Overview

The 2026 New York Fed cross-sectional DCF and 2009 ECB affine stock–bond approaches ask a common horizon question but estimate different objects from different evidence. One uses an annual U.S. firm panel from 1980–2024, forecast cash flows, and cross-firm variation; the other uses monthly U.S. index, bond, payout, inflation, and latent-state data from 1983–2008. Their rates differ in horizon, benchmark, cash-flow concept, frequency, conditioning, and weighting, so co-movement or slope differences are not contradictions until those labels are matched.

Learning outcomes

The two methods share a horizon question but identify different objects: a maturity-specific discount curve versus a conditional expected excess return for an index.
A DDM rate can co-move with horizon-specific estimates because both use valuation information, but correlation does not prove they recover the same risk price.
Compare curves only after matching horizon, benchmark, asset population, period, expectations, reinvestment, conditioning, and assumptions.

Keywords

Equity risk premiumNew York FedECBDCFPricing kernelDividend discount

Two Ways to Ask the Same Horizon Question

This chapter compares the two main approaches to the equity-premium term structure: the New York Fed’s cross-sectional discounted-cash-flow method and the ECB’s affine, arbitrage-free stock-and-bond model. The comparison is meant to improve interpretation, not to choose a universal winner. By separating their shared economic question from their different data, assumptions, and conditioning, we can tell whether two estimated curves genuinely disagree or simply describe different objects.

Same economic question, different evidence

Both approaches begin with the same underlying fact: equity is a claim on uncertain cash flows arriving at multiple dates. Both therefore ask how much compensation is associated with equity exposure over a specified horizon, and both compare that compensation with a risk-free investment. The common subject is the schedule of expected or required excess compensation. Neither approach reduces the problem to a realized-return average alone, and neither treats a single constant discount rate as a complete description of every date in the cash-flow stream.

The New York Fed method starts with observed equity values for many firms. It projects each firm’s shareholder cash flows, assigns discount rates to different maturities, and estimates a structured curve that makes model values line up with observed prices across the panel. The source of identification is cross-sectional: firms have different expected cash-flow patterns, so the same candidate curve affects their prices differently. Observable characteristics can alter the level and slope, allowing the method to represent systematic differences among firms rather than forcing all firms onto one curve.

The affine method starts with a common stochastic discount factor and the state variables that drive it. It uses a broad stock index alongside inflation, payout yield, and nominal bond yields. Two latent factors complete the state by representing the real short-rate environment. The same pricing kernel must account for bond prices and stock returns, so no-arbitrage restrictions link the equity premium to the bond term structure and the evolving economic state. Its source of identification is therefore structural and time-series based rather than primarily cross-sectional.

These are not merely two data sets applied to the same formula. The cross-sectional method asks: given the different cash-flow clocks of firms, what maturity curve best explains their observed prices? The affine method asks: given the current state of the stock-and-bond system and a common pricing rule, what excess return is implied at each horizon? The first is designed to uncover heterogeneity across firms and portfolios. The second is designed to trace conditional variation through time.

The differences in scope are material. The New York Fed study uses an annual U.S. firm panel from 1980 through 2024, with June valuation dates and an analyst-merged final sample of 56,126 firm-year observations. It excludes financial and utility firms and other observations that are difficult to fit under its fundamental valuation setup. The ECB study uses monthly U.S. data from January 1983 through December 2008, including the S&P 500, payout or dividend yield, inflation, real ex-dividend returns, and nominal zero-coupon yields at one- to ten-year maturities. It studies one broad index rather than a broad cross-section of firms.

The cash-flow concepts also differ. The New York Fed model explicitly forecasts shareholder free cash flow from earnings, book-equity changes, return on equity, and sales growth. The ECB model’s stock pricing uses a payout-yield factor, and payout includes buybacks as well as dividends. The New York Fed curve is separated from a maturity-varying risk-free rate inside the DCF calculation. The ECB curve is the difference between expected stock returns and model-implied real bond yields generated by the same pricing system.

Published by DeepMox

FAQ

What is the main difference between the New York Fed and ECB approaches?

The New York Fed approach uses cross-firm cash-flow timing to estimate a maturity-specific discount curve, while the ECB approach uses a common pricing kernel to infer conditional index premia from the evolving stock–bond state.

Why can two equity-premium curves move together without measuring the same thing?

They can co-move because prices, expected cash flows, and market conditions affect both, but correlation does not remove differences in levels, horizons, expectations, or conditioning.

When is a difference between two term-structure estimates a real disagreement?

It becomes a substantive disagreement only after horizon, real or nominal basis, cash-flow definition, asset population, period, and conditioning information are matched and specification differences remain.

Previous12 min2009 ECB Equity Risk Premium Model

Next14 min2026 New York Fed Equity Risk Premium Evidence

2026 New York Fed vs 2009 ECB Equity Risk Premium Models

Two Ways to Ask the Same Horizon Question

Same economic question, different evidence

FAQ