Germany's summer 2026 labour market: 3.06 million unemployed, real wages up 1.5%

The Bundesagentur für Arbeit’s release for August 2026 carries a sentence and a number that pull in different directions. Under the heading Arbeitskräftenachfrage it states that “die Arbeitskräftenachfrage bleibt weiter schwach” — labour demand remains weak — and in the same paragraph reports 656,000 vacancies registered with the agency in August, 25,000 more than a year earlier. The day before, the Statistisches Bundesamt had reported that real earnings in the second quarter of 2026 were 1.5% higher than in the same quarter of 2025. Neither statement is wrong. Neither explains the other, and neither publisher attempts to. ^{1 5}

What the August release established

Registered unemployment stood at 3,061,000 in August 2026, 54,000 higher than in July; seasonally adjusted, the rise was 4,000, so the adjustment accounts for 50,000 of the 54,000. Against August 2025 the level was 36,000 higher — on an implied year-earlier level of 3,025,000, a rise of 1.2%. The rate rose by 0.1 percentage points to 6.5%. ¹

Destatis carries the same 6.5% in its monthly indicator table for the BA-definition rate — a convenient cross-check, and a place to pull the series month by month without parsing the agency’s PDFs. That table also flags a break: from the January 2009 reference month, the published rates relate to all civilian labour force rather than to dependent civilian labour force alone. A chart running straight through that point is not a like-for-like series. ⁷

On the demand side, 656,000 vacancies were registered with the BA in August — vacancies reported to the agency, which is not the same thing as total labour demand. Against 3,061,000 registered unemployed in the same month and from the same source, that is 4.7 registered unemployed per registered vacancy. The description of demand as weak is the agency’s own, and the figures to hand do not show what the vacancy stock was in earlier years, so this piece cannot place the August level on a longer series. What the release does show is a stock 25,000 higher than a year ago, set down in the same paragraph as that description. ^{1 3}

The employment figure in the same release is not an August number. Socially insured employment was 34.81 million and 73,000 lower than a year earlier — about 0.2% — but the reference month is June 2026, and the seasonally adjusted change reported alongside it, plus 1,000, is May to June. Quoting it next to the August unemployment count puts a June figure beside an August one. ¹

The other unemployment count is rising seven times faster

Germany publishes a second unemployment figure on a different concept. On 28 August, Destatis reported 1.84 million people unemployed in July 2026 on the ILO concept measured through the Arbeitskräfteerhebung, 149,000 or 8.8% more than in July 2025. The unadjusted Erwerbslosenquote was 4.2%, against 3.9% in July 2025. Adjusted for seasonal and irregular effects, the count was 1.74 million, 11,000 or 0.6% above June 2026. The 4.2% is the unadjusted rate; the adjusted rate is a different number and must not be quoted as though it were the same one. Total employment on the Inländerkonzept was 45.48 million in July. ⁶

Germany's two unemployment measures, most recent published month of each

Measure	Reference month	Value
Registered unemployed (BA)	August 2026	3,061,000
Change on previous month, unadjusted	August 2026	+54,000
Change on previous month, seasonally adjusted	August 2026	+4,000
Change on same month a year earlier	August 2026	+36,000
Registered unemployment rate (BA)	August 2026	6.5%
Unemployed, ILO concept, unadjusted	July 2026	1,840,000
Unemployed, ILO concept, seasonally adjusted	July 2026	1,740,000
Change on a year earlier, unadjusted	July 2026	+149,000 (+8.8%)
Unemployment rate, ILO concept, unadjusted	July 2026	4.2%
Unemployment rate, ILO concept, unadjusted	July 2025	3.9%

Real earnings rose anyway

In the second quarter of 2026, nominal earnings were 4.1% higher than a year earlier, consumer prices 2.5% higher, and real earnings 1.5% higher. ⁵

The distribution is the more useful part of the release. Among full-time employees, the lowest-earning fifth recorded nominal growth of 6.7% and the highest-earning fifth 3.3%, against 3.9% for full-time employees as a whole. The top quintile therefore grew more slowly than full-time employees generally, not merely more slowly than the 4.1% headline; the spread is 3.4 points, and the bottom fifth grew roughly twice as fast as the top. The release reports the pattern and offers no explanation for it. It is evidence that the bottom of the full-time distribution moved fastest, and evidence of nothing about why: an average can also shift because the composition of employment changes rather than because anyone’s pay changed, and these figures do not separate the two. ⁵

Earnings and prices, second quarter 2026 against second quarter 2025

Series	Change
Nominal earnings, whole economy	+4.1%
Consumer prices	+2.5%
Real earnings, whole economy	+1.5%
Nominal, full-time employees overall	+3.9%
Nominal, lowest quintile of full-time employees	+6.7%
Nominal, highest quintile of full-time employees	+3.3%

What those percentages are worth against actual levels

For a monthly frame, Destatis publishes a separate table: average gross monthly earnings of full-time employees in April 2025, excluding special payments, of €4,784, on paid weekly working time of 39.1 hours and gross hourly earnings of €28.19. Twelve times that is €57,408, some €7,033 below the annual figure — the two are not the same measure, and the annual one includes payments the April snapshot excludes. Both are averages, not medians. The sector spread is wide: financial and insurance services stood at €91,678 in the 2025 annual table against the whole-economy €64,441, so a benchmark taken from the national average and applied to one sector is wrong in a predictable direction. ^{9 8}

What to watch next

Three things are worth putting in a monthly briefing rather than a headline.

First, the divergence between the two unemployment counts. If the ILO measure keeps rising at several times the rate of the registered one, the choice of series stops being a footnote and starts changing the conclusion. Both publishers release monthly, so each month brings a fresh pair of readings to watch it against.

Second, the lag inside the BA release. Socially insured employment arrives about two months behind the headline unemployment figure, so an August release describes June employment. Workforce plans built in September on “the latest employment number” are working with early-summer data.

Third, the detail beneath the headline. The BA’s monthly report carries the tables, time series and regional breakdowns behind the national figure, and its vacancy statistics hub carries the regional and occupational cuts of registered vacancies. For staffing specifically, the agency’s temporary agency work statistics publish semi-annually and include a pay breakdown for agency workers in socially insured employment — the place where the sector’s own earnings can be read from official data rather than inferred from a national average. ^{2 3 4}

Figures in this piece reflect the sources as at the last editorial review on 2 September 2026.