What does seasonally adjusted mean? Season and calendar terms in official data

Seasonal adjustment is a statistical step that estimates the rises and falls which recur in the same months every year, and removes them from a series so one month can be compared with the month before. Calendar adjustment removes working-day, holiday and leap-year effects. TÜİK publishes Türkiye's tourism statistics with neither adjustment.

A coastal resort sells out in August and closes two floors in February. Every month a release arrives with a figure described as seasonally and calendar adjusted, and the owner asks the obvious question: why does nobody adjust hotel numbers the same way? This article defines the handful of words behind that label, says which series a hotelier reads are adjusted and which are not, and shows how to quote a figure so the next reader can check it.

What do seasonal effect and calendar effect mean?#

A seasonal effect is a rise or fall that comes back in the same months every year; a calendar effect comes from how the calendar falls in a given month, such as which weekdays it contains or where a public holiday lands.

TÜİK defines seasonal effects as periodic rises or falls that recur from year to year in certain months or periods, in its Mevsim ve Takvim Etkilerinden Arındırma Metaveri Dokümanı (opens in a new tab) (22 January 2026). Calendar effects arise from the calendar make-up of a month, period or year: working-day, trading-day, leap-year and holiday effects. For Türkiye, TÜİK lists the calendar effects as weekends; the fixed public holidays of 1 January, 23 April, 1 May, 19 May, 15 July, 30 August and 28-29 October; the moving Ramadan and Kurban holidays together with their eves; and 29 February.

The practical difference is ownership. A seasonal effect belongs to the business: the sea is warm in August. A calendar effect belongs to the month: this August had five Saturdays, last August had four.

Adjustment means estimating these effects statistically and, where they are significant, removing them from the data. The phrase "where they are significant" does real work, and publishers apply it case by case. TÜİK's consumer confidence index is a working example: it is adjusted for seasonal effects only, because its calendar effects are not statistically significant.

What is the difference between adjusted and unadjusted data?#

Unadjusted data are the figures as collected; adjusted data are the same figures with the estimated seasonal and calendar effects taken out, published next to the raw ones rather than instead of them.

The Turkish labels follow the same logic. "Mevsimsellikten arındırılmış" removes the seasonal effect, "takvim etkisinden arındırılmış" removes the calendar effect, and "mevsim ve takvim etkisinden arındırılmış" removes both.

The unadjusted series is the one that matches your own records. If you want to check a published figure against your ledger, use the unadjusted number. The adjusted series is built for comparison instead: it answers a different question, whether activity rose, given that this month always looks different from the last one. Trend is a reading, not usually a separate published figure: it is the direction that remains once the recurring effects are removed, and it is only as good as the adjustment behind it.

In a report, name the version you are quoting. "Output rose 1.1% on the previous quarter" means one thing; "output rose 1.1% on the previous quarter, seasonally and calendar adjusted" means something a reader can check.

What does it look like in practice? A resort's year, with invented figures#

With invented figures, the raw month-to-month change measures the calendar and the season; an adjusted version of the same series would compare August with July on a like-for-like footing.

For example, a resort sells 6,000 room-nights in August and 1,200 in February. The unadjusted change from February to August is plus 400%, which tells you nothing about whether the business improved. A seasonal adjustment would estimate the recurring August uplift from several years of the same series and remove it, so August is compared with July after both months are put on the same footing.

What the adjustment cannot do: it works on a series, not on a property. A national adjusted figure says nothing about whether your own August beat your own plan. The hotel's manual version of the same idea is the same-month comparison: August this year against August last year, which holds the season roughly constant without any statistics.

The two are not identical. A same-month comparison still carries the calendar: if a public holiday moved, part of the difference you see is the calendar, not demand. Keep invented illustrations labelled as such in your own reports too. A worked example with round numbers is a teaching device, not evidence.

Which official series are adjusted, and which are not?#

It depends on the series, and each publisher states it in its metadata: TÜİK's tourism statistics are not adjusted at all, while several economic indices are published both ways.

TÜİK's metadata for the Departing Visitors Statistics and the Citizen Entry Statistics (updated 11 May 2026, attached to the Q3 2026 bulletin of 30 October 2026) states under item 18.6, "Düzeltme", that no seasonal adjustment is applied, as the tourism statistics metadata (opens in a new tab) shows. The Household Domestic Tourism Statistics metadata (updated 12 May 2026, attached to the Q2 2026 bulletin of 22 October 2026) says the same and adds that the data are not revised.

Among TÜİK's 2026 releases, the industrial production index and the services production index give annual changes from calendar-adjusted indices and monthly changes from seasonally and calendar adjusted indices. Quarterly GDP gives a seasonally and calendar adjusted change on the previous quarter and a calendar-adjusted change on a year earlier; labour force statistics headline seasonally adjusted rates. The consumer price index is a special case: its metadata describes seasonal adjustment under the special-coverage indicators, and the overall index plus the B, C, goods and services series are published adjusted for seasonal effects only, in a separate table.

The rule that follows: never assume a series is adjusted because a neighbouring one is. Open the metadata and look for the adjustment note before you quote the number. The same discipline applies to a Ministry series: if its metadata does not say the data are adjusted, treat them as unadjusted. If you cite these numbers in reports, the habits in how to read and cite official tourism statistics apply here too.

Why should a hotel compare the same month a year earlier?#

Because a hotel's month-to-month change mostly measures the season, and the same month a year earlier holds the season roughly constant without any statistical treatment.

The comparison is only as good as the checks behind it. Write down what you adjusted for, so the next reader can repeat it.

Even the same-month comparison fails when the religious holidays move. Ramadan and Kurban shift earlier each year, so a given month can hold a bayram one year and none the next; the planning questions around this are covered in bayram and public holidays: planning 2027 rates and restrictions early. Fixed public holidays also drift across weekdays, which changes the number of sellable weekend nights in the month. A one-off event, a renovation, a closed floor or a change in your own room stock will dominate a same-month comparison, and no adjustment will remove it.

When the comparison still looks wrong, the answer is usually in the checks, not in the data.

How do you read monthly and yearly change in an adjusted series?#

Read the monthly change from the seasonally adjusted or seasonally and calendar adjusted series, and the yearly change from the calendar-adjusted series, and say which one you are quoting.

This is TÜİK's own recommendation: calendar-adjusted data for comparisons with the same month or period a year earlier, seasonally adjusted or seasonally and calendar adjusted data for comparisons with the previous month or period. A worked case from TÜİK's 2026 releases: quarterly GDP gave a seasonally and calendar adjusted change of plus 1.1% on the previous quarter and a calendar-adjusted change of plus 2.3% on a year earlier.

Never mix the two. A monthly change taken from an unadjusted series set beside a yearly change from an adjusted one produces a comparison that means nothing. An adjusted figure is an estimate, so quote it with the series name and the release date; when a later release revises it, the earlier quotation should still be traceable.

Is the hotel's own seasonal index the same thing?#

No. A seasonal index is your own management tool, built from your own history; seasonal adjustment is a statistical treatment a publisher applies to a national series.

A hotel index is usually built by averaging each month's share of the year over several years, then using the result for budgets, staffing, rate fences and cash planning. It carries no significance test and rests on few observations, so one unusual year moves it a lot. It also breaks when the market changes: a new flight route, a renovation, a shift in source markets or a change in your own room stock all reset the pattern.

The published adjusted series tells you about the market; your index tells you about your property. Both are useful, and they answer different questions. Keep them apart in reports: if a chart shows one, label it, because a reader who assumes the wrong one will draw the wrong conclusion about the month. The honest limit is that an index tells you what normally happens, not what should happen next, and it is not a forecast. To see the pattern itself, reading Türkiye's tourism season month by month from the official data is the companion piece.

What to do next#

Before you quote any published figure, open its metadata and look for the adjustment note, then write into your monthly report template which version you used and what you compared it with. The terms in this article are one idea in different clothes: a number only means something next to the number it is compared with, and the comparison is the reader's job as much as the publisher's.