Building a Six Factor Model Using Intangible Value with Kai Wu

Episode Summary

Executive Summary: Kai Wu argues that value investing has struggled because traditional metrics ignore intangible assets such as IP, brand, human capital, and network effects, which now drive much of corporate value. He presents an intangible value factor (IHML) that is weakly correlated with traditional factors, works in both long-short and long-only formats, and may complement rather than replace classic value by capturing innovation-driven firms before they fully become quality stocks.

Main Topics: Why traditional value has struggled (Priority: 5/5): Wu explains that price-to-book and similar value measures underperform because modern businesses create value through intangibles rather than physical capital, leaving high-intangible firms like tech underpriced by old accounting frameworks. Why accounting-based capitalization only partially fixes the problem (Priority: 4/5): The discussion covers attempts to capitalize R&D and advertising, but Wu argues these methods rely on arbitrary assumptions and only recover part of the lost signal because intangible investment has a weak input-output relationship. Defining the four pillars of intangible capital (Priority: 5/5): Wu outlines his framework for measuring intangible value through intellectual property, brand equity, human capital, and network effects, each capturing a different source of modern moat creation. Building the intangible value factor (IHML) (Priority: 5/5): Wu describes a yield-style factor construction that ranks firms by intangible assets per dollar of market value, using machine learning and alternative data to quantify patents, talent, culture, and other intangible signals. How IHML compares with the Fama-French factors (Priority: 5/5): The paper positions intangible value as a potential sixth factor with low correlations to market, size, momentum, quality, and traditional value, suggesting it adds distinct explanatory power. Long-only implementation and portfolio construction (Priority: 4/5): Wu shows how the factor can be used in real-world long-only portfolios, where a modest factor tilt can improve returns while maintaining acceptable tracking error. Relationship to growth and innovation ETFs (Priority: 4/5): The paper argues that many growth/innovation ETFs are effectively backdoor bets on intangible value, but they also introduce unwanted exposures to low quality and expensive valuation; IHML captures innovation more cleanly.

Key Arguments: Traditional value failed largely because accounting treats intangibles as expenses rather than assets, so value metrics systematically mismeasure modern firms. Capitalizing intangibles is directionally useful but only partially resolves the issue because intangible output is not tightly linked to historical spending. A robust intangible framework should measure multiple sources of moat creation, not just R&D or advertising. IHML is built as a genuine value factor: it ranks firms by intangible assets relative to price, not by absolute innovation intensity. The factor is distinct from existing factors; correlations are near zero, and its value correlation has changed over time as markets have increasingly priced intangibles. Half the factor’s alpha comes from the long side and half from the short side, making it relevant for both long-only and long-short investors. IHML can improve portfolios without forcing investors to choose between traditional value and innovation exposure; it may serve as a hedge across different market regimes. Many growth ETFs capture intangible value indirectly, but they also bring negative exposure to value and quality, which can dilute their attractiveness. Companies investing heavily in intangibles may see lower profitability initially, but profitability tends to improve in subsequent years as the investments mature. The most attractive opportunities may be firms that are still building moats: not yet high-quality incumbents, but clearly on the path to becoming them.

Data Points: Value-factor underperformance period: ~15 years - Wu notes that traditional value strategies, especially low price-to-book, have struggled since roughly 15 years ago. Intangible value factor correlation range: -14% to +9% - Reported rolling correlations of IHML to other factors were generally very low. Correlation with quality: ~11% - IHML had a slightly positive average correlation with the quality factor. Average correlation with traditional value: ~9% - The mean correlation of intangible value to HML/value was positive overall but varied across subperiods. Capitalize-intangibles fix recovers: ~30% - Wu says capitalizing intangible investment only gets investors about 30% of the way back toward the desired result. Model backtest history: 1990s to present; tests also back to the 1970s - Wu says the signal is reasonably robust from the 1990s onward and has been tested farther back. Portfolio allocation example: 50% index funds / 50% factor funds - Used as a long-only illustrative portfolio construction example. Tracking error from modest factor tilt: ~3 percentage points - A 25% factor tilt in the illustrative long-only portfolio produced about 3% tracking error versus the benchmark. Tracking error from aggressive factor tilt: ~6% to the S&P - An all-factor version with equal weights to five factors would raise tracking error to about 6%. Backtest excess return from modest tilt: ~1.7% - The illustrative long-only factor tilt generated about 1.7% excess return in the backtest. Exposure contribution from intangible value in growth ETFs: ~20% of return contribution - Wu says IHML accounted for about 20% of the return contribution in growth ETFs in his decomposition. Amazon example: Years of reinvestment before profits - Used qualitatively to illustrate how firms can sacrifice current earnings to build future intangible moats. QQQ outperformance since 2010: ~70 percentage points over the S&P - Cited as an example of the strong decade for innovation/growth exposure.

Pivotal Quotes: "Value missing should work. It’s just that we’re mismeasuring value because we’re focused on this tangible capital, which, you know, was once, you know, in Ben Graham’s day, 100% of what mattered, and today is a vanishingly small amount." — Kai Wu: Explaining why traditional value has struggled in an intangible-heavy economy. "The point being that, like, any sort of investment, whether it’s in R&D or, you know, think differently in advertising, right? ... there’s not really a strong link between the historical input costs and then the resulting output." — Kai Wu: Why accounting capitalization is only a partial fix for measuring intangible assets. "I think intangible value can be viewed as a way of kind of homing in, extracting the good parts of these innovation ETFs while throwing out some of the less desirable parts like their negative exposure to quality and value." — Justin Carboneau: Summarizing the portfolio role of the intangible value factor relative to growth funds.

Implications: Listeners should view intangible value as a complementary factor, not just a replacement for classic value. For investors, it offers a way to capture innovation and modernization while reducing unwanted growth-style baggage. For the industry, it suggests factor models may need to expand to better reflect today’s intangible-driven economy.