Greece's unemployment rate is 7.6%, so why are 655,023 people on the jobless register?

Greece published two very different unemployment figures for September 2026. On 30 October 2026, ΕΛΣΤΑΤ (Ελληνική Στατιστική Αρχή, Elliniki Statistiki Archi), the Hellenic Statistical Authority, put the seasonally adjusted unemployment rate at 7.6%, or 359,609 people.¹¹² On 20 October 2026, ΔΥΠΑ (Δημόσια Υπηρεσία Απασχόλησης, Dimosia Ypiresia Apascholisis), the public employment service, reported 655,023 people on its register for September 2026.⁴ Greek outlets call the difference «χαμένοι» ("missing") unemployed, or ask where they «χάνονται» ("go missing").⁸⁶ In fact the two figures count different things under different rules.¹⁴

What does ELSTAT count?

ELSTAT's figure comes from the Labour Force Survey, a household survey harmonised with EU law under Regulation (EU) 2019/1700.¹ Under ELSTAT's definition, the unemployed are people aged 15 to 74 who were not employed, were immediately available for work, and had either looked actively for work in the previous four weeks or found a job starting within three months.¹ΕΛ Anyone who worked for at least one hour for pay or profit in the survey's reference week counts as employed.¹ The definition sets no limit on how long someone has been out of work, although Alexis Mitropoulos, in a signed article on Alfavita on 20 October 2026, claims the surveys leave out people unemployed for 12 months or more.¹¹¹

What does ΔΥΠΑ's register count?

ΔΥΠΑ's figure is a count of everyone on its register, not a survey.⁴ The body of ΔΥΠΑ's September 2026 release calls them «εγγεγραμμένοι στο μητρώο», people on the register; it uses «άνεργοι» (unemployed) only for people receiving benefit, although its title speaks of «εγγεγραμμένη ανεργία» (registered unemployment).⁴ Οικονομικός Ταχυδρόμος (ot.gr), a business paper, reports that active job search and availability on the register are checked "not systematically, only sporadically", in our translation; Sofokleousin, a financial news site, makes the same point.⁷⁹

People also stay on the register for a long time. In September 2026, 359,951 people, or 55.0% of ΔΥΠΑ's register, had been registered for 12 months or more.⁴

Registration brings more than a cash benefit. ΔΥΠΑ paid unemployment benefit to 169,838 people in September 2026, about 26% of the register by our calculation, so about three in four people on the register received no benefit payment that month.⁴ Registered unemployed people can also obtain, at no cost through ΔΥΠΑ's online services, a certificate for medical and pharmaceutical care, once they have been registered for a minimum period that depends on their age.¹⁰ Some ΔΥΠΑ payments depend on long registration, such as the one-off €300 for people registered for more than five years, and the planned new unemployment benefit may change how the benefit works.

The two counts overlap only in part. Someone who works a few hours a week is employed in ELSTAT's survey but may stay on ΔΥΠΑ's register.¹⁷ ELSTAT's unemployed stop at age 74, while ΔΥΠΑ's register includes 61,398 people aged 65 and over.¹⁴

Where does the 367,126 gap come from?

On 5 October 2026, Newsit, a news site, set ΔΥΠΑ's August 2026 register of 717,926 against ELSTAT's August count of 350,800 unemployed, a gap of 367,126.⁶ The arithmetic is correct, and both inputs match the August releases.⁴⁵ ot.gr repeated the 367,126 gap on 26 October 2026, and Το Βήμα, a national daily, used it in a piece republished by CyprusNews.eu on 30 October 2026.⁷⁸

By then ΔΥΠΑ had published its September register, on 20 October 2026.⁴ On September 2026 figures, the gap is 295,414 people and the register is 82.1% larger than ELSTAT's count, by our calculation.¹⁴ ELSTAT has also revised August: its September release implies 352,100 unemployed in August 2026, by our calculation, rather than the 350,800 first published.¹⁵

On September 2026 figures, the gap between ΔΥΠΑ's register and ELSTAT's unemployed is 295,414, not 367,126, by our calculation.¹⁴

ot.gr also wrote that ΔΥΠΑ's August 2026 count was 95.55% higher than ELSTAT's.⁷ΕΛ By our calculation, 717,926 is 104.65% higher than 350,800, and Newsit's 104.6% is right; 95.55% is a different ratio, 350,800 divided by the 367,126 gap.⁴⁵⁶ Newsit itself warned that the register cannot be used to claim that real unemployment is 14% rather than 7.4%.⁶

Did unemployment rise or fall in September 2026?

Both, depending on the comparison. ELSTAT's rate was 7.6% in September 2026, against 7.4% in August 2026 and 8.8% in September 2025.¹ ELSTAT's count of unemployed rose by 7,509 (2.1%) on August 2026 and fell by 57,627 (13.8%) on September 2025.¹ ΔΥΠΑ's register moved the other way month on month, falling by 62,903 (8.8%) on August 2026, and it was down 65,016 (9.0%) on September 2025.⁴ In August 2026 the monthly directions were reversed: ot.gr reports that ELSTAT's unemployed fell by 25,708 while the register rose by 2,598.⁷

ELSTAT warns that its monthly estimates rest on a small sample.¹ Estimates of around 10,000 people for the whole country carry a coefficient of variation of at least 15%, and ELSTAT says more accurate estimates are better based on its quarterly data.¹ΕΛ ELSTAT publishes no margin of error for the monthly change, but given that warning, a one-month rise of 7,509 people (0.2 points on the rate) is weak evidence that unemployment actually rose.¹

How does Greece compare with the euro area?

In September 2026, ELSTAT put unemployment at 17.3% for people aged 15 to 24, down from 20.0% a year earlier.¹³ Unemployment among women was 10.5% in September 2026, against 5.2% among men.¹³ Because the survey follows the EU definition, Greece's 7.6% can be compared with the euro area, where Eurostat put the September 2026 rate at 6.5%.¹²