Salience and Proportional Thinking in Reaching for Yield

Salience and Proportional Thinking in Reaching for Yield

The salience and proportional-thinking mechanism for reaching for yield rests on Bordalo-Gennaioli-Shleifer ($2013$) salience theory: investors over-weight features of an option that are salient relative to alternatives. The risky asset’s expected return becomes more salient (and therefore more attractive) when the ratio of the risky asset’s return to the safe asset’s return is large.

Salience formalization. Lian-Ma-Wang ($2019$) write the investor’s effective optimization as a salience-weighted mean-variance problem:
${\max }_{\varphi \in [0,1]}\delta \mathbb{E}[r_p]-\frac{\gamma }{2}\text{Var}(r_p),$
where $\delta \ge 1$ is a salience-weight on the expected return that increases with the ratio $(r_f+\mathbb{E}[x])/r_f=(\mathbb{E}[r])/r_f$. As the ratio rises (because $r_f$ falls or $\mathbb{E}[x]$ rises), the expected-return component receives greater weight in the investor’s objective, increasing the optimal allocation to the risky asset.

Worked numerical comparison.

• High-rate condition: safe $5\%$, risky $10\%$. Ratio: $10/5=2\times$.
• Low-rate condition: safe $1\%$, risky $6\%$. Ratio: $6/1=6\times$.

Both conditions have an excess return of $5$ percentage points and an identical Sharpe ratio. But the ratio is much higher in the low-rate condition. Salience theory predicts that the low-rate ratio attracts more attention to the risky asset’s expected return relative to its risk, and the investor allocates more.

The gross-return framing test (Test 3). Lian-Ma-Wang’s sharpest test of salience operates on framing. Net-return framing (“the risky asset’s average return is $6\%$”) emphasizes the ratio $6\%/1\%$. Gross-return framing (“for every Franc you invest, you get $1.06$ Francs on average”) removes the ratio salience by expressing returns as fractional changes. Under the salience mechanism, gross-return framing should weaken the reaching-for-yield effect:

Condition	Baseline (net)	Gross	Difference
High rate ($5\%$/$10\%$)	$57.13\%$	$52.65\%$	$4.47$ pp
Low rate ($1\%$/$6\%$)	$64.51\%$	$54.44\%$	$10.06$ pp
Difference (High–Low)	$7.38$ pp	$1.79$ pp	—

The reaching-for-yield effect is large under net framing ($7.38$ pp) and small under gross framing ($1.79$ pp). Gross-return framing dampens the effect by approximately $76\%$. The result confirms that salience and proportional thinking are operating; the effect is not solely driven by reference dependence (which would not be sensitive to net-vs-gross framing).

The two mechanisms operate jointly. The reference-dependence and salience mechanisms predict overlapping but separable patterns. Both predict reaching-for-yield. Reference-dependence predicts history dependence (Test 2 confirms). Salience predicts framing-sensitivity (Test 3 confirms). Both predict non-linearity (Test 1 confirms). The full empirical picture supports both mechanisms operating jointly; neither alone fully accounts for the observed pattern.

Intuition

The salience mechanism is closer to a perception-side bias than to a preference-side one. Whereas reference dependence changes how the investor evaluates outcomes (loss-aversion over reference deviations), salience changes how she attends to the available options. The risky asset has a higher mean return than the safe asset; the “by how much” is a ratio question, and ratios are more salient at low denominators.

The everyday intuition is familiar from consumer-pricing examples. A \10$savingsona$$50$purchase($20%$off)feelsmoreattractivethana$$10$savingsona$$5{,}000$purchase($0.2%$off),eventhoughthedollaramountisthesame.Theproportionalthinkingembeddedinsaliencetheorycapturesexactlythisasymmetry.Appliedtoportfoliochoice,thesamelogicpredictsthata$5$-ppexcessreturnlooksmoreattractivewhenthesaferateis$1%$thanwhenitis$5%$ — the ratio is bigger.

The gross-return framing test is the key empirical separator from reference dependence. Reference dependence operates on absolute return levels (where the reference rate $r_{\text{ref}}$ is a level); reframing returns as gross multiplicative factors does not change the absolute levels and should not affect a reference-dependent investor. Salience operates on ratios; reframing returns as multiplicative factors makes the difference between (e.g.) $1.06$ and $1.01$ look small (it is just a $5$-pp difference between two near-unity numbers) and dampens the salience boost. The data confirm the distinct prediction of salience by showing that gross-return framing dampens the effect substantially.

Properties

(Salience-Weighted Mean-Variance) $\delta$ inflates the perceived expected return.

${\max }_{\varphi \in [0,1]}\delta \mathbb{E}[r_p]-\frac{\gamma }{2}\text{Var}(r_p).$
The salience weight $\delta \ge 1$ multiplies the expected-return component of the standard mean-variance objective, increasing the optimal allocation to the risky asset. The standard benchmark corresponds to $\delta =1$. The Bordalo-Gennaioli-Shleifer mechanism makes $\delta$ depend on contextual ratios: $\delta$ rises when the risky asset’s return is salient relative to the safe asset.

(Ratio-Driven Salience) $\delta$ rises with $(r_f+\mathbb{E}[x])/r_f$.

The salience weight is increasing in the ratio of the risky asset’s gross return to the safe asset’s gross return: $(r_f+\mathbb{E}[x])/r_f=\mathbb{E}[r]/r_f$. As $r_f$ falls (holding $\mathbb{E}[x]$ fixed), the ratio rises and $\delta$ rises. The ratio mechanism is what generates the prediction that low rates increase risk-taking — not the level of the risky-asset return per se, but the ratio of that return to the alternative.

(Comparison Numbers) $6/1=6\times$ vs $10/5=2\times$ — same excess return, different ratio.

The two Lian-Ma-Wang conditions have identical excess returns ($5$ pp) and identical Sharpe ratios. They differ only in the ratio of risky to safe returns: $6\times$ in the low condition, $2\times$ in the high condition. Salience predicts that the larger ratio in the low condition triggers a larger $\delta$ and therefore higher allocation. The data confirm: allocations are $\sim 8$ pp higher in the low condition, despite identical excess returns.

(Gross-Framing Test — Test 3) Reframing returns as gross multiplicative factors weakens the effect.

Lian-Ma-Wang’s Test 3 manipulates the framing of the same underlying lottery:

• Net framing: “The average annual return is $6\%$.” (Emphasizes the ratio $6\%/1\%$.)
• Gross framing: “For every Franc you put into Investment B, you get $1.06$ Francs on average.” (Removes the ratio salience.)

The reaching-for-yield effect is $7.38$ pp under net framing and only $1.79$ pp under gross framing — a reduction of approximately $76\%$. The result is the cleanest evidence that salience is operating: a purely cosmetic reframing of the same underlying lottery cuts the reaching-for-yield effect by three-quarters.

(Distinct From Reference Dependence) Framing-sensitivity is the discriminating signature.

Reference-dependence predicts that the reaching-for-yield effect should be present regardless of framing — what matters is the absolute level of $r_f$ relative to the reference $r_{\text{ref}}$, not how the returns are presented. The fact that gross-return framing dampens the effect by $76\%$ shows that something beyond reference dependence is at work; that something is salience. The two mechanisms are therefore identified by separate empirical tests: reference dependence by history dependence (Test 2), salience by framing sensitivity (Test 3).

(Non-Linearity Predicted by Salience) Effect grows non-linearly as $r_f$ falls.

Both reference dependence and salience predict non-linearity in the rate-allocation relationship. The salience mechanism predicts that the effect grows with the ratio $\mathbb{E}[r]/r_f$, which itself grows non-linearly as $r_f$ falls. As $r_f\to 0^{+}$, the ratio diverges and the salience effect should be very large. The empirical pattern of allocations rising sharply at very low rates is consistent with the salience prediction (and also with the reference-dependence prediction, so the non-linearity test alone does not separate the mechanisms).

(Joint Operation) Both mechanisms together explain the full pattern.

The reaching-for-yield literature has converged on the view that both mechanisms operate jointly. Reference dependence explains why the effect appears below a threshold and why it depends on the path of past rates. Salience explains why framing the same returns differently changes the effect’s magnitude. Together they account for the observed non-linearity, history dependence, and framing sensitivity. Neither alone is sufficient; the empirical pattern requires both behavioral inputs.

Examples

$1\%$ vs $6\%$ versus $5\%$ vs $10\%$

Two pairs of investments with identical excess return:

• Pair A: safe $5\%$, risky $10\%$. Ratio: $2\times$. Excess return: $5$ pp.
• Pair B: safe $1\%$, risky $6\%$. Ratio: $6\times$. Excess return: $5$ pp.

A salience-driven investor perceives the risky asset in Pair B as much more attractive than in Pair A because the ratio is much higher, even though the excess return is identical. She allocates more to the risky asset in Pair B. The mean-variance benchmark would predict identical allocations because excess return and variance are identical.

The gross-framing reframing

The same Investment B in the low-rate condition can be described two ways:

• Net frame: “Average annual return is $6\%$, with possible outcomes ranging from $-25\%$ to $+40\%$.” Investor compares $6\%$ to safe $1\%$: a $6\times$ ratio. Risky asset looks attractive.
• Gross frame: “For every Franc you invest, you get $1.06$ Francs on average, with outcomes ranging from $0.75$ to $1.40$ Francs.” Investor compares $1.06$ to safe $1.01$: a $1.05\times$ ratio. Risky asset looks much less attractive.

Both frames describe the identical lottery in mathematical terms. The salience-driven investor responds to the framing, allocating much less to the risky asset under gross framing. The empirical $76\%$ reduction in the reaching-for-yield effect under gross framing is the clean evidence for this mechanism.

Why salience matters for monetary policy

The Federal Reserve’s setting of the federal funds rate determines the safe asset’s return for U.S. retail savers. When the Fed cuts rates from (say) $5\%$ to $0.5\%$, the salience mechanism predicts a substantial reaching-for-yield response from retail investors that would not occur under a pure mean-variance benchmark. The $0.5\%$ rate makes the comparison ratio against any reasonably-yielding risky asset extremely large, and the risky asset’s return becomes salient. This is the mechanism behind the “risk-taking channel of monetary policy” that several Fed officials (Bernanke $2013$; Stein $2013$; Powell $2017$) have flagged as a side-effect of low-rate environments. The salience mechanism implies that the risk-taking response is partly an information-presentation phenomenon, not just a fundamental change in investor preferences.

See Also

• 8.15. Reaching for Yield — parent hub.
• 8.15.1. Reference Dependence in Reaching for Yield — sibling: complementary behavioral mechanism operating jointly with salience.
• 8.6. Belief-Based Biases — broader category of belief-side cognitive biases that salience belongs to.
• 8.6.5. Representativeness Heuristic — closely related cognitive bias affecting probability assessment.
• 8.5. Prospect Theory — the broader behavioral framework that includes both salience and reference dependence.