Reading the cycle: indicators, lags, and the yield curve

Economic indicators are conventionally grouped by their timing relative to the business cycle.

• Leading indicators turn before the cycle. They signal where the economy is heading. Examples: yield-curve slope, manufacturing new orders, building permits, consumer confidence, stock market levels, initial unemployment claims.
• Coincident indicators turn with the cycle. They measure where the economy is now. Examples: real GDP, industrial production, retail sales, non-farm payrolls.
• Lagging indicators turn after the cycle. They confirm what already happened. Examples: the unemployment rate, core inflation, business loan delinquencies, prime rate.

The taxonomy is structural, not absolute — particular indicators can move slightly earlier or later in particular cycles. The framework's usefulness is operational: when forecasting, weight leading indicators more heavily; when describing current conditions, use coincident indicators; when assessing the durability of a turn, look for lagging confirmation.

The rest of this lesson examines the most informative entries in each category.

The yield curve plots government bond yields against maturity. Its slope — typically measured as the 10-year yield minus the 3-month or 2-year yield — has the longest track record of any single leading indicator.

The empirical regularity: in major economies, an inverted yield curve (long yields below short yields) has preceded most recessions by 6–24 months. Each US recession since the 1960s has been preceded by an inversion of the 10-year minus 3-month or 10-year minus 2-year spread. Not every inversion produces a recession (false signals exist), but the rate of true positives has been high.

The structural reasons:

• Expectations channel. Long rates are roughly the average of expected future short rates plus a term premium. An inverted curve implies the market expects short rates to fall — which historically happens when the central bank cuts rates in response to a slowdown.
• Credit channel. A flat or inverted curve compresses bank net interest margins (banks borrow short, lend long). Lending becomes less profitable, banks tighten standards, credit contracts, demand falls.
• Information channel. The bond market aggregates the views of large numbers of professional investors with money at stake. The signal embeds their collective forecast.

The yield curve is read together with the term premium — a decomposition of yields into expected-future-rate components and risk premiums. Term-premium estimates (NY Fed, Adrian-Crump-Moench) help distinguish signal from noise.

The Purchasing Managers' Index (PMI) is a monthly diffusion index produced from surveys of purchasing managers at firms. Variants exist for many countries: ISM Manufacturing and ISM Services (US), S&P Global PMI (many countries), Caixin (China), NBS official PMI (China).

The construction: respondents report whether activity in each of five subcomponents (new orders, output, employment, supplier deliveries, inventories) has improved, deteriorated, or stayed the same since the previous month. Each subcomponent is converted to a diffusion index:

$\text{index} = \% \text{improved} + 0.5 \times \% \text{unchanged}.$

The index is above 50 when more respondents report improvement than deterioration; below 50 when more report deterioration. A composite index combines the subcomponents.

The structural value:

• Timeliness. PMI is published days after the month closes; GDP is published weeks after the quarter ends. PMI gives a real-time read of momentum.
• Granularity. Subcomponents tell you what is improving or deteriorating. Falling new orders with stable employment hints at near-future employment cuts.
• Cross-country comparability. PMI methodologies are similar across countries, enabling rapid comparison of cyclical position.

PMI is read as a level (above/below 50) and as a change (PMI trend, PMI relative to recent history). A persistent decline below 50 across several months in the new-orders subcomponent is a stronger signal than a single print.

Labor market data is both a coincident and lagging indicator, with different series providing different information.

Initial unemployment claims — a leading indicator. Weekly counts of newly filed unemployment insurance claims rise within weeks of employer layoff decisions. The four-week moving average is the standard read; rapid rises above the prevailing trend often precede broader weakening.

Non-farm payrolls — coincident. Monthly count of jobs added or lost across the economy, from the establishment survey. The headline number is volatile and revised; the three- or six-month average gives a cleaner signal.

Unemployment rate — lagging. From the household survey. Rises after layoffs are reflected in the establishment data and after labor-force participation adjusts. The unemployment rate's trough usually marks the late stage of an expansion.

Job openings and quits — both informative. The JOLTS data (US; similar series elsewhere) reports unfilled openings and voluntary quits. High quits indicate worker confidence in finding alternative employment (a hot labor market); falling quits typically precede broader labor-market weakening.

Wage growth — typically lagging. Average hourly earnings, employment cost index, Atlanta Fed wage tracker. Wages respond to labor-market conditions with lags; sustained wage acceleration usually appears after several quarters of tight labor markets.

The practical read: combine the leading (claims), coincident (payrolls), and lagging (unemployment) series. A coherent cyclical turn shows up across them in sequence over months.

Two definitions of recession coexist with different operational uses.

Technical definition (commonly cited). Two consecutive quarters of negative real GDP growth. Used in some country definitions and widely repeated in press coverage. Simple to apply, but real GDP is revised substantially after first release and the 'two consecutive quarters' criterion fails to capture short or shallow recessions.

NBER definition (US). A significant decline in economic activity, spread across the economy, lasting more than a few months. The National Bureau of Economic Research's Business Cycle Dating Committee determines peaks and troughs by examining real GDP, real income, employment, industrial production, and retail sales. Determinations are announced with substantial lags — often 6–18 months after the peak — and supersede the technical rule.

The NBER framework prioritizes depth, duration, and breadth over the strict GDP rule. The 2020 US recession was declared and dated by NBER despite being only two months long, because the depth and breadth across indicators satisfied the criteria.

Most developed-economy statistical agencies have their own dating committees with similar frameworks. The lesson for practical use: a 'recession is/isn't happening' headline that cites two GDP prints is using the simpler technical rule; a definitive call usually comes months later from a dating committee.

Market participants and policy makers often diverge from official dating in real time, calling 'recession risk' based on leading indicators while official datemakers wait for confirmation across multiple coincident series.

Inflation — particularly core inflation — is one of the most-watched lagging indicators of the cycle. The lag has structural causes from the previous lessons in this cursus.

• Sticky prices mean inflation responds to demand changes with 12–24 month lags. A recession's downward pressure on inflation appears well after the trough; an expansion's upward pressure appears well after the peak.
• Expectations anchoring means inflation responds slowly to most shocks if expectations are anchored. Cyclical turns produce small inflation movements relative to large output movements.
• Wages are sticky downward — workers resist nominal wage cuts even when unemployment is high. Wage and unit-labor-cost inflation lags employment.

A practical consequence: inflation is the wrong variable to use as a real-time cyclical signal. Inflation tells you about the cycle's past state, not its present or future. Central banks responding to inflation alone tend to overshoot — tightening into a slowdown because past inflation is still high, easing into an expansion because past inflation is still low. The mandate to forecast forward addresses this directly, but the political accountability for inflation outcomes pulls in the other direction.

The yield curve, PMI, and labor-claims data are far more cyclically informative in real time than CPI prints. Reading them together is the operational skill.

A coherent cyclical reading combines indicators across categories. A stylized example: imagine the following set of conditions.

• Yield curve: 10y–3m spread has been inverted for 9 months.
• PMI: Manufacturing PMI at 47 (below 50) for 3 months; new orders subcomponent at 44.
• Initial claims: four-week average has risen 20% from cyclical lows.
• Payrolls: three-month average has slowed to 100k/month from 250k.
• Unemployment rate: 3.8%, near multi-decade lows.
• Wage growth: 4.5% year-over-year, slightly below peak.
• Core inflation: 3.2%, above target but trending down.

Reading this set:

• Leading indicators (yield curve, PMI new orders, claims) all flash slowdown/recession risk.
• Coincident indicators (payrolls) confirm slowing momentum but not contraction.
• Lagging indicators (unemployment rate, wage growth, core inflation) still reflect the prior expansion.

The pattern is consistent with the late-expansion-to-early-slowdown phase: leading indicators have rolled over, coincident indicators are decelerating, lagging indicators still strong. A forecaster would treat recession probability as elevated; a central banker would weigh the leading-indicator signal against the strong-but-lagging inflation read.

This kind of multi-indicator reading is the operational synthesis of the entire cursus. The frameworks established in earlier lessons — what money is, how inflation responds, what the central bank's tools do, real vs nominal, fiscal–monetary interaction — let you interpret each signal correctly and combine them coherently.

Six lessons of macroeconomic framework:

• What money is. Base, broad, and the velocity equation. Most modern money is created by commercial banks through loans, not by central-bank reserve expansion.
• Inflation. Measured by several indices, driven by demand and supply pressures, fundamentally shaped by expectations. Persistent inflation is harder to address from the supply side than from the demand side.
• Central-bank toolkit. Policy rate, balance sheet, forward guidance — different instruments operating on different channels with different lags. Modern central banks have moved from corridor to floor systems and added balance-sheet policy.
• Real vs nominal. The discipline of dividing by the price level. Most economic decisions are real, most contracts are nominal, and inflation reallocates real wealth between them.
• Fiscal vs monetary. Two levers, two channels, one interaction. Fiscal multipliers depend on cyclical state and monetary regime. Debt sustainability depends on $r - g$ more than on headline deficits.
• Reading the cycle. Leading, coincident, lagging indicators. The yield curve and PMI are particularly informative; inflation is the wrong variable for real-time signal.

The framework does not predict the future and does not advocate particular policies. It gives you a structural reading of current data and a vocabulary for distinguishing claims about mechanism from claims about specific outcomes. The mechanism claims are durable; the outcome claims belong to forecasters and policymakers.