expectations, or a change in the macroeconomic environment (like rising interest rates) causing the market to be willing to pay a lower valuation multiple for their earnings, or their massive investments in AI failing to generate the expected returns. This is a fundamentally different and potentially less destructive type of risk. Therefore, although it is appropriate for Marks to compare the current situation to the "early stages" of a bubble from an investor psychology perspective, the two eras are not comparable in terms of the intrinsic properties of financial risk. A potential crisis today is more likely to be a painful asset re-pricing rather than a devastating blow to the survival of an entire generation of tech companies, as was the case in 2000-2002.

Part III: A Tale of Two Markets: Concentration Risk and the Valuation of the Broader Stock Market

Marks's analysis delves into the internal structure of the market, pointing out that the real concern is not the "Magnificent Seven" with the highest valuations, but that the "other 493 stocks" have also been assigned high valuations relative to historical levels. This observation reveals a core feature of the current market: the spread of high valuations and the extreme market concentration. After expressing concern about the overall market valuation level, Marks further points out a key internal divergence: the difference between the few tech giants represented by the "Magnificent Seven" and the "remaining 493 companies" in the S&P 500 index. He is more worried that high valuations seem to be spreading to "more ordinary companies." This section will deeply analyze this market divergence phenomenon, quantify the impact of market concentration, and use the S&P 500 Equal Weight Index as an analytical tool to test whether high valuations are an exception for a few companies or have become a widespread phenomenon.

3.1 The Few and the Many: Quantifying Market Concentration and Growth Divergence

The weight of the "Magnificent Seven" in the S&P 500 index has reached an unprecedented high, which in itself is a significant risk signal. Increased market concentration means that the index's Beta is increasingly driven by the idiosyncratic risk of a few companies, rather than broad market macroeconomic factors. This violates the basic assumption of the Capital Asset Pricing Model (CAPM) that the market portfolio should be fully diversified.

The CAPM formula is: E(R_i) = R_f + β_i(E(R_m) - R_f), where β_i = Cov(R_i, R_m) / Var(R_m). When the market portfolio R_m is dominated by a few stocks, Var(R_m) will more heavily reflect the volatility of these few companies, thus distorting the risk pricing of the entire market.

We can use the Herfindahl-Hirschman Index (HHI) to quantify market concentration. For the S&P 500 index, its HHI is calculated as:

HHI = Σ_{i=1 to 500} (w_i)^2

Where w_i is the weight of the i-th company in the index. A higher HHI value indicates higher market concentration. Calculating the HHI time series for the S&P 500 index over the past few decades clearly shows that it is currently at a historic high, even surpassing the peak of the tech bubble in 2000.

The risk brought by high concentration is that once these leading companies encounter growth bottlenecks, regulatory crackdowns, or technological disruption, the entire market index will face a sharp correction, and the vast number of passive investors will have nowhere to hide. In recent years, one of the most significant features of the U.S. stock market has been the sharp rise in market concentration. The performance of a few mega-cap companies has played a decisive role in the direction of the entire index.

First, from the perspective of return contribution, this concentration is vividly demonstrated. In 2023, when the S&P 500 index achieved a 24.2% return, the "Magnificent Seven" as a group saw their stock prices rise by an astonishing 75.7% on average. This trend continued into 2024, with these seven companies alone accounting for 53.7% of the S&P 500's 25.0% total return. This type of rally, driven by a very small number of stocks, is known as having narrow "market breadth" and is a potential warning sign for the health of the market. In fact, data shows that 2023 was one of the years with the narrowest market breadth since 1995.

Second, from the perspective of index weights, concentration has also reached a historic high. As of recently, the combined weight of the top ten companies in the S&P 500 index has exceeded 33%, the highest level since the 1970s. The weight of the "Magnificent Seven" alone accounts for about 28% to 30% of the index. This high weight means that the volatility and risk of the index are increasingly tied to the fate of these few companies. This phenomenon is consistent with the academic theory of the rise of "superstar firms," where globalization and technological change are driving a few of the most efficient companies to capture an ever-larger share of the market and profits, which is reflected in the structure of the capital markets.

What is more worrying is the huge gap in fundamental growth. Data shows that in the second, third, and fourth quarters of 2024, the net profit growth of the remaining 493 companies in the S&P 500, excluding the "Magnificent Seven" (the S&P 493), is expected to be only 2-3%. This growth rate is below the rate of inflation, meaning that the vast majority of companies in the market are experiencing a contraction in real earnings. In contrast, the net profit growth of the "Magnificent Seven" for the same period is expected to range from 36.8% to 13.7%. This stark contrast reveals a harsh reality: the market's growth is almost entirely driven by a few giants, while the vast majority of "ordinary companies" are stagnating or even regressing.

3.2 The Evidence in the Weights: A Comparison of Market-Cap Weighted and Equal-Weighted Indices

To more clearly isolate the impact of these few giants and observe the true condition of the "remaining 493 companies," we can compare two different construction methods of the S&P 500 index: the traditional Market-Cap Weighted Index and the Equal Weight Index (EWI). In the EWI, every company, regardless of its market capitalization, is given the same weight (about 0.2%). Therefore, the performance of the EWI is more representative of the condition of the "ordinary" or "average" company in the index.

From a valuation perspective, the difference between the two is extremely significant. The forward P/E ratio of the market-cap weighted index is about 22x, while the P/E ratio of the equal weight index is only 17x. This directly quantifies the effect of the high valuations of the "Magnificent Seven" pulling up the overall index. A P/E ratio of 17x, although still slightly above the historical average, is far from alarming, suggesting that the valuations of the vast majority of "ordinary companies" are relatively reasonable.

To deeply understand the root of the performance difference between the EWI and the market-cap weighted index, we need to turn to modern asset pricing theory. An equal-weight strategy is not just a different weighting scheme; it inherently introduces a systematic deviation (or "factor exposure") to specific risk factors. According to the five-factor model proposed by Eugene Fama and Kenneth French, the expected return of a stock can be explained by five factors: market risk, size, value, profitability, and investment patterns. The model's expression is:

R_i - R_f = α_i + β_{i,MKT}(R_M - R_f) + β_{i,SMB}SMB + β_{i,HML}HML + β_{i,RMW}RMW + β_{i,CMA}CMA

Where:

• R_i is the return of asset i, and R_f is the risk-free rate.

• R_M - R_f is the market risk premium.

• SMB (Small Minus Big) is the size factor, representing the excess return of small-cap companies over large-cap companies.

• HML (High Minus Low) is the value factor, representing the excess return of companies with a high book-to-market ratio (value stocks) over those with a low book-to-market ratio (growth stocks).

• RMW (Robust Minus Weak) is the profitability factor.

• CMA (Conservative Minus Aggressive) is the investment pattern factor.

Compared to the market-cap weighted index, the equal weight index naturally has a positive exposure to the "size factor" (SMB) because it gives higher weight to smaller-cap companies. At the same time, since value stocks are usually not the largest companies by market cap, an equal-weight strategy also tends to lean towards the "value factor" (HML) and correspondingly underweights the "momentum factor." Therefore, the long-term outperformance of the EWI in history can be largely attributed to the size and value factors providing a positive risk premium for most of the time. Its recent underperformance is due to the market style being extremely skewed towards large-cap (negative SMB exposure) and growth (negative HML exposure) stocks, represented by the "Magnificent Seven."

Table 3: Comparison of Characteristics Between S&P 500 Market-Cap Weighted and Equal Weight Indices

Source: Compiled from various sources. Returns and valuations are recent representative data and fluctuate over time.

Metric	S&P 500 Market-Cap Weighted Index	S&P 500 Equal Weight Index
Long-Term Annualized Return (1989-2023)	~9.30%	~10.35%
Recent 1-Year Annualized Return	25.1%	11.9%
Forward P/E Ratio	~22x-26x	~17x
Top 10 Holdings Weight (%)	~33.5%	~2.0%
Implied Factor Exposure	Biased towards large-cap, growth, momentum	Biased towards small-cap, value

3.3 Evaluating the "Remaining 493 Companies"

Marks is more concerned that "more ordinary companies" are also enjoying high valuations. There are several possible driving factors behind this phenomenon:

1. The Rise of Passive Investing: The huge inflows of capital into index funds and ETFs lead to the indiscriminate purchase of all constituent stocks within an index, thereby systematically raising the valuations of all stocks, regardless of their fundamentals. This "a rising tide lifts all boats" effect masks the quality differences between companies.

2. The Effect of Interest Rate Suppression: In a long-term low-interest-rate environment, investors, in search of higher returns, are forced to accept higher risks and lower risk premiums. This causes capital to flow from safe fixed-income assets into the stock market, pushing up stock valuation levels across the board. Even as interest rates have begun to rise, the inertia of this risk appetite and valuation habit persists.

3. The Contagion of Optimism: The optimism surrounding the disruptive technologies represented by the "Magnificent Seven," such as AI, can spill over to other related and unrelated industries. Investors may mistakenly believe that the wave of technological revolution will benefit all companies, thus assigning them unrealistic growth expectations.

Marks's concern is profound. While the high valuations of the "Magnificent Seven" can be partly explained by their extraordinary profitability, market position, and growth prospects, applying similarly optimistic valuations to "ordinary companies" with shallower moats and mediocre growth is a clearer sign of a bubble. This suggests that the market's pricing mechanism may be malfunctioning and that investors' ability to identify risk is becoming dulled.

We can quantify this problem by analyzing the valuation level (such as P/E and P/B ratios) of the "S&P 493" (the S&P 500 index excluding the "Magnificent Seven") and its deviation from the historical mean. If it is found that the valuation of the "S&P 493" is also above the 90th percentile historically, then Marks's concern would be strongly supported by empirical evidence.

By using the valuation level (forward P/E of about 17x) and earnings growth (negative in real terms) of the EWI as a proxy for "ordinary companies," we can more precisely evaluate Marks's assertion. His concern—that high valuations are spreading to "more ordinary companies"—is correct to some extent, because even a P/E ratio of 17x is not cheap for a group whose real earnings are shrinking. However, the core of the problem is not a general bubble in valuations, but the extreme scarcity of growth. The market is paying a high premium for the certainty of growth provided by the "Magnificent Seven" because, outside of them, it is almost impossible to find any meaningful sources of growth.

This extreme market concentration has, to some extent, caused the S&P 500 market-cap weighted index to deviate from its traditional role as a diversified representation of the "market." By purchasing such an index fund, investors are inadvertently no longer getting a broad, diversified investment in the U.S. economy, but are making a large-scale, concentrated bet on the continued success and high growth expectations of a few tech giants. The actual risk exposure of such a portfolio has quietly shifted from a diversified market risk (beta) exposure to a highly concentrated, active investment biased towards large-cap growth and momentum factors. This poses a serious challenge to traditional asset allocation models, such as the 60/40 portfolio, because the intrinsic diversification of its equity component has been greatly reduced.

Part IV: The Search for Defensive Yield: A Prudent Assessment of Corporate Credit

Facing a stock market with high valuations, Marks proposes a practical strategic recommendation: turn to credit (corporate debt) as a more defensive investment. After expressing a cautious attitude towards the highly valued stock market, Marks turns his attention to corporate credit, viewing it as a more defensive investment option in the current environment. He emphasizes that credit provides contractually promised payments and a knowable return. This recommendation is based on a deep understanding of the risk-return characteristics of different asset classes. This section will conduct a rigorous evaluation of this view, going beyond his macro-level assertion to deeply analyze the composition of the credit risk premium, and combining current data on credit spreads, default rates, and recovery rates to determine whether credit assets truly offer compensation commensurate with their risks.

4.1 The Fundamental Difference Between Stocks and Credit

From the perspective of a company's capital structure, credit (debt) holders and stock (equity) holders are in completely different positions.

• Priority of Claim: In the event of a company's liquidation, debt holders are paid before equity holders. This gives credit a natural "safety cushion."

• Return Structure: The return on credit is a fixed, contractually guaranteed interest payment (coupon) and the return of principal. Its upside is limited (at most, one can recover principal and interest), but its downside risk is also relatively controllable (as long as the company does not default). The return on stocks, on the other hand, comes from the company's future earnings growth and changes in valuation, with unlimited upside potential and the downside risk of total loss.

This structural difference can be elegantly described using option theory. According to the structural credit risk model proposed by Robert Merton in 1974, a company's equity can be viewed as a European call option on the total value of the company's assets, V, with the face value of the debt, F, as the strike price.

Equity = max(V - F, 0)

And the company's debt can be viewed as risk-free debt minus a European put option sold to the shareholders, with the company's assets as the underlying and the face value of the debt as the strike price.

Debt = F - max(F - V, 0)

This put option represents the right of the shareholders to choose to default (i.e., not to exercise). Therefore, buying corporate credit is essentially selling a put option on the value of the company's assets. As long as the company's asset value V does not fall below the face value of the debt F, the credit investor will receive the full principal and interest.

Marks's advice is not to abandon stocks completely, but to make a tactical asset allocation adjustment by increasing the weight of credit to enhance the defensiveness of the portfolio. This is consistent with classic portfolio theory. When constructing the efficient frontier, investors should decide on the optimal asset allocation weights based on their judgments of the future expected returns, volatilities, and correlations of various assets. Marks's judgment is that the expected Sharpe ratio of stocks ([E(R_stock) - R_f] / σ_stock) is no longer attractive relative to the expected Sharpe ratio of credit ([E(R_credit) - R_f] / σ_credit).

He emphasizes that investing in high-yield bonds in 1998 would likely have "done very well" over the subsequent 27 years. This is not a prediction of the future, but an articulation of a probabilistic issue. The core of credit investing is to obtain a contractually promised return that is highly probable, while taking on manageable risk. In a world full of uncertainty, this "certainty" itself is of extremely high value.

4.2 Deconstructing the Credit Risk Premium: A Theoretical Framework

First, it must be clear that the yield investors receive from corporate bonds above that of risk-free government bonds of the same maturity, known as the credit spread, is not entirely a "risk premium." The credit spread must compensate for two components: one is the Expected Loss (EL), which is the mathematical expectation of the loss of principal and interest due to the issuer's default; the other is the Credit Risk Premium (CRP), which is the additional compensation for bearing non-diversifiable, systematic default risk.

The relationship can be expressed as:

Spread ≈ EL + CRP

The expected loss itself is determined by two key parameters: the Probability of Default (PD) and the Loss Given Default (LGD), so EL = PD × LGD. Therefore, investors can only earn a positive risk premium when the credit spread is significantly higher than the expected average loss.

Modern financial theory provides a deeper insight into the determinants of the credit risk premium. Within the framework of structural models (like the Merton model), a company's credit risk is viewed as an option on its asset value, closely related to the company's leverage and asset volatility. In a broader asset pricing theory context, the size of the credit risk premium depends on the covariance of default losses with the Stochastic Discount Factor (SDF). In short, when defaults are more likely to occur in bad economic conditions, where the marginal utility of investors is higher (i.e., default has systematic risk), investors will demand a higher risk premium as compensation. Empirical studies have also found that the credit risk premium is significantly related to macroeconomic growth, consumer sentiment, overall market distress, and liquidity, among other factors.

4.3 An Empirical Examination of High-Yield Bond Spreads

Marks's argument is made in a specific market context, so we must examine whether the current price—the credit spread—is attractive. According to data from the Federal Reserve Bank of St. Louis (FRED), the Option-Adjusted Spread (OAS) of the ICE BofA US High Yield Index has recently been hovering around 2.95%, or about 295 basis points.

Placing this value in a historical context reveals its extremely low level. In 1998, the reference point used by Marks, high-yield bond spreads were also tight, but they subsequently widened sharply during the Russian debt crisis and the collapse of Long-Term Capital Management (LTCM). At the peak of the 2008 Global Financial Crisis, the spread soared to an extreme level of 2182 basis points (21.82%). And during the market panic triggered by the COVID-19 pandemic in 2020, the spread briefly touched 1087 basis points (10.87%). Compared to these historical crisis periods, the current spread of less than 300 basis points means that the compensation the market offers for bearing the credit risk of high-yield bonds is at a historic low. In fact, data shows that high-yield bond spreads have been below the current level only 10% of the time over the past 15 years.

4.4 The Ultimate Determinants of Return: Default Rates and Recovery Rates

The level of the spread alone is not fully indicative; we must compare it with the expected loss.

• Default Rate Analysis: According to data and forecasts from rating agencies like Moody's and S&P, the long-term average annual default rate for speculative-grade (high-yield) companies is typically between 4.1% and 4.5%. For 2024-2025, the forecasts of mainstream institutions generally fall within the range of 3.4% to 4.6%. This indicates that despite a relatively stable economic environment, future default rates are expected to return to or be slightly above their long-term historical average, not at an abnormally low level.

• Recovery Rate Analysis: The Loss Given Default (LGD) is another key component of expected loss, calculated as 1 minus the Recovery Rate. The long-term average recovery rate for senior unsecured bonds is approximately between 37% and 49%, corresponding to an LGD of 51% to 63%. However, a noteworthy recent trend is the change in the composition of defaults. According to S&P data, a significant portion of default events in 2023 were "distressed exchanges," rather than traditional bankruptcy liquidations. The recovery rates for these types of defaults are usually significantly higher, pushing the overall recovery rate for speculative-grade bonds in 2023 to 66%.

• Estimation of the Implied Risk Premium: We can now synthesize this data to make a rough estimate of the true risk premium offered by the current market.

  • Current Spread (OAS): ~2.95%

  • Expected Default Rate (PD): ~4.0% (taking the midpoint of the forecast range)

  • Expected Recovery Rate: ~66% (based on recent data)

  • Expected Loss Given Default (LGD): 1 - 66% = 34%

  • Expected Loss (EL): EL = PD × LGD = 4.0% × 34% = 1.36%

  • Implied Credit Risk Premium (CRP): CRP = Spread - EL = 2.95% - 1.36% = 1.59%

This risk premium of about 1.59% (159 basis points) is quite meager by historical standards. It poses a serious challenge to Marks's view that credit assets are defensively attractive at present.

The "defensive" nature of credit assets is essentially a function of price, not an inherent, unchanging attribute. Marks's logic—that credit offers contractual returns and is less risky than stocks—is theoretically correct, but the investment value of this logic is highly dependent on the purchase price, i.e., the level of the credit spread. In today's low-spread environment, the compensation offered by credit for bearing default risk is very limited. Its risk-return profile shows a clear asymmetry: there is little room for spreads to tighten further, but in a recession or a period of heightened market risk aversion, there is enormous room for spreads to widen. A widening of spreads would directly lead to a fall in bond prices, and this capital loss could easily wipe out the meager coupon income. Therefore, buying high-yield credit at the current price is not a defensive deployment to guard against future risks, but rather a gamble that the economy will remain stable, for which the reward is quite low. This seems to diverge from the spirit of prudence and defensiveness that Marks advocates for when markets are at a high.

A deeper risk lies in the possibility that the recent change in the composition of defaults may be masking underlying vulnerabilities. The current high recovery rate (66%) is largely due to the high proportion of distressed exchanges among default events. This form of default typically occurs in a decent economy where companies still have room for restructuring. However, in the event of a severe economic recession, the nature of defaults could quickly shift to more destructive "hard defaults," such as bankruptcy liquidations. Historically, the recovery rates for such defaults are much lower, averaging only 40% to 50%. If we recalculate the expected loss using a more conservative recovery rate of 40% (i.e., a 60% LGD), which is more in line with a recession scenario, the result would be vastly different:

EL_recession = 4.0% × 60% = 2.40%

At this point, the implied risk premium would plummet to:

CRP_recession = 2.95% - 2.40% = 0.55%

This calculation reveals a disturbing possibility: the risk compensation offered by the current market spread is almost entirely built on the assumption that the economy will continue to do well and the pattern of defaults will remain benign. Once the economic environment deteriorates and returns to more traditional patterns of default and recovery, the true risk premium contained in the current spread will approach zero. The perceived "safety" of the market may just be an illusion based on the current benign, but perhaps not stable, state of defaults.

Table 4: Comparison of High-Yield Market Conditions at Key Time Points

Source: Spread data from FRED; default and recovery rate data compiled from annual reports by Moody's and S&P. 1998 data is approximate.

Metric	Mid-1998	Late 2008 (GFC Peak)	March 2020 (COVID Peak)	Present
High-Yield Index OAS (bps)	~350-400	~2182	~1087	~295
Subsequent 1-Year Spec-Grade Default Rate (%)	4.9% (1999)	13.4% (2009)	8.8% (2020)	3.4%-4.6% (forecast)
Average Spec-Grade Recovery Rate (%)	~40-50 (hist. avg.)	33.8%	~40-50 (hist. avg.)	66% (recent)

Part V: The Global Investment Dilemma: A Comparison of the U.S. and Other Developed Markets

In the final part of the interview, Marks broadens his perspective globally, offering a dialectical assessment of the United States' status as an investment destination. On one hand, he affirms that the U.S. "is still the best place in the world to invest," while on the other, he notes that its relative attractiveness may be declining. Marks uses a vivid analogy to describe the global investment landscape: the U.S. is a "good car that's expensive," while the rest of the world may offer "less-good cars that are cheaper." This analogy brings up a core question in asset allocation: Should investors pay a premium for quality, or should they search for opportunities in undervalued areas? This section will quantify the valuation gap between the U.S. and international developed markets (represented by the MSCI EAFE index) and break down the drivers of their long-term relative performance to assess whether the discount on international stocks is justified.

5.1 A Tale of Two Valuations: Quantifying the U.S. Premium

The long-term outperformance of the U.S. market can be attributed to a series of structural advantages, including a spirit of innovation, free markets, the rule of law, and deep capital markets. These factors together create an environment capable of continuously generating "alpha," where even investing in the entire market index can yield excess returns relative to the markets of other countries.

However, Marks's analogy—the U.S. is a "good car that's expensive," while the rest of the world offers "less-good cars that are cheaper"—gets to the heart of the matter: price. No matter how good an asset is, if the price is too high, its future investment return will be greatly diminished.

This touches on a core issue in international asset pricing. The traditional International Capital Asset Pricing Model (ICAPM) posits that the optimal portfolio for a global investor should be the global market portfolio, and the expected return of each country's assets is determined by its sensitivity to global market risk (its global beta). However, a large body of empirical research shows that country-specific factors also play an important role in explaining return differences.

The long-term premium of the U.S. market can be seen as compensation for its superior macroeconomic fundamentals and corporate governance. But when this premium expands to extreme levels, investors need to consider whether it has already over-reflected its advantages and even incorporated overly optimistic expectations for the future.

The valuation premium of the U.S. stock market relative to the rest of the world has been a significant and persistent feature of the global market in recent years.

By comparing the P/E ratios of the MSCI USA Index and the MSCI EAFE (Europe, Australasia, and Far East) Index, we can clearly see this trend. Over the past decade (2013-2023), the average P/E ratio of the MSCI USA Index was about 20.6x, while the average P/E of the MSCI EAFE Index was only 16.7x. Interestingly, this is the exact opposite of the situation in the preceding decade (2002-2012), when the EAFE index's valuation (21.0x) was actually higher than that of the U.S. index (16.9x). This indicates that the current U.S. premium is not a permanent phenomenon, but a specific product formed in the post-Global Financial Crisis era.

As of early 2024, this valuation gap has widened to one of its broadest levels in recent decades. Data shows that the discount of the MSCI EAFE index's P/E ratio relative to the S&P 500 index has exceeded 40%. This huge valuation chasm provides solid quantitative evidence for Marks's observation that international markets are being "sold at a discount."

Marks's viewpoint guides us to conduct a global relative value analysis. We can assess the relative attractiveness of different national markets by comparing their valuation metrics (like the CAPE ratio), expected earnings growth, and macroeconomic outlooks.

For example, we can construct a simple country attractiveness scoring model:

Score_c = w_1 × (1 / CAPE_c) + w_2 × E(g_c) - w_3 × σ_c - w_4 × RiskPremium_c

Where c represents the country, CAPE is the cyclically adjusted P/E ratio, E(g) is the expected earnings growth rate, σ is the market volatility, and RiskPremium is a premium reflecting political, institutional, and other risks. Through this model, we can quantitatively compare the investment value of the U.S. with other markets (such as Europe, Japan, and emerging markets).

It is highly likely that at the current juncture, although the U.S. scores high on E(g) and RiskPremium, its extremely low 1 / CAPE term (i.e., its extremely high valuation) will pull down its total score, making some "less-good" but "cheaper" markets more attractive on a composite score basis.

This is not to suggest that investors should completely exit the U.S. market, but to emphasize the importance of global asset allocation. Adding some assets that are "for sale at a discount to the U.S." to a portfolio can serve to diversify risk and improve risk-adjusted returns, even if the long-term growth potential of these markets is not as good as that of the U.S. The founder of modern portfolio theory, Harry Markowitz, proved that as long as the correlation between assets is not +1, diversification can reduce the overall risk of a portfolio. In today's increasingly interconnected global markets, finding assets with low correlation has become more difficult, but also more important. The valuation difference itself is a significant source of future return differences and a key entry point for achieving effective global diversification.

5.2 Drivers of Regional Performance Divergence: A Decomposition of Returns

To understand why such a huge valuation gap has emerged, we must deconstruct the fundamental reasons for the U.S. market's long-term outperformance of international markets since the Global Financial Crisis. A return attribution analysis by J.P. Morgan provides a clear answer. From June 2008 to the end of 2024, the U.S. market's 11.9% annualized return was mainly composed of two parts: 6.3% from earnings growth and 3.3% from valuation multiple expansion (i.e., an increase in the P/E ratio). In contrast, of the EAFE market's 3.6% annualized return, the contribution from earnings growth was only 1.6%, and the contribution from valuation expansion was 1.3%. What's worse, it also suffered an annualized drag of -2.1% due to the depreciation of its currencies against the U.S. dollar.

This huge difference in earnings growth is largely due to the vastly different industry structures of the two regions. The superior earnings growth of the U.S. market was mainly driven by the technology and consumer discretionary sectors, whose earnings growth far outpaced that of their international counterparts. The U.S. market has a significantly higher weight in these high-growth, high-profit-margin industries, while the EAFE index is more skewed towards traditional "old economy" or value-oriented sectors like financials, industrials, and materials. This structural difference is the core reason explaining the long-term performance and valuation divergence between the two markets.

5.3 Evaluating the "Cheaper, Less-Good Cars"

Despite its historical underperformance, investing in international markets still has its strategic value. First is diversification. Due to different industry compositions and economic cycles, international markets can offer sources of risk and return that are not perfectly correlated with the U.S. market. Second is higher yield. The dividend yield of the EAFE index is about 3.4%, much higher than the S&P 500's approximately 1.5%, which is attractive to income-seeking investors. Finally, there is the potential for mean reversion. History shows that the relative performance of U.S. and international markets exhibits long-cycle rotations. The current decade-plus of U.S. outperformance, along with a significant valuation premium, may have set the stage for a relative mean reversion in favor of international markets in the future.

Of course, the reasons supporting the U.S. market's premium are also ample. As Marks acknowledges, the U.S. has an unparalleled spirit of innovation, free markets, the rule of law, and vibrant, great companies. Its superior earnings growth is not a statistical illusion but a reflection of real economic vitality. For investors, the core questions are: Is this superior performance sustainable in the future? And has it already been overpriced by the market?

A deep analysis of the divergence between the U.S. and EAFE markets reveals that this global asset allocation choice is, to a large extent, an amplification and projection of the internal style divergence within the U.S. market. The decision on global allocation implicitly involves a bet on specific economic sectors and investment styles. As we saw in Part III, the dominant force in the U.S. market is a few large-cap tech/growth companies. The composition of the MSCI USA Index also naturally has a high weight in sectors like Information Technology and Communication Services. Meanwhile, the MSCI EAFE Index has a higher weight in classic "value" sectors like Financials, Industrials, and Materials. Therefore, the decision to overweight EAFE and underweight the U.S. is highly similar in its underlying logic to the decision to overweight the S&P 500 Equal Weight Index and underweight the market-cap weighted index. Both are questioning the continued leadership of U.S. mega-cap growth stocks and are betting on a recovery in value and the broader market. This perspective links the seemingly independent Part III (domestic market structure) and Part V (global market comparison), revealing their common factor exposure: the "expensive good car" is expensive because it is a growth/tech sports car; and the "cheaper car" is cheap because it is a value/industrial sedan.

Furthermore, when evaluating international investments, currency is a crucial but often overlooked long-term driver of returns. Since 2008, currency movements have created a massive drag of -2.1% per year on the U.S. dollar-denominated returns of the EAFE index. This means that nearly a quarter of the 8.3% total return gap between the EAFE and U.S. indices was caused by currency factors. For most of the post-financial crisis era, the U.S. dollar has been in a long-term bull market. However, like asset valuations, currency trends are also cyclical and subject to mean reversion. Therefore, a key potential rationale for investing in international markets is not just that EAFE stocks are cheap, but also that the U.S. dollar itself may be expensive. Once the long-term appreciation cycle of the dollar reverses, even if the fundamental performance of EAFE constituent stocks is mediocre, U.S. dollar-based investors could still achieve significant excess returns due to a currency tailwind. This greatly elevates the importance of analyzing macro currency cycles in asset allocation decisions.

Table 5: U.S. vs. EAFE Markets—A Comparison of Relative Valuation and Fundamental Drivers

Source: Compiled from various sources. Valuations and weights are recent representative data and fluctuate over time. Return attribution data is from J.P. Morgan.

Metric	MSCI USA	MSCI EAFE
Average P/E Ratio (2002-2012)	16.9x	21.0x
Average P/E Ratio (2013-2023)	20.6x	16.7x
Current Forward P/E Ratio	~23x	~14x
Dividend Yield (%)	~1.5%	~3.4%
Major Sector Weights (%)
- Information Technology	~30%	~9%
- Financials	~13%	~18%
- Industrials	~9%	~15%
Annualized Return Attribution (2008-2024)
- Contribution from Earnings Growth	6.3%	1.6%
- Contribution from Valuation Expansion	3.3%	1.3%
- Contribution from Currency Impact	0.0%	-2.1%
Total Annualized Return	11.9%	3.6%

Marks's analysis, from beginning to end, is permeated with a profound skepticism and a reverence for cycles. He does not attempt to predict short-term market movements but, through a deep analysis of valuations, psychology, and history, provides investors with a navigation framework based on long-term rational judgment in this challenging and contradictory environment. Every one of his points finds an echo in modern financial theory and rigorous academic research, which is the source of the depth and enduring vitality of his ideas.

Disclaimer:

The information and analysis presented in this article are for informational and educational purposes only and do not constitute financial, investment, legal, or tax advice. The content is not intended to be a recommendation to buy, sell, or hold any security or to make any investment decision.The views and opinions expressed in this article are solely those of the author and are subject to change without notice. They do not represent the opinions of any other entity with which the author may be affiliated. While the author has made every effort to ensure the accuracy and completeness of the information derived from the source material, no warranty is made as to its accuracy. All investment decisions should be made with the guidance of a qualified professional.