The legislative reform

The EU’s current legislative order mixes two partly competing representative principles, member-state representation and popular representation, without assigning each of them a clean institutional form. It also blurs the line between legislature and executive, because one of the Union’s co-legislators is composed of national ministers.

The Council is supposed to represent the member states in the legislative process, but under qualified majority voting a measure must be backed both by a threshold of member states and by states representing a threshold of the Union’s population. The Council therefore does not represent the member states on a purely state basis. Under qualified majority, the population represented by the supporting states also matters, importing an element of the popular principle into a body supposed to represent states.

Because the Council is composed of national ministers, it is also not distinctly legislative in character. It is a forum of executive officeholders performing a legislative function, which blurs the line between legislative and executive power. This structure tends to channel political conflict through executive bargaining rather than ordinary legislative politics. Where unanimity applies, that logic becomes sharper still, because state protection takes the form of a national veto, allowing a single government to block legislation. The European Parliament is supposed to represent citizens, but because seats are allocated through degressive proportionality between member states, citizens do not enjoy equal representational weight across the Union. That compromises the Parliament’s role as the Union’s chamber of popular representation.

That hybrid design is exacerbated by an executive-heavy legislative process. Neither co-legislator has a general right of legislative initiative. The European Commission alone has the power to propose laws. Under the ordinary legislative procedure, both co-legislators can amend proposals and both must approve the final text. But special legislative procedures reduce the Parliament to consultation or consent without amendment, unanimity preserves national government vetoes in some fields, and much of the Council’s decision-making still takes place through less public and less traceable executive bargaining than ordinary parliamentary proceedings.

The proposals in this essay amount to a treaty-level redesign and a constitutional refounding of the Union’s legislative order. The claim is that if the Union is going to become a democratic federal polity, it should separate the representation of citizens from the representation of member states by giving each a distinct institutional form. It should also improve the separation of powers by vesting state representation in a distinct federal legislative chamber rather than in national governments directly.

The current treaties define the European Parliament as a degressively proportional assembly of Union citizens, the Council as a body of ministers representing states, and Commission initiative as the ordinary basis of legislation. They treat electoral harmonization as a matter for the existing Parliament, not as a basis for constructing a new bicameral legislature. So this proposal cannot be reached by generous interpretation of the current settlement. Its legislative core needs treaty amendment at the level of institutional architecture.

This essay argues for a proposal shaped by five considerations: population equality in the popular chamber, equality of member states in the upper chamber, legislative governability, electoral accountability, and legal-political feasibility. No design can maximize all five at once. The aim is to make those compromises explicit in the structure of the legislature, so that the role of each chamber is clear and the choices involved can be argued about openly.

The proposal would reconstitute the collective legislature as the Parliament of the European Union. Its lower chamber, the Chamber of Representatives, would be elected by the people and designed to come as close to population equality as the Union's federal structure allows. Its upper chamber, the Senate, would represent the member states as equal constituent polities rather than as weighted population blocs.

Both chambers would be co-equal in ordinary legislation. Each would be able to introduce legislation, and both would have to approve the same text before it became law. In a federation built from pre-existing states, equal state representation in the legislature forms part of the constitutional bargain by which states enter the federation. The Senate gives institutional expression to that bargain by guaranteeing the member states a place in Union lawmaking as continuing constituent polities. The Chamber of Representatives provides direct popular representation and Union-level democratic accountability. The Senate is democratic as well, but on a different representative basis. Its role is to represent the member states within the federal legislature.

If either chamber were confined to exceptional questions while the other handled ordinary legislation, one representative principle would be pushed to the margins of Union lawmaking. That remains true even if the asymmetry applied only in particular fields, as it often does now. The limited chamber would be reduced to a reserve check instead of taking part in ordinary lawmaking on equal terms. This proposal rejects that model. Conflict between popular and member-state majorities should play out inside a public legislature, not be displaced into Commission agenda control, special procedures, and executive bargaining. Symmetry means that ordinary law must pass through both chambers on equal terms. Conflict between popular and member-state majorities is then worked through inside the legislature, rather than displaced into executive bargaining, special procedures, or agenda control.

Symmetry also requires a clear deadlock rule. If the two chambers cannot agree on a common text, the proposal would move to a public conciliation stage tasked with producing a single compromise draft. If conciliation succeeds, both chambers vote on that text. If it fails, the proposal falls. Disagreement would have to pass through a visible parliamentary sequence.

The daily proceedings, debates, and votes of both chambers would be public and recorded, with narrow exceptions for sensitive or classified information. Both chambers would also have investigative powers, including the power to summon witnesses and compel testimony from Union officers. This would make legislative responsibility more public, more legible, and easier to trace than it is under the current settlement.

Legally, this proposal lies beyond secondary legislation, interinstitutional adjustment, or broad interpretation of the existing treaties. A bicameral Parliament of the European Union, a Senate replacing the Council, and a redistribution of legislative initiative all belong to constitutional settlement. This is not a plan for minor institutional repair. It is a case for redesign at the level where the Union’s representative problems could actually be resolved.

Political feasibility does not turn on whether governments, parties, and publics would accept this proposal now. They likely would not. The real question is whether it identifies a coherent federal settlement around which future conflict, bargaining, and constitutional choice could be organised. On that standard, the question is whether the proposal is coherent and politically conceivable, not whether it is immediately acceptable. A serious federal proposal has to specify who is represented, how legislative conflict is handled, what requires treaty change, and what can be left to later law. This essay argues that such a settlement can be stated clearly, even if present political conditions are hostile to it.

The settlement of constitutional essentials should be distinguished from implementing detail. The treaties should entrench the two chambers, the replacement of the Council by a Senate, the powers of each chamber, including the redistribution of initiative rights, and the basic representative principle of each chamber. They should also fix the core electoral guarantees that give those principles practical effect.

The treaties should require the size of the Chamber of Representatives to remain consistent with both representational equality and legislative governability. The chamber should be large enough to permit a high degree of population equality within the constraints of the Union’s federal structure, but not so large as to impair its effective functioning as a deliberative and legislative body. For the purposes of this essay, the reference model assumes a chamber of 720 members, matching the size of the European Parliament in the 2024-2029 term. The law may determine the precise number of seats within those constitutional limits.

For the Chamber of Representatives, the treaties should also entrench direct election by Union citizens aged 18 or over who are ordinarily resident in the constituency in which they vote, and eligibility for election for any Union citizen aged 21 or over, both subject to limited legal disqualifications consistent with the constitutional order. Membership of the chamber should be incompatible with executive office, judicial office, or membership of another legislative chamber.

The treaties should also entrench the apportionment of seats first to member states, a guaranteed one-seat minimum for every member state, the rule that no district may cross a member-state border, the chamber’s fixed five-year term structure, the prohibition of dissolution before the end of a term, periodic reapportionment between member states to preserve equality of representation in light of population change, and the requirement that representation be made as equal among citizens as the Union’s federal structure allows. If a seat falls vacant during a term, the chamber should be required to order a by-election to fill that seat for the remainder of the term, and polling day should take place no later than 90 days after that decision.

The treaties should also state that districting should, where possible, respect the territorial integrity of sub-state geographic polities recognised under the constitutional or legal order of the member state concerned, so that regional and local political identities are not ignored without reason while member states remain free to reconstitute their internal political geographies in accordance with their own constitutional order.

For the Senate, the treaties should entrench the equal representation of member states, the allotment of three senators to each member state, election by the lower or sole chamber of each member-state parliament, the chamber’s basic six-year staggered term structure, and incompatibility with executive office at either national or Union level. If a Senate seat falls vacant before the end of its term, the lower or sole chamber of the member-state parliament concerned should elect a replacement within 90 days for the remainder of the unexpired term.

Electoral timetables, counting mechanics within those constitutional bounds, the use and design of member-state districting commissions, the detailed districting hierarchy, the ordinary procedures of electoral districting, and the detailed procedure for Senate replacement and lower-chamber by-elections can then be left to implementing law. Judicial review of districting and lower-chamber elections should begin in the courts of the member state concerned, subject to review at Union level where questions of treaty compliance or federal electoral principle arise. Not every electoral detail needs to be constitutionalised, provided ordinary legislation cannot undo the system’s core constitutional logic.

Abolishing the Council as part of the legislative order would not end executive coordination. National executives could still coordinate administratively and intergovernmentally before and after legislation. What would end is their place inside the constitutional definition of the legislature and the conversion of executive bargains into legislative outcomes. The details of executive coordination therefore fall outside this essay, which is about legislative refoundation, not a full redesign of the executive.

At the level of actors, the main institutional resistance would likely come from national executives, because the reform strips the Council and the European Council of their central role in executive bargaining over Union law. Some smaller states would likely also resist, despite Senate equality, because the lower chamber would be far more population-based than the current European Parliament. Nationalists and Eurosceptics would likely resist because the proposal assumes a federal transformation of the Union rather than rejecting one. The coalition most likely to support it would come from actors already committed to federal constitutional reform, not defenders of the present settlement.

How the two chambers are chosen

The proposed bicameral legislature does not offer two versions of the same representative claim. It turns two different principles into two different selection rules. The Chamber of Representatives is elected by the public under a much more population-based apportionment. The Senate rests on the equality of the member states as constituent polities.

Representatives would be elected by the public in territorial districts for 5-year terms using the single transferable vote, or STV. Most districts would elect between 3 and 6 representatives, because that is the range where STV can combine meaningful proportionality with a recognisable territorial link. In the small number of fallback districts that return only one representative, the same ranked ballot would operate as instant-runoff voting: ranked choice remains, proportionality does not. The reference district model in this paper is meant to show that such a lower-chamber map can be drawn while keeping districts inside member-state borders. That border rule, together with the narrow one-seat floor in apportionment, reflects the federal premise of the proposal: the Union does not rest on a unitary demos apart from its member states, and the lower chamber should therefore be as population-based as the federal order allows rather than organized as a borderless electorate.

Senators would be elected indirectly by the parliaments of the member states, not by governments and not by direct popular vote. This distinction matters because member states, member-state governments, and member-state parliaments are not the same thing. The Senate is supposed to represent states as constituent units of the Union. Governments are therefore the wrong occupants: a chamber of ministers would simply reproduce the previously discussed executive bargaining in parliamentary dress. Direct election would blur the chamber's logic in a different way by turning it into a second popular chamber with only weaker foundations. Election by national parliaments gives state representation a parliamentary rather than executive form. The states remain the represented units; national parliaments are the electing bodies because they provide democratic authorization without collapsing the upper chamber back into cabinet diplomacy.

Each member state would elect 3 senators serving 6-year terms, with terms staggered so that one Senate seat from each state comes up every 2 years. In unicameral systems, the sole chamber elects the senator; in bicameral systems, the lower chamber does. The lower or sole chamber is the institution most consistently tied across the Union to general electoral accountability. Many domestic upper chambers, by contrast, are territorially weighted, indirectly constituted, politically weak, or otherwise constitutionally idiosyncratic. Because the Union Senate already secures equal state representation at federal level, there is no strong reason to import those divergent domestic second-chamber formulas into the selection process. Lower-chamber election keeps the electing body parliamentary and democratically visible while keeping heterogeneous national territorial logics out of a chamber whose equality is already fixed at Union level.

The minimum election rule should also be set at constitutional level. Any member of the electing chamber could nominate a candidate. On the first ballot, election would require an absolute majority of all members. For the next 2 weeks, further ballots could be held on the same basis. If no candidate had been elected by the end of that period, a final ballot would be decided by plurality. The rule is therefore majoritarian, not proportional to the parliamentary balance. It will often reflect the governing majority or governing bargain, though not always. In minority-parliament situations, or where coalition discipline is weaker, other candidacies may emerge.

The Senate is not meant to reproduce, state by state, a miniature proportional portrait of domestic party pluralism. Its purpose is to give the member states, as enduring constitutional units, a parliamentary and non-executive form of representation in federal legislation. On that baseline, the relevant comparison is the current Council, whose members are national governments themselves. Election by the national lower or sole chamber does not worsen that condition. It parliamentarizes and partly democratizes it, while fixed terms, staggered renewal, and incompatibility with executive office stop the chamber from collapsing into a day-to-day extension of current governments. The plurality fallback is a last-resort deadlock rule, not the ordinary democratic principle of Senate selection.

Senators would hold independent terms. Early dissolution of the electing chamber would not terminate a Senate mandate. If a Senate seat fell vacant before the end of its term, the lower or sole chamber of the member-state parliament concerned would elect a replacement within 90 days for the remainder of the unexpired term, so that the staggered cycle remained intact.

Eligibility for election should extend to any Union citizen aged 35 or over, subject to limited legal disqualifications consistent with the constitutional order. A senator may be elected from within the national parliament, but could not hold both offices at once. Senate membership should be incompatible with executive office, judicial office, or membership of another legislative chamber.

The choice of 3 senators per state follows from the same logic. Three is the smallest equal allotment that lets one seat from each state come up every 2 years while leaving two continuing senators in place. One senator per state would remove any internal plurality. Two would make each renewal replace half a delegation. Three therefore gives a minimal but real balance between continuity and turnover. It preserves the upper chamber as a body of equal states without letting it swell into a second, indirectly chosen popular assembly. The core constitutional choices are the represented unit, the equal allotment, the electing body, the election rule, and the term structure. Administrative details can remain matters of implementing law, but the Senate’s representative principle should not.

Representation across both chambers

The Chamber of Representatives is much more population-based than the current European Parliament. The Senate, by contrast, is a state chamber in which every member state gets three senators. The system as a whole therefore still gives smaller states more weight than larger ones. That is a deliberate aspect of the constitutional design.

To compare those consequences, this section uses a country-level representation quotient, or RQ. An RQ of 1.000 means exact population parity within a chamber, measured against that chamber’s own average population per seat. Values above 1.000 indicate relative over-representation; values below 1.000 indicate under-representation. These are chamber-level country measures, not measures of district equality within member states. The appendix therefore uses a different metric, the district parity quotient, or DPQ, for lower-chamber district equality.

The combined figure reported below is only a heuristic. It averages the two chamber-level RQs under equal formal weighting. It does not measure total political influence, which would depend on agenda control, interchamber bargaining, and the practical use of bicameral power. Its purpose is narrower: to show the distributive pattern that follows from the proposal’s premise that both chambers have equal formal legislative status.

The lower-chamber electoral model

Once the lower chamber is understood as the more population-based chamber, its electoral system becomes a constitutional choice rather than a technical afterthought. This proposal adopts the single transferable vote, or STV, because it can combine candidate choice, medium-scale proportionality, and recognisable territorial districts better than a single at-large electoral arena. The district model is built to support that system.

STV is the multi-member form of ranked-choice voting. Voters rank candidates in order of preference, and seats are filled through successive counts and transfers rather than by treating a single party list as the only meaningful unit. That makes it attractive for a federal lower chamber: it preserves a direct link between representatives and places while allowing outcomes that are more proportional than plurality systems usually produce.

Under this proposal, voters would cast a single ranked candidate ballot. Parties could nominate multiple candidates in a district, and independents could run under a common signature requirement. Seats would be allocated by the Droop quota, with surpluses and eliminations transferred by next usable preference under a harmonized Union-wide counting rule. The point of harmonization is not to centralize election administration. It is to stop materially different national variants from being treated as if they were one system. Those rules belong in the electoral law of the revised chamber, not in treaty text.

STV makes the most sense when compared with the plausible alternatives. Relative to open-list proportional representation, it gives up some simplicity and some list-level proportionality in exchange for more voter control over candidate order and less dependence on party lists as the sole link between voters and representatives. Relative to mixed-member designs, it avoids splitting the chamber into district and list tiers or layering a compensation mechanism onto a Union that is already apportioning seats by member state. For a lower chamber meant to be territorial, candidate-centered, and more population-based within a federal union, that combination is the better fit.

In medium-sized districts, STV can widen voter choice, allow competition both within and across parties, and reduce some forms of vote splitting because lower preferences still matter. But those effects depend on district magnitude, nomination practices, party coordination, and the exact counting rules. STV should be defended for what it can do under these conditions, not treated as a cure-all.

This matters because the map relies heavily on 3- to 5-seat districts, with additional 2-seat districts and a very small number of single-seat fallbacks. Those magnitudes are compatible with STV, but they do not produce the very low effective thresholds associated with large-list proportional systems. The proposed lower chamber is therefore better described as moderately proportional and territorially grounded than as maximally proportional. In the narrow fallback cases where the model produces only one seat, the same ranked ballot becomes the single-winner form of ranked-choice voting, often called instant-runoff voting. Ranked choice remains. Proportionality does not.

The arithmetic behind that description is simple. Under STV, the notional threshold for winning one seat is the Droop quota, approximately 100 / (M + 1) for a district of magnitude M. For a list-based proportional system with the same district magnitude, the guaranteed one-seat threshold is broadly similar. The decisive variable is therefore usually district magnitude, not the label STV or proportional representation by itself. Table 4 translates the proposed magnitude structure into percentage terms.

In most member states, the current European Parliament election is held either in a single national district or in a small number of large regional districts. That makes the current threshold environment much looser than the one proposed here. Ireland and Malta already use STV, so the electoral formula itself is not alien to Union practice. The real discontinuity lies in district size and seat allocation. Belgium, Italy, and Poland are the closest current comparators because they already regionalize the vote, but even there the proposed model usually works with smaller district magnitudes than the current European Parliament election.

In Table 5, each comparison cell reports the electoral model, district structure, and approximate one-seat threshold. Structure is stated consistently as territorial units first, then seats attached to each unit. Where current systems use labels such as colleges or constituencies, those labels are retained. What matters for comparison, however, is the number of electoral units and the seat magnitudes attached to them, not the vocabulary each system uses. PR is shorthand for proportional representation.

The lower-chamber districting sequence

The lower-chamber districting sequence in this reference model starts from four constraints. Two follow directly from the constitutional design already set out: seats are first apportioned to member states, and no district may cross a member-state border. A third follows from the requirement that districting should, where possible, respect existing recognised territorial units within member states. In this reference model, that requirement is operationalised through the nested NUTS territorial hierarchy rather than through freehand line drawing. The remaining constraints are implementing choices made for this model rather than treaty essentials. Because the chamber is built for STV, the model treats 3 to 6 seats as the preferred district range, allows 2-seat and 7-seat districts only at the edge, and reserves 1-seat districts for narrow fallback cases. For the purposes of this reference model, 720 seats are allocated among the member states using base-1 Webster apportionment: each member state gets one seat first, and the rest are then apportioned by population under the Webster formula.

The lower chamber is the more population-based chamber, but it is not built as a borderless Union-wide electorate. Member-state boundaries and the one-seat floor remain federal constraints. They keep every member state present in the people’s chamber and confine districting to politically recognisable territories, even at the cost of some residual deviation from pure population equality.

Districting within each member state follows an ordered sequence rather than an open-ended balancing exercise. First, the member-state border is absolute: no district may cross it. Second, districts should come as close as practicable to the population target implied by the state’s seat total while staying within the preferred 3- to 6-seat range wherever possible. Third, the model favors territorial inheritance: it keeps the highest viable NUTS unit whole where it can. Fourth, if a unit is too large, the search moves down to the next NUTS level instead of splitting it arbitrarily; if a unit is too small, it is merged first with contiguous siblings inside the same parent shell, not across that shell. Fifth, contiguity is required throughout. Sixth, below-NUTS3 local review is allowed only after those higher-ranked rules are exhausted, and even then only to choose among otherwise admissible options. Seventh, territorial legibility matters, but only inside that narrowed space; it is not a free-standing override. The point is to reduce discretion by making district construction follow a visible sequence rather than freehand optimization.

The sequence also needs an administering authority. Under subsidiarity, the reference model does not rely on a single Union-level map-drawing body. It assumes instead that districting would be carried out within each member state under Union legislation setting the common criteria, the review cycle, transparency requirements, and the available avenues of challenge. Whether member states use independent commissions, courts, or another institution is left here to implementing law. The constitutional point is narrower: the governing criteria must be federal even if the primary institutions applying them remain state-based.

Exceptions come last and remain limited. Islands, remote territories, exclaves, embedded-polity cases, and autonomy cases are treated as explicit exception classes because they raise recurrent problems that ordinary NUTS logic cannot solve cleanly. Linguistic and minority considerations enter chiefly through those inherited territorial units and explicit autonomy exceptions, not through an open-ended community-of-interest override. A general minority or linguistic override would reopen exactly the kind of discretion the rule hierarchy is meant to constrain. The model is a reference model in the strict sense: it offers a rule-governed demonstration of how a Union-wide map can be generated and defended, not a claim that every boundary should be frozen forever.

Districting is only a limited instrument of minority accommodation. It can recognize minority and linguistic claims when they are territorially legible, constitutionally salient, and compatible with the model's broader commitments to equality, contiguity, and administrability. It cannot serve as a general compensatory device for every dispersed or non-territorial claim. Where minority or linguistic communities coincide with viable territorial units or autonomy structures, the model can respect that fact. Where they are too small, too dispersed, or territorially non-viable, districting cannot plausibly solve the problem without sacrificing the model's commitments to population equality, contiguity, and administrability. The reference treatment of Åland makes the point. The archipelago is not assigned its own district because, within a relatively small member state and under the general seat-range rules, it is too small to form a viable district of its own. That does not worsen the present European Parliament arrangement, which likewise gives Åland no separate seat. A member state could justify a different fallback exception if it reasoned it openly within the governing criteria, but the reference model does not treat separate district status as an automatic entitlement. Minority protection in a federation therefore has to rest not only on districting, but also on the wider liberal-democratic constitutional order: civil rights, linguistic guarantees, autonomy where applicable, and judicial protection.

Two concrete examples show how the ordinary sequence works before any exception rule is invoked. At NUTS 2 level, the model resolves BE22 (Prov. Limburg (BE)) inside BE2 (Vlaams Gewest) by merging it with sibling BE21 (Prov. Antwerpen), producing a combined quota of about 4.56 seats rather than looking across the NUTS 1 boundary into BE3 (Walloon Region), most obviously toward neighboring BE33 (Prov. Liège). At NUTS 3 level, inside DEA1 (Düsseldorf), an early local merge joins DEA18 (Remscheid, Kreisfreie Stadt) to the touching sibling DEA1A (Wuppertal, Kreisfreie Stadt) rather than jumping to non-touching DEA11 (Düsseldorf, Kreisfreie Stadt). The examples are meant to show that the search is rule-governed before it becomes discretionary.

Limited exceptions

Most districts can be built through the ordinary regional hierarchy alone. Departures from that sequence are kept narrow and explicit. Instead of opening a broad escape hatch for convenience or visual neatness, the model limits them to a short named list: islands, remote territories, embedded-polity cases, exclaves, and autonomy cases.

Seven areas requiring closer local review

Seven large metropolitan or regional cases needed closer local review below NUTS 3. They are included to show how the rule hierarchy works once the ordinary NUTS sequence runs out, not to license freehand redrawing of large urban areas. In these places the reference model used bounded judgment, informed where possible by finer local boundaries and existing electoral geography, but only after the higher-ranked criteria had already narrowed the admissible options. Because there is no common Union-wide hierarchy below that level, and because authoritative local polity information is not equally well defined across all cases, these areas could not be resolved programmatically in the same way as the higher-order sequence. Where usable local political geography existed, the model relied on it. Where it did not, it had to lean more heavily on contiguity and physical geography, which are weaker guides. A knowledgeable districting authority acting under subsidiarity should be able to apply the same principles on the ground with better local knowledge. These cases are evidence about workable subregional groupings, not independent criteria for drawing lines.