When Did American Deindustrialization Begin?

When Did American Deindustrialization Begin?

LEARN / REINDUSTRIALIZATION

When Did American Deindustrialization Begin?

There are three defensible answers, because manufacturing employment, manufacturing’s share of employment, and manufacturing’s share of national output each turned at different times. Any source giving a single confident date has chosen a series without telling you.

Key points

Why the question has three answers

Deindustrialization is not a single event with a date attached to it, and the disagreement about when it started is mostly a disagreement about what is being counted. A country can employ fewer people in factories while producing more goods than it ever has. It can hold factory employment steady while the rest of the economy grows around it, so that manufacturing shrinks in relative terms without losing a single job. Each of those describes a different phenomenon, each is measured by a different federal series, and each produces a different starting date.

Three candidate answers follow, drawn from three separate government sources. They do not contradict one another. They answer different questions, and the reason the public argument stays unresolved is that participants rarely say which question they are answering.

Answer one: June 1979, if you count workers

The Bureau of Labor Statistics has tracked manufacturing payroll employment through the Current Employment Statistics program since January 1939. That series rises through the wartime buildup and the postwar decades, reaches its all-time high in June 1979, and never recovers to that level through five subsequent recessions.

Data12,611,000
Manufacturing payroll employment, July 2026, seasonally adjusted. Against the June 1979 peak of roughly 19.5 million, that is a decline of approximately 35 percent. Source: BLS Current Employment Statistics, series CES3000000001.

Series and provenance
Agency
U.S. Bureau of Labor Statistics
Program
Current Employment Statistics
Series
CES3000000001 (FRED: MANEMP)
Measure
All Employees, Manufacturing
Units
Thousands of persons
Adjustment
Seasonally adjusted
Period
January 1939 to July 2026
Latest
12,611 thousand, July 2026
Source tier
Primary
Retrieved
August 9, 2026

Establishment survey. Counts payroll jobs rather than people, so a worker holding two jobs is counted twice.

Open the source

This is the answer most people mean when they date the beginning of decline, and it is the one that carries the most political weight, because a job is something a person holds and loses. It is also the answer most exposed to the objection in the next section, since a fall in employment tells you nothing on its own about whether output fell with it.

Answer two: the early postwar decades, if you count share

Manufacturing’s share of total employment peaked well before its absolute level did. Total nonfarm employment kept growing after the war while manufacturing employment grew more slowly, so factories accounted for a falling fraction of American work long before the number of factory workers began to fall. On this measure, relative decline had been underway for roughly a quarter century by the time the absolute count turned over in 1979.

Policy consequences follow from the distinction, because the two measures imply different causes. A falling share is consistent with an economy adding jobs faster elsewhere, which is what growth looks like. A falling absolute count is consistent with capacity leaving. Treating the first as evidence for the second is one of the most common errors in the public argument.

Answer three: the output series cannot tell you

The third way to ask the question is whether manufacturing’s contribution to national output has fallen. The Bureau of Economic Analysis publishes manufacturing value added as a share of GDP in its industry accounts, and the quarterly series most often cited falls steadily across every year it covers.

Data13.1% → 9.4%
Manufacturing value added as a share of GDP, first quarter 2005 to first quarter 2026. Source: U.S. Bureau of Economic Analysis, Value Added by Industry, series VAPGDPMA, retrieved via FRED.

Series and provenance
Agency
U.S. Bureau of Economic Analysis
Program
GDP by Industry accounts
Series
VAPGDPMA
Measure
Value Added by Industry: Manufacturing as a Percentage of GDP
Units
Percent of GDP
Adjustment
Quarterly, not seasonally adjusted
Period
Q1 2005 to Q1 2026
Latest
9.4 percent, Q1 2026
Source tier
Primary
Retrieved
August 9, 2026

This quarterly series begins in 2005. BEA publishes annual GDP-by-industry estimates for 1947 forward, with a classification break between historical SIC-based and current NAICS-based vintages.

Open the source

That quarterly series begins in 2005, which is twenty-six years after the employment peak, so the chart above cannot by itself date anything earlier than its own first observation. The limitation belongs to the series rather than to the evidence. BEA publishes annual GDP-by-industry estimates on a NAICS basis for 1947 forward, and SIC-based annual estimates for 1947 through 1997, which do reach the relevant period.

A measurement note on the peak figure

Measurement noteThe annual BEA series that reaches 1947 is not directly comparable across its whole length. BEA notes that historical SIC-based industry accounts are not consistent with the latest updates to the annual accounts, and the NAICS-based estimates cover 22 industry groups for 1947 to 1976 against 65 from 1977. Longer-run comparisons are possible and require the classification break to be stated.

Measurement noteBLS publications give the June 1979 peak as both 19.6 million and 19.4 million, depending on the article and the series used. Both figures appear in official BLS publications and neither is an error. They reflect different series, different seasonal adjustment, and in some cases annual averages rather than monthly readings.

Share figures show the same pattern. BLS has published manufacturing’s peak share of total nonfarm employment as 32 percent in May 1953 in one publication and 38 percent in 1943 in another, again because the two are measuring against different denominators over different periods.

In practice this means that the size of the decline holds regardless of which figure is used. Measured from 19.6 million, employment has fallen about 36 percent. Measured from 19.4 million, about 35 percent. The conclusion survives the choice of series, which is the test worth applying whenever two official numbers disagree.

Institute analysisA source that reports one confident figure without naming its series is providing less information than it appears to. Where two federal publications disagree, the useful response is to test whether the conclusion depends on the choice. If it does not, the disagreement is a footnote. If it does, the disagreement is the finding.

The word is older than the argument

Vocabulary for industrial decline was in circulation long before the employment series turned. The Oxford English Dictionary traces “deindustrialization” and its related forms to the 1930s, when American mill towns were already losing the textile capacity that had built them. Lowell, Massachusetts, had watched its mills move south for two decades by then, which is worth holding alongside any account that treats industrial decline as a phenomenon of the 1970s or the 1990s.

Regional deindustrialization has been a recurring feature of American industrial history rather than a single national episode. New England lost textiles to the South. The South later lost them abroad. The national aggregates conceal these movements because a job lost in Massachusetts and gained in South Carolina nets to zero in a national series while transforming both places.

Common misconceptions

That the decline began with a specific trade agreement. The employment peak precedes NAFTA by fourteen years and permanent normal trade relations with China by twenty-one. Trade policy can be argued to have accelerated a decline already in progress. It cannot be argued to have started one that began before it.

That falling employment proves falling production. Output per worker rose across the same period, so employment and production can move in opposite directions. Establishing that production fell requires the output series, not the employment series, and that argument is taken up separately.

That the 1979 peak is a target. The composition of industrial work changed, and rebuilding productive capacity would not reconstitute the 1979 employment structure. A policy measured against that benchmark is being measured against something nobody is offering.

What the evidence says, and where it is contested

ContestedThe dates above are not in dispute. What they mean is. One reading holds that a falling employment share through a period of rising output is the ordinary signature of a productive economy, and that dating decline from it mistakes success for failure. A second holds that the absolute turn in 1979 marks the point at which specific capabilities began leaving the country, and that aggregate output conceals which ones. A third holds that the question is regional rather than national, because the national series nets out movements that determined the fate of particular places.

These readings disagree about what should be measured, so no additional data resolves them. The Institute’s position, stated as a position, is that the capability question is the decision-relevant one: whether a specific thing can be produced at required volume within a required time tells a government more than any aggregate, and it is the question none of the three series above was built to answer.

Related Institute research

What Is Reindustrialization?
The definition, and why reshoring and reindustrialization answer different questions.

The Production Gap That’s Costing America the Technology Race
The distance between what the country designs and what it can build.

Who Trains the Next Generation? The Workforce Training Commons Problem
Why the workforce constraint persists even where firms want to hire.

Sources

Reference entry maintained by the Institute for American Manufacturing & Technology. Figures are drawn from primary sources and cited above. Where the Institute states a position rather than a finding, it is marked as such.

Private Professional Network

Institute for American Manufacturing & Technology

Join the IAMT Network. Connect with researchers, policymakers, and industry leaders working on American manufacturing, energy, and technology policy.

Apply for Membership →