The referendum cannot literally be unheld
The referendum cannot literally be unheldWaadaa Collage

The referendum parliament cannot ignore

A decisive February vote approved sweeping constitutional reforms. Now Bangladesh faces a harder question: how far can Parliament depart from what the electorate explicitly endorsed?
Updated on: 

Bangladesh held two votes on February 12, 2026.

One elected a new Parliament. The other asked voters to approve a package of constitutional reforms that had emerged from more than a year of political upheaval, negotiations and attempts to redesign the institutions of the state.

The same electorate cast both ballots, at the same polling stations, under the same Election Commission. Now the two results appear to be on a collision course.

The referendum passed comfortably. According to the Election Commission’s corrected gazette, 47,225,980 voters — 68.26 percent of valid votes — voted Yes. Turnout was 60.26 percent.

The Bangladesh Nationalist Party had campaigned for the referendum. Its chairman (now the Prime Minister Tarique Rahman) urged supporters to vote Yes. The party also won the parliamentary election by a landslide, taking more than two-thirds of the seats with 49.97 percent of the popular vote, or about 3.75 crore votes.

But the Yes side in the referendum received roughly one crore more votes than the BNP did.

That difference now sits at the core of an unusually consequential constitutional dispute: When voters expressly approve constitutional reforms in a referendum, how much freedom does the Parliament they elect on the same day retain to rewrite them?

The government appears to have one answer. It plans to introduce a constitutional amendment Bill in January 2027 reflecting positions on which the BNP had previously recorded dissent, and then submit relevant amendments to another referendum under Article 142(1A).

The opposing legal argument is more fundamental. Parliament may decide how to translate the February mandate into constitutional language, it says, but it cannot legislate the opposite of what voters approved. Nor can a second referendum be used simply to obtain a different answer to a question the electorate has already settled.

At stake is not merely the architecture of Bangladesh’s next Constitution. It is a more basic question about where constitutional authority ultimately resides: with Parliament, or with the people from whom Parliament derives its power.

A referendum born out of a political rupture

The dispute begins in August 2024.

After a mass uprising brought down the government, Parliament was dissolved and an interim administration took office following an advisory opinion from the Appellate Division under Article 106 of the Constitution.

The National Consensus Commission subsequently brought political parties into negotiations over institutional reform. Those talks produced the July National Charter, signed on October 17, 2025, by 24 political parties, including the BNP. Two more joined later.

The Charter contained 48 constitutional proposals. Thirty had full agreement. On several others, parties entered “notes of dissent.”

On November 13, 2025, the president issued the July National Charter (Constitution Reform) Implementation Order, 2025. The Referendum Ordinance, 2025, followed under Article 93.

The arrangement was unusual. If the referendum passed, newly elected lawmakers were also to sit as a Constitution Reform Council, taking a separate oath and completing the reforms within 180 working days of the council’s first sitting.

Voters were presented with a single Yes-or-No question covering the Order and four propositions.

The first concerned the caretaker government during elections, the Election Commission and other constitutional bodies, which were to be formed according to processes set out in the Charter.

The second proposed a 100-member upper house allocated according to each party’s share of the national vote. A majority in that chamber would be required for constitutional amendments.

The third said winning political parties would be “bound” to implement 30 agreed reforms, including changes involving women’s representation, presidential powers and the prime minister’s tenure.

The fourth said the remaining reforms would be implemented according to each party’s commitments.

The electorate said Yes. The thing that happened afterward was less straightforward.

BNP lawmakers declined to take the Constitution Reform Council oath. In April, Parliament allowed the Referendum Ordinance to lapse. Then, on September 27, the government announced plans for its own amendment Bill, incorporating the BNP’s dissents and, where required, putting it to another referendum.

The question is whether those subsequent political decisions can alter the constitutional significance of what happened on February 12.

When the people speak directly

Article 7 provides the starting point for the argument that they cannot.

Article 7(1) declares that all powers of the Republic belong to the people. Article 7(2) describes the Constitution as “the solemn expression of the will of the people” and makes it the supreme law.

That language matters because Parliament’s power to amend the Constitution is not unlimited.

In Anwar Hossain Chowdhury v Bangladesh, the landmark 1989 Eighth Amendment case, the Appellate Division distinguished the constituent power of the people from the derivative amending power exercised by Parliament under Article 142.

Parliament can amend the Constitution. It cannot use that authority to destroy its basic features, including democracy and the sovereignty of the people. The February referendum presents a difficult extension of that principle.

If Parliament’s constitutional authority derives ultimately from the people, what happens when the people themselves have expressed a direct and specific preference about the provisions Parliament is about to amend?

The argument for the referendum’s binding force is that Parliament retains considerable freedom over implementation — wording, thresholds, timing and legislative form — but not over the fundamental choices voters made. An upper house based on national vote share cannot simply become one based on parliamentary seats. A requirement for upper-house consent to constitutional amendments cannot simply disappear.

To do so, under this argument, would turn a delegated power against its source: Parliament would be using authority derived from the people to defeat the people’s expressly recorded decision.

The government has a substantial answer.

Article 7(1) itself says that all powers must be exercised “under, and by the authority of” the Constitution. The Constitution does not expressly provide for a binding referendum of the kind conducted in February outside the mechanism established in Article 142(1A).

But the referendum was not conducted in a legal vacuum.

It was authorised by an ordinance promulgated under Article 93 while Parliament was dissolved. An ordinance has the force of law while in effect. The Election Commission administered the vote, and the broader transition itself rested on the Appellate Division’s advisory opinion under Article 106.

The argument therefore does not require treating the referendum as a free-standing source of constitutional law. It treats the result instead as a constraint on Parliament’s derivative amending power.

Article 142(1A) adds another complication.

It requires certain constitutional amendments — involving the Preamble and Articles 8, 48, 56 and 142 — to receive approval in a referendum even after securing the necessary parliamentary majority.

The constitutional hierarchy is striking. On these particularly important provisions, even a two-thirds parliamentary majority is insufficient. The electorate gets the last word.

That safeguard has itself received judicial protection. On December 17, 2024, the High Court Division held that its repeal was inconsistent with the Constitution’s basic structure. On July 9, 2026, the Appellate Division upheld the ruling by dismissing the appeals.

The judgment also observed that the caretaker-government system, incorporated into the Constitution in response to public aspirations, had acquired basic-structure significance.

The February vote touched directly on matters falling within Article 142(1A), including presidential powers under Article 48 and the constitutional amendment process under Article 142.

The government can argue that Article 142(1A) applies only to a referendum conducted after Parliament has passed an amendment Bill.

But the broader constitutional question remains awkward: If the Constitution considers a direct vote of the people indispensable before Parliament can alter certain fundamental provisions, what constitutional weight should be given to a direct vote in which the people have already expressed their preference on those provisions?

Can a referendum simply expire?

There is also the question of the ordinance itself.

Article 93 allowed the president to promulgate ordinances while Parliament was dissolved. The Referendum Ordinance did not itself amend the Constitution. It authorised a vote.

That vote took place. The Election Commission counted the ballots and gazetted the result. Parliament later allowed the ordinance to lapse. Does that erase what happened while it was valid?

The argument against that proposition rests partly on the language of Article 93(2): an ordinance “shall cease to have effect.” That ordinarily describes what happens going forward. It does not necessarily reverse completed acts undertaken while the law was operative.

The distinction is crucial here. The referendum cannot literally be unheld. Nearly 70 million valid votes cannot be returned to ballot boxes because the statute authorising their collection later ceased to operate.

An analogy can be found in the Indian Supreme Court’s decision in Krishna Kumar Singh v State of Bihar in 2017, which considered whether actions taken under an ordinance could survive after that ordinance expired. The court looked, among other things, to constitutional necessity and public interest.

A completed national referendum presents a strong version of that problem.

There is a further challenge. On March 3, 2026, the High Court Division issued Rules concerning the matter, leaving aspects of the referendum’s legal status sub judice.

But a Rule is not a final judgment. Unless and until a court invalidates the relevant legal instruments or their consequences, the referendum result remains an official, gazetted act of the state.

The Constitution Reform Council presents a separate issue.

Its legal foundation has been questioned because the Implementation Order relied on Article 48, which does not itself confer legislative power; because the Order purported to claim precedence over the Constitution; and because neither the council nor its separate oath appears in Article 148 or the Third Schedule.

Those objections could prove formidable without necessarily disposing of the referendum itself.

Article 7(2) says an inconsistent law is void only “to the extent of the inconsistency.” It is therefore possible, at least conceptually, to sever an invalid council procedure or precedence clause from the substantive propositions voters approved.

There is also a practical peculiarity: the members of the proposed Constitution Reform Council were to be the same people elected to Parliament.

If the Constitution requires those lawmakers to implement reforms through Article 142 rather than through a separately constituted council, the institutional route changes. The underlying referendum result does not.

One electorate, two ballots

The most politically uncomfortable aspect of the dispute may be the simplest. The Parliament now considering whether to depart from the February referendum was created by precisely the same electoral process that produced the referendum.

The two ballots were cast on the same day, at the same polling stations, under the same interim administration and Election Commission.

Since taking office, Parliament has approved more than 100 ordinances issued by that interim government.

The BNP, moreover, signed the July Charter, albeit with notes of dissent. It campaigned for a Yes vote. At a Jan. 30 rally, its chairman urged supporters to vote Yes on the referendum ballot just as they voted for the party on the parliamentary ballot.

The government can reasonably distinguish the two legal instruments. The parliamentary election rested on the Constitution and the Representation of the People Order, 1972; the referendum rested on an ordinance that subsequently expired.

Yet both emerged from the same transitional legal order.

Bangladesh has confronted analogous problems of legal continuity before. After the Fifth and Seventh Amendment judgments invalidated periods of martial-law rule, Parliament passed two laws in 2013 to preserve ordinances enacted during those periods. The practical lesson was unavoidable: dismantling a legal order does not automatically erase every legal consequence it produced.

The BNP’s notes of dissent raise a different question. The party can argue that when it signed the Charter, it expressly reserved disagreement with certain proposals. But the wording of the referendum complicates that position.

Propositions (a) and (b) were presented separately. Proposition (d), by contrast, expressly left “other reforms” to each party’s commitments.

If the parties’ earlier reservations automatically qualified propositions (a) and (b), there would have been little reason to distinguish them from proposition (d).

The manifesto argument encounters a similar problem.

The BNP proposed an upper house allocated according to parliamentary seats and then won more than two-thirds of the seats. But voters do not cast separate ballots for every promise in a manifesto. They vote for a package containing hundreds of positions.

The referendum question on the upper house was specific. It proposed allocation according to national vote share. The electorate approved it on the same day — and the Yes side received about one crore more votes than the BNP.

The constitutional conflict is therefore not easily reduced to Parliament versus referendum. Both possess democratic legitimacy. The harder question is what happens when a general electoral mandate and a specific referendum mandate point in different directions.

The government's proposed second referendum makes that question more urgent.

Article 142(1A) was designed as a safeguard: for certain fundamental amendments, Parliament's supermajority must also obtain the people's consent.

But if a new amendment Bill departs from decisions already approved in February and is then bundled with popular reforms into a single Yes-or-No referendum, the mechanism could effectively be used to ask voters to reconsider an answer they have already given.

Alternatively, provisions outside Article 142(1A) could be separated into another Bill and passed by a two-thirds parliamentary majority without another direct vote.

The legal argument against both routes is essentially the same: constitutional power cannot be used indirectly to accomplish what it cannot lawfully accomplish directly.

None of this means Parliament has no role. Quite the opposite. The Constitution still requires constitutional amendments to proceed through Article 142. Parliament must draft the text, debate it and pass it by the required majority. Any amendment involving the provisions listed in Article 142(1A) must satisfy the referendum requirement as well.

Nor does the argument necessarily require a court to order Parliament to legislate.

Article 78(1) protects parliamentary proceedings, but not necessarily the constitutional validity of the laws those proceedings produce. Bangladesh’s courts have reviewed constitutional amendments before, including in the Eighth and Fifteenth Amendment cases.

The potential judicial remedy would therefore be one of restraint rather than legislation: a declaration that Parliament cannot enact an amendment that reverses the substance of the February result, or structure another referendum to circumvent it.

The threshold for such judicial intervention would be high. But that only brings the dispute back to the question with which it began.

Bangladesh’s Constitution says all powers of the Republic belong to the people. It also entrusts Parliament with the power to amend the Constitution.

On February 12, the people exercised both forms of democratic authority at once. They chose the Parliament. And on a separate ballot, they told that Parliament what constitutional reforms they wanted.

The coming dispute will determine whether the first decision permits Parliament to rewrite the second.

–

Asif Shahriyar Sushmit is the Executive Director, Institute of Policy Dynamics

Daily Waadaa
dailywaadaa.com