Unemployment Rate: Definition, Measurement, Okun’s Law and Why the Statistic Misleads

A person is unemployed if, during the reference period, they are:

1. Without work — not in paid employment or self-employment;
2. Currently available for work; and
3. Actively seeking work — having taken specific steps in the preceding four weeks.

All three conditions must hold. Fail any one and you are not in the labour force.

Labour Force Participation Rate (LFPR) = (Labour Force ÷ Working-Age Population) × 100.

The unemployment rate and the participation rate must always be read together. Either one alone can be systematically misleading.

Employment fell by two people. Four people stopped searching and left the labour force. The unemployment rate improved from 10% to 8.3%. This is not a hypothetical; it is a recurring feature of deep recessions, and it is why the employment-to-population ratio is often the more honest headline number.

• Discouraged workers — want a job, available, stopped searching because they believe none exists.
• Marginally attached workers — searched within the last year but not the last four weeks.
• Involuntary part-time workers — want full-time hours, counted as fully employed.
• The underemployed — a qualified engineer driving a taxi is employed. Full stop.
• Those on disability benefit — in some economies, a substantial reservoir of hidden non-employment.

The US publishes a spectrum from U-1 (long-term unemployed only) to U-6 (adding marginally attached and involuntary part-time). U-6 is typically close to double the headline U-3 rate. When a politician quotes “the unemployment rate,” ask which one.

Natural Rate of Unemployment (u_n) — the rate that prevails when cyclical unemployment is zero. It equals frictional plus structural unemployment. Often used interchangeably with NAIRU (the Non-Accelerating Inflation Rate of Unemployment), though the concepts differ subtly.

“Full employment” does not mean zero unemployment. Zero unemployment would mean nobody was ever between jobs — a sign of a sclerotic economy, not a healthy one.

u_n cannot be measured directly. It is inferred from the behaviour of inflation — which makes any statement of the form “unemployment is below the natural rate, so inflation will rise” close to circular. Estimates of the US natural rate have been revised repeatedly and substantially, always after the data proved them wrong. Treat any specific number with suspicion, and say so in an essay.

If a 1% rise in unemployment meant 1% fewer workers, output should fall roughly 1% (less, actually, since labour’s share is under 1). Why 2%? Because when demand falls, firms also cut hours, reduce shifts, hoard labour at lower effort, and workers exit the labour force. Each of these reduces output without raising the measured unemployment rate. Okun’s coefficient captures all the margins the unemployment rate misses — which is, again, the theme of this entire topic.

Blanchard, O. J., & Summers, L. H. (1986), “Hysteresis and the European Unemployment Problem,” NBER Macroeconomics Annual 1

The puzzle: European unemployment rose sharply in the late 1970s and, unlike in the US, simply stayed high for a decade after the shocks that caused it had passed. The natural-rate hypothesis says the economy should return to u_n. It did not.

The argument: The natural rate itself depends on the history of actual unemployment. Mechanisms:

• Skill atrophy. Long spells out of work erode human capital.
• Employer screening. Firms use unemployment duration as a signal of unobserved quality, so the long-term unemployed are not interviewed.
• Insider–outsider dynamics. Employed “insiders” bargain for their own wages, with no weight on unemployed “outsiders,” so wages do not fall enough to re-employ them.
• Capital scrapping. A recession destroys the capital stock that workers would have used.

Why it matters: Under hysteresis, a temporary demand shock produces a permanent rise in unemployment. The distinction between cyclical and structural unemployment dissolves: cyclical unemployment, if allowed to persist, becomes structural. This transforms the cost of a recession, and it is the strongest argument for aggressive and rapid stabilisation policy.

The modern echo: The unusually strong policy response to COVID-19 — furlough in Europe, transfers in the US — was justified explicitly in hysteresis terms. Preserving the employment match was the objective, not merely supporting income.

Movement along the curve = the business cycle. Demand rises, vacancies rise, unemployment falls.

Outward shift of the curve = deteriorating matching efficiency. High vacancies coexisting with high unemployment means the jobs and the workers exist, but cannot find each other. This is the signature of structural unemployment — skills mismatch, geographic immobility, or search frictions. An outward shift is bad news that a falling unemployment rate can conceal.

• Compute the unemployment rate and the participation rate from a population table. Watch the denominators — they are different.
• Classify a given worker: a laid-off factory worker whose skills are obsolete is structural; one laid off in a recession who will be rehired is cyclical; a graduate searching for a first job is frictional.
• Know that full employment ≠ zero unemployment, and that at full employment frictional and structural unemployment persist.
• Connect to the Phillips curve: unemployment below u_n puts upward pressure on inflation.

“Evaluate the usefulness of the unemployment rate as a measure of labour market health.”

1. Define the ILO criteria precisely — all three conditions.
2. Show, with the worked arithmetic above, that the rate can fall while employment falls.
3. Introduce the participation rate and the employment-to-population ratio as complements.
4. Discuss the excluded groups: discouraged, marginally attached, underemployed, involuntary part-time.
5. Reference U-6 as an attempt to capture them.
6. Introduce hysteresis (Blanchard & Summers) to argue that the cyclical/structural distinction is not fixed.
7. Use the Beveridge curve to distinguish demand deficiency from matching failure.
8. Conclude: no single statistic suffices; the rate must be read alongside participation, hours, and vacancies.

Exercise 1 — Can Policy Change the Natural Rate?

The natural rate hypothesis holds that monetary policy cannot affect unemployment in the long run — it can only choose the inflation rate. Hysteresis holds that a prolonged demand shortfall permanently raises the natural rate through skill atrophy and employer screening.

If hysteresis is real, then a central bank that tolerates a deep recession to fight inflation has permanently raised the economy’s floor for unemployment. Does this imply that central banks should run economies “hot,” accepting above-target inflation to lower u_n? Construct both sides of the argument, and identify what evidence would settle it. Consider carefully whether hysteresis operates symmetrically — that is, whether a boom lowers the natural rate as much as a slump raises it.

📄 Read: Blanchard, O. J., & Summers, L. H. (1986). “Hysteresis and the European Unemployment Problem.” NBER Macroeconomics Annual, 1, 15–78. Pay attention to the insider–outsider mechanism, and ask whether it survives in economies with weak unions.

Exercise 2 — Should Unemployment Be Measured at All?

The unemployment rate depends on a behavioural search test that is self-reported, unverifiable, and sensitive to benefit-conditionality rules. Two countries with identical labour markets but different benefit regimes — one requiring proof of job search, one not — will report different unemployment rates.

Would the employment-to-population ratio, which requires no behavioural test at all, be a superior headline statistic? Consider what is lost: the ratio cannot distinguish a retiree from a discouraged worker, a student from a disabled person. Is a statistic that requires a subjective test but tracks the concept we care about preferable to one that is objective but conceptually blunt?

📄 Read: Krueger, A. B. (2017). “Where Have All the Workers Gone? An Inquiry into the Decline of the U.S. Labor Force Participation Rate.” Brookings Papers on Economic Activity, Fall 2017, 1–87. Note his findings on disability, pain medication, and the men who are neither employed nor counted as unemployed.