The 1-over-n Heuristic

The 1-over-n Heuristic

The $1/n$ heuristic is the empirical regularity, due to Benartzi and Thaler ($2001$), that retirement-plan participants tend to allocate roughly equal fractions $1/n$ of their contributions across the $n$ funds offered by the plan, regardless of those funds’ risk-return characteristics. The heuristic is naive diversification: the investor diversifies across whatever is in front of her, with no consideration of what each option actually contains.

Predictive content. If participants used $1/n$, then aggregate equity allocations should depend mechanically on the share of equity-type funds in the menu. Benartzi and Thaler ($2001$) test this prediction in two ways:

• Laboratory experiment. Subjects asked to allocate retirement contributions between (i) a bond fund and a stock fund, (ii) a bond fund and a balanced fund, and (iii) a stock fund and a balanced fund. The implied equity allocations are $54\%$, $35\%$, and $73\%$ respectively — exactly tracking the equity content of each two-fund menu (bond+stock has $50\%$ equity content via $1/n$; bond+balanced has $30\%$; stock+balanced has $75\%$). A rational portfolio chooser should not respond to which two funds are bundled together.
• Field evidence. A sample of $170$ large U.S. $401(k)$ plans. Sorting plans by the fraction of equity-type funds offered, the mean equity allocation rises monotonically: low-equity-share plans average $48.6\%$ in equities, medium-share plans $59.8\%$, high-share plans $64.1\%$ ($p<0.01$). Plan menu composition mechanically determines participant equity exposure.

Mechanism. The heuristic is menu-anchored: participants receive the menu as a “given” choice set and respond by spreading their dollars across it. They do not hold a target equity allocation in mind and back out a portfolio that achieves it; they treat each fund as a category and apply equal weights. The result is a diversification illusion: participants feel they have diversified by holding many funds even when those funds load on the same factors.

Policy implication. Plan menu design becomes a de facto asset-allocation instrument. Changing the share of equity funds in the menu changes participants’ aggregate exposure with no individual making any active decision about that exposure. This is one of the canonical results motivating the nudge agenda in retirement-plan design.

Intuition

A rational portfolio chooser starts with a target asset allocation — say, $60\%$ equities and $40\%$ bonds — and then selects specific funds to implement that allocation. The choice of which funds appear on the menu is irrelevant to the implementation; the investor would simply put $60\%$ of her contributions into whichever fund vehicle delivers equity exposure most cheaply. A naive diversifier proceeds in the opposite direction. She sees a menu of $n$ funds and asks not “what allocation do I want?” but “how should I split my money across these options?” The default answer, with no further reasoning, is equally. The asset-allocation outcome is then a side-effect of the menu rather than a chosen target.

The two laboratory conditions in Benartzi-Thaler are diagnostic precisely because a rational chooser would behave identically under all three: facing a bond+stock menu, a bond+balanced menu, or a stock+balanced menu, she would arrive at the same equity allocation by mixing the available funds. The naive diversifier, applying $1/n$, produces an equity allocation that mirrors the equity content of the menu. The $54\%$/$35\%$/$73\%$ progression in the lab and the corresponding $48.6\%$/$59.8\%$/$64.1\%$ progression in the field both fit the naive prediction tightly.

The result is striking because retirement allocation is a high-stakes decision — the asset mix at age $30$ has enormous consequences at age $65$ — and yet the data show that participants make this decision by applying a heuristic that depends on the arbitrary composition of the menu. The cognitive cost of computing the rational allocation is high enough, and the perceived stakes immediate enough, that participants substitute the easy heuristic.

Properties

(Equal Weighting Across Available Funds) Allocation is $1/n$ regardless of fund characteristics.

The defining property of the heuristic: facing $n$ funds, the participant invests roughly $1/n$ of contributions in each. The fund’s expected return, volatility, fees, or asset class does not enter the decision. This is the strongest version of the heuristic; weaker versions allow some adjustment toward “obvious” categories (more in equities for younger participants, less for older) but retain $1/n$ as the dominant rule of thumb.

(Menu-Composition Determines Exposure) Equity allocation rises with the equity share of the menu.

The empirical implication of $1/n$: aggregate equity exposure across plan participants is a near-linear function of the fraction of equity-type funds offered. Benartzi-Thaler’s $170$-plan sample confirms a near-monotonic relationship; subsequent work (Huberman-Jiang $2006$; Choi-Laibson-Madrian $2009$) replicates the basic pattern across larger samples while documenting departures at extremes (when the menu has very few or very many funds, simple $1/n$ breaks down).

(Diversification Illusion) Participants feel diversified holding many factor-redundant funds.

The $1/n$ allocator can hold $10$ different equity funds and feel “well diversified” even when those funds all load on the same market factor. This generates a true under-diversification result: nominal fund count is high but factor exposure is concentrated. The illusion arises because the heuristic operates on the menu’s surface structure (number of distinct funds) rather than on the underlying factor exposures.

(Non-Linearity at Extreme $n$) The heuristic breaks down when $n$ is very small or very large.

When the menu contains only $2$ funds, $1/n$ implies a strict $50$/$50$ allocation, but observed allocations show some adjustment. When the menu contains $20+$ funds, participants typically pick a smaller subset rather than spreading across all. Huberman and Jiang ($2006$) document that participants cluster at $3$–$5$ chosen funds even in plans with $20+$ options. The pure $1/n$ rule is therefore best understood as a description of the typical allocation behavior over moderate menu sizes, not a literal allocation algorithm.

(Default Funds Reduce $1/n$ Pull) When a plan offers a target-date fund as a default, participants who accept the default escape $1/n$ entirely.

The post-$2006$ rise of target-date funds (TDFs) as default plan investments has weakened the empirical $1/n$ pattern in U.S. data. Participants who simply accept the TDF default hold a single fund with a professionally-managed allocation, bypassing the menu-spreading instinct entirely. This is a clean example of why default design matters (see nudges and default options): the same participant who would otherwise apply $1/n$ accepts whatever single allocation the default specifies.

(Plan-Sponsor Implication) Menu design is a de facto asset-allocation policy.

Because participants apply $1/n$ to whatever menu they receive, the plan sponsor’s choice of menu composition has direct welfare consequences. Adding three new equity funds to a previously bond-heavy menu mechanically raises participants’ equity allocation, even though no individual was asked to rebalance. This is the converse of the rational-chooser benchmark, in which menu composition is irrelevant. The result is one of the most-cited justifications for the choice-architecture literature.

Examples

The Benartzi-Thaler ( $2001$) lab experiment

Three groups of subjects are asked to allocate hypothetical retirement contributions between two funds.

Group	Fund A	Fund B	$1/n$ implied equity allocation	Observed equity allocation
1	Bond fund	Stock fund	$50\%$	$54\%$
2	Bond fund	Balanced fund ($60\%$ equity)	$30\%$	$35\%$
3	Stock fund	Balanced fund ($60\%$ equity)	$80\%$	$73\%$

The three observed equity allocations track the $1/n$ predictions remarkably closely, despite the fact that all three groups face the same underlying choice problem (allocate retirement contributions between equity and fixed income). A rational chooser would not respond to the cosmetic difference of which funds appear on the menu; an $1/n$ chooser does.

Field test on $170$ $401(k)$ plans

Benartzi and Thaler ($2001$) classify each plan by the share of equity-type funds in its menu (low, medium, high). Mean equity allocation across participants in each plan tier:

Equity-fund share	Mean equity allocation
Low	$48.64\%$
Medium	$59.82\%$
High	$64.07\%$

The ANOVA $p$-value is $0.01$. The relationship is not driven by selection (employees do not choose their employer based on plan menu), so the cross-sectional variation in equity exposure is mechanically driven by menu design rather than by participant preferences.

A modern menu redesign

A plan sponsor in $2020$ replaces three previously-included sector equity funds (technology, health care, real estate) with a single broad-market index fund. Net result: one new equity fund replaces three. By $1/n$, participants who were spreading evenly across the original menu would have allocated $3/n$ to equities through these three funds; after the change, they allocate $1/n$. Aggregate equity exposure falls by approximately $2/n$ of plan assets without any individual making an active decision about the change. The plan sponsor has reduced participant equity exposure by changing the menu’s surface structure.

See Also

• 8.11. Bounded Rationality — parent hub.
• 8.10.2. Under-Diversification — closely related anomaly; $1/n$ is one mechanism that generates it.
• 8.12.3. Nudges and Default Options — policy companion: how default funds and plan design exploit the heuristic.
• 8.10. Behavioral Household Finance — broader hub on household-portfolio anomalies.
• 8.11.1. Limited Attention and PEAD — sibling under-attention bias on the asset-pricing side.