How many people work in Türkiye’s accommodation sector? Reading TÜİK and SGK employment data

Two official series answer this. TÜİK's paid-employee statistics for August 2026 count 508,260 paid employees in accommodation (NACE 55), an activity category, and 1,512,615 across accommodation and food service together. SGK's August 2026 insured statistics count 545,260 compulsory insured persons in accommodation. The gap is scope, not error.

A headline says employment in tourism rose again. The number behind it may be mostly about restaurants, not rooms. Türkiye tourism employment statistics come from two publishers with two different methods, and they answer different questions. After reading this, you will be able to take any employment figure in the news, identify which series it came from, decide whether it describes your business, and quote it without misleading a board or an investor.

Which official series count employment in Türkiye's accommodation sector?#

Three sources are quoted most often, and each answers a different question. Knowing which one a number came from is the whole skill.

TÜİK's Paid Employee Statistics are built from administrative records: SGK monthly premium and service declarations, GİB withholding tax returns and, for finance, BDDK records. They count paid employees in workplaces with at least one employee. The same release excludes apprentices, interns, working owners, partners and unpaid family workers, so it measures a wage relationship, not everyone present in a building.

TÜİK's labour force survey works differently: it is a household survey that counts people, including the self-employed and unpaid family workers, and it is the source for the employment and unemployment rates you hear in the news. SGK's monthly insured statistics come from the same declaration system as TÜİK's inputs, but are published as workplace and insured-person counts by NACE division, which lets you see workplaces as well as people.

The three do not reconcile with each other by design. No publisher issues a table that makes them agree, so a writer who mixes them in one sentence creates a number that exists nowhere. A practical rule: choose the series that matches your sentence. "People paid by accommodation businesses" is TÜİK paid employees. "Insured persons and workplaces declared" is SGK. "People who say they work" is the survey.

Public data. No endorsement implied.

How many people work in accommodation in Türkiye?#

In August 2026, TÜİK counted 508,260 paid employees in accommodation (NACE 55), and SGK counted 545,260 compulsory insured persons in the same activity. Both are single-month figures, unadjusted.

The TÜİK paid-employee release for August 2026 (No. 57956, 15.10.2026) (opens in a new tab) puts accommodation at 508,260 paid employees, against 507,179 in August 2025, a rise of 0.2%. The same release puts accommodation and food service together (NACE section I) at 1,512,615, against 1,464,982 a year earlier, up 3.3%. Food and beverage service activities (NACE 56) account for 1,004,355 of that total, against 957,803 in August 2025, up 4.9%.

The SGK insured statistics for August 2026 (opens in a new tab), Table 13, show accommodation with 26,971 workplaces and 545,260 compulsory insured persons, of whom 324,056 were men and 221,204 women. For scale, the same table counts 161,441 workplaces and 948,987 compulsory insured persons in food and beverage service, and 2,350,093 workplaces and 17,269,532 compulsory insured persons across all activities.

Read the accommodation line as flat and the food service line as the one that moved. A 0.2% year-on-year change in a single unadjusted month is not a trend by itself.

Why do the TÜİK and SGK figures differ?#

Because they count different units from partly different records, on different publication timetables. A gap of a few percent between them is expected, not a mistake in either.

The unit differs: TÜİK counts paid employees, SGK counts compulsory insured persons. A person can hold more than one declared job, and a count of declarations is not a count of people. The exclusions differ too: TÜİK's series leaves out apprentices, interns, working owners, partners and unpaid family workers, while SGK's compulsory insured count is built from premium declarations on a different basis.

The inputs and timing differ as well. TÜİK combines SGK declarations with GİB withholding tax returns and, for finance, BDDK records; SGK's own bulletin reports its declaration base. SGK's metadata states that its insured statistics refer to the month two months before the current month, so a bulletin read in October describes August.

Neither publisher reconciles the two series, and neither is designed to be subtracted from the other. When two official numbers disagree, the honest sentence names both, gives each its source and month, and says they measure different things.

Why do headlines merge hotels with restaurants?#

Because the standard activity classification puts accommodation and food service in one section, and the combined figure is the one that gets quoted.

NACE is the activity classification used by both TÜİK and SGK. Section I covers accommodation and food service; division 55 is accommodation, division 56 is food and beverage service. A business is classified by its main activity, not by who its customers are: a restaurant that mostly serves local residents sits in division 56 exactly like a hotel restaurant.

Division 55 is not a hotel count either. It covers accommodation as an activity, so a figure from it is not a count of hotels, and Türkiye's certified facilities are not all hotels. In August 2026 food and beverage service was roughly twice the size of accommodation in both series, so a combined headline is driven by restaurants, cafes, bars and catering.

The combined figure is legitimate for a question about the whole hospitality trade and misleading for a question about rooms, beds and front desks. When you see "tourism employment" in a headline, look for the activity code before you repeat the number. If the code is section I, the sentence is about accommodation and food service together.

Why does the number peak in summer, and how should a hotel use it?#

The series is unadjusted, so it carries the season inside it: August includes the staff a summer coast needs and January does not. That shape is useful for planning, provided you compare like with like.

Unadjusted means no seasonal correction has been applied. Month-to-month moves therefore mix real hiring changes with the calendar, and only a same-month year-on-year comparison isolates the change. For a hotel, the useful reading is the shape of the curve across the year, not the size of any single month: it tells you when the labor market around you is tightest.

Recruitment timing follows the curve. If the market peak and your own peak coincide, you are competing for the same people at the worst moment, so hiring and onboarding move earlier. The drop after the peak is a retention question as much as a hiring one: which roles you keep through the shoulder months, and which you rebuild each season.

Two operational tools do more for a seasonal property than any statistic: seasonal staff onboarding: a first-week training checklist by department, and hotel shift planning: covering the desk, floors and nights fairly. Use the national series as context, never as a target. Your own applicant flow, offer acceptance and first-month attrition are the numbers that decide your season.

What can these series not tell you?#

They cannot give you a hotel-only headcount, hours worked, contract types, occupations or the size of undeclared work. Each of those needs a different source.

There is no official hotel-only employment count in these releases. Division 55 is the closest available unit, and it is an activity category, not a property type. Undeclared work is invisible by construction: a series built from premium and withholding declarations cannot count people who are not declared.

Hours and contract type are not in the paid-employee count. A full-time permanent contract and a short seasonal one both appear as one paid employee. Occupation, nationality and turnover are absent: the series tells you how many were paid in a month, not who they were or how long they stayed.

SGK's bulletin also prints average daily earnings subject to premium, TRY 2,252.69 for accommodation in August 2026. That is a premium base with ceilings and exclusions, not a wage, and it must never be presented as a salary. None of this makes the series weak. It makes it specific: use it for level and direction in an activity, and go elsewhere for everything about the job itself.

How do you cite these numbers correctly?#

Name the series, the publisher, the reference month and the activity code, and say the figure is unadjusted. That sentence survives scrutiny; a bare number does not.

Write the reference month, not only the release date: August 2026 data, released 15 October 2026. Name the activity: NACE 55 for accommodation, NACE section I for accommodation and food service together, and never write "hotel employees" over either. Give the bulletin identity where you can: TÜİK No. 57956 for the paid-employee release, SGK Table 13 for the insured statistics. For the wider set of tourism series and their traps, the reading guide on official Türkiye tourism statistics covers the rest.

What to do next#

Open the two bulletins yourself before the number goes into a board deck: TÜİK's paid-employee release and SGK's insured statistics, both for the same month. Write down which series you are quoting, which activity code it uses and whether it is unadjusted, then keep that sentence intact wherever the figure travels. The parts of this habit fit together: separate the sources, separate the activities, respect the season, and state the limits before someone else does.