Eight years of voter migration in İstanbul and Ankara: the registered-voter map redrawn neighborhood by neighborhood from 2015 to 2023 — and the surprisingly small effect this migration had on voting preferences.
When we assess an election we usually focus on a single question: How did the parties' votes change? Yet before vote shares change, something quieter often shifts first — the electorate itself, that is, where the people who will go to the ballot box actually live. In this piece we set parties aside and look only at that base. How was the voter map of the two largest cities, İstanbul and Ankara, redrawn over the eight years from the 7 June 2015 parliamentary election to the 14 May 2023 parliamentary election? The short answer: as the city grew, the center emptied and the periphery filled; yet this profound relocation shaped voting preferences less than one might think.
Over eight years İstanbul's registered electorate rose from 10.24 million to 11.37 million — a net gain of 1.13 million, or 11 percent. In Ankara the pace was even higher: the electorate climbed from 3.71 million to 4.29 million, up 15.7 percent. But this growth was not spread evenly. Beneath the totals lie two opposite geographies: a filling periphery and an emptying center.
The fastest-growing districts are almost all on the city's outer ring, clustered in new-housing and estate zones. Sancaktepe's registered electorate grew 54 percent over eight years, from 222,000 to 343,000. Arnavutköy rose 52 percent, Başakşehir and Beylikdüzü 44 percent, and Çekmeköy 41 percent. In absolute numbers the champion was, by a wide margin, Esenyurt: it alone added 185,000 new voters, taking its count from 458,000 to 643,000 — an increase equal to the entire electorate of a mid-sized district.
In the historic, central districts the needle turned the other way. The starkest example is Fatih: its registered electorate fell 12.7 percent, from 303,000 to 265,000, losing 38,600 voters in eight years. It was followed by Güngören, Beşiktaş, Zeytinburnu, Beyoğlu and Şişli. Established centers such as Kadıköy, Bakırköy and Üsküdar effectively stood still within a growing city.
Percentage change. Green bars are growing peripheral districts; red bars are shrinking central districts.
In Ankara the periphery grew even more sharply, but no central erosion took place. Etimesgut grew 30 percent over eight years, adding 108,000 voters on its own; Pursaklar, Gölbaşı and Kahramankazan grew around 32 percent, and Akyurt 51 percent. By contrast the large central districts did not shrink: Çankaya rose 5.8 percent and Keçiören 9.5 percent. Voter loss appeared only in rural districts such as Nallıhan, Haymana and Ayaş. So Ankara's story reads not as "the center emptied" but as "the gap between periphery and countryside widened" — the central erosion is a phenomenon specific to İstanbul.
Percentage change. Peripheral districts grew while rural districts shrank; the center (Çankaya, Keçiören) rose moderately.
This movement in registered voters lines up with official population data — but with an instructive detail. Population growth and voter growth do not advance at the same pace in every district. In Başakşehir the two are almost identical: both population and electorate grew 44 percent. In Sancaktepe, by contrast, the electorate (54 percent) grew markedly faster than the population (39 percent). The reason is demographic: in a district like Sancaktepe, which filled with young families after 2008, the children of those years turned eighteen within eight years and became voters. Indeed the district's voter-to-population ratio rose from 63 to 70 percent. In short, peripheral districts did not merely attract people; as they matured, their voter rolls grew even faster than their population. On the emptying side the picture is symmetrical in reverse: Fatih lost both 63,000 residents and 38,000 voters.
| District | Registered voters 2015→2023 | Population 2015→2023 |
|---|---|---|
| Esenyurt | +40% · 458k→643k | +32% · 743k→978k |
| Sancaktepe | +54% · 223k→343k | +39% · 355k→493k |
| Başakşehir | +44% · 221k→318k | +44% · 353k→510k |
| Etimesgut (Ankara) | +30% · 355k→463k | +17% · 528k→617k |
| Çankaya (Ankara) | +6% · 678k→717k | +2% · 923k→938k |
| Beyoğlu | −4% · 175k→168k | −10% · 242k→219k |
| Fatih | −13% · 303k→265k | −15% · 419k→356k |
Population figures: TÜİK Address-Based Population Registration System. Ankara's central district Çankaya held steady in both population and electorate; İstanbul's Fatih lost both. Central erosion thus remains specific to İstanbul.
Everything so far has concerned where the voters are. But did this profound relocation change the result at the ballot box to the same degree? Intuition says yes: in an Esenyurt whose electorate was nearly half renewed, we would expect the vote distribution to be upended too. The data gives a more measured answer.
From 2015 to 2023 the AK Party's vote fell in almost every neighborhood of İstanbul; moreover this decline stayed within a narrow band of four to seven points — in the booming periphery and the stagnant center alike. In other words the AK Party's loss was largely a uniform, nationwide wave, not the work of new voters arriving in particular neighborhoods. At the neighborhood level, the relationship between voter growth and the change in the AK Party vote is only weak: fast-growing neighborhoods swung slightly further from the party, but the difference is small.
Where the loss went is a story in itself. In the growing peripheral districts all four major parties of 2015 lost ground: the AK Party fell about 5 points, the MHP 3.7, and the HDP-lineage Green Left Party 4.3; even the CHP dropped 1 point. The gainers, by contrast, were parties that were not on the ballot at all in 2015 — the İYİ Party took 6.6 points, New Welfare (Yeniden Refah) 3.6, the TİP 3.6 and the Zafer Party 2.7. So the vote shed by the AK Party did not flow to the CHP; it dispersed largely to New Welfare and Zafer on the right and the İYİ Party on the center-right. This picture is less a rupture peculiar to the peripheral districts than the local reflection of the general fragmentation Turkish politics underwent over these eight years.
The real lesson is not to pin the outcome on a single cause. When party votes change in a district, we can read it neither purely as "voters changed their minds" nor purely as "the population changed, so the vote changed." In the İstanbul case the decisive factor, beyond both explanations, was a political trend running in common across the country. The voter geography was profoundly redrawn; yet voting preferences moved less with this new map than with the general course of national politics. These comparisons rest on the collective average of neighborhoods, not of individual voters; they therefore show a general tendency, not an individual profile.
Over eight years the voter map of İstanbul and Ankara was profoundly redrawn: historic central districts lost voters while the new-housing zones on the periphery each gained as many voters as a mid-sized district. This is a real, verified population movement. Yet over those same eight years, voting preferences changed less with this dramatic map than with the shared national trend. The city was rearranged; but the dimension of the ballot box that really changed was its location, not its color.
Data and method. Source: Supreme Election Council (acikveri.ysk.gov.tr), neighborhood-level registered-voter counts; population from the TÜİK Address-Based Population Registration System. The comparison was made between two elections of the same type — the 7 June 2015 and 14 May 2023 parliamentary general elections — so that the distortion arising from the different voter universes of local and general elections is excluded. More than 99 percent of neighborhoods in İstanbul and Ankara match with the same identity across the two elections; the measured change comes from real voter movement, not administrative boundary shifts. District totals match the official YSK bulletins to within a few thousandths (owing to overseas and mobile ballot boxes).
Limits. Inferences about voting tendency rest on neighborhood averages and cannot pin down individual voting behavior (the fundamental limit of ecological inference). "General fragmentation" is an interpretation; the numerical finding is the rise in the combined share of the new parties. This page is part of the regular "Analysis of the Week" series.
© Sandık Atlası · Impartial, sources visible.