Dividend policy

2.1 The mechanical anatomy of a dividend: and the rule that quietly changed

A dividend moves through four dates, in this order, and precision about each one is not pedantry, real money changes hands on the wrong side of the wrong date. The declaration date is when the board formally approves the dividend and announces the amount and the other three dates. The record date is the date on which the company's books are checked: whoever is the registered shareholder of record on that date gets paid. The ex-dividend date is the first date a share trades without the right to the upcoming dividend attached, buy on or after the ex-date and the dividend goes to the seller, not you. The payment date is when cash (or new shares, for a stock dividend) actually moves.

Here is the fact to hold precisely, because it is a genuine, recent, mechanical change, exactly the kind of thing a stale mental model gets wrong with total confidence. Under the older T+2 settlement cycle (a trade settles two business days after execution), the rule taught for decades was "the ex-dividend date is one business day before the record date", a trade executed on the record date itself would not settle until after it, so the last day to buy and still be the owner of record was one business day earlier, and the ex-date fell the day after that. Both India and the United States have since moved to T+1 settlement (India across 2022–2023; the US SEC mandating it effective May 2024), and the consequence is exactly what settlement-timing logic predicts: the ex-dividend date is now the same calendar day as the record date, in both markets. DTCC's own T+1 dividend-processing guidance states it directly: under T+2 "the Ex-Date would be one day prior to Record Date," while under T+1 "the Ex Date would be the same day as Record Date." India's exchanges apply the identical logic. An AI assistant, or a colleague, reciting "ex-date is one day before record date" today is repeating a rule that stopped being true when settlement moved to T+1, precisely the stale-but-fluent statement this module warns about. Always confirm the current settlement cycle before trusting either version.

2.2 Cash dividend, stock dividend/bonus issue, and stock split: same shareholder arithmetic, different mechanics

A cash dividend is the plain case: a stated amount per share, paid in cash, reducing retained earnings and cash by the same amount. Two alternatives change the share count instead of (or alongside) sending out cash, and they are frequently conflated despite doing different things to the accounts.

A bonus issue (India's term) or stock dividend (the US term for the same shareholder-facing mechanic) capitalizes existing reserves into new, fully-paid shares distributed to holders in a fixed ratio, "1 bonus share for every 2 held" is a 1:2 bonus. No cash moves and total shareholders' equity is unchanged, but it is reallocated within equity: reserves/retained earnings fall, and share capital (plus, for a small US stock dividend under 20–25%, additional paid-in capital at fair value) rises by the same amount. The shareholder ends up with more shares at a proportionally lower theoretical price, total holding value unchanged. Worked Example 3 computes this exactly.

A stock split does something mechanically different: it merely subdivides the existing par/face value and share count, with no entry through reserves or retained earnings at all. A 4-for-1 split turns one ₹10-face-value share into four ₹2.50-face-value shares (India expresses this as a face-value subdivision on the exchange filing); the US equivalent multiplies the share count and divides the price, with no par-value note if the stock is no-par. Same shareholder arithmetic as a bonus issue (more shares, proportionally lower price, unchanged total value) but a different accounting event underneath it. The practical tell: a bonus issue/stock dividend consumes reserves; a split does not. A company with thin reserves can still split its stock; it may not sustain repeated bonus issues.

2.3 Dividend reinvestment plans (DRIPs)

A DRIP lets a shareholder elect to receive new shares instead of cash for a declared dividend, administered either by the company itself (new shares issued directly, sometimes at a stated discount to market as an incentive) or by a broker/transfer agent buying shares in the open market on the shareholder's behalf. The arithmetic is simple and worth being exact about, because the fractional-share remainder is where sloppy computation creeps in: shares purchased = total cash dividend ÷ reinvestment price, and the reinvestment price is not always the plain market price, a company-administered DRIP frequently applies a small discount (commonly 1–5%, disclosed in the plan document) as an incentive to reinvest. Whatever fraction of a share results is either credited as a fractional entitlement (the now-standard convention at most brokers) or paid out as cash-in-lieu. Worked Example 3 runs a full DRIP computation, fractional remainder included. One framing worth carrying deliberately: a DRIP is not a bonus. It is the shareholder's own after-tax (where applicable) dividend cash buying more of the same stock, economically identical to taking the cash and manually reinvesting it, minus a brokerage commission and, in a company-administered plan, plus whatever discount the plan offers.

2.4 Why dividend policy is a real decision, not an accounting afterthought

Start from the austere baseline: Modigliani and Miller (1961) proved that, in a frictionless world (no taxes, no transaction costs, no asymmetric information, full market access) dividend policy is irrelevant to firm value. A shareholder who wants cash today and gets no dividend can sell a few shares ("homemade dividends"); one who gets an unwanted dividend can reinvest it. Value comes from the investments the firm makes, never from how it slices the resulting cash flows into "paid out" versus "retained." This baseline matters because every real-world reason dividend policy does seem to matter is a named friction on top of it:

• Signaling. Management knows things shareholders don't. Because cutting a dividend reads as a strong, costly signal of trouble, managers smooth dividends deliberately rather than let them track earnings volatility one-for-one. Lintner's (1956) partial-adjustment model formalizes exactly this, and Worked Example 4 computes it on a real earnings path. A dividend increase signals confidence the higher level is sustainable; a cut signals the opposite so strongly that boards will borrow, or sell assets, before cutting one if they can avoid it.
• Agency / free-cash-flow hypothesis (Jensen, 1986). Cash sitting inside a company that management controls can fund empire-building or inefficiency. A dividend commitment forces the company back to capital markets more often, subjecting it to investor and lender scrutiny, a blunt but real discipline on managers whose incentives are not perfectly aligned with owners'.
• Clientele effects. A retiree living off portfolio income wants cash dividends; a high-bracket taxable investor may prefer buybacks (a deferred, often lower-taxed capital gain); an index fund is indifferent. A policy attracts a "clientele" that prefers it, and a sharp policy change can force an unwelcome, price-affecting clientele rotation even when value-neutral on paper.
• Tax-preference argument. Whenever dividends and capital gains are taxed differently (different rates, different timing (a capital gain is taxed only on realization)) the tax code creates a real preference for one payout channel, and that preference can flip with a single Finance Act or Budget. §2.5 works this out concretely for India and the US, since it has shifted materially and recently in India.

2.5 The capital-return choice: dividends versus buybacks, and the tax mechanics that actually decide it

M3.07 owns the value-creation arithmetic of a buyback, the conservation-of-value proof that a buyback only creates value for remaining shareholders when executed below intrinsic value, and destroys it above intrinsic value; go there for that math, not here. What belongs here is the tax-and-mechanics layer that frequently dominates the real-world choice regardless of the value math, because it has changed direction in India twice since 2024.

India, three regimes since 2024. Before 1 October 2024 the company, not the shareholder, paid a buyback distribution tax under Section 115QA of the Income-tax Act, 1961, and shareholders received the proceeds free of further tax. From 1 October 2024 to 31 March 2026, under the Finance (No. 2) Act, 2024, a new clause (f) inserted into Section 2(22) redefined "dividend" to include any payment by a company for the buyback of its own shares: the entire buyback consideration a shareholder received was a deemed dividend, taxed at the shareholder's own slab rate, not a capital gain. Alongside it, Section 46A deemed the consideration nil for capital-gains purposes, so the original acquisition cost was not netted against proceeds at all but survived as a capital loss, carried forward against future capital gains; TDS of 10% applied for resident shareholders with a valid PAN. From 1 April 2026 the rule turned again. The Income-tax Act, 2025 replaced the 1961 Act from that date, and the Finance Act, 2026 omitted its deemed-dividend clause for buybacks (Section 2(40)(f), the successor of Section 2(22)(f)), so buyback proceeds are a capital gain again: consideration less cost (Section 69 of the 2025 Act, the successor of Section 46A). A non-promoter pays the ordinary rates on listed shares, 12.5% on long-term gains above the ₹1.25 lakh annual exemption and 20% on short-term gains. A promoter pays an additional tax that lifts the effective rate to 22% for a domestic-company promoter and 30% for any other promoter, before surcharge and cess. Worked Example 5 runs one ₹500 buyback against a ₹200 cost basis under both rules: ₹150 of tax on the full ₹500 for a 30%-slab resident under the 2024 rule, ₹37.50 on the ₹300 gain for a long-term non-promoter holder under the 2026 rule. Confirm whether, and at what rate, tax is withheld on the payment under the current rule against the Act, since TDS provisions move with Budget cycles. The direction has flipped twice since 2024, so a stale source (human or AI) will confidently give you the wrong regime; confirm the effective date against the Finance Act in force on the transaction date (as of October 2026).

The United States. A qualified dividend (one meeting the holding-period test below) is taxed at the same preferential rates as a long-term capital gain, materially below ordinary income rates for most taxpayers; an unqualified dividend is taxed as ordinary income. The test, per the IRS's own Form 1099-DIV instructions, is anchored to the ex-dividend date from §2.1: the shareholder must not have held the stock for fewer than 61 days during the 121-day period beginning 60 days before the ex-dividend date. Get the ex-date wrong (the stale T+2-era rule, say) and you will misjudge whether a real position actually qualifies. Long-term capital gains receive the same preferential treatment against ordinary income, the mechanism buybacks lean on in a taxable US account, since a shareholder who does not sell realizes no tax event, and one who does sell pays capital-gains rates only when they choose to realize the gain.

Side by side: India's 2024 shift made buybacks less tax-favourable to a resident shareholder (full proceeds taxed as dividend, versus previously tax-free), and the 2026 shift restored part of the case for a non-promoter (only the gain is taxed, at capital-gains rates) while charging promoters a higher effective rate; the US qualified-dividend/capital-gains preference continues to favour deferral-friendly payout, holding-period permitting. Neither is stable forever, verify the currently effective provision before advising anyone on this section's strength alone.

2.6 India versus US, side by side

{
  "skill": "Getting a corporate-action date or a jurisdiction's current tax mechanic right, without treating the model's answer as the source",
  "use": "Use an assistant to draft the mechanical sequence of a corporate action (which date follows which, what typically happens on each) and to point you at the specific statutory section or exchange circular currently in force, rather than asking it to simply state the current rule.",
  "tools": ["Chat assistants - Claude, ChatGPT, Gemini", "Research/browsing tools that can fetch the exchange's or tax authority's own current page", "The company's own investor-relations FAQ or circular, when available (companies routinely publish accurate plain-language tax notes on their own IR pages)"],
  "prompt": "For <CORPORATE ACTION / TAX QUESTION> in <India/US>, do not state the current rule from memory. Instead: (1) name the exact statute, section, or exchange circular that governs it, (2) tell me whether this rule has changed materially in the last three years (settlement-cycle changes and Finance Act/Budget changes are both common triggers), and (3) tell me exactly where I should read the current, effective text myself.",
  "verify": "Trace to the named primary source - the exchange (NSE/BSE, DTCC), the tax authority's own Act/circular, or the company's own investor-relations page - and confirm the effective date of the specific provision predates or covers your situation (Guardrail #4, Date-check, is doing most of the work in this entire module). Never treat 'ex-date is one day before record date' or a specific tax rate as durable without checking the settlement cycle and the Finance Act/Budget currently in force.",
  "diy": "You must already know the four dates in sequence and why the ex-date/record-date relationship is mechanically tied to the settlement cycle (Section 2.1) - without that, you cannot even evaluate whether an AI's answer sounds structurally right before you go check it.",
  "market": "IN"
}