The One Factor That Explains the Struggles of Value, International and Small-Cap Stocks | Kai Wu

Episode Summary

Executive Summary: Kai Wu argues that value investing isn’t dead—it’s mismeasured. Traditional book-value-based models ignore intangible assets like IP, brands, talent, and network effects, causing systematic underweights to innovative firms. He explains how machine learning and NLP can quantify intangibles, why these signals are largely uncorrelated with traditional value, and how the approach improves both U.S. and international portfolios.

Main Topics: Why Traditional Value Has Struggled (Priority: 5/5): The conversation frames the recent underperformance of value as a measurement problem: classic value factors emphasize tangible assets and miss the growing economic importance of intangibles. How Intangible Assets Are Measured (Priority: 5/5): Wu describes using alternative data and NLP/AI tools to extract signals from patents, LinkedIn, Glassdoor, trademarks, and other unstructured sources. The Four-Pillar Intangible Framework (Priority: 4/5): His framework groups intangibles into intellectual property, brand equity, human capital, and network effects, then combines them into a composite score. Factor Performance and Portfolio Construction (Priority: 5/5): The podcast discusses backtests showing intangible value works as a standalone factor and improves multi-factor portfolios because it is largely uncorrelated with traditional factors. International and Emerging Markets Applications (Priority: 5/5): Wu extends the framework globally, arguing that weak international performance is largely due to lower intangible investment and that selecting high-intangible firms narrows the U.S.-international gap. AI’s Role in Investing Workflows (Priority: 4/5): The discussion closes with a broader view of AI: best used to structure unstructured data and automate discrete analyst tasks, not to blindly optimize historical factor relationships.

Key Arguments: Value investing is not dead; the problem is that conventional measures like price-to-book ignore intangible capital and therefore misclassify modern businesses. Accounting expensing of intangibles creates a bias against firms that invest in R&D, brand building, talent, and network effects. Capitalizing intangibles helps conceptually, but historical cost data alone is too noisy and nonlinear to fully capture their economic value. Alternative data plus NLP/LLMs make it possible to convert unstructured information into scalable intangible-value signals. The four pillars—IP, brand, human capital, and network effects—capture most economically relevant intangibles in a parsimonious way. Intangible value has little or no correlation with traditional value, size, momentum, market, or quality, so it can diversify a portfolio rather than replace other factors. Companies investing heavily in intangibles may look less profitable today but often exhibit a J-curve, with stronger profitability years later. International and EM underperformance is explained largely by weaker growth, which in turn is linked to lower intangible investment. High-intangible international stocks can match or exceed broad U.S. returns relative to local benchmarks, especially after accounting for valuation and currency effects. AI is most useful in finance for structuring unstructured data and automating repetitive analyst tasks, not for directly mining historical return patterns for predictive factors.

Data Points: Traditional value factor performance history: ~80 years of consistent performance, followed by ~15 years of weakness - Wu contrasts the long historical success of value with the recent period of underperformance. Correlation between intangible value and traditional value: Basically zero - Wu says the two factors are largely uncorrelated over the full sample. Drawdown after adding capitalized intangibles to Fama-French value: 67% drawdown vs. 65% drawdown - Including intangibles improved the factor only modestly in backtests. U.S. stock return advantage over international: About 8 percentage points per year - He attributes much of the divergence between U.S. and international equities to growth differences. Real EPS growth, U.S.: 5.5% per year - Part of the decomposition of U.S. vs. international performance. Real EPS growth, international: -0.1% per year - International stocks were roughly stagnant in U.S. dollar terms over the sample. Valuation contribution to U.S.-international gap: About 2.5 percentage points per year - The remainder of the performance gap after earnings growth differences. Country-level correlation between average intangible value and future EPS growth: ~54% - A scatterplot across countries showed strong predictive association. Weighted country-level correlation: ~70% - When weighting larger countries more heavily, the relationship strengthened. Top patent holders in U.S. patent data: 4 of top 10 are U.S. companies; 6 are non-U.S. - Used to show that global intangible leadership is not exclusively American. Excess return of high-intangible international portfolio: About 4.5 percentage points per year - Relative to the international index, before currency adjustment. High-intangible international portfolio vs. U.S. index after currency adjustment: Basically the same returns since 2010 - Shows that stock selection within international markets can close the U.S. gap. Contribution to international strategy outperformance: ~2/3 stock selection, ~1/3 sector selection - Attribution of returns for the developed international intangible portfolio. Correlation between U.S., developed international, and EM intangible portfolios: ~25% to 50% - Indicates diversification benefits across regions despite similar methodology. U.S. market concentration: About 35% of the S&P 500 is in 7 stocks - Used to motivate diversification outside the U.S. U.S. valuation discount for non-U.S. stocks: Around 50% on Shiller P/E - Supports the case for international exposure. AI task split in finance roles: About half of analyst tasks better done by LLMs and half by humans - From Wu’s work on AI financial analysts.

Pivotal Quotes: "Value investing cannot be dead, right? ... The challenge is more how you measure it and what you're measuring." — Kai Wu: Opening explanation of why traditional value has struggled. "Intrinsic value equals book value plus or tangible value plus intangible value." — Kai Wu: Core framework for understanding why book-value-based value investing misses modern firms. "It's not about U.S. versus international. It's about intangible versus tangible." — Kai Wu: Summarizes the international investing thesis and the source of performance divergence.

Implications: Investors may need to update factor models, accounting assumptions, and portfolio construction to include intangibles. High-quality, innovation-heavy firms can appear expensive on traditional metrics yet still offer strong long-term returns, domestically and abroad.