The releases that move the tape: PMI, employment, retail sales and CPI
Four families of number produce most of the movement on any calendar. Read in the order they arrive they are not four releases — they are one chain, and each link predicts the next.
Lesson 10 of 12.
The chain, and why the order matters
These four are usually taught as a list. They are better understood as a sequence, because each one is an early read on the one after it.
PMI is a survey, so it arrives first — managers reporting what they are seeing before it reaches any official statistic. Employment is the first hard data, and it either confirms the survey or contradicts it. Retail sales say whether the wages from that employment were actually spent. CPI says what the spending did to prices — and that is the number the central bank of lesson nine is acting on.
Which is why a PMI surprise can move a currency more than its subject deserves. It is not news about factories. It is the first evidence about a rate decision three months away.
PMI: a survey, and the only level that matters is fifty
The purchasing managers index asks people who buy things for a living whether orders, output, hiring, deliveries and inventories are better or worse than last month. The answers are combined into one number, and it is constructed so that 50 is the dividing line: above 50 a majority report improvement, below 50 a majority report deterioration.
Two consequences follow from it being a survey rather than a measurement. It is fast, arriving weeks before the hard data it anticipates. And it measures direction rather than magnitude — 60 does not mean twice as good as 55; it means more respondents said "better".
The source runs four consecutive American prints against EUR/USD — previous figures of 57.5, 60.5, 58.7 and 64.7, forecasts of 56.6, 60.0, 58.8 and 65.0, actuals of 60.7, 58.7 and 60.8. Notice what that sequence does: two big beats, one clear miss, then another beat. A run like that is exactly the situation described in lesson five — a market with no strong prior left, where the next release matters more than usual.
Deck eight adds the European side of the same comparison: previous 64.4, forecast 63.0, actual 64.9 — a beat against an expectation of deterioration, landing on the same session as a German IFO reading of 99.2, a US retail figure and testimony from the chair of the Federal Reserve. Sessions stack like that routinely, and when they do the reaction belongs to the release the market cared about most, not to the one that printed first.
Employment: the hard data, and the part that is really about wages
Employment arrives as a family rather than as one figure — the change in jobs, the unemployment rate, and average earnings. They can disagree, and when they do the market usually follows earnings, because wages are the mechanism by which employment turns into inflation.
The source gives two clean examples of the unemployment rate surprising in both directions, both Canadian, read against CAD/CHF:
- Previous 9.4 per cent, forecast 9.2, actual 8.2. A very large beat — far more people in work than expected.
- Previous 7.5 per cent, forecast 7.8, actual 8.1. A miss in the other direction, and a reminder that this series moves both ways.
- And on Australian employment change against AUD/JPY, a run of previous readings of 50.0K, 29.5K and 88.7K — a series volatile enough that the forecast is doing real work, because no trend in it is obvious.
Retail sales, and why the core number is the one to read
Retail sales measure what consumers actually spent, broken down by sector. The source lists the components, of which apparel, department stores, food and beverage, electronics and appliances, and furniture are the legible ones.
The important distinction is between the headline and core retail sales, which strips out motor vehicles. One person buying a car is worth hundreds of people buying clothes, so a handful of vehicle sales can swing the headline without saying anything about consumer behaviour. Core is the cleaner signal, and it is the one to compare against its forecast.
The Canadian sequence against USD/CAD shows why forecasting this series is hard: previous readings of 1.9, −0.2 and 2.9 per cent, against forecasts of 0.1, 0.3 and −2.4. A series that swings from +2.9 to a forecast of −2.4 is not one anybody has a firm view on, and releases like that produce outsized reactions for exactly that reason.
CPI: the number the central bank is actually watching
The consumer price index measures the average change in prices consumers pay for a basket of goods and services over time. The source calls it the most widely used measure of inflation and the representative one, and that is fair — when a central bank talks about hitting a target, this is usually the series it means.
The same headline-versus-core distinction applies and matters more here than anywhere. Core CPI removes food and energy, the two most volatile components. That sounds like removing the parts people actually feel, and it is — but a central bank cannot change the oil price by moving an interest rate, so it sets policy against the part of inflation that responds to policy.
The American prints in the source are previous 0.6 per cent against a forecast of 0.2 with an actual of 0.8, and previous 0.8 against a forecast of 0.4 with an actual of 0.6. Both are substantial upside surprises, and deck eight reads them across three instruments at once — the dollar index, the S&P 500 and the copper price — which is the right way to read an inflation surprise, because it reaches all three through different routes.
Reading them together
The practical value is in the disagreements.
- PMI strong, employment weak — the survey is early or wrong. Wait for the second month before treating either as a trend.
- Employment strong, retail weak — income is being saved rather than spent. That is disinflationary and it argues against the rate rise the jobs number implied.
- Retail strong, CPI weak — real demand without price pressure, which is the combination central banks like most and currencies react to least.
- CPI strong, everything else weak — the worst case, and the one where a currency can fall on a rate rise, because the market stops reading the rise as strength and starts reading it as a policy error being corrected.
What to take from this one
Three things:
- The four families are a chain: survey, then jobs, then spending, then prices. Each is an early read on the next, which is why a survey can move a currency at all.
- Fifty on PMI, and core rather than headline on retail sales and CPI. Both are one-line rules that prevent the most common misreadings of these releases.
- The information is in the disagreements between them. Four releases pointing the same way tell you what you already knew from the first.
Next: why being right is not enough
Why a good number can be sold, what the ratio of open long to open short positions is telling you, and why it is read against the crowd.