Leading, coincident and lagging: when an indicator speaks

Why it matters

Two records on Kitalpha describe the same labour market on different clocks. Initial jobless claims is weekly: as of 2026-09-05 it counted 206,000 people filing a first claim for unemployment insurance in the week. The unemployment rate is monthly, and its latest print is 4.1%. A reader who takes every release as a snapshot of today would treat the two as equally current. They are not: one tends to move months before the other. Which, and why, is the question this lesson answers.

The concept

Economic indicators are conventionally sorted into three groups by their timing relative to the economy as a whole. Leading indicators tend to turn before activity does. Coincident indicators move with it. Lagging indicators turn after it. The sorting is empirical, based on how each series has behaved around past turning points, and it is a tendency rather than a law; but the reasons behind the tendencies are concrete, and knowing them is what makes a release readable.

A leading series usually records a decision taken early in a change of direction. Weekly initial jobless claims are the clearest labour-market example: a person files a first claim in the week a job ends, so the count registers job losses almost as they happen, and it does so every week rather than once a month. Orders placed with factories and permits issued for new housing lead for the same reason: they are commitments made before the work is done. A coincident series moves with activity itself: payroll employment, industrial production, retail sales. A lagging series changes only after the effects have worked through. The unemployment rate lags because it is a stock, the number of people still looking, and that stock keeps rising for months after new job losses slow, since people who lost work take time to be re-hired and since the count of those looking shifts as participation changes. Consumer prices lag for a related reason: costs and wages adjust with a delay.

Two consequences follow for reading a calendar. First, a release dated today can describe a turn that began months ago. A rise in the unemployment rate in the latest print is not the first news of weakness; it confirms weakness that weekly claims registered earlier. Second, no indicator is a forecast. A leading series is early, not prophetic; it can rise for a few weeks and subside, which is why weekly claims are read against a four-week moving average, the mean of the last four prints, so that a single week’s holiday or weather effect does not pass for a turn.

The pairing in this lesson is Kitalpha’s substitute for the textbook pairing of purchasing-manager surveys against unemployment, which the site does not carry; it keeps the same logic with two series the site records every week and every month. The chart below plots two years of weekly claims. Look past the week-to-week jaggedness at the level over several months: that level is what the four-week average tracks and what the unemployment rate later confirms.

Worked example

Take the claims record as of 2026-09-05 and the four-week average published beside it. The steps below read the latest weekly print and the average of the last four, take the gap between them, and express that gap as a share of the average. Every figure comes from the named record on the date shown.

Read the gap in the third line as a distance from the recent run, not as a signal in itself. A single week can sit thousands of filings above or below the average because of a holiday, a storm or a reporting quirk, and the fourth line shows how small such a gap is in proportion to a weekly total in the hundreds of thousands. What a leading series offers is direction over several weeks: when the average itself drifts one way for a couple of months, the flow of new job losses has changed, and the slower measures have not yet caught up. Nothing in these steps says what happens next; they say where the fastest-moving labour-market count stood this week relative to its recent past.

Faded example

Now the lagging series: the unemployment rate as of 2026-08-01. The latest print and the print from a year earlier are given. Complete the last step: the change over the year in percentage points.

Retrieval check

Mark your confidence before each answer. Every option carries an explanation; read the ones you rejected too.

Your summary