Catering Theory of Dividends

Catering Theory of Dividends

The catering theory of dividends (Baker-Wurgler $2004$) is the proposition that firms initiate dividends when investor sentiment for dividend payers is high and omit dividends when investor sentiment for non-payers is high. The mechanism is the same as in equity issuance market timing: managers respond rationally to time-varying mispricing of a salient category — here, the dividend payer category.

The dividend premium. Catering is operationalized through the dividend premium, the difference between the average market-to-book ratio of dividend-paying firms and the average market-to-book ratio of non-payers:
${\text{DivPrem}}_t=\log \left(\frac{(M/B{)}_t^{\text{payers}}}{(M/B{)}_t^{\text{nonpayers}}}\right).$
A high dividend premium means investors are willing to pay more for dividend-paying firms relative to comparable non-payers — a signal that sentiment for the “dividend payer” category is currently positive. A low (or negative) dividend premium means non-payers trade at higher multiples than payers — sentiment is currently against dividends.

The central empirical finding. Baker and Wurgler ($2004$) document that the rate of dividend initiations — the fraction of non-payer firms that switch to paying dividends in year $t+1$ — is strongly positively correlated with the dividend premium in year $t$. The correlation in the U.S. data ($1962$–$2000$) is approximately $0.4$. Dividend omissions show the symmetric pattern: payers omit dividends more frequently when the dividend premium is low.

Why the firm responds. A non-paying firm whose comparable payers are trading at a premium can capture the premium by initiating dividends — its market-to-book multiple should rise toward the payer-category average. A paying firm whose comparable non-payers are trading at a premium can capture the symmetric premium by omitting dividends. The decision to initiate or omit therefore tracks the cross-sectional dividend premium without any change in the firm’s underlying cash flows or dividend-policy fundamentals.

Mechanism — categorical thinking and time-varying sentiment. The catering theory rests on two behavioral primitives. First, investors categorize firms by whether they pay dividends and form sentiment at the category level (see categorical thinking). Second, the category-level sentiment varies over time, perhaps in response to macroeconomic conditions, salient news events, or generational shifts in investor preferences. The combination produces time-varying mispricing of the payer category, which managers then exploit by changing their firms’ category membership.

Generality. Catering as an explanation is not specific to dividends. The same logic applies to closed-end fund initiations (when CEFs trade at premia), to merger waves (when stock-financed mergers are favored), and to corporate name changes (firms add or drop “dot com” suffixes in response to investor sentiment for that category). The dividend application is the cleanest because the dividend-payer category is salient, well-defined, and binary.

Intuition

Why would a firm change its dividend policy in response to time-varying investor enthusiasm for dividends? The catering theory’s answer is that the firm captures a valuation premium by joining (or leaving) the favored category. The mechanism does not require the firm’s underlying cash-flow capacity to change; it requires only that investors price the category differently from time to time and that the firm can switch categories at low cost.

The contrast with classical dividend theories is sharp. In the Miller-Modigliani frictionless benchmark, dividend policy is irrelevant; in the signaling theory, dividend changes communicate private information about cash flows; in the agency-cost theory, dividends discipline managers’ free cash flow. None of these frameworks predicts that the firm should initiate dividends in response to aggregate investor sentiment for dividends — yet the data show exactly that pattern. The catering theory was developed precisely to explain the dividend-premium correlation that the classical theories cannot.

The behavioral input is not investors who specifically want dividends in some periods and not in others — it is investors who categorize firms by dividend status and whose category-level sentiment varies. The firm’s decision to initiate or omit is the rational response to this category-level sentiment, the same way that issuing equity when the firm is overvalued is the rational response to firm-level sentiment.

Properties

(Definition) The dividend premium is the log market-to-book gap between payers and non-payers.

${\text{DivPrem}}_t=\log \left(\frac{\overline{(M/B{)}_t^{\text{payers}}}}{\overline{(M/B{)}_t^{\text{nonpayers}}}}\right).$
A positive premium means the market values dividend-paying firms at higher multiples than otherwise-comparable non-paying firms. Baker-Wurgler ($2004$) construct the premium for U.S. nonfinancial firms over $1962$–$2000$. The series is highly time-varying: it is large positive in the $1970$s, declines through the $1980$s and $1990$s, and turns negative around the dot-com peak of $1999$–$2000$. The variation gives the cross-sectional regression substantial power.

(Initiation Correlation) High dividend premium predicts high initiation rate.

The headline result: the rate of dividend initiations among previously-non-paying U.S. firms in year $t+1$ is strongly positively correlated with the dividend premium in year $t$, with a correlation of approximately $0.4$ in the long sample. Time-series visualization (Baker-Wurgler $2004$, Figure $2$) shows the two series tracking closely: the early-$1970$s initiation peak coincides with a peak in the dividend premium; the late-$1990$s collapse in initiations coincides with a sharp negative dividend premium.

(Omission Symmetry) Low (or negative) dividend premium predicts high omission rate.

Symmetric to initiations: when the dividend premium is low or negative, payers omit dividends at elevated rates. The mechanism is the same — by omitting, the firm switches out of a disfavored category. The combined initiation-omission pattern produces a long-run drift in the fraction of public firms that pay dividends, which has fallen from about $66\%$ in the late $1970$s to about $20\%$ by the late $1990$s, with the catering channel contributing meaningfully to the decline.

(Mechanism) Categorical thinking + time-varying category sentiment.

The catering theory rests on two behavioral inputs. First, investors form beliefs at the category level rather than the firm level (see categorical thinking). Second, category-level sentiment shifts over time, perhaps in response to macroeconomic conditions (a recession may make the safety of dividends salient) or to salient news (a high-profile dividend cut may shift sentiment against payers). The two inputs together generate time-varying mispricing of the dividend-payer category, which the firm exploits by switching membership.

(Distinct From Classical Dividend Theories) Catering is not signaling, not clientele, not agency.

Catering is conceptually distinct from each of the classical dividend theories (see the dividends hub for the full classical picture). Signaling models predict that dividend changes communicate private information about future cash flows; catering predicts that dividend changes track aggregate investor sentiment with no private-information content. Clientele models predict that the same firms always pay dividends and serve the same investors; catering predicts that firms switch in and out of the dividend-paying group as sentiment shifts. Agency models predict that dividends discipline free cash flow; catering predicts initiations among firms with no obvious free-cash-flow problem.

(Generality) The same logic applies to closed-end funds, mergers, and name changes.

Baker-Wurgler argue that catering is not a dividend-specific phenomenon. Closed-end funds are launched in waves coinciding with high CEF premia (low or negative discounts) — fund sponsors cater to investor demand for CEF-style products when sentiment for that category is high. Merger waves cluster in periods of stock-financed deals being favored. Firms added “dot com” to their names in $1999$–$2000$ to capture sentiment for the technology category, then dropped it post-$2000$. The same mechanism — manage the firm’s category membership in response to time-varying category sentiment — is at work in each.

(Cross-Reference With Dividends Hub) Catering is one of several behavioral dividend theories.

The dividend chapter (8.14. Dividends) covers a richer menu of behavioral dividend theories: catering (the firm-side market-timing story), bird-in-hand (investors prefer cash now over capital-gain promises later), self-control and mental accounting (investors use dividends to pace consumption), and inertia / valuation-yardstick theories (dividends as a social norm and as a signal of valuation discipline). Catering is the most cleanly testable of these because it generates a sharp time-series prediction tied to the dividend-premium measure.

Examples

The post- $2000$ collapse in dividend initiations

From the early $1990$s through $2000$, the dividend premium turned negative — non-payers (mostly tech firms) traded at much higher market-to-book ratios than payers. The rate of new dividend initiations collapsed correspondingly, falling to historic lows. After $2002$, as tech-stock sentiment cooled and the dividend premium rose, dividend initiations recovered. The Baker-Wurgler theory predicts exactly this pattern: when the dividend-payer category is disfavored, no firm has the incentive to switch into it; when the category becomes favored, switching resumes.

Microsoft's $2003$ dividend initiation

Microsoft initiated its first regular dividend in early $2003$. The firm had been profitable for over a decade with no obvious need to start paying dividends; its cash position was enormous and growing. The Baker-Wurgler dividend premium had risen sharply from $2000$ through $2002$ as investor sentiment shifted toward dividend-paying firms after the tech crash. From a catering perspective, Microsoft initiated to capture the now-positive dividend premium — its market-to-book multiple should rise toward the payer-category average. The theory does not require that Microsoft needed to pay dividends in any classical sense; it requires only that the firm could capture a category-membership premium by switching status.

The Citizens Utility Company case

Long Citizens Utility Company is the textbook example of investor preference for cash dividends. The company had two share classes that differed only in whether they paid cash or stock dividends. Despite the higher tax burden of cash dividends, the cash-dividend share class consistently traded at a premium to the stock-dividend class. The case demonstrates that the category “cash-dividend stock” carries an investor preference that defies tax efficiency, supporting the categorical-thinking microfoundation that catering rests on.

See Also

• 8.13. Behavioral Corporate Finance — parent hub.
• 8.14. Dividends — companion hub on the broader dividend-policy puzzle.
• 8.14.4. Behavioral Theories of Dividends — places catering alongside bird-in-hand, self-control, and mental-accounting theories.
• 8.13.1. Market Timing of Security Issuance — sibling: the equity-issuance analogue of catering.
• 8.11.5. Categorical Thinking and Comovement — the cognitive microfoundation that catering rests on.
• 8.4. Noise — the limits-to-arbitrage pillar required for the dividend premium to persist.