Economic Depression

An economic depression is a contraction in economic activity severe enough, or long enough, that observers reach for a stronger word than recession. That is the whole of it: the term marks a judgment about degree, not a category with a test. The National Bureau of Economic Research, whose Business Cycle Dating Committee sets the official-in-practice dates for US recessions, addresses the question directly and declines to draw the line. "The NBER does not separately identify depressions in its business cycle chronology," it says. "The period between a peak and a trough is a contraction or a recession, and the period between the trough and the peak is an expansion." The entry on recession covers who does that dating and how long it takes.

The display name here is "economic depression" rather than the bare word for a reason worth a sentence. "Depression" on its own is ambiguous in ordinary English, most commonly meaning the clinical condition, and in American usage it frequently means one specific historical episode rather than the general category. The economic sense needs the qualifier to be unmistakable, and in the record that qualifier is doing real work, because the general category is precisely the thing NBER says it does not maintain.

The interesting part of NBER's answer is what it says after the refusal. Having declined to define the term, NBER goes on to describe how the term is actually used, and reports two usages that do not agree with each other. "The term depression is often used to refer to a particularly severe period of economic weakness," it says. "Some economists use it to refer only to the portion of these periods when economic activity is declining. The more common use, however, also encompasses the time until economic activity has returned to normal levels."

Read that carefully, because it is the most consequential sentence on this subject. The two usages measure different things. Under the narrow one, a depression ends at the trough, which is the same moment a recession ends, so the word is only a comment on severity. Under the common one, it continues through the recovery until activity is back to normal, which is a different and much later date, and one nobody publishes. A speaker using the first sense and a speaker using the second can describe the same period and disagree by years without either being wrong.

The popular threshold is not in the dating body's material. The rule of thumb that circulates, that a depression means a fall in output of about 10 percent or a downturn lasting three years or more, appears nowhere in NBER's business-cycle dating pages. Its dating-procedure answers name no percentage and no duration for the term, and the landing page for the chronology does not use the word at all. That absence is the practical point rather than a curiosity, because an unsourced threshold gives a reader no way to check a claim and no trigger a plan could respond to.

The 1930s is the reference case, and NBER gives it dates. It identifies the 1930s as "the most recent episode in the United States that is generally regarded as a depression," and supplies the turning points: "a peak in economic activity occurred in August 1929, and that a trough occurred in March 1933," followed by "a second peak in May 1937 and a trough in June 1938." NBER describes both contractions as severe, and says the one starting in 1929 "is widely acknowledged to have been the worst in US history." Note what those dates are: they are recession dates, produced by the same procedure NBER applies to every other cycle. The word depression is doing descriptive work on top of them, not replacing them.

What follows for a household is a negative and it is genuinely useful. Because the term has no criteria, no dater and two spans, "are we in a depression?" is not a question with an answer to look up, and a plan cannot be conditioned on one. The events that would actually reach a household in such a period, a job loss, a fall in a portfolio, a credit market that stops lending, are each specific and each addressed elsewhere. The label adds severity to the description and nothing to the response.

There is no procedure, because there is nothing to run. NBER's committee reviews monthly and quarterly indicators, identifies peaks and troughs, and publishes contractions and expansions. No further classification happens, and no institution takes up the label afterward.

A hypothetical is not needed to show what the two usages do, because NBER's own dates do it. Take the 1929 episode and apply NBER's counting convention, under which the first month of a contraction is the month following the peak and the last month is the month of the trough.

• The narrow usage, covering only the period when activity was declining, gives the contraction from September 1929 through March 1933. That is 43 months.
• The common usage, which NBER says also encompasses the time until activity returned to normal levels, extends past March 1933 by an amount nobody publishes. What is on the record is that a second contraction followed, from June 1937 through June 1938 on the same convention, or 13 months, and that NBER describes both as severe.

So the 1930s depression is either a 43-month contraction or a span of most of a decade containing two severe contractions and an interrupted recovery between them. Both answers come from the same set of NBER dates. The difference is entirely which of the two usages the speaker meant, and the record contains no basis for preferring one.

For a reader trying to evaluate a claim in the present tense, the operational version of all this is short. Ask which usage is meant. Ask what threshold is being applied and who published it. If the answer to the second question is nobody, the claim is a description of severity, which may well be apt, and not a finding.

What the term conveys usefully

• It communicates severity in one word, and severity is real information when a contraction is far outside the ordinary range.
• NBER's own dates for the 1930s give the category a concrete anchor, so a comparison to that episode can be made against published turning points rather than against a feeling.
• The absence of a definition is itself informative. It tells a reader that no data release will ever settle the question.

Why it is a weak basis for any decision

• The body that dates US recessions declines to identify depressions, so there is no published turning point to verify a claim against.
• The two usages NBER records cover different spans, so two people can use the word correctly and mean periods years apart.
• The familiar 10 percent and three-year thresholds are not in NBER's dating material, which means a claim resting on one cannot be checked against the chronology.
• Because the label attaches only in hindsight and by consensus, it can never be a timely signal, and a plan built to react to one has no trigger to react to.