In 2016, the Hungarian government held an anti-refugee referendum. It failed. Here’s how Budapest voted

As Hungary’s parliament readies itself to pass a draconic set of laws to make life harder for NGOs and volunteers who try to help asylum-seekers, this seems like a good time to look back at the time when the government’s anti-immigrant propaganda suffered a rare defeat.

On October 2, 2016, the Hungarians went to the polls to vote in a referendum the government had invested an enormous amount of money in. Or at least some did. The government got the ‘No’ vote it wanted, but turnout didn’t reach the legal threshold of 50% — not by a long shot.

Four ways to visualize voter behaviour: from a basic turnout percentage to specifying those who voted ‘No’ (as the government wished); voted ‘Yes’; cast an invalid vote; or abstained. [1]

What was the referendum all about?

The aims and abuses of the referendum were described at great length at the time, and I’m not qualified enough to offer an analysis of my own. But here’s my stab at a summary.

The government asked Hungarians to say ‘No’ to the prospect of the European Union settling refugees in Hungary without the Hungarian parliament’s say-so.

Keep in mind: the number of refugees Hungary was being asked to accept was puny; nobody was going to send refugees into Hungary without the government’s cooperation; the proposal already seemed politically dead in the larger EU anyway; and the legal basis for submitting a European Commission plan to a plebiscite is extremely shaky.

Throughout the summer, the government blanketed the country with billboards, posters and leaflets in a massive propaganda campaign that cost as much as HUF 11.3 billion (€37 million) or even €48.6 million — more than both sides of the Brexit campaign spent together, but on the tax-payer’s dime.

Fear-mongering about “no-go zones” on the tax-payer’s dime

The billboards instilled fear of refugees with texts like “Did you know? Almost one million immigrants want to come to Europe from Libya alone” and “Did you know? Since the start of the immigration crisis, sexual harassment of women has increased in Europe”.

The leaflets propagated long-discredited myths about “hundreds of no-go zones” in European cities.

Public broadcaster M1 and other government-friendly media joined in as well, with the M1 evening news programme leading with the refugee issue and referendum 93% of the time, and devoting 95% of its airtime on the subject to echoing the government’s position.

The major left-liberal opposition parties called for a boycott, while the far-right Jobbik criticized the government’s wasteful ways but supported its call for a No vote.

The satiric “Two-Tailed Dog Party,” meanwhile, called for casting creatively invalidated ballots — and managed to crowdfund an impressive poster campaign of its own in the final weeks.

Prime Minister Viktor Orban responded to the referendum’s failure by staging a press conference without journalists and saying that he will proceed regardless.

Budapest view with Two-Tailed Dog Party posters

How did Budapest vote?

I had lived in Budapest for over eleven years by the time of the referendum, so I was especially interested in how the capital city’s residents would respond to the referendum campaign.

Would they heed the government’s call? Take the ‘doggy’ way and invalidate their ballot? Stay home as an active act of protest? Or stay home out of a general sense of disgust, disinterest and fatigue?

There’s no way, on the basis of the electoral data alone, to distinguish between the latter two things, but the results did reveal how Budapest differed from the rest of the country: [2]

In Budapest, overall turnout was 39.4%, some four points lower than nationally. Moreover, those who did turn out were less likely to follow the government’s lead and vote ‘No’. Instead, 11.8% of them cast invalid ballots — almost twice as much as the 6.3% who did so nationally (and multiple times the 0.9% of Budapest voters who cast an invalid vote in the 2014 parliamentary elections).

That means that 4.6% of the eligible voters of Budapest made the trek to the polling station only to deliberately spoil their ballot in protest. A success for the satirists — especially since images of creatively invalidated ballots made for a handy visual tool for reporters and social media users, and increased the visibility of the opposition message. The number of ‘Yes’ votes, meanwhile, was roughly the same in Budapest (2.1% of ballots cast, representing 0.7% of eligible voters) as nationally (1.7%/0.7%).

So how many people actually voted ‘No’? 86% of those who went out to vote cast a valid ballot saying ‘No’. That adds up to 34.0% of Budapest’s eligible voters, compared to 39.7% nationally.

It’s a relief but not wholly unexpected that Budapest turned out a little more resistant to the government’s campaign than the rest of the country. The city is certainly no bulwark of the opposition — thanks in part to ingeniously rewritten municipal election laws, for example, government party Fidesz/KDNP has 20 of the 33 city council’s seats. Still, in the 2014 parliamentary elections, Fidesz/KDNP got 45% of the vote and almost twice as much support as the left-liberal opposition alliance nationally, but it barely outpaced the left-liberals in Budapest, 38.5% to 36.8%. In this year’s election, the chasm between the city and the rest of the country only grew.

Buda vs Pest; downtown vs suburbs

There are stark differences within the city, however. The traditional divide runs along the Danube, pitting the more prosperous (and traditionally Fidesz-friendly) Buda against the generally poorer Pest. But there are also major cultural differences between the liberal downtown areas, the high-rise neighbourhoods further out, and the quiet outer suburbs.

Those differences showed up in this referendum too:

The government’s referendum message resonated more in Buda than in Pest, and to the extent that it met with support in Pest it was mostly in the outer suburbs.

The highest shares of ‘No’ voters were in outlying districts XVI and XXII. Just like district XVII, where the referendum fared relatively well too, those districts consist largely of small, single family homes and semi-detached houses with gardens, with mostly middle-income residents, and have an almost rural feel to them (or at least to an inner city dweller like me!). It doesn’t seem surprising that politically, too, they are more similar to small-town Hungary.

The share of ‘No’ voters was also relatively high, but not by as much, in two outlying districts with lower incomes: XVIII and XXIII

.The third-highest share of ‘No’ voters, however, was in District XII, or ‘Buda Hills’. It’s Budapest’s wealthiest district, and the two other districts that are almost as rich — I and II — both saw an above-average share of eligible voters go to the polls and vote ‘No’ as well.

That may seem strange. With far-right parties in one European country after another rivaling or overtaking social-democrats among working class electorates, we don’t tend to associate the success of xenophobic messages with wealthy voters. But there are two things that might help explain it.

Income, turnout and Fidesz support — or why the wealthy of Buda Hills voted ‘No’

First, traditionally speaking this is Fidesz territory. When it comes to Budapest districts, Fidesz support has been correlated with higher incomes, and the party received over 45% of the vote in the 2014 elections in districts I and XII.

Fidesz did almost as well in 2014 in the equally leafy district II and the very downtown area of Pest, district V, with its shops and (government) offices. So it’s not that surprising that the government’s referendum received a fair amount of loyal support in these districts as well: each of them had an above-average share of ‘No’ voters.

In this year’s elections, Fidesz suffered eye-catching declines in these same, wealthier neighbourhoods. Fidesz candidates got 3–4% less than in 2014 in four parliamentary constituencies that overlap with the higher-income districts, even as they made gains across much of the country outside Budapest. It seems intuitive enough to suppose that the government’s obsessively xenophobic and populist campaign didn’t play as well there. But the anti-refugee referendum showed that plenty of loyal Fidesz voters there were receptive to anti-immigrant rhetoric.

Second, higher incomes generally mean higher turnout in any election, and this year’s parliamentary election showed again that Budapest and Hungary are no exception. It was easier for the government to get the turnout it wanted in these wealthier parts of town than in districts where turnout is already relatively low in no matter what election, and Fidesz needed to not just persuade the locals of its message but also overcome a greater prevalence of political apathy.

That said, income levels and referendum turnout only aligned to some extent, and the same goes for Fidesz support and referendum turnout. It all depends on the type of neighbourhood! More on that later.

Across inner Pest, few people turned out. But the reasons why probably differ

The government’s message motivated fewest residents to go and vote ‘No’ in the inner city areas immediately surrounding the downtown district: districts VI, IX, and in particular VII, VIII and XIII.

Each of those latter three districts comes with its own stereotypes. District XIII is a mix of quite different neighbourhoods — from the hip inner parts of Újlipótváros to the towerblocks of Vizafogó and the scattered high-rise, residential and semi-industrial parts of Angyalföld — but together they make up easily the ‘reddest’ district of Budapest. Even when the Socialist Party was wiped out across the country in the 2010 elections, this district elected its only remaining two constituency MPs. District VII, while poorer than XIII, has its own share of hipsters and encompasses the so-called “party neighbourhood” with the city’s greatest concentration of pubs and clubs. In 2014 it was the single best city district for the green ‘Politics Can Be Different’ party.

Finally, District VIII (colloquially called ‘Nyócker’) is the ‘bad part of town’ Hungarians will warn you about, though it’s come a long way since the 1990s. While it’s started to gentrify and even Vogue is writing about it now, it’s still the poorest district of the city, and includes a high percentage of Roma residents. Along with District X, this district is also home to a relatively high number of immigrants, by Hungarian standards, but whether that hurt or helped the government’s anti-refugee message among their Hungarian neighbours is probably a dicey question, and their numbers are modest in any case.

The abstention map cannot reveal who stayed home in protest, and who stayed home because they are just generally disengaged from politics. There will have been plenty of the latter — even in national elections, almost 40% of Hungarians stayed home in 2014, and abstention was still 30% in 2018. But they’re obviously not evenly spread.

In the 2014 parliamentary elections, turnout was 10% higher in District XIII than in District VIII. In the referendum, these two districts had identical abstention rates — the highest of the city. But they likely reflect a very different mix of motives.

By one measure, the referendum “underperformed” especially in district V

Comparing the turnout in the referendum with turnout in the 2014 elections helps reveal which districts generally have low turnout anyway, and which districts were marked by unusually low turnout now.

In 2014, abstention was 38.2% nationally and anywhere between 26% and 43% in eleven of Pest’s city districts,[3] with some half a percent of eligible voters casting an invalid vote. Abstention + invalid votes in district XIII clocked in at 33.7% of eligible voters — well below the national average; in district VIII at 44.1%, higher than the national average. So it’s unsurprising that district VIII had the highest abstention rate in the referendum too, but in district XIII high abstention was out of character, and presumably to a greater extent a reflection of protest and disgust.

Back in 2014, districts XIII and XIX (Kispest) had close to the same percentage of non-voters, but that was very much not the case this time. The referendum message must have played better in Kispest.

Reviewing the proportion between the share of voters abstaining or casting invalid votes in the referendum and in 2014 suggests skepticism about the referendum, specifically (as opposed to electoral politics in general) was actually highest in downtown district V. There, abstention+invalidation was 2.4 times as high as in 2014. In middle-class district XIV (Zugló) and district XIII it was 2.1–2.2 times as high.

Vice versa, abstention was ‘only’ 1.6–1.8 times as high as in 2014 in districts VIII, IX, X and XX. Country-wide, abstention and invalidation was just 1.5 times as high in the referendum as in 2014.

Where did most people cast invalid ballots? A silver lining in Buda

The map of abstention rates obviously looks a lot like a negative image of the ‘No’ vote map… but not entirely! Abstention wasn’t the only thing keeping the ‘No’ vote down, after all. In perhaps the most encouraging part of the referendum, a significant number of voters did come out, but only to cast an invalid ballot in protest.

Since it was primarily the Two-Tailed Dog Party which called for invalidating your ballot, you might expect their number to have been highest in the inner Pest districts with a bit of a countercultural slant. But no! It was upscale districts I, II and XI that had the largest share of eligible voters who actively took the time to go to the polling station only to invalidate their ballot:

That’s right: upscale districts I, II and XI made up half of the districts where the government succeeded most in turning out its “No” vote (37–38%), but they were also the districts with the largest minority (6–7%) of active resisters.

In fact, turning out to cast an invalid ballot was pretty much an upper/middle class thing: the higher the income levels in a district, the more people cast an invalid ballot — though you were more likely to do so if you lived in inner Pest:

The map of invalidated ballots looks a little different, though, when measuring invalid votes purely as percentage of ballots cast:

From this angle, it’s actually district XIII which tops the list (15.8%), very closely followed by districts I, II, V, VI and VII, all with at least 15.0%. To simplify how this works: in district XIII, few people went to the polls for this referendum, and of those who did, quite a few cast an invalid ballot. In district XII, an even larger minority of eligible voters cast an invalid ballot, but they were swamped by their Fidesz-supporting neighbours.

On the other end of the scale, however, the data visualized in these two maps show the same things: district XX (Pesterzsébet) is at the bottom of the list, followed by district XV (Rákospalota) and district XXI (Csepel). Each of these three districts is marked by low incomes and high support for the far-right Jobbik party (but low support for Fidesz).

Finally, there was only one, tiny opposition group— Gábor Fodor’s Liberal Party — which called for a ‘Yes’ vote. It was criticized by all the other opposition parties for doing so, since every ‘Yes’ vote would merely help the government get its referendum to the validity threshold. It’s not the first time that tiny, nominally oppositional political parties appear to be advancing the government’s cause, and for a small party with just one MP the Liberals sure didn’t seem to be lacking in funds to pay for its own billboards.

Vanishingly few voters took the bait, however:

Digging deeper: Outlier precincts

Who doesn’t love some precincts with extreme results? This time, you’re going to have to download the map separately and view it in full size to be able to read it.

In Fidesz orange: the precincts with the highest share of voters who turned out, and voted “No” as the government wanted. They’re pretty spread out, mostly located in the outer suburbs. Five of the ten precincts where the government most succeeded in mobilizing the “No” vote were in the leafy and affluent districts II and XII, though not in their most luxurious parts. Appropriately enough, two of those were located close to one another near the Orbánhegy (Orbán hill) neighbourhood. The other half were in outer suburbs of Pest, where modest family houses line the street.

György Aladár utca, District XII

In green: the precincts with the highest share of voters who just stayed home. A lot of these are places where turnout is always low. No fewer than six of the ten were all bunched up together in the back of Nyócker, the part of the eighth district where gentrification hasn’t reached, with a large Roma population and high poverty rates. Two were on the same street, Diószeghy Sámuel utca.

Another two were located in the post-industrial zone of outer-Ferencváros, Kén Street being one of the city’s poorest. The Aszódi utca precinct is interesting because, although it had low turnout in the 2014 election too, Jobbik did extremely well. At least some of those Jobbik voters didn’t bother to turn out. One of the two precincts in the XIIIth in this list, meanwhile, is a high-rise Socialist bulwark.

Diószeghy Sámuel utca. For more of my photos from the eighth district, see Flickr.

In blue: the precincts with the highest share of voters who took the effort of turning out, only to invalidate their ballot. In four of these precincts, over 10% of eligible voters undertook this act of civic resistance.

Seven of the top 10 precincts are in Buda, on the Castle Hill tourists know well or Naphegy right behind, at walking distance behind Déli station or Széll Kálmán Square, or out near the woods on the edge of the city. But the three precincts in the Újlipótváros neighbourhood in District XIII stand out, and not just because they’re all right next to each other and one of them is located on a street named after Raoul Wallenberg. In these three, the combination with low overall turnout meant that over 30% of those who turned out, invalidated their ballot!

Raoul Wallenberg utca

More on correlations: Income, Fidesz support, and the ‘No’ vote

Generally, as noted, it seemed like the higher the average income and 2014 Fidesz support was, the more people turned out to vote “No” in this referendum. But to the extent there was a correlation, it played out almost on two different tracks for, on the one hand, districts in Buda and inner Pest, and on the other hand, districts in the largely high-rise districts around inner Pest and its quiet outer districts.

In both parts of the city, higher income levels and higher Fidesz support (as percent of all eligible voters) aligned to some extent with a higher “No” vote in the referendum; but with just two exceptions, the referendum played about 5% better in the latter part of the city:

Why does this chart only include 11 city districts? See note [3] at the bottom of this post…

On the national map of the referendum results, some similar patterns could be made out. Turnout was highest in the western provinces of Győr-Moson-Sopron and Vas, along the Austrian border, which are among the country’s most prosperous regions. Other prosperous regions — in particular Budapest and surroundings — had low turnout when compared to national results.

The correlation between income and turnout seemed much stronger in the 2014 parliamentary elections, suggesting that the ‘No’ vote underperformed in some (if not all) wealthier regions. Győr-Moson-Sopron and Vas are also among the provinces where Fidesz has traditionally done well and did well in 2014 too, though the relation doesn’t seem wholly consistent on that count either.

The ‘No’ vote and Jobbik support

While a relation between Fidesz vote and ‘No’ vote in the referendum seems predictable enough, it’s interesting that there was no such alignment between the 2014 vote for the far-right Jobbik and the ‘No’ vote.

Jobbik did best in some of the Budapest districts with the lowest incomes — X, XV, XX, XXI and XXIII, mostly in the city’s south — as well as the lower-middle income districts XVII, XVIII and XIX. The share of voters who turned out to vote ‘No’ was generally pretty mediocre in those districts. These neighbourhoods included few voters who invalidated their ballot in protest, but abstention was quite high. Consequently, the mobilization of the ‘No’ vote was no higher in core Jobbik neighbourhoods like districts X and XX than it was in districts V and XIV, which have very low Jobbik support.

The same was true nationally. Turnout in the Jobbik-inclined northeast was mediocre. Index.hu already pointed out partway through election day that turnout was low in Jobbik bulwarks Érpatak (where it ended up at 36%) and Ózd (31%) — though an above-average 47% ended up turning out in Ásotthalom, whose Jobbik mayor became Internet-famous with his anti-refugee publicity stunts. But again, it may well be that turnout is generally low in many of these Jobbik bulwarks, and that the low turnout in this referendum is mostly just a reflection of that general apathy.

Add up the 2014 vote for Fidesz and Jobbik, and get a near-perfect parallel with the referendum’s ‘No’ vote

So the Fidesz-KDNP vote only aligns partly with the ‘No’ vote in the referendum, and the Jobbik vote doesn’t align with it at all. But what if you add them up? In Pest, at least, you get a measure that aligns almost perfectly with the percentage of people who turned out to vote ‘No’:

This seems like a good closing point. The data here is limited, of course, to just a dozen Budapest city districts. But it might still be a testament to how the referendum succeeded in mobilizing the core right-wing share of the vote, with the exception of district V. Alternatively, it’s a testament to how this referendum campaign, for all the money that was spent on it, failed to reach beyond the core right-wing vote.

The small print

[1] Re: “Did not cast a vote”: As of October 4, 3:30 PM, the official results listed 3,599,160 eligible voters who “turned out as voters or returned a valid postal voting document”, but only 3,561,734 “ballots cast at polling stations and ballots included in valid postal voting documents”. The discrepancy of 37,426 (0.42% of all eligible voters) was mostly made up by 34,336 uncounted postal votes. Trying to understand, I gathered tentatively that these were cases where the “valid postal voting document” was deemed to not contain a ballot, e.g. because the ballot was judged to have not been put in the envelope in the prescribed way. It seemed like an odd thing, affecting no less than 37% of all postal voting documents, and I might have misinterpreted — in any case, by January 2017 the numbers on the Election Office website had changed and this category was down to just 3,279 votes.

[2] With the exception of the chart at the top of this post, all charts and tables reflect results as they were registered on the National Election Office website on October 4, when the vote was marked as “99.98% processed”. The final results as they were online by January 2017 were slightly different, but nationally none of the numbers changed by more than 1%. By now, the website has been controversially overhauled, and while most numbers remain as in 2017, the share of invalid votes in the referendum no longer appears to be listed anywhere. The results by district do not include mail-in ballots.

[3] I never found aggregated data anywhere about 2014 turnout by city district; the National Elections Office site only provides that data by individual precinct and electoral district (of which there are 18 in Budapest, which only partly correspond with city districts). It also never provided a data download for the 2014 results, the way it did for a while at least for the referendum’s results — but I mind-numbingly copied precinct results for 11 city districts (and parts of several more) in Pest into a spreadsheet, so for those I could calculate the turnout myself.