A stock is a claim on a stream of future cash flows, and its value is the present value of that stream discounted back to today. The interest rate — specifically the risk-free Treasury yield plus an equity risk premium — sits in the denominator of that calculation. When rates rise, every future dollar is worth less today, so prices fall; when rates fall, the opposite. Equity duration is the formalization of this idea borrowed from bond math: it measures how sensitive a stock's price is to a change in the discount rate. The core tension is that this discount-rate channel is theoretically airtight but, in the real world, fights constantly with the numerator — rates usually move because the economy is changing, which also changes expected earnings — so the observed rate-stock relationship is far messier than the textbook says.

How it's formed (the mechanics)

The dividend-discount / DCF identity is the foundation: Price = Σ (CFₜ / (1 + r)ᵗ), where r ≈ risk-free rate + equity risk premium. Two assets with identical near-term cash flows but different timing respond very differently to a change in r:

• Short-duration ("value") stocks — mature firms with front-loaded, cash-rich earnings (banks, utilities-by-payout, energy, deep-value). Most of their value arrives soon, so it is discounted by r only a few times.
• Long-duration ("growth") stocks — firms whose profits sit decades out (unprofitable tech, biotech, hypergrowth). More of their value is subject to the compounding of the discount factor over many years.

Lyn Alden's worked DCF example illustrates the asymmetry cleanly: cutting the discount rate from 12% to 8% raised a 5%-growth value stock's fair value by ~44.7%, a high-growth stock's by ~61.8%, and a delayed-profitability "hypergrowth" stock's by ~89.9% — same rate move, very different price response. That differential sensitivity is equity duration in action.

Quantitatively, equity duration is far longer and noisier than bond duration. The estimate is highly model-dependent — it hinges on assumed growth and terminal value, which are exactly the inputs that move with rates. Dechow, Sloan & Soliman's (2004) implied equity duration averages ~15 years across firms (SD ~4 years), while Dividend-Discount-Model-based estimates of the broad S&P 500's duration run higher — commonly cited in the ~15–25-year range, well above the duration of most government bonds. The spread between those methods is itself a signal of how sensitive the number is to assumptions.

How it's used in practice

1. Sector/style rotation framing. The single most common practical use is explaining style leadership: in falling-rate or low-rate regimes, long-duration growth tends to outperform; in rising-rate regimes, short-duration value and rate-beneficiary financials tend to lead. The 2020–2021 (zero rates → growth mania) vs. 2022 (aggressive hikes → growth crash, value relative strength) episode is the canonical case study. 2. Valuation context, not a signal. Comparing the market's earnings yield (E/P) to the 10-year Treasury — the "Fed model" / equity-risk-premium lens — is used to gauge whether stocks are cheap or rich relative to bonds. Lyn Alden flags the 2000 dotcom peak as the extreme: a cyclically-adjusted earnings yield of under 2.5% against ~6.5% Treasuries, a deeply negative equity risk premium (note: a trailing E/P at the 2000 peak was nearer ~3%; the sub-2.5% figure is the CAPE-based yield). 3. Event reaction. Around FOMC meetings, traders treat the surprise component of a rate move — not the move itself — as what hits prices. Anticipated hikes are already in the price.

Adoption, debate & evidence

The discount-rate logic is universally accepted in theory. The empirical magnitudes are genuinely contested.

• The headline number. Bernanke & Kuttner (2005, Journal of Finance) found an unanticipated 25-bp cut in the fed funds target is associated with roughly a 1% rise in broad indexes. This is the most-cited estimate, but note it applies to unexpected moves, not announced ones.
• What drives it — open dispute. Bernanke-Kuttner attributed most of the reaction to the equity risk premium (risk appetite). A 2024 reexamination using dividend-futures data (Nagel & Xu, "Bernanke-Kuttner Redux," NBER) argues most of the move is explained by bond-yield changes themselves, not the equity premium — a meaningful reframing of the transmission mechanism. Treat the "why" as unsettled.
• The Fed model is descriptive, not predictive. Cliff Asness's "Fight the Fed Model" (2003) shows the model describes how P/Es are set (investors anchor stock yields to nominal bond yields) but does so via money illusion — comparing a real earnings yield to a nominal bond yield. Cross-country work by Estrada ("The Fed Model: The Bad, the Worse, and the Ugly") finds earnings yields and bond yields are cointegrated in only 2 of 20 countries, and plain P/E beats the Fed model at forecasting real returns in 18 of 20; separate ECB cross-country evidence (WP 515) reaches the same skeptical conclusion. Do not use the Fed model to forecast.
• Equity duration as a return predictor is promising but not bond-grade robust; estimates swing with model assumptions, and the "short-duration premium" overlaps heavily with the value, profitability, and quality factors rather than being clearly independent (Weber, 2018, documents that low-duration stocks are high-value, high-profitability, high-payout stocks). Treat any standalone "low-duration edge" with skepticism.

Strengths & limitations

Works best as a directional, cross-sectional lens — explaining why one cohort of stocks reacts more than another to the same rate shock, and as valuation context at extremes.

Fails when used mechanically because:

• Rates and earnings move together. Rates often rise because growth/inflation is strong, lifting earnings and offsetting the discount-rate drag. This is why hikes are sometimes greeted positively. The sign of the rate-stock correlation is regime-dependent, not constant.
• The discount rate is unobservable. The equity risk premium is a residual, so any duration estimate inherits the analyst's growth and terminal-value assumptions.
• #1 misuse: treating "rates up → stocks down" as a deterministic trading rule, or using the Fed model as a buy/sell trigger. Both have weak out-of-sample records.

Sources

• Bernanke & Kuttner, "What Explains the Stock Market's Reaction to Federal Reserve Policy?" Journal of Finance 60(3), 2005 — the ~1% per 25-bp surprise estimate.
• Nagel & Xu, "Movements in Yields, not the Equity Premium: Bernanke-Kuttner Redux," NBER WP 32884 (2024) / Becker Friedman Institute summary — disputes the equity-premium mechanism.
• Asness, "Fight the Fed Model" (SSRN, 2003) — money-illusion critique of the Fed model.
• Estrada, "The Fed Model: The Bad, the Worse, and the Ugly" — cross-country evidence (2 of 20 cointegrated; P/E beats Fed model in 18 of 20). ECB Working Paper 515 — separate cross-country evidence the Fed model lacks predictive power.
• Lyn Alden, "Interest Rate Effects on Equities" (lynalden.com) — DCF worked examples (44.7% / 61.8% / 89.9% fair-value lift for value/growth/hypergrowth on a 12%→8% discount-rate cut); 2000 CAPE-earnings-yield (<2.5%) vs ~6.5% Treasury comparison.
• Dechow, Sloan & Soliman, "Implied Equity Duration: A New Measure of Equity Risk" (2004) — implied duration ~15-year mean. Gormsen & Lazarus, "Interest Rates and Equity Valuations"; Weber (2018), "Cash Flow Duration and the Term Structure of Equity Returns" — duration mechanics, the short-duration premium and its factor overlap.
• U.S. Bank, "How Do Changing Interest Rates Affect the Stock Market?" — multi-channel (borrowing, spending, competing yields) overview.

Disputes flagged: (1) the transmission mechanism — equity risk premium vs. yields — is unresolved (Bernanke-Kuttner vs. Nagel-Xu); (2) the Fed model is descriptive only and fails as a forecaster; (3) equity-duration return predictability overlaps with the value factor and is model-sensitive.