Skip to content

01 · Macro Data Reading: Expectation Gaps, Structure Breakdown, and a Complete Interpretation Template

The same "CPI 3.2%" — why do some shout bearish while others shout bullish? Because the number itself means nothing; the gap between the number and expectations is what matters. This article builds the "expectation gap" mindset, then unpacks reading details for CPI, PMI, nonfarm payrolls, and the unemployment rate one by one, closing with a replicable complete template for interpreting macro data.


1. Expectations and Expectation Gaps: Lesson One of Data Reading

Expectation gap: prices react only when actual ≠ forecast

Markets are arenas of trading against expectations: asset prices already embed the market's shared expectations of the future; on release, price reacts only to the part where "actual ≠ forecast". That gap is the expectation gap.

Numeric Example: Is CPI 3.2% Bullish or Bearish?

Suppose the Fed is in a "rate-cut debate" phase and the market expects this month's CPI YoY at 3.0%:

ScenarioActual releaseExpectation gapMarket readingEquity reaction (reference)
A3.0%0 (in line)Business as usual, no new informationMuted
B2.8%0.2pp below forecastInflation falling faster than expected → rate-cut hopes riseBullish for stocks
C3.2%0.2pp above forecastThe "prices returning to 3.0%" expectation fails → cuts delayedBearish for stocks

Key conclusion: in scenario C, 3.2% isn't high in itself, but it is "above expectations" — the easing path the market had bet on is interrupted, and assets get priced as bearish. Bullish and bearish are never a function of the number's level but of the gap between "actual vs forecast".

Lesson One of Data Reading

Bullish and bearish are never a function of the number's level but of the gap between "actual vs forecast". 3.2% isn't high in itself, yet "0.2pp above expectations" makes assets price as bearish — trade the expectation gap, not the number.

Where Do Expectations Come From?

  • Median of Bloomberg/Reuters economist surveys (consensus forecast)
  • Institutional nowcasting models (Atlanta Fed GDPNow, New York Fed Nowcasting, etc.)
  • Rate market pricing (hike/cut probabilities implied by fed funds futures)

💡 In-Line Data Almost Never Moves Markets

Data matching forecasts almost never produces a move; big deviations from forecasts are the source of volatility. Many beginners chase right at the release only to watch price barely twitch because the data matched forecasts — they were trading the "number", not the "expectation gap".


2. Breaking Down the CPI Structure

Core CPI vs Headline CPI

IndicatorDefinitionUse
Headline CPIFull basket of goods and services pricesNews headlines; shapes public perception
Core CPIExcludes food and energyPolicy reference; less volatile, better reflects inflation trend

Food and energy prices are heavily influenced by weather, geopolitics, hog cycles and other external factors, swing violently monthly, and easily mask trends. Central banks watch trends, which is why core CPI is the number with the strongest policy implications.

How to Read the Components: Shelter, Energy, Food

  • Shelter: about 1/3 of US CPI weight, the largest single component. Rent is extremely sticky — only when the shelter component peaks and turns down is US inflation's substantive decline confirmed.
  • Energy: transmits directly from oil prices with large monthly swings — MoM beats YoY for reading it; oil surges/slumps often distort headline CPI beyond recognition.
  • Food: same logic — driven by weather and supply events; single-month swings say nothing about trend.

The "Hog Cycle": A Distortion Common Knowledge in China's CPI

Pork carries a significant weight in China's CPI; the hog cycle (breeding-sow inventory → slaughter supply 10-14 months later → pork price) periodically pulls/drags CPI:

  • During pork upcycles, CPI may be pushed higher even if other prices are stable;
  • During pork downcycles, CPI can be pressed to extremely low or even negative readings — a low CPI then does not mean deflation pressure; it's just the hogs helping.

So when reading China's CPI, always pull the pork component out separately: CPI excluding pork is the reading that reflects real demand.

Why the Fed Watches PCE Instead of CPI

The Fed's preferred inflation gauge is the core PCE price index (the Fed's 2% target definition), not CPI. Differences:

DifferenceCPIPCE
Compiled byBureau of Labor Statistics (BLS)Bureau of Economic Analysis (BEA)
Data sourcesConsumer expenditure survey (small sample)Business sales data (broad coverage)
WeightingFixed basketDynamic weights (auto-reflects shifting consumption structure)
HealthcareMostly consumer out-of-pocketIncludes employer-paid medical spending
Level tendencyUsually ~0.3-0.5pp above PCEMilder; the policy target gauge

💡 CPI Is the Dress Rehearsal; PCE Grades the Final Exam

After every CPI surprise, markets turn to the same month's PCE released two to three weeks later — CPI is the rehearsal; PCE is the Fed's official grading.


3. Reading the PMI

Boom-Bust Line 50: The Passing-Mark Logic

PMI uses 50 as its boom-bust line:

PMIMeaning
Above 50Manufacturing expanding MoM
Equal to 50Flat vs previous month
Below 50Manufacturing contracting MoM

💡 The 50 Line Is Not a Good/Bad Boundary

50 only divides "expanding vs contracting MoM"; it is not a good/bad boundary. 50.5 counts as lackluster-but-stable, 49.5 as slight contraction. What truly matters is trend and components, not any single month's print.

New Orders − Finished-Goods Inventory: The "Inventory Cycle" Signal

PMI is a survey-based forward-looking indicator; its most valuable part is component combinations. Watch the relationship between two core components:

New ordersFinished-goods inventorySignal
RisingFallingDemand improving, inventory being digested → precursor to active restocking, bullish
RisingRisingDemand warming but firms restocking cautiously, or demand overheating superficially
FallingRisingDemand weakening, inventory piling up → likely future discounting to destock, bearish
FallingFallingFirms actively destocking and cutting output → digestion phase before a bottom

The "new orders − finished-goods inventory" spread works as a simple leading index: when the spread bottoms and turns up, it often leads industrial profits and equity turning points by several quarters (see Industry Data Reading for the full inventory cycle).

Caixin vs Official PMI: Sample Differences

China has two manufacturing PMIs:

AspectOfficial (CFLP/NBS) PMICaixin PMI
Compiled byChina Federation of Logistics & Purchasing + National Bureau of StatisticsCaixin + S&P Global
SampleLarge/medium/small firms, weighted toward large-medium/SOE namesWeighted toward small-medium/private/export-oriented firms
ReleaseLast day of each month (9:30)First business day of each month (9:45)
Use as referenceAggregate, policy orientationSME and foreign-trade health

💡 The Two PMIs Are Convincing Only When They Agree

The two datasets are convincing only when directionally consistent; divergence (official weak, Caixin strong) usually signals "big firms weak, small firms strong" or vice versa — structural differentiation rather than an aggregate signal.


4. Reading Nonfarm Payrolls

The NFP report (US Department of Labor, first Friday of each month, 20:30/21:30 Beijing time) contains a trio: nonfarm job additions, the unemployment rate, and average hourly earnings growth. The three numbers often contradict each other.

How to Weigh the Three Signals

SignalMeaningMarket impact (reference)
Job additions beatEconomy resilientBullish dollar and stocks; but if it delays cuts, good news turns bad
Unemployment rate unexpectedly risesLabor market weakeningBearish dollar, bullish cut expectations
Hourly earnings rise YoYInflation pressure persistsBearish for cut expectations

Weighing Three Contradictory Signals in One NFP Report

Numeric example: one month's NFP report —

  • Job additions 250k (forecast 180k, a large beat)
  • Unemployment 4.3% (previous 4.0%, unexpectedly up 0.3pp)
  • Hourly earnings +3.9% YoY (forecast 4.1%, below expectations)

Surface contradictions: jobs strong → yet unemployment rose; wage growth slowing → wage-price pressure easing. How does the market weigh them?

  • First look at the driver: if the unemployment uptick is driven by "rising labor force participation" (more people entering to look for work), that's supply-side improvement and actually benign; only if driven by falling employment does demand weaken.
  • Then look at wages: below-expectation earnings mean the wage-price spiral is cooling, relieving the "strong jobs → sticky inflation → no cuts" transmission chain.
  • Reference conclusion: such a combination is often read as "resilient economy but easing inflation pressure" — neutral-to-positive for risk assets.

Common knowledge: the first wave after NFP often takes direction from whichever number looks most extreme, but the 30-60 minute post-open direction is professional money's true verdict after weighing. Don't chase the first wave.

The First Wave After NFP Is a Trap

The 30-60 minute post-open direction is professional money's true verdict after weighing. The first wave after release is usually just an emotional reaction to "whichever number looked most extreme"; chasing it means using retail money to fill institutional orders.

The "Revisions" Trap in Employment Data

  • The NFP initial print gets heavily revised up or down over the following two months as samples are supplemented; revisions exceeding 100k are routine — e.g., the March 2024 initial print was later cumulatively revised down by hundreds of thousands.
  • Common knowledge: the initial value is a provisional number treating "probable" as "certain". Heavy betting on initial prints means building decisions on numbers that will be rewritten.
  • Right approach: follow the three-month moving average and cumulative revision trend; don't fixate on a single month's initial print.

5. Details of the Unemployment Rate

U3 vs U6

MeasureDefinitionNumerical relation
U3Official unemployment: unemployed and actively seeking / labor forceThe headline rate, most quoted
U6Broadest measure: U3 + marginally attached workers + involuntarily part-time for economic reasonsUsually ~1.5-3pp above U3

U3 falling while U6 rises says the surface unemployment rate looks fine, but "hidden unemployment" (people giving up the search, those forced into part-time) is growing — labor-market quality deteriorating.

Labor Force Participation

Labor force participation rate = labor force / working-age population; it explains the unemployment rate's "illusions":

  • Falling participation (people exiting the labor force) passively lowers the unemployment rate — the denominator shrinks, the number improves, but employment hasn't.
  • Post-pandemic US participation has stayed persistently below pre-pandemic levels, so read claims of "historically low unemployment" with a discount.

6. The "Time-Lag" Awareness of Data

Nearly all macro data passes through a initial → revised → final revision chain:

StageTraitsTrading implication
Initial (preliminary)Incomplete sample, released fastestBiggest market reaction, least reliable
RevisedAmended after supplementary samplesOften ignored but closer to truth
FinalFully revisedOnly affects historical series, not current trading

⚠️ Don't Build Long-Term Conclusions on Initial Prints

GDP, NFP, and CPI all undergo systematic revision. Don't build long-term conclusions on initial prints; even in short-term data trades remember — you're trading "market sentiment about the initial print", not "economic truth".


7. The Complete Macro Data Interpretation Template

For any macro data release, walk six steps (also fixable as your own notes template):

text
① What is the data      → name, definition, coverage period (CPI YoY? core? seasonally adjusted?)
② What was expected     → what was consensus? what was the market betting on?
③ What was actual       → how far off the forecast? how many standard deviations?
④ What's the structure  → who drove the gap? break down by component/region/industry
⑤ First market reaction → which assets moved at release? direction and magnitude? consistent with intuition?
⑥ Sustainability        → single-month noise or trend turning point? which follow-ups (PCE/PMI/minutes) will verify?

Template quick-fill (CPI example):

StepEntry
① DataUS CPI YoY (headline + core)
② ExpectedCore CPI forecast 3.2%; market pricing a 70% chance of a September cut
③ ActualCore 3.0%, 0.2pp below forecast
④ StructureSlowing shelter was the main driver; energy dragged
⑤ First reactionTreasury yields fell, gold jumped, Nasdaq gapped up
⑥ SustainabilityWait for PCE in two weeks; if PCE falls in the same direction, the cutting logic is confirmed

Risk Warning

⚠️ Risk Warning

Macro data releases are often accompanied by violent gaps, widened bid-ask spreads, and vanishing liquidity; heavy positions or market orders can fill at terrible prices. The "market expectations" underlying expectation-gap judgments shift dynamically, and initial prints are frequently revised substantially — decisions based on initial data carry extra uncertainty. All indicator definitions, release arrangements, and numeric examples here are teaching references only; defer to the latest official definitions and latest market conditions. This article is not investment advice.

Further Reading

For study and research only — not investment advice. Markets are risky.