Lecture 6: Guest Speaker David Swensen

What this lecture covers

David Swensen, Yale's chief investment officer since 1985, joins Professor Shiller's class to defend the "Yale Model" of endowment management against a 2008 Barron's article that blamed it for being too aggressive during the financial crisis. Swensen uses the criticism as a starting point to explain the three tools available to any investor - asset allocation, market timing and security selection - and to show, with Yale's own numbers, why a diversified, equity-heavy, alternatives-heavy portfolio has outperformed over the long run.

The lecture sits alongside the course's earlier sessions on portfolio theory and gives a practitioner's view of ideas Shiller has been building up in lecture, including diversification and risk versus return. After watching, you should be able to explain why asset allocation dominates investment returns, why market timing and security selection are negative-sum activities on average, and why illiquid, less efficiently priced markets reward active management more than bond or large-cap stock markets do.

Key ideas

• Asset allocation: the decision of how much of a portfolio to hold in each asset class; Swensen argues it explains the vast majority, and in aggregate more than 100%, of investor returns because timing and selection are net-negative activities.
• Market timing: short-term deviation from long-run target weights; Swensen cites mutual fund data showing dollar-weighted returns consistently trail time-weighted returns because investors buy after gains and sell after losses.
• Security selection: betting on individual securities within an asset class; it is a zero-sum game before costs and a negative-sum game after fees, commissions and taxes.
• Market efficiency and dispersion: the spread between top-quartile and bottom-quartile manager returns is small in efficient markets like bonds and large-cap stocks, and much wider in venture capital, buyouts and real estate, which is where active management can add the most value.
• Diversification as a "free lunch": spreading a portfolio across many asset classes can raise expected return for a given risk level, or lower risk for a given return, an idea Swensen traces to Harry Markowitz and Jim Tobin.
• Equity orientation: Ibbotson's long-run data (1925-2009) show stocks and small stocks vastly outperforming bonds and bills over decades, which justifies a long-horizon investor holding heavy equity exposure despite short-run crashes.
• Diversification still fails in a panic: in 1987, 1998 and 2008 nearly all risky assets sold off together as investors fled to Treasuries, but Swensen argues holding a large permanent Treasury position is too costly over a full cycle.
• Survivorship bias: fund performance databases that exclude funds that closed understate how hard it is to beat the market, since failed funds are disproportionately the losers.

Walkthrough

Barron's criticism and the Yale Model's origins (0:02)

Shiller introduces Swensen and his record growing Yale's endowment from under $1 billion in 1985 to $16.7 billion by mid-2010. Swensen recounts how a 2008 Barron's article, written after the Lehman Brothers collapse, blamed the "Yale Model" for over-relying on illiquid alternative assets and under-owning stocks and bonds. He uses this as a frame for the rest of the talk: he will explain what the Yale Model actually is and test whether Barron's criticism holds up. He recalls arriving at Yale in 1985 and finding that most college endowments held roughly 50% domestic stocks, 40% bonds and cash, and 10% alternatives, a mix he judged undiversified and overweight in low-return bonds given endowments' very long time horizons.

The three tools of a portfolio manager (11:25)

Swensen lays out the framework he uses for the rest of the talk: asset allocation, market timing and security selection are the only three ways a portfolio's returns can differ from a passive benchmark. He explains why security selection and market timing are zero-sum before costs, and negative-sum after fees, commissions and the cut taken by Wall Street, which is why asset allocation ends up explaining more than 100% of the variance in institutional returns once those leakages are subtracted out.

Why asset allocation dominates (15:32)

Using Roger Ibbotson's long-run data from 1925 to 2009, Swensen shows that a dollar in Treasury bills grew 21 times, in Treasury bonds 86 times, in large stocks about 2,592 times, and in small stocks over 12,000 times. This is the evidence for holding heavy equity exposure when a portfolio has a long horizon. He balances this with the 1929-1932 crash, where a small-stock portfolio lost 90% of its value peak to trough, to explain why even long-horizon investors need diversification rather than a single high-return asset class.

Market timing and security selection in practice (29:54)

Swensen turns to evidence that investors are systematically bad at timing markets: Morningstar data show dollar-weighted mutual fund returns trailing time-weighted returns in every category, because investors pile into funds after strong performance and pull out after weak performance; he illustrates this with internet funds during the dot-com bubble, where investors lost 72% of invested capital despite a positive average annual return. On security selection, he cites a study putting the odds of beating the market after fees and taxes at about 14%, and explains that survivorship bias in fund databases makes even that figure too optimistic, since roughly a third of funds in a bias-free database had failed and disappeared.

Where inefficiency creates opportunity (40:09)

Swensen presents the spread between top-quartile and bottom-quartile manager returns across asset classes as a rough measure of market inefficiency: about half a percentage point in bonds, two points in large-cap stocks, four in foreign stocks, and much larger gaps of 7 to over 40 percentage points in absolute return, real estate, buyouts and venture capital. This dispersion is why Yale concentrates its active management effort in less efficient, illiquid markets rather than trying to beat the bond or large-cap stock market.

Revisiting Barron's and Q&A (45:18)

Swensen returns to the Barron's criticisms directly: diversification does fail in a panic, as it did in 1987, 1998 and 2008, but holding a large permanent allocation to Treasuries to hedge against rare panics carries a high opportunity cost over a full cycle, and he points to Japan's Nikkei, which lost 73% of its value over 20 years, as a reminder that even equity-biased diversified portfolios can disappoint. He closes with Yale's 10- and 20-year performance figures relative to average college endowments, then takes student questions covering the principal-agent problem of hiring outside managers, why individuals should generally use index funds rather than try to replicate Yale's approach, valuation and technology investing, the effect of the growth of hedge funds and private equity fees on investment opportunity, and why he avoids using the Sharpe ratio to describe Yale's risk-adjusted performance.

Before you watch

• It helps to have followed the course's earlier lectures on portfolio theory, diversification and risk-return trade-offs, since Swensen assumes familiarity with these ideas rather than defining them from scratch.
• A basic sense of what stocks, bonds, hedge funds, private equity and venture capital are will make the discussion of asset classes easier to follow.

Check your understanding

1. Why does Swensen argue that asset allocation can explain more than 100% of an institutional portfolio's returns?
2. What evidence does he give that individual and institutional investors tend to time markets badly?
3. Why does the spread between top-quartile and bottom-quartile manager returns differ so much between bonds and venture capital, and what does Yale do with that information?
4. According to Swensen, why doesn't a large permanent allocation to Treasury bonds make sense for a long-horizon investor, even though it would help in a panic?
5. Why does he avoid citing the Sharpe ratio when describing Yale's performance?