Last week, I outlined the views of Joe Davis, Vanguard’s Chief Economist, who argues in his recent book that the economic and market outlook for the next decade largely hinges on whether AI delivers meaningful productivity gains.

In the bull case, AI performs as hoped. Productivity improves, economic growth remains durable, markets benefit, and the US is better able to manage the risks associated with large and growing deficits and debt.

The bear case is more sobering. AI proves to be more like social media, with high engagement but limited productivity. In that scenario, debt and deficits take center stage and become a heavier drag on growth, and market returns suffer accordingly.

As I noted last week, in Davis’s framework, both stocks and bonds perform well in the “AI Transforms” scenario and struggle in the “AI Disappoints” scenario. This week, I want to highlight some of his more nuanced views on market forecasts and risk.

Looking at the decade beginning in 2027, Davis focuses on a classic US 60/40 stock-bond portfolio. He does so in part because he believes US investors tend to be under-diversified with respect to foreign stocks.

Under the AI Transforms scenario, he estimates a median annual return of 7.6 percent. That figure is before inflation and represents the midpoint of a wide distribution of outcomes. In this favorable scenario, he assigns a five percent probability to a negative real, or inflation-adjusted, return.

Under the AI Disappoints scenario, the outlook deteriorates meaningfully. Davis estimates an annualized real return of 3.6 percent over the same period before inflation, accompanied by a 38 percent probability of a negative real return.

Because it is impossible to know whether AI will ultimately transform or disappoint, Davis constructs what he calls a “Megatrends-Aware” portfolio, designed to perform more resiliently across both scenarios.

Davis’s Megatrends-Aware portfolio has three primary elements.

First, he recommends adding non-US stocks, anywhere from 16 to 40 percent of the equity allocation. A typical Acropolis portfolio allocates about 25 percent of equities to non-US stocks, putting us comfortably within his suggested range.

Second, Davis argues for a meaningful tilt toward value stocks, on the order of 25 to 40 percent of equities. Because value can be defined and implemented in different ways, it isn’t easy to make a precise comparison to our own positioning.

The broader point, however, is clear: he does not expect a strong growth tilt to be as profitable in the next decade as it was in the last.

He reasons that for AI to transform the economy, profits must spread beyond a small group of technology firms that are building the infrastructure. Productivity gains would need to flow across industries, leading to more broadly distributed earnings growth rather than continued tech-stock concentration.

Finally, Davis suggests modestly increasing bond exposure, even if yields rise in the AI Disappoints scenario. Higher yields, he argues, ultimately benefit long-term investors through improved income and reinvestment opportunities, much as they did following the 2022 rate shock.

By taking these three steps, Davis estimates that the Megatrends-Aware portfolio would generate an annualized real return of 5.0 percent in the AI Disappoints scenario. That represents a meaningful improvement over his 3.5 percent estimate for an unaware portfolio, while reducing the probability of a negative real return to 19.4 percent.

The trade-off, of course, is modestly lower upside in the more favorable scenario. Davis estimates that the Megatrends-Aware portfolio would center around a 7.1 percent return in the AI Transforms case, roughly half a percent below the non-aware portfolio. Even so, the probability of a negative real return declines further, from five percent to just 2.6 percent.

None of us knows what the future holds, and Davis is clear that many outcomes lie outside the scenarios he presents. Davis’s framework underscores a reality investors often overlook: the greatest risk is not being wrong about the future, but building portfolios that depend on being right.

That view is consistent with how we think about long-term investing, and why we believe our clients are well-positioned regardless of how the AI story ultimately unfolds.