Week 10: Economic Cycles — Expansion, Peak, Contraction, Trough, and What They Do to Assets

---

1. Why This Is Important

The economy does not move in a straight line, and neither does any asset that ultimately depends on the economy for its cash flow. It expands, it overheats, it contracts, it recovers, and then it does the whole thing again. The names for the four stops on that loop — expansion, peak, contraction, trough — are over a hundred years old. The pattern is older than that.

You need to take cycles seriously for four reasons.

The cycle decides which asset wins. Stocks lead in early

expansion, commodities lead in late expansion, long bonds lead into and through recession, cash leads only briefly at the very top. A portfolio built for one phase can spend a year or two doing nothing right and a year doing everything wrong. You do not have to time the cycle precisely; you do have to know which phase you are in roughly, because some assets are quietly cheap at every point in the cycle and some are quietly expensive, and the difference is usually larger than the equity risk premium itself.

Recessions arrive on a schedule, even if the timing is fuzzy.

Since 1945 the United States has had thirteen recessions. The average expansion is about five years and the average contraction is about eleven months. Anyone who tells you "the business cycle is dead" is forgetting that the same thing was said in 1929, 1968, 1999, and 2007. The cycle does not disappear. It just changes shape.

**The yield curve has called every recession since 1955 with one

ambiguous exception.** That is an extraordinary track record for a single indicator, and it is freely published. If you ignore only one signal in this entire course, do not let it be the 10y minus 2y Treasury spread.

**Cycle awareness is what separates a thoughtful 60/40 from a

complacent one.** Week 4 showed that the long-run statistics on 60/40 are excellent. This week shows the statistics underneath the statistics: 60/40 is a regime bet, and the regime can change. The regime-shift point — passive index investing has worked for forty years, and that does not make it permanent — lives or dies on the cycle.

This lesson covers NBER dating, the layered short / medium / long cycles, the three families of indicators, the yield curve in detail, the asset-class playbook by phase, and what 2008, 2020, 2022, and the 2026 setup teach about how the textbook fails when the cycle stops cooperating.

---

2. What You Need to Know

2.1 NBER Dating — Who Decides a Recession Has Happened

The popular shorthand "two consecutive quarters of negative GDP" is not the official US definition. The actual arbiter is the **Business Cycle Dating Committee of the National Bureau of Economic Research (NBER)** — eight academic economists who, after the fact, declare a peak month and a trough month. Their working definition of a recession:

A significant decline in economic activity that is spread across the economy and lasts more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales.

Three things follow.

• Recessions are dated by month, not quarter. February 2020 is

the peak; April 2020 is the trough. Two months long — the shortest US recession ever. The "two-quarter" rule would have missed it entirely.

• Recessions are declared late. The committee waits until

revisions have settled, which is typically six to eighteen months after the peak. By the time NBER officially calls the start of a recession, the stock market has often already bottomed and begun rallying. This is the single biggest practical reason that retail investors who "wait until the recession is confirmed" buy late and miss the recovery leg.

• The criteria are plural. Real GDP can be negative for two

quarters without an NBER recession (the technical-recession of the first half of 2022 was not dated as one, because employment and industrial production kept rising). And NBER can call a recession with one positive quarter mixed in (1980).

The full US post-1945 recession list, with peak → trough months:

#	Peak	Trough	Length
1	Nov 1948	Oct 1949	11 mo
2	Jul 1953	May 1954	10 mo
3	Aug 1957	Apr 1958	8 mo
4	Apr 1960	Feb 1961	10 mo
5	Dec 1969	Nov 1970	11 mo
6	Nov 1973	Mar 1975	16 mo
7	Jan 1980	Jul 1980	6 mo
8	Jul 1981	Nov 1982	16 mo
9	Jul 1990	Mar 1991	8 mo
10	Mar 2001	Nov 2001	8 mo
11	Dec 2007	Jun 2009	18 mo
12	Feb 2020	Apr 2020	2 mo

Twelve recessions in eighty years. Average expansion length between them: about sixty-three months. The longest expansion ever recorded — June 2009 through February 2020 — was 128 months and ended only because of an exogenous shock (COVID), not internal overheating. That fact matters: not every cycle ends from the inside.

2.2 Cycles Within Cycles — Kitchin, Juglar, Kondratieff

The textbook usually presents one cycle. The honest picture is that there are several, layered on top of each other.

• Kitchin (~3–5 years). Inventory cycle. Firms over-order →

inventory glut → cut orders → inventory drawdown → re-order. This is what the manufacturing PMI moves on. Most of the noise in monthly economic data is Kitchin.

• Juglar (~7–11 years). Capex cycle. Firms invest in fixed

capital (factories, equipment, software). Capacity overshoots demand, capex contracts, recession ensues, the cycle resets. This is the recession cycle most investors mean when they say "the business cycle." Average post-war US expansion length sits squarely inside the Juglar window.

• Kondratieff (~40–60 years). Long wave. Driven by

technological revolutions, demographic regimes, monetary regimes. Steam, rail, electrification, the auto, semiconductors, the internet — each anchored a long wave. **This is the wave the regime-shift point is talking about.** The 1981–2021 disinflationary bull market in stocks and bonds was one Kondratieff up-leg. The next one will not look identical.

We do not trade Kitchin / Juglar / Kondratieff as numerology ("the next 9.6-year cycle peaks in October 2027"). The numbers above are not deterministic. We use the intuition — that there are short, medium, and long cycles superimposed, and that the phase you are in depends on which cycle you are looking at. The 2022 inflation shock was a Juglar peak inside a possibly-cresting Kondratieff. That is a different setup from 2008, when both medium and long waves were still on the up-leg of disinflation, and it explains why the same playbook (Fed cuts → bonds rally, stocks rally) worked in 2008 and did not work in 2022.

The four-quadrant framework — growth up/down crossed with inflation up/down — is the practical version of this. We use it in §2.5.

2.3 Leading, Coincident, and Lagging Indicators

Every economic data series falls into one of three categories according to its timing relative to the cycle.

• Leading indicators turn before the economy does. The

yield-curve slope, building permits, the ISM new-orders sub-index, initial jobless claims, the S&P 500 itself, and the Conference Board's Leading Economic Index (LEI) are the classics. Lead time: typically six to eighteen months.

• Coincident indicators turn with the economy. Industrial

production, real personal income excluding transfers, real manufacturing and retail sales, and especially **non-farm payrolls**. These four are roughly the variables NBER actually uses to date the cycle. Lead time: zero.

• Lagging indicators turn after the economy does. The

unemployment rate, core CPI, the prime lending rate, the duration of unemployment, the ratio of consumer credit to personal income. Lag time: three to twelve months.

The single most common retail mistake is to treat lagging indicators as decision-grade. "The unemployment rate is 3.8% — the economy is fine" is a 2007 sentence. Unemployment was 4.7% in November 2007, the month the previous expansion peaked, and it peaked at 10.0% twenty-three months later. By the time unemployment confirms a recession is underway, the stock market has already lost a third of its value. The yield curve, by contrast, told you there was a problem in mid-2006.

The honest hierarchy: **price data first, soft data second, hard data third, lagging data never as a primary signal.** Price data is the yield curve, credit spreads, equity sector leadership, the copper-to-gold ratio. Soft data is the surveys (ISM, consumer confidence). Hard data is the official prints (payrolls, GDP, production). Each is roughly a quarter slower than the one before. The cycle's leading edge lives in markets, not in the data release calendar.

2.4 The Yield Curve — The Master Leading Indicator

The single most reliable leading indicator of US recessions is the slope of the Treasury yield curve, conventionally measured as the 10-year yield minus the 2-year yield (or the 10-year minus the 3-month). When that spread goes negative — long rates below short rates — the curve is inverted.

Why does inversion predict recession?

• It compresses bank profitability. Banks borrow short and

lend long. When short rates exceed long rates, the spread on every new loan they write is squeezed. Banks pull back lending, credit creation slows, and the most cyclical sectors (housing, capex, autos, small business) starve first.

• It is the bond market's verdict on Fed policy. A 10-year

yield below the 2-year yield is the long end saying *the Fed has raised short rates too far, growth and inflation will both fall, and rates will be lower in the future*. The bond market is collectively wrong less often than any other market on macroeconomic forecasts of this kind.

• Inversion → expectation → behavior. Once corporate

treasurers, bank credit officers, and the financial press are all staring at an inverted curve, the anticipation of recession itself causes the recession. Capex gets deferred, hiring slows, layoffs get planned. The signal is partly self-fulfilling.

Lead time from first inversion to NBER peak in the post-1976 era:

First inversion	NBER peak	Lead
Aug 1978	Jan 1980	17 mo
Sep 1980	Jul 1981	10 mo
Dec 1988	Jul 1990	19 mo
Feb 2000	Mar 2001	13 mo
Jan 2006	Dec 2007	23 mo
Aug 2019	Feb 2020	6 mo (then COVID)
Jul 2022	TBD 2026?	22+ mo

Two warnings before you turn this into a personal market-timing system. First, the lead time is variable — you cannot use inversion as a "sell now" signal without giving up two to three years of late-cycle equity returns. Second, the un-inversion (when the curve re-steepens after being inverted) is historically a closer-to-the-event signal than the inversion itself. The re-steepening in late 2024 / early 2025 is the part of this playbook to watch through 2026.

2.5 Asset Performance by Regime

The single best summary of how asset classes behave across the cycle uses a four-regime framework based on the direction of real GDP growth and proximity to a recession.

• Early-cycle (recovery). Recession just ended; growth is

rebounding; inflation is still falling; rates are low. **Stocks > corporate bonds > Treasuries > cash > commodities.** Equity beta wins; cyclical sectors lead; small-caps and high-yield bonds outperform. This is the highest-Sharpe phase of the cycle for risk assets — historical S&P 500 returns in the first twelve months out of an NBER trough average roughly +35%.

• Mid-cycle. Growth is steady, inflation is contained, rates

are normalising. **Stocks > corporate bonds > commodities > Treasuries > cash.** Equities still win but the leadership rotates from cyclicals to growth/quality. This is the longest phase by duration and the easiest to be passively long.

• Late-cycle. Growth is still positive but decelerating;

inflation is rising; the central bank is tightening. **Gold / commodities > cash > stocks > corporate bonds > Treasuries.** Real assets lead; energy and materials outperform; long-duration bonds get punished as yields rise. This is the phase where 60/40 starts to fail because both legs come under pressure simultaneously — 2008 and 2022 both belong in this bucket.

• Recession. Growth is negative; inflation is falling (the

exception is stagflation, see §2.6); the central bank is cutting. Long Treasuries > gold > cash > stocks > commodities *on the way down*, but recession-tagged years usually contain a recovery rally too (1933, 1954, 2009), which lifts the realised full-year stock average. Duration wins; defensive equity sectors (staples, health care, utilities) outperform cyclicals; gold does well as real rates fall.

A static bar chart of the four-regime average annual return for the four major asset classes across 1929–2024:

Read the chart with three mental rules:

• The regime gap is large for the volatile assets. Stocks vary

about 8 percentage points across regimes; gold varies about 12; long bonds vary about 4. Even crude phase detection moves the expected return by more than a full year of equity-risk-premium.

• Recession-year stock returns look surprisingly positive.

That is because NBER recession years often contain the recovery rally — 1933 (+50%), 1954 (+53%), 2009 (+26%) all sit inside recession years. The bear-market damage is concentrated in the late-cycle tag (which catches 2008 and 2022), not in recession itself.

• No single asset wins every regime. That is why the four

tranches in the framework include a barbell of equities and long Treasuries plus a real-asset sleeve. The leg the cycle is hostile to now is exactly the leg that wins when the cycle changes.

The interactive Regime Explorer below lets you click any year from 1929 forward and see (a) which regime it was tagged with, (b) the realised one-year forward return of each of the four asset classes from that year, and (c) the average of all years sharing that regime. The exercise that does the most for your intuition: click 2007, then 2008, then 2009. Watch the sequence flip from late-cycle to recession to early-cycle, and watch the optimal asset class flip with it.

2.6 2008, 2020, 2022, and 2026

Four recent setups, each useful for a different reason.

2008 — the disinflation playbook works. Late-cycle imbalance (housing credit), recession, Fed cuts to zero, inflation collapses, long Treasuries rally violently (+20% in 2008), equities bottom in March 2009, the textbook 60/40 recovery unfolds across 2009–2010 almost on schedule. The cycle obeyed the textbook.

2020 — the exogenous shock case. A global health shock arrives, NBER calls the shortest recession in history (two months), the Fed and Treasury combine for the largest stimulus in peacetime. Long Treasuries, gold, and stocks all rally together because the response was overwhelming. This was not a cycle in the Juglar sense; the imbalances had not built up. Importantly, that response left a gigantic monetary overhang that became the input for 2022.

2022 — the textbook fails. Inflation surges to 9% on the back of the 2020–21 stimulus and supply shocks. The Fed hikes 425 bp in nine months. Long Treasuries lose 18%. Stocks lose 18%. 60/40 has its second-worst calendar year in history. The cycle was not a normal late-cycle / recession arc; it was a late-cycle inflation shock without a recession, and the disinflation playbook stopped working. The regime-shift point — *forty years does not make passive permanent* — earned its keep that year.

2026 (the live setup as of this writing). The yield curve inverted in July 2022 and re-steepened over 2024–25. Headline inflation is back near 3% but services inflation is sticky. Unemployment has drifted from a cycle low of 3.4% to roughly 4.5%. The S&P 500 hit a new high in late 2024 driven mostly by megacap AI capex. The leading indicators — yield curve re-steepening, ISM new-orders sub-50, jobless claims drifting up — are pointing at a 2026 slowdown. Whether it gets dated as a recession is a coin flip. **The lesson is not which way to bet but which signals to watch:** payroll diffusion, the credit spread index, and the ratio of cyclicals to defensives in the S&P 500. Those will turn before the price index does.

2.7 The Forty-Year Frame — Where the Regime Question Lives

Each Juglar cycle is about a decade. Each Kondratieff is about forty years. The forty-year disinflationary bull market that ran from August 1981 (Volcker peak) through the 2021 zero-rate top was one Kondratieff up-leg. Within it there were five Juglar recessions (1990, 2001, 2008, 2020) and roughly ten Kitchin inventory cycles. Every one of those Juglar recessions was followed by a stronger up-leg, because the Kondratieff backdrop was still disinflation and falling rates.

The 2022 inflation shock was the first material data point suggesting the Kondratieff backdrop may have flipped. We do not know yet — these things resolve over a decade, not a quarter. But the possibility of the flip is what makes the discipline of cycle-aware allocation matter more in the next ten years than it did in the last forty. When the backdrop is a smooth disinflationary tailwind, "buy and hold the index" is right ~95% of years. When the backdrop is uncertain, the cycle is the difference between making 8% real and losing 4% real.

---

3. Common Misconceptions

"A recession means the stock market has crashed." No. The

stock market typically peaks six to nine months before the NBER peak and bottoms six to nine months before the NBER trough. By the time a recession is confirmed, the market has usually done most of its damage and is partway through its recovery. Investors who wait for the headline announcement to re-enter buy the recovery rather than the bottom.

"Two negative quarters of GDP equals a recession." A

useful headline but not the official rule. NBER uses depth, diffusion, and duration across multiple series. The first half of 2022 was negative real GDP and was not dated as a recession because employment kept rising.

"The yield curve has stopped working." This is said before

every recession ("a new economy"), and was said again in 2023 when the recession people expected within a year of inversion did not arrive. The signal's lead time has always been wide (six to twenty-four months); a twenty-month delay is normal, not a failure.

"Unemployment is low so the economy is fine." Unemployment

is the most lagging of the lagging indicators. It is the worst summary statistic for current cycle position and the best summary statistic for the previous cycle position.

"The Fed will save the market like it always does." The Fed

saves the market when inflation is below target. When inflation is above target, as in 2022, the Fed and the market are on opposite sides of the trade — that is the entire point of the dual mandate. Conditioning on the post-1995 era of Greenspan / Bernanke / Yellen / Powell put expectations is an example of recency bias.

"Cycles are too noisy to use — you can't time them." False

dichotomy between precise timing (impossible) and *broad phase awareness* (clearly value-additive in the historical record). The four-regime asset performance gap is large enough that you do not need to be early or right at every turn. Getting one phase shift correct per decade pays for the discipline.

"60/40 is broken because of 2022." 2022 was the worst

60/40 year in eighty-five years and the second-worst in ninety-five. That is a statistically uncomfortable observation, not a structural one. Across the 95-year record, 60/40 has produced positive nominal returns in roughly 80% of calendar years. The 2022 lesson is that the stock-bond correlation is regime-dependent — a refinement, not a refutation.

"Commodities always hedge inflation." They hedge the kind

of inflation that runs through tradeable goods (energy, metals, food). They do not hedge service-sector inflation (rents, wages, healthcare). The 2022 inflation shock was tradeable; the 2024–25 stickiness is service-sector. Different hedge needed.

"Gold only works during inflation." Gold's performance is

primarily driven by real interest rates. Falling real rates help gold; rising real rates hurt it. That happens to coincide with inflation surprises sometimes, but the cleaner driver is the real-rate path, which is also why gold rallied from $1,800 in late 2023 to over $3,000 by 2025 even as headline inflation moderated.

"This time it's different." Said before every cycle peak.

The technology, the leadership stocks, and the policy details are different every time. The cycle structure — credit expansion, capex overshoot, central-bank tightening, slowdown, cuts, recovery — is not. Both halves of that statement matter.

---

4. Q&A Section

**Q1. If recessions are dated only after the fact, how is the information useful in real time?** A. It is useful in two ways. First, the historical pattern lets you put present readings (yield curve, payroll diffusion, ISM, high-yield credit spreads) on a calibration curve and assign rough probabilities. Second, the response-side discipline does not depend on knowing the start date precisely: when the leading indicators turn negative, you trim equity beta and add long-bond duration gradually. That is far cheaper than waiting for the NBER announcement, which arrives after the easy money has been made.

Q2. Should I sell stocks the moment the yield curve inverts? A. No. The lead time from first inversion to recession peak has ranged from six to twenty-four months. Selling on day one of inversion has historically meant giving up six to twelve months of further equity gains, often the best of the cycle. A more defensible playbook: at first inversion, stop adding equity risk, lengthen bond duration, and let payrolls / claims / credit spreads tell you when the actual rollover begins.

Q3. Why does cash do so badly across most regimes? A. Cash earns roughly the short rate, which is roughly inflation plus a small real premium in the long run. After taxes and inflation, the long-run real return on cash is close to zero. The only regime where cash strictly wins is the very brief window at the cycle peak when yields are high and risk assets are falling — and that window is usually six to twelve months, not the multi-year holdings most retail "cash on the sidelines" positions become.

**Q4. What is the difference between a recession and a bear market?** A. A recession is an economic event (output, employment, sales, income falling). A bear market is a price event (a major index falling 20% or more from its peak). They overlap but are not the same: 2018's S&P 500 drawdown was 19.8% with no recession; 2008–09 had both. The bear market is usually the bond market's and the stock market's anticipation of the recession to come plus the reaction once it arrives.

Q5. Are emerging markets and the US on the same cycle? A. Loosely. US monetary policy is the gravitational center — when the Fed tightens, dollar liquidity tightens for the world — but emerging market cycles can lead or lag the US by quarters because of commodity exposure, debt structure, and capital flows. We do not invest in non-US equity directly under this course's investable-universe rule; if you want exposure to the EM cycle, do it through US-listed multinationals or US-listed EM ETFs.

**Q6. What is "stagflation" and where does it fit in the four-quadrant framework?** A. Stagflation is the diagonal: growth falling and inflation rising. Late-cycle into recession with inflation refusing to roll over. Asset performance under stagflation is genuinely difficult because both legs of 60/40 lose: stocks fall on growth, long bonds fall on inflation. Commodities, gold, short-duration inflation-protected bonds, and equity sectors with pricing power (energy, staples) are the survivors. 1973–75 and 2022 are the two clearest US examples of the post-war era.

Q7. How long is the average expansion really? A. Post-1945, US expansions have averaged about sixty-three months from trough to next peak. Six of the twelve post-1945 expansions ran longer than five years. The two longest — 1991– 2001 and 2009–2020 — both ended with a one-time event (tech bubble unwinding plus 2001 recession; COVID for 2020), and without those endings each would have run further. Expansions do not die of old age; they die of imbalance plus a trigger.

**Q8. Where does the Federal Reserve's balance sheet fit in this framework?** A. As a liquidity layer underneath the cycle. The textbook cycle is about the real economy. Quantitative easing and quantitative tightening change the price of risk assets even when the real-economy cycle has not changed. The 2009–2014 QE era and the 2020–21 QE pulse are both clearer in equity index prices than in GDP data — which is part of why headline P/E ratios have drifted upward without earnings growth keeping pace. We come back to this in the central bank week.

**Q9. Should retirees and accumulators read the cycle the same way?** A. No. An accumulator with twenty years to compound benefits from buying into recessions because they are dollar-cost averaging at lower prices. A retiree with a five-to-ten-year spending horizon is on the wrong side of the cycle if the sequence of returns is bad early. The barbell (long Treasuries + equities) and the four-tranche structure exist mostly for the retiree case; an accumulator can run a simpler 60/40 if they have the discipline to keep buying through a 2008.

Q10. Is the 2026 setup more like 1973–75, 2008, or 2020? A. None of them cleanly. The closest analogue is probably the late 1960s into the early 1970s — the end of one Kondratieff disinflationary regime, sticky service inflation, an asset bubble in a narrow leadership group (Nifty Fifty then, AI megacaps now), and a fiscal trajectory that constrains how much the Fed can ease without re-igniting inflation. The lesson from that era is not "buy gold and run for the hills" — equity markets returned roughly nothing in real terms over the 1965–1982 stretch, but the years inside it were highly tradeable on a phase-aware basis. The discipline this lesson is teaching matters a great deal more in that kind of decade than it did in 2010–2020.