Ten US economic indicators

Three groups

Leading indicators move before the cycle turns, coincident indicators turn with it, and lagging indicators follow afterwards. The grouping follows the established convention in business cycle work; nothing is invented here, and the ten series are never combined into a score.

Leading

The 10-year minus 2-year spread was +0.41% (2026-08); consumer sentiment 55.2 (2026-07); building permits 1,394k units (2026-08) and housing starts 1,275k units (2026-08), the last two seasonally adjusted at annual rates.

Coincident

Real GDP grew +1.48% at an annual rate in the quarter beginning 2026-04; retail sales were +6.04% above a year earlier (2026-08) and industrial production +1.42% (2026-08).

Lagging

The unemployment rate was 4.1% (2026-08); core CPI was +2.45% above a year earlier (2026-08); nonfarm payrolls changed by +162k on the month (2026-08, +603k over the last twelve months).

Sample

On this page the ten series span 2015-08 to 2026-08. Each cell draws at most 132 observations; quarterly real GDP has 40. Missing observations are left blank — never carried forward, never filled with zero.

Updates and disclaimer

Every series is refetched automatically once each weekday from FRED, which redistributes official statistics from the Bureau of Labor Statistics, the Bureau of Economic Analysis, the Census Bureau, the Federal Reserve Board and the University of Michigan. This is a personal project with no affiliation to any of them; the page is a data monitoring tool and does not constitute investment advice.

Indicators that change before the wider economy turns. Most come from the parts of the economy that react first to interest rates and expectations: how the bond market is priced, how households feel, and the permits issued before any ground is broken. A turn here does not guarantee a recession; it means the pressure shows up here first.

Indicators that turn at roughly the same time as the business cycle. Their job is to confirm where the economy is now rather than to predict. Output, sales and production are exactly the kinds of data the NBER weighs when it dates a recession.

Indicators that move only after the turning point. Firms take time to adjust headcount and prices pass through more slowly still, so employment and inflation usually turn later than output. That is not a flaw: their use is to confirm that the turn really happened.

What do leading, coincident and lagging indicators mean?

The three labels describe when a series moves relative to the wider business cycle. Leading indicators change before the economy turns; most of them come from the parts of the economy that react first to interest rates and expectations, such as the Treasury yield spread, household sentiment, and the permits that must be issued before any ground is broken. Coincident indicators turn at roughly the same time as the cycle, so their job is to confirm where the economy is now rather than to predict: real GDP, retail sales and industrial production belong here, and they are exactly the kinds of data the NBER weighs when it dates a recession. Lagging indicators move only after the turning point, because firms take time to adjust headcount and prices pass through more slowly still, which is why unemployment, core inflation and payrolls usually turn later than output. It is worth being clear that the grouping is about timing, not reliability: a turn in a leading indicator does not guarantee a recession, and a lagging indicator is not useless — confirming that the turn really happened is precisely what it is for. This site follows the established grouping and does not invent one, nor does it combine the ten series into any single score.

How should I read a year-over-year rate against a level?

A level is simply what the indicator measured in that period — an unemployment rate of 4.1%, or 1,394 thousand building permits. A year-over-year rate is the percentage change from the same period twelve months earlier, which strips out seasonality and most of the long-run trend. Each series here is stored in whichever form it is most commonly quoted: core CPI, retail sales and industrial production as year-over-year rates; the unemployment rate, the yield spread, consumer sentiment, building permits and housing starts as levels; real GDP as a quarterly change at an annual rate; and nonfarm payrolls as a monthly change in thousands of jobs. Every chart caption states in words which transform it is drawing and what the source units are, because the same number means wildly different things under different transforms — the core CPI reading of 2.45 is a percentage change, not an index value. One trap is worth naming: a falling year-over-year rate means prices or sales are rising more slowly, not that they are falling. For that, the rate has to go below zero.

Why is GDP quarterly, and why does it arrive so late?

Real GDP is compiled by the Bureau of Economic Analysis, which has to reconcile household consumption, business investment, government spending, inventories and trade into one consistent set of accounts. The source data behind those pieces arrive at monthly, quarterly and annual frequencies, and the whole thing then has to be deflated to get a figure in real terms. That reconciliation cannot be done every month, which is why GDP is a quarterly series. The release schedule follows from the same constraint: an advance estimate about a month after the quarter ends, then a revision in each of the next two months, with annual and comprehensive revisions reaching further back after that. In other words, by the time the latest quarter is published, the period it describes ended one to four months ago. That is exactly why it sits among the coincident indicators here rather than being treated as a live reading: it confirms where the economy was, accurately but slowly. For something timelier, monthly retail sales and industrial production cover parts of the same activity and are published far sooner.

How often is this page updated, and where does the data come from?

Automatically, once every weekday. All ten series come from FRED at the Federal Reserve Bank of St. Louis, and the agencies that compile them are the Bureau of Labor Statistics (the unemployment rate UNRATE, core CPI CPILFESL and nonfarm payrolls PAYEMS), the Bureau of Economic Analysis (real GDP GDPC1), the Census Bureau (building permits PERMIT, housing starts HOUST and retail sales RSXFS), the Federal Reserve Board (industrial production INDPRO and the 10-year minus 2-year spread T10Y2Y), and the University of Michigan (consumer sentiment UMCSENT). The series themselves are monthly or quarterly, but those agencies publish on dates spread right across the month — the employment report early, CPI mid-month, building permits around the 17th, GDP at the end — so refetching daily gets new observations live as soon as they appear, and a run that finds nothing new changes nothing. The full fetched dataset, every observation included, is published as JSON at /data/indicators.json.