In a research paper published in August 2026, Inigo Fraser Jenkins of AllianceBernstein makes a provocative argument: the defining feature of ultra-long horizon investing1 is not a specific asset allocation. It is governance. The observation is deliberately provocative. Jenkins acknowledges the irony of mapping a century-long investment framework at precisely the moment when AGI, climate change, and geopolitical disorder have combined to produce "a greater level of path uncertainty than that experienced perhaps ever in the period of modern investing." And yet he makes the case that the longer the horizon, the more urgently investors need a structured framework to guide decisions across radically different states of the world.
The Long Run Is Not the Short Run, Compounded
The paper opens by dismantling a persistent misreading of investment time horizons. "The long run is not just a concatenation of short runs; the process of investing should differ." The distinction matters because short-horizon and long-horizon investing pull in genuinely different directions — on benchmarks, on risk metrics, on the role of liquidity, and especially on what diversification is actually supposed to achieve.
For Jenkins, the traditional architecture of long-only, market-cap-weighted, 60/40 investing reflects a period — roughly 1980 to 2022 — that was anomalously benign for financial assets. That period featured falling inflation, rising real returns, and a reliable negative stock-bond correlation that made bonds a credible diversifier. None of those structural tailwinds can be assumed to persist. On the longer sweep of history, a 60/40 portfolio beating inflation is, in his assessment, essentially a coin flip once the time horizon extends far enough.
"Different This Time" — And Possibly Correct
The phrase "it's different this time" is usually a red flag. Jenkins deploys it carefully. Two civilizational-level forces — AI and climate change — represent genuinely unprecedented dispersion of outcomes, not just elevated headline risk. The Dallas Fed chart reproduced in the paper illustrates this starkly: the range of plausible AI-driven GDP per capita paths spans from continued trend growth to singularity-level expansion to human extinction. For climate, the key implication is not a downward revision to the central growth forecast but the dramatic widening of the distribution around that central case. Nonlinear tipping points become more probable past the 1.5 to 2.0 degree warming range, and those tipping points carry implications for migration, political stability, and capital market covariance that no model can fully price.
The investor's response to this regime of wider distributions is not to seek "lower risk" in the traditional sense. "Lower risk does not mean a shift to the traditional safe-haven assets of the last 40 years." Instead, the appropriate response is diversification across economic regimes — building portfolios that survive in materially different states of the world.
Real Assets, Real Returns
The allocation conclusion that flows from this analysis is striking in its directness. Jenkins argues that an ultra-long horizon portfolio should be biased "strongly, perhaps exclusively, toward real assets." Equities qualify as real assets, but so do directly held real estate, farmland, timberland, and infrastructure. The case for illiquid assets is explicitly stronger when governance permits true long-horizon measurement. Gold earns a role as well, with Jenkins noting "strong evidence that over long horizons all fiat currencies dramatically depreciate against gold" and citing its potential hedge value against system-collapse scenarios.
Bonds, by contrast, require a fundamental rethinking. Over a 26-year holding period, the annualized volatility of US equities and 10-year government bonds actually converges — a counterintuitive result that dissolves the received wisdom of stocks as inherently riskier. The real risk hierarchy depends on the investment horizon, and for very long horizons, the probability of sustaining a real loss from equities declines sharply relative to nominal bonds.
The Perils of Survivorship Bias
Among the paper's sharpest observations is a warning about extrapolating from cap-weighted equity history. In 1899, the US and UK equity markets were the dominant global weights — and they delivered positive real returns. The next six largest markets at the time subsequently fell to zero. That history is rarely incorporated into strategic return assumptions, which remain heavily anchored to US outcomes. "There is therefore a danger in extrapolating data based on US returns too far into the future, or for a global forecast." Jenkins is not dismissive of the US structural case — he explicitly endorses continued US strategic exceptionalism in demographics, corporate profitability, and energy security — but he flags geographic concentration as a genuine long-horizon risk that cap weighting obscures.
Fundamentals Over Valuation at Truly Long Horizons
The paper's treatment of return forecasting is nuanced. Valuation metrics, including the Shiller CAPE, gain predictive power as the time horizon extends toward 10 years. But past a certain point, their efficacy declines sharply — by 30 years, the Shiller PE has almost no predictive power for forward equity returns. What matters then is a fundamentals-based return model: income yield plus real dividend growth, decomposed into GDP growth per worker, working-age population growth, and changes in profit share of GDP. On those inputs, Jenkins projects long-run real equity returns of approximately 4.5% for the US and 4.3% for the developed world — below historical averages, but positive.
5 Key Takeaways for Advisors and Investors
- Governance is the strategy. For ultra-long horizon investors, the investment process and benchmark structure matter more than any particular asset allocation decision. Inflation protection, not market-relative performance, should be the governing benchmark.
- The 60/40 does not reliably beat inflation over very long horizons. The benign decades from 1980 to 2022 are the exception, not the template. Advisors anchoring long-term planning to that period should reassess the baseline.
- Real assets should anchor the portfolio. Equities, real estate, farmland, timberland, infrastructure, and gold all fit the mandate of preserving purchasing power across diverse economic regimes. Nominal long-duration bonds do not carry the same structural case.
- Volatility is the wrong risk metric for long-horizon investors. The probability of sustaining a real loss over extended periods is the more relevant gauge. On that measure, equities outperform nominal bonds decisively once the horizon spans multiple business cycles.
- Survivorship bias inflates long-run return assumptions. Cap-weighted global equity history flatters the outcome because the US happened to survive and thrive. Meaningful geographic diversification beyond current benchmark weights is a prudent discipline for truly long-horizon mandates.
Footnote:
1 Fraser Jenkins, Inigo, Alla Harmsworth, Robertas Stancikas, and Maureen Hughes. "The 100-Year Portfolio: A State of Mind Rather Than an Allocation." AllianceBernstein, Aug. 2026, https://www.alliancebernstein.com/content/dam/global/insights/insights-whitepapers/100-year-portfolio.pdf.
