Bank of America September Fund Manager Survey: Cash rebounds, why is risk still high

Fund managers have begun to pull back some risk, yet there is no consensus on a defensive allocation. Cash levels are rising and equity overweights are declining, accompanied by an even deeper underweight in bonds and continued crowding in semiconductors. The market’s key divergence is shifting from whether growth can be sustained, to whether growth, interest rates, and financing costs can all be maintained simultaneously.

Sentiment is cooling, but risk allocations remain considerable

The Bank of America September Global Fund Manager Survey shows average cash ratios rose from 3.5% in August to 3.9%, global equity net overweight dropped from 56% to 49%, and the composite sentiment index fell from 8.0 to 7.0, the lowest in three months. Compared to the high-risk allocation in August, the clearest change in September was that institutions began to increase their buffer and reduce some risk exposure.

This survey was conducted from September 4 to 10 and released on September 15. There were 170 global respondents overseeing $470 billion in assets; including regional surveys, the total sample was 190 people. It reflects judgment and allocations during the survey period and should not be seen as real-time positioning for each daily market move post-publication.

The three figures cited are not directly comparable. 3.9% is the average portion of assets in cash; the 49% equity net overweight is the percentage reporting an overweight minus those reporting an underweight— it does not mean the funds have allocated 49% of assets to equities. The sentiment index synthesizes growth expectations, cash, and equity allocations, and cannot be directly converted into buy or sell amounts.

This limits the inference one can make from monthly changes. The August global survey had 180 people managing $525 billion in assets; both the number of participants and assets changed in September. The difference in asset scale between surveys is not redemption amount, and a decline in net overweight does not prove every fund cut exposure. The survey can detect a shift in consensus, but cannot substitute for actual fund flow data.

Nevertheless, the cooling is meaningful in itself. A 0.4 percentage point increase in cash provides more room for portfolio adjustments. While equity net overweight fell, it is still approximately 0.9 standard deviations above its historical average. Thus, September reflects a pullback from a relatively high risk level, rather than entering an overall pessimistic, low-exposure environment.

Why reduce equities but not rotate into bonds?

The traditional defensive strategy is to reduce equities and increase bonds, but the September survey shows a different pattern: net bond underweight widened from 39% to 48%, real estate net underweight widened from 7% to 21%, and net commodity overweight fell from 24% to 19%. The rising appeal of cash has not brought a consensus preference for other defensive assets.

Institutional outlook on the economy helps explain this allocation. 55% of respondents expect a “no landing” scenario for the global economy in the next year, with only 2% choosing hard landing; in another macro trend question, 50% selected “stagflation,” meaning below-trend growth and above-trend inflation coexisting. The definitions differ, so these responses don’t directly contradict each other.

Continued economic growth does not require growth to exceed trend. Meanwhile, if inflation remains elevated, central banks will find it hard to provide monetary easing quickly. Corporates may still achieve nominal revenue growth, maintaining a net overweight to equities. Bonds, however, face rate and inflation constraints. In the survey, the net proportion expecting short-term rates to rise reached 36%, the highest since September 2022.

The choice of greatest tail risk further reinforces this view: 33% picked disorderly bond yield spikes, 28% selected the AI bubble, and 24% chose a second wave of inflation. These represent the proportion of respondents choosing sources of risk, not probabilities of these events, nor can they be simply summed to compute market downside probability.

Yield changes impact several pricing chains at once. Rising yields increase the hurdle rate for discounting future profits, raise corporate financing costs, and also heighten the relative attractiveness of holding cash. If profits continue to grow but growth is insufficient to offset these changes, equities could come under pressure alongside bonds. Thus, September’s caution centers on interest rates and pricing conditions, not entirely on recession fears.

Bearishness on bonds could itself become crowded. Shorting US Treasuries is the second most crowded trade, picked by 18%. Should inflation, growth, or bond supply developments fall short of market fears, this consensus could face a reversal. Thus, concentrated risks include both an exit from long equities and a squeeze on short bonds. High concern over rising yields does not mean yields can only move in one direction.

AI investment grows more certain, but credit concerns are mounting

Views on AI are equally two-sided. 79% of respondents expect that major AI players won’t announce capex cuts in 2026, up from 71% in August. Institutions still believe the investment boom will last, and that demand for related equipment, construction, and services has not turned broadly bearish.

Meanwhile, the net proportion seeing corporate overinvestment reached 33%, matching February’s level; 42% of respondents cited AI majors’ capex as the most likely source of a systemic credit event, up from 38% in August. Here, 42% refers to the proportion choosing that source of risk, not to a belief in a 42% probability of a systemic crisis.

The denominators for these two risk questions must be considered separately. 28% picked an AI bubble as their primary tail risk across multiple options; 42% chose AI capital spending as the most likely origin of a credit event. The same respondent could select both; the difference in proportions does not prove risk has shifted from equities to credit markets.

All three responses can be true at once. Firms must keep investing to remain competitive, yet not every investment will recoup its cost quickly. Construction spending occurs first, while revenue and cashflow returns may lag; if the payback period stretches and financing costs continue to rise, the strain on balance sheets will intensify.

Thus, the divergence is shifting from “will investment continue” towards “can investment deliver adequate returns.” For upstream industrial chains, sustained capex can support orders; for funders and creditors, the key is how much spending is sustained by operating cashflow versus borrowing, and how soon new businesses generate cash revenue. The same investment growth may not bring the same return to every stage of the value chain.

The most crowded trade remains global semiconductor longs, chosen by 53%— unchanged from August; the tech sector net overweight is also steady at 30%. These data show that AI-related conviction has not markedly faded. At the same time that credit risk concerns rise, core tech allocations are retained, indicating that institutions want to participate in growth but are increasingly wary of its cost.