What the Evidence Says About Markets, Firms, and Stress

This chapter reads the 2026 New York Fed estimates as evidence about the shape of the equity-premium term structure, not as a single headline rate. The central task is to separate what the long U.S. panel supports about the aggregate market from what changes when we condition on recessions, firm characteristics, or modeling choices. That separation shows why an upward aggregate curve can coexist with flatter or inverted curves underneath it, and why robustness improves confidence in a pattern without making the result model-free.

The aggregate curve changes in bad times

The benchmark estimate for the full firm panel is upward sloping. The 15-year risk premium is about 8.3%, and the normalized 15-year-minus-5-year slope is about 7.5 percentage points. Both are economically meaningful and statistically significant in the study’s benchmark. The result says that, under the paper’s cash-flow forecasts, weighting, terminal assumptions, and curve form, later cash flows carry more implied compensation than intermediate cash flows in the aggregate U.S. sample.

That statement is stronger than the individual estimates at the very short end. The first few maturity points are estimated less precisely because many firms’ cash flows have duration well beyond one or two years. In the unrestricted baseline, the extrapolated short-end premia can be negative and have wide confidence bands. The authors argue that these points have limited influence on valuation and test alternatives that impose a flat first five years or require premia to be nonnegative. Those restrictions leave the aggregate curve upward sloping, beginning near zero in the nonnegative version and reaching a long-term level just below the unrestricted benchmark.

The aggregate result is therefore best stated as a shape claim with a precision boundary. The evidence is not that every short-horizon expected equity return is negative, nor that the 8.3% long-term level is a universal cost of equity. It is that the fitted schedule rises from the five-year anchor toward the long end under a range of reasonable specifications. The level moves more readily than the slope when long-run cash-flow growth assumptions change. For example, the paper reports a long-term premium near 5% when the steady-state growth rate is set to 3% instead of 6%, while the qualitative slope is much less affected.

The shape also changes with the economic state. The authors classify bad times in three ways: a year with at least three NBER recession months, a year in the bottom quartile of historical equity returns, or a year in which at least 25% of days fall in a drawdown period. All three definitions produce the same qualitative result. Using the NBER measure, the normalized slope falls from roughly 7.5% in good times to 3.8% in bad times. The 15-year premium falls from approximately 8.3% to 6.3%.

The curve does not invert in these estimates. Its flattening comes mainly from the short and intermediate horizons. During bad times, those discount rates rise markedly, while the long-term premium falls slightly.