Leading Economic Indicators

A leading economic indicator is a measured series that historically changes direction before broad economic activity does. The term names a category rather than a single product, and the category has two siblings: coincident indicators, which move at roughly the same time as the economy, and lagging indicators, which confirm a turn after it has happened. The Conference Board, a not-for-profit research organization, publishes a composite index for each branch, and its own wording sets out what the first two are for: "The Leading Economic Index (LEI) provides an early indication of significant turning points in the business cycle and where the economy is heading in the near term. The Coincident Economic Index (CEI) provides an indication of the current state of the economy."

The naming trips people up, because the same organization uses both a generic label and a branded one. "Leading economic indicators" is the category, and any series with the right timing properties can belong to it. "The Conference Board Leading Economic Index for the US" is one specific composite index built from a fixed list of ten of them. Coverage often shortens the second to the first, which makes a single publisher's product sound like the whole field.

What is actually inside the leading index is the part most worth knowing, because the composite is usually described as though it were a market indicator. Its ten components, as the publisher lists them, are average weekly hours in manufacturing; average weekly initial claims for unemployment insurance; manufacturers' new orders for consumer goods and materials; the ISM Index of New Orders; manufacturers' new orders for nondefense capital goods excluding aircraft orders; building permits for new private housing units; the S&P 500 Index of Stock Prices; the Leading Credit Index; the interest rate spread between 10-year Treasury bonds and the federal funds rate; and average consumer expectations for business conditions. One of the ten is a stock-market series. The rest are order books, hiring hours, jobless claims, building permits, credit conditions, the term spread and survey expectations.

The coincident index is the quiet link back to the official recession record. Its four components are payroll employment, personal income less transfer payments, manufacturing and trade sales, and industrial production, and The Conference Board states that these "are included among the data used to determine recessions in the US." So the coincident composite is assembled from the same kind of monthly activity series that the National Bureau of Economic Research weighs when it dates turning points. That is why a coincident index describes the present while a leading index makes a claim about the near future.

The leading index also comes with a published decision rule, which is unusual and makes it falsifiable. The publisher calls it the 3Ds: "Duration refers to how long the decline has lasted. Depth denotes the size of decline. Duration and depth are measured by the rate of change of the index over the most recent six months at an annualized rate. Diffusion is a measure of how widespread the decline is among the LEI's component indicators—on a scale of 0 to 100, a diffusion index reading below 50 indicates most components are weakening." The rule itself: "The 3Ds rule signals an impending recession when: 1) the six-month diffusion index lies at or below 50 ... and 2) the LEI's six-month growth rate (annualized) falls below the threshold of −4.1%." The publisher separately describes the index as a tool that "anticipates—or 'leads'—turning points in the business cycle by around seven months," which is its claim about its own product rather than an independent finding.

The asymmetry between signal and confirmation is the honest lesson, and it is visible in two documents. In its release of September 18, 2025, The Conference Board reported: "Widespread weakness among the LEI's components and a negative growth rate over the past six months triggered the recession signal in August." The same document also stated that the publisher was "not forecasting recession currently." NBER's chronology, read in September 2026, still shows its most recent peak as February 2020, and NBER describes its own approach as retrospective, waiting "until sufficient data are available to avoid the need for major revisions to the business cycle chronology." So a leading index produces a dated, checkable signal while events are still unfolding, and the body that assigns the recession label does not. That gap is structural. It means a reader can know what a leading indicator said last month and cannot know, inside any useful horizon, whether the label it pointed toward will ever be applied.

Each composite is built by combining its component series into one index number scaled to a base year, so movements are read as changes rather than levels. The Conference Board's summary table states the convention directly: "Indexes equal 100 in 2016." An index of 100 therefore means "the same as the 2016 average," which carries no information about whether that was a good year or a bad one. Only the direction, the rate of change and the diffusion across components carry meaning, which is exactly what the published decision rule uses.

A hypothetical worked through the rule shows how the two conditions interact. Suppose a leading index has fallen at an annualized rate of 5.0 percent over the most recent six months, which is below the stated threshold of −4.1 percent, and that its six-month diffusion index reads 40, which is below 50. Both conditions are met, so the rule signals. Now change one number: keep the same 5.0 percent decline but suppose the diffusion index reads 65, meaning most components are still improving and the fall is concentrated in a few. The depth condition is met and the diffusion condition is not, so the rule does not signal. The design is deliberately unwilling to call a downturn on a large move in a narrow part of the economy.

What the category is good for

• It is forward-looking by construction, which is what most other economic statistics are not.
• A composite spreads the reading across several series, so one bad month in one industry does not move it much.
• The Conference Board publishes its components and its decision rule with numeric thresholds, which makes its signals checkable after the fact rather than a matter of interpretation.

Where it misleads

• A signal is a probability statement, not a dated event, and the publisher can and does issue one while declining to forecast a recession.
• The official U.S. recession chronology is assigned retrospectively, so a leading signal can never be confirmed or refuted within the period a reader is living through.
• The index level is meaningless on its own, because the base year is an arbitrary calendar year rather than a neutral condition.
• Component lists are revised over time, so a long chart of one index is not a chart of one unchanging recipe.
• Nothing in the category is a market-timing tool. Only one of the ten leading components is a stock-market series, and the index is about economic activity rather than prices.