Judicial review of the electoral process is always a fraught thing. Depending on the outcome of a case, it will be welcomed by some as blessed relief from unfair legislation, denounced by others as activist overreach. And then there are cases that can just be surprising and, frankly, baffling.
Such was the decision Pl. ÚS 44/17 of the Czech Constitutional Court on February 2, 2021, which struck down key elements of that country’s electoral system. Departing from the tendency of most prior cases to defer to legislators’ choices, eleven of the fifteen judges ruled that the combined use of the D’Hondt highest average method, fourteen multi-seat districts of varying size, and a steep additive threshold for electoral alliances made it harder for some parties to win seats in line with their share of the vote and thereby violated provisions of the Czech constitution.
I’m not trying here to do justice to the history of the case, which started in December 2017 as a complaint by senators from STAN, a party of mayors and independents, and represents the climax of a quarrel that started even longer ago between the judge rapporteur, Jan Filip, and one of the architects of the electoral system, Zdeněk Koudelka (see their 2000 exchange here). I also can’t go into all the issues addressed in the 63-page majority opinion or the trenchant dissent. For now, I will just identify a few things that I’m struggling with as I process the court’s reasoning — is this primarily about the electoral system’s failure to be proportional, or about its violation of equality? Both? — and its implications for what the Czech parliament has to do before the elections scheduled for October.
Article 18 of the Czech constitution stipulates that the 200 seats in the Chamber of Deputies shall be elected by secret ballot on the basis of equal voting rights and according to the “principle of proportional representation”. Since none of the commonly-used electoral systems can translate vote shares into a precisely identical share of seats, some deviation from proportionality is inevitable. The question for judges is how much is tolerable.
Czech judges had hitherto accepted the disproportionality generated by the electoral system, with one major exception twenty years ago: The constitutional court in January 2001 struck down several amendments pushed by the two major parties (Social Democrats and Civic Democrats) that would, if implemented, have resulted in such extreme disproportionality as to no longer comply with Article 18. The electoral system being judged in this year’s case was put in place in 2002, designed to take into account the court’s objections. (For details, see ch. 4 of Embodying Democracy and Jakub Charvát’s Politika volebních reforem v ČR po roce 1989.)
As noted by JD Mussel but not by the court, there is a standard metric for measuring disproportionality, Gallagher’s least squares index. (Yes, it has its critics, but let’s leave that aside.) The Czech system in place since 2002 has produced disproportionality scores of variable but not extreme range.
While higher than in some countries using PR (such as Austria, Denmark, or Sweden), Czech scores are comparable to plenty of others, such as in Chile, Costa Rica, Croatia, Poland and Spain. What is most striking is that it tracks fairly closely to Slovakia’s scores, even though Slovakia avoids most of the things the court found fault with in the Czech system: Slovakia has a single national district for all 150 seats, uses Hagenbach-Bischoff instead of D’Hondt, and has a less severe additive threshold for alliances (7% for a two- or three-party alliance, and 10% for four or more; the Czech system since 2002 sets the threshold at 10% for two parties, 15% for three and 20% for four or more).
In other words, Slovakia operates a system similar in many respects to that used in the Czech Republic in the 1990s, which the Czech constitutional court signalled would be acceptable, yet Slovak elections result in much the same degree of disproportionality.
The Czech electoral amendments struck down by the court in 2001, by contrast, would conceivably have resulted in disproportionality on an entirely different scale. By carving the country into 35 districts with an average magnitude of 5.7 while applying a modified initial D’Hondt divisor (the square root of 2), it might have resulted in disproportionality scores over 20, and thus more akin to France under its two-round majority-plurality system. From the constitutional point of view, it was a slam dunk. The same cannot be said of the system being struck down now.
STAN was just the latest in a line of parties that had tried since the mid-1990s to get the court to strike down electoral rules on the grounds that they violated the various equality guarantees of the constitution and its charter of rights. Before the decision was announced, I did not expect the challenge to succeed, owing to its similarity to a previous one brought after the 2013 elections by the Pirate Party (STAN’s prospective partner in a 2021 electoral alliance). That case had been framed around largely the same constitutional issues — alleged violations of equality in general, of equal voting rights under proportional representation, of free competition between political parties, and of the protection of minorities.
The court had rejected the Pirates’ complaint on the grounds that, as per a 2007 precedent, the equality at issue was relative: while one person was entitled to one vote, there was no right to an equal impact on the outcome of the election. Furthermore, the court argued then, the electoral system had to not only represent, but also integrate, ensuring that the resulting legislature was functional and able to produce a working majority and stable government. The opinion in the Pirates case stated that “it truly is not the ambition of the Constitutional Court to change the current electoral system”, since it “in no way departs from the model approved in various democratic states”.
Now, it is perfectly normal for courts to change their minds over time, but when they do, it is also normal for them to explain why. Anticipating complaints about out-of-control unelected judges , the opinion in the STAN case does go to great length (especially in paragraph 209) to justify the court’s role in the electoral process, and relies in particular on a Harvard Law Review article by David Pozen on the need to act against “bad faith” lawmaking. In paragraph 91, the point at which the opinion finally draws a bead on its target, it announces that the changes made to the electoral system in 2002, no less than the ones enacted in 2000 and struck down by the court in 2001, were a “significant step backwards (and an exemplary example of ‘bad faith’[…]) as opposed to efforts to achieve the best reflection of the proportion of political parties’ strength across the state […]”.
The problem for me is that the court sidesteps the fact that those 2002 changes were upheld in the Pirates case in 2014, since which only six years had elapsed and only one general election had taken place. There is no direct confrontation with the reasoning in the Pirates case, no grounds offered for why the court now understood equality in ways qualitatively different from when it decided in 2014 that the challenged ensemble of electoral institutions did not violate it.
Thus, in Pl. ÚS 44/17, we are given neither an argument for why the Pirates case was correctly decided but different from the STAN case, nor for why the Pirates case was wrongly decided.
It is hard to avoid the conclusion that the only thing that had changed in the meantime was that since late 2019 there has been a bill filed by Christian Democrats (Chamber file 649) to modify the electoral system, which the legislature would act on only if moved by the court. It may also matter that the court’s own composition had changed, with the addition in particular of Vojtěch Šimíček, a leading expert on election law from his time as chair of the Supreme Administrative Court’s bench for disputed elections. (He is also notable for his charmingly informal bio on the court’s website.) This allowed the judge rapporteur, Filip, to build a majority around the position staked out in a dissent in the Pirates case by Judge Kateřina Šimáčková.
And at root, decision Pl. ÚS 44/17 is in fact not all that radical a departure from precedent. It continues to subscribe (in paragraphs 115 and 147–155) to the essential tenet of the pathmarking decision Pl.ÚS 25/96 from 1997, in a challenge brought by a minor party, the Democratic Union, to the 5% threshold that barred them from eligibility for seats. In that case, the court upheld the threshold and its infringement of equality as a necessary “integrative” element to prevent an excessive number of parties in the legislature “and with it the threat to the functioning, capability and continuity of the parliamentary system”. I have argued that the Czech court endows the 5% threshold with an unduly talismanic force, as if it were the thin line safeguarding fragile democracy from the fate of Weimar Germany or the French Fourth Republic. The court applied this in a 2004 case to local elections, and has clung to it for elections to the European Parliament even while influential counterpart courts, such as the German, have deemed it unjustified (Judge Šimáčková, again, dissented in that case).
Probably the least well explained part of the STAN case is section IX.f, dealing with the steep additive threshold for electoral alliances. It was introduced in the 2000 package of reforms, questioned but allowed by the court in 2001, and retained in the 2002 legislation. As the section unfolds, it seems that the court will continue to allow it, even if it was suspected to be the product of “bad faith” lawmaking and had not performed its ostensible function of discouraging party fragmentation. After side tours into other countries’ experience with apparentment and STV, the section quite abruptly concludes that the current rules on coalitions are “disproportionate and inconsistent” and need to be struck down. They make clear, however, that this judgement does not apply to the less severe scale that was in effect before 2000.
Which brings me to the issue of what kind of fix can be made.
I agree with the court’s majority that this decision still arrives in time for the legislature to craft a remedy and does not fall within a pre-election blackout when rules should not change (the opinion imports the US Supreme Court’s Purcell principle). If the court is not going to advance a more demanding conception of equality, then relatively minor adjustments should suffice. For example, following McLean and Johnston’s critique of the use of D’Hondt for a six-seat district in European Parliament elections, even just shifting to the use of Sainte‐Laguë divisors (1, 3, 5…) would bring about a significant reduction in the largest party’s bonus. Taking the 2017 results for one of the two smallest districts, Liberec, in which Andrej Babiš’s ANO won four of the eight seats, using Sainte‐Laguë would reduce their number to three. While ANO’s 37.5% of the district’s seats would still be above their share of the district vote (29.8%), it would be much less extreme than the 50% of the seats they took under D’Hondt. (Sainte‐Laguë would have allowed the Communists to pick up one seat.)
Jakub Charvát, however, warns that Sainte‐Laguë would still not take care of all the problems of district magnitude, and would have some perverse results of its own.
Many other fixes are possible, such as reverting to the Hagenbach-Bischoff system used before 2002, with a restored second tier for seats not filled at the district level. Charvát and other Czech experts on electoral systems briefed the Czech Senate in a seminar on February 16; the transcript (in Czech) can be found here and the video here. Charvát summarizes a couple of the options:
It would be paradoxical to dispense entirely with the regional districts and move to a single national district at a time when Slovaks are publicly questioning theirs. It was introduced there in 1998 by Vladimír Mečiar’s Movement for a Democratic Slovakia, in the hope it would protect the ruling party from defeat at the hands of a coalescing opposition. It didn’t, but Babiš’s ANO may be interested in a national district for the same reason, and perhaps to enhance the central leadership’s control over the candidate list and tweak it to focus on the party’s strongholds.
In the end, what is the problem that this is all trying to fix? Is it the immediate question of accurate representation, or the hope that a different electoral system will miraculously produce better governance after multiple crises of government formation and survival that had little to do with the allocation of seats in the legislature?
I will try to return to this post and revise it when the parliament passes amendments that are currently being negotiated. Any comments or corrections are welcome in the meantime.