Lessons from a Century of Fed Hiking Cycles: Priced Like 1999, with a Chance at 1994

Signposts: toward 1994 or 1999

Valuation is what sets 1999 apart from 1994

What followed each first hike

Policy separated 1994 from 1999

The stakes are higher than in 1994 or 1999

Long yields have risen faster since the Sep. 2026 hike

At the September 16 press conference, Chair Warsh gave three reasons long yields had risen: a stronger economy; "competition for capital," with hyperscalers "out in the market raising funding"; and geopolitics.

Markets since the Sep. hike

Gains are concentrating in the largest stocks

Stress is building at the bottom of credit

Correlation and index volatility are near their lows

Before the hike: capex news and the July deleveraging

Capex news has tended to lift suppliers more than spenders

Six days chosen for the size of the market reaction: an illustration, not a statistical test. On July 28 a warning about returns on the spending reversed the pattern. Same-day returns. Suppliers: average of 12 chip, server, networking and power-equipment stocks. Megacap tech: Microsoft, Amazon, Alphabet, Meta, Oracle. Credit: HYG and LQD.

July: deleveraging hit factor bets while the index stayed calm

Full comparison: each first hike against today

What followed each condition

How to read: each cell covers the US episodes where that condition matched today (●), partly (◐) or not (○). The top figure is their average real S&P 500 return over the next 24 months; below it, how many of them saw a bear market begin. For example, growth matched today in 6 episodes, returns averaged +15%, and 2 of the 6 had a bear market.

Each cell: the average real return on the S&P 500 over the 24 months after the first hike, and "bear in k of n": how many of those n episodes saw a bear market (a 20% fall on daily closes) begin within 24 months, highlighted when it was half or more. For comparison, a bear market began within 24 months of any month 42% of the time since 1928. US episodes; samples are small. Only 1999 fully matched today's valuation, so the ● column for valuation is a single episode; on excess CAPE yield, 1965 and 1967 also sit with today. The partial and non-matches are the real comparison: +6% across 8 episodes where valuation partly matched today, +21% across 9 where stocks were cheaper. Fiscal points the other way: bear markets followed 6 of the 8 first hikes with a surplus or small deficit, against 3 of the 10 with larger deficits, so loose fiscal policy may have helped postpone sell-offs. But several of the larger-deficit first hikes (1983, 2004, 2022) came in recoveries from recent recessions, when returns tend to be good for cyclical reasons, so this is not evidence that deficits caused the better outcomes.

Appendix

Resemblance score: each first hike against today

● matches today's reading, ◐ partly, ○ not; † marked by applying the thresholds to FRED and Shiller data. Score (of 8): how many conditions match today (◐ counts half), a summary of resemblance rather than a forecast. 12m and 24m: real total return on the S&P 500 after the first hike. Columns: valuation, rates and curve, growth, inflation, tech story and earnings, financing, fiscal, leverage. *Nominal share-price index (OECD). Conditions are shown separately: they point in different directions, and only some have tended to come before weak returns.

Returns after each first hike

Real total return on the S&P 500 (Shiller; spliced with SPY after September 2023). Largest fall on monthly average prices, with the daily-close figure where it differs by more than 5 points. Bear market: a 20% fall on daily closes beginning within 24 months (Ned Davis Research dates; Dow for 1919). Recession: NBER peak month within 24 months. *Nominal share-price index (OECD).

Nine of the 18 US first hikes were followed by a bear market within two years. Since 1928, a bear market began within two years of any given month 42% of the time (35% since 1946). 1999 and 1994, the two closest matches, ended on opposite sides.

Method notes

Anchor: the month of each cycle's first increase in the discount rate (before 1955) or the federal funds rate. Single hikes that did not start a cycle, such as March 1997, are excluded. Curve judged from 12 months before to six months after; inflation over the following six months. Each condition is marked ● (matches today), ◐ (partly) or ○ (not):

• Inflation: ● core 2–3% and flat or falling; ◐ core rising, below 2% or 3–5%; ○ above 5%.
• Rates and curve: ● long yields up about a point with the curve parallel or steepening; ◐ long yields up with the curve flattening; ○ long yields falling, the curve inverted within six months, or rates pegged.
• Growth: real GDP growth over the past year against CBO's estimate of potential growth. ● 0–1.5 points above potential; ◐ below potential or 1.5–3 above; ○ more than 3 above, or recession.
• Tech story and earnings: ● a new technology leads and earnings growth is strong and concentrated in it; ◐ secondary story, or strong broad earnings; ○ neither. For † rows, earnings growth above 5% counts as ◐.
• Financing: ● corporate debt growing 4–7% and accelerating, with heavy stock issuance; ◐ one of the two, or debt growth of 10% or more; ○ neither. For † rows, issuance is not observed, so the maximum is ◐.
• Fiscal: ● deficit 5–7% of GDP outside recession, a rising debt ratio and net interest of 2.5% of GDP or more; ◐ some of these; ○ surplus or small deficit with a falling debt ratio.
• Valuation: ● CAPE of 30 or more and a trailing earnings yield within 2 points of today's gap with the 10-year; ◐ one of the two; ○ neither.
• Leverage: ● margin debt up 30% or more and stocks 35% or more of household financial assets; ◐ one of the two, or margin debt up 10–30%; ○ neither. Margin debt: NYSE and FINRA from 1997, Fed broker loans before.

FactorPulse factors are long/short portfolios built after controlling for each other: size (large against small), beta (most against least market-sensitive), leverage (most against least indebted), Treasury sensitivity (stocks that move with long Treasuries against those that don't), morning activity (stocks traded most heavily early in the session), residual volatility (most against least stock-specific volatility). See the methodology.

The valuation condition uses CAPE and the trailing earnings yield against the 10-year:

Sources

• History: Shiller monthly market data · FRED (CPI, core CPI, 10-year, 2-year, 1-year, fed funds, discount rate, real GDP, potential GDP, nominal GDP, nonfinancial corporate equities, deficits, debt, net interest, household equities, profits, corporate debt, broker margin loans, recessions) · Ned Davis Research bear-market dates via Hartford Funds · Yardeni · Dow Theory record · OECD share prices for Japan and China · FINRA margin statistics
• Today: BLS CPI · FRED (2-, 10- and 30-year yields, real 10-year, monthly, VIX, ICE BofA HY, CCC and IG spreads) · CBO · BEA · FactSet Earnings Insight, Oct 2 · State Street SPY holdings · Brookings on AI financing · SEC filings on leases not yet commenced: Microsoft 10-K, Meta, Amazon and Alphabet 10-Qs for June 30, 2026, Oracle 10-K for May 31, 2026, and earlier 10-Qs and 10-Ks · LPL, What history says about Fed hikes and stocks, Sep 14, 2026 · Alphabet 8-K on its equity raise · Oracle 10-Q, quarter ended Aug 31, 2026 · BLS Employment Situation, September 2026 · Fed statement, Sep 16, 2026 · Chair Warsh's press conference transcript, Sep 16, 2026 · Summary of Economic Projections, Sep 16, 2026 · UBS CIO, Feb 18, 2026 · Renaissance Capital
• FactorPulse: factor returns, 20-day correlation and volatility decomposition, ETF and stock returns