Indicators / Industrial production
Real economy

Industrial Production, Year over Year

Physical output from factories, mines and utilities, measured against the same month a year earlier. It is the most direct read on the goods-producing economy.

Updated monthly · Source: Federal Reserve G.17 via FRED, computed by us
Current
—
as of …
1-month change
—
—
Percentile
—
050100
Status
—

History

Loading…

What it covers

Industrial production measures physical output from three sectors: manufacturing, mining, and electric and gas utilities. Manufacturing is the bulk of it, roughly three quarters by weight. The index is built from physical quantities rather than revenues, which is what makes it different from most economic statistics: it counts tons, barrels, kilowatt hours and assembled units, then aggregates them with value-added weights.

The sector is smaller than it once was. Manufacturing is about eleven percent of GDP and eight percent of employment, down from roughly a quarter of both in the 1970s. But small sectors can carry large cycles, and this one does: capital goods are the first thing firms stop buying when demand softens and the first thing they resume when it stabilises. Manufacturing output typically turns a quarter or two before GDP does, in both directions.

The index is expressed against a base of 100 in 2017. A reading of 104 means output is four percent above the 2017 average. The headline on this page is the twelve-month change, which strips out the trend and shows the cycle.

Two extremes that are not what they look like

The maximum in the record is plus 62 percent, in July 1933, and the minimum is minus 34 percent, in February 1946. Neither is a business cycle. The 1933 figure is measured from a 1932 base that had collapsed to roughly half of its 1929 level, so the percentage is large because the denominator was destroyed. The 1946 figure is the post-war conversion from military to civilian production, when tank plants stopped making tanks and had not yet started making cars.

Anyone reading a chart of this series back to 1920 without that context will draw the wrong conclusion about volatility. The relevant modern range is narrower: since 1950, twelve-month changes have mostly stayed between minus fifteen and plus fifteen percent, and the largest post-war moves were 1974, 1982, 2009 and 2020, all recessions or oil shocks.

What the flat decade means

The more interesting feature of the last fifteen years is not volatility but the absence of growth. US industrial production in the mid-2020s is roughly where it was in 2007, having fallen sharply in 2008 and 2009, recovered through 2014, and then stalled. Mining has moved with the shale cycle and utilities with weather and data centres, but manufacturing has been broadly flat for a decade.

Two forces explain most of it. The first is import competition, which shifted assembly abroad while leaving high-value design and production at home. The second is productivity: US manufacturing output per hour has kept rising, so the same output needs fewer people. Employment in the sector fell by roughly a third between 2000 and 2010 and has been flat since, while output recovered. A flat output index and a flat employment index together mean productivity is not improving much either, which is the part that does not fit the standard story.

How to use it against other indicators

Industrial production is a coincident-to-leading indicator of goods demand, so it pairs naturally with the labour market. A rising twelve-month change with a falling unemployment rate is an expansion; a falling change with a rising unemployment rate is the start of a downturn. The two rarely disagree for long, and when they do, the production series usually turns first.

It also pairs with the consumer price index on the supply side. The 2021 and 2022 inflation episode had a clear goods component, visible as a surge in production followed by a collapse in delivery times and then in prices. Watching output alone will not tell you about inflation, but watching output against demand will.

How we calculate it

We take INDPRO from FRED, the seasonally adjusted total industrial production index, and divide each month's level by the level twelve months earlier. The result is rounded to two decimals. The index level is shown in the summary line so the base is visible.

The percentile is the share of monthly observations since January 1920 that sit below the current reading. That is a long and structurally different history, so the percentile is a coarse guide: the economy of 1930 is not comparable with the economy of 2026, and the distribution reflects that.

The source is a Federal Reserve release in the public domain, the G.17 statistical release, distributed through FRED. The percentage change is computed here.

Distribution since 1920

Each row is the reading at that point in the historical distribution. The sample spans the Depression, the war economy and the modern period, so treat the tails as historical rather than comparable.

Important

Not investment advice. This page is published for information and education only. It is not a recommendation, solicitation, or offer to buy or sell any security. Market data may be delayed, incomplete, or revised. Indicators describe the past; they do not forecast returns. Past performance does not guarantee future results. Do your own research or consult a licensed financial professional before acting on anything you read here.

Questions

Is industrial production the same as manufacturing output?

No. Manufacturing is about three quarters of the index by weight. The rest is mining, which moves with energy extraction, and utilities, which moves with weather and electricity demand. The mining component can dominate short-term moves even though it is small.

Why was the 1933 reading so extreme?

Because it was measured against a 1932 base that had collapsed. Industrial production fell by roughly half between 1929 and 1932, so the recovery year produced a huge percentage change from a destroyed denominator. It is a genuine number describing an artificial situation.

Does falling industrial production always mean recession?

Not always, but it usually means a goods-sector slowdown. The sector is cyclical enough that a sustained decline has preceded most post-war recessions. A single weak month is noise, particularly when utilities are involved.

Why is the latest month locked?

The full record back to 1920 and the methodology are free. The most recent month, which is the part that shows whether output is turning now, is part of the member plan.

Related