On 5 July 2015, 61.31% of valid votes in Greece’s referendum were cast for No. The number is clear. What becomes much less clear, looking at the assistance agreement reached later that summer, is what the refusal was able to change.^[1]

Two headlines can turn this into a very short story: the public rejected the terms; the government subsequently accepted a bailout. But the distance between those headlines contains the difficult questions. Which proposal was rejected? What was still open to negotiation? What would refusing the next agreement have meant?

The ballot concerned a proposal presented to the Eurogroup on 25 June, comprising a reform document and a preliminary debt-sustainability analysis. It was not a direct vote on leaving the euro, nor a permanent rejection of every possible future assistance package.^[2]

Accuracy about that question is necessary. It is not sufficient. Saying that the eventual agreement was legally different cannot by itself settle what happened to the political refusal. A country may retain the formal power to decline an offer while finding the range of survivable alternatives painfully narrow.

The calendar outside the polling station

Before the vote, banking pressure was already acute. On 28 June, the European Central Bank maintained the ceiling on emergency liquidity assistance to Greek banks at its 26 June level. Holding a ceiling unchanged was not the same as withdrawing all existing assistance. It was also not an assurance that assistance would expand to meet further outflows.^[3]

Against the background of deposit withdrawals, the Greek authorities imposed a bank holiday and capital controls. Those were domestic measures, not simply an ECB order closing Greek banks. Yet locating the formal domestic decision does not remove the surrounding constraints. The European Court of Auditors’ later chronology records how these developments interacted.^[4]

A ballot does not stop that calendar. Debts mature, payment systems need liquidity and negotiations consume time. These are not reasons to disregard a vote. They are reasons why saying No in a crisis can mean something very different from selecting No on a form with no consequences until the answers have been collected.

The previous European programme expired on 30 June. Greece requested a new ESM loan on 8 July; the new programme began on 19 August. There was a new request, negotiation and approval process, not merely the implementation under a new cover of the two documents on the referendum ballot.^[5]

Nevertheless, procedure does not make the disappointment disappear. One can judge continued negotiation to have been the less damaging course and still ask why the space for negotiation had become so limited. Acceptance under severe pressure is a decision. It is not necessarily satisfaction with the conditions, and it cannot retrospectively turn a refusal into enthusiasm.

Other societies were implicated too. Governments providing assistance had to answer to their own publics and legislatures about resources and risk. A Greek No could not automatically become an obligation on everyone else to continue lending on unchanged terms. No single national electorate could give the last answer for all the participants.

That observation does not make their bargaining positions equal. The ability to wait, endure an interruption or find another source of finance can differ enormously. Consent still matters in such a situation. What it cannot do is end the inquiry into the circumstances under which it was obtained. A signature records a decision; it does not measure the freedom of every alternative.

Money is not one thing

Looking back to 2010 reveals another reason to avoid a single undifferentiated villain or rescuer. The first European package used bilateral euro-area lending through the Greek Loan Facility, alongside IMF lending. The second involved the EFSF; the third, in 2015, used the ESM. It was not one unchanged institution operating for eight years.^[5]

The acronyms matter only because powers and responsibilities matter. Who can release a tranche? Who proposes a condition? Who legislates it at home? Who can reconsider it? Calling every decision ‘Europe’s demand’ can make responsibility simultaneously immense and elusive: everyone participates, yet each actor appears responsible only for a fragment.

The third programme had a maximum envelope of €86 billion. The ESM actually disbursed €61.9 billion before the programme ended on 20 August 2018. A ceiling is not a disbursement. These were loans, not a grant whose entire face value could simply be divided among households.^[6]

Money used to meet a maturing debt can prevent an immediate break in payments without appearing as an equivalent increase in family income. Conversely, the absence of a direct household transfer does not establish that the financing provided no household benefit. The connection between sovereign payments, banks and ordinary life needs explaining, rather than being settled by the word ‘rescue’.

Stability can consequently refer to different distances from the same event. For a financing institution, it may concern the next payment. For a budget, it may concern the required balance. For someone living on wages or a pension, the question is how life can be arranged after an adjustment. These perspectives can support one another. They need not coincide automatically.

Nor did recovery arrive as originally expected. The IMF’s 2013 evaluation of the first programme acknowledged a downturn much deeper than projected and unemployment above expectations. It examined fiscal contraction, implementation, confidence and other factors. This was a participant’s retrospective evaluation, not a finding that all losses had a single cause.^[7]

A failed forecast is more than a number to correct quietly in the next document. If immediate sacrifices were justified partly by an expected recovery, a delayed recovery changes the proposition people were asked to accept. The reasons must be explained again. An updated spreadsheet cannot do that work on its own.

There is no costless alternative history waiting to be inserted here. Earlier restructuring, slower adjustment or a different tax mix each requires evidence about effects and risks. Asking whether another course was possible should make alternatives more specific. It should not give hindsight the privilege of proposing choices that never have to encounter the world.

A pay-slip becomes a legal question

Alongside the aggregate figures there is a documented individual challenge. Ioanna Koufaki contested her pay reduction; the public-service union confederation ADEDY also challenged austerity measures. On 7 May 2013, the European Court of Human Rights declared the relevant complaints inadmissible as manifestly ill-founded.^[8]

This court was neither the ESM lender nor the European Union’s Court of Justice. It addressed particular measures under the Convention’s property protection. Its decision cannot be expanded into a collective institutional declaration that every economic policy adopted across eight years was either lawful in every respect or the best available policy.

The court considered the exceptional crisis, the legislature’s latitude in social and economic policy and the burden on the applicants. It did not undertake to choose the best national budget. The application failed, but the record still matters: a specific person had made a specific reduction a question requiring a legal answer, rather than remaining merely an example in someone else’s account of public hardship.^[8]

It would therefore be inaccurate to say that individuals had no remedy of any kind. Having a route to challenge a pay measure is nevertheless different from being able to reopen the entire assistance negotiation or alter a disbursement date. The reach of a legal proceeding cannot tell us that a person’s whole difficulty has been heard.

A further question survives even meticulous procedure. If adjustment lasts longer than anticipated, how can those least able to bear it bring that fact into the next decision? Courts, parliaments, negotiations and public argument do different work. None should be expected to replace all the others merely because it has delivered an answer within its own remit.

Keeping that question open does not require deleting what assistance did provide. Loans supplied real financing; the third programme eventually ended. Programme completion is a verifiable institutional event. It is not evidence that every household had recovered, or that debt and oversight ceased to matter afterwards.^[5][6]

The Greek case resists being made into a fable about either obedience or irresponsibility. Shared financial risks were real. So were revisable forecasts and uneven burdens. Governments made commitments, while individuals had to live with consequences they had not personally negotiated. Understanding that combination takes more work than identifying which side signed last.

The referendum’s significance was not that a No could immediately alter every external condition. It made refusal unmistakably part of the record. What followed still required an account: why an agreement was accepted, what remained contested, and which costs could not be declared irrelevant simply because finance had arrived.

Perhaps the most persuasive reason for a shared arrangement is that it should offer more room than confronting a crisis alone. Whether it did so cannot be judged only by a signed agreement or a completed procedure. We also need to ask what alternatives remained after refusal, and what questions could still be raised after acceptance.

Most valid votes said No. Another agreement was subsequently signed. Both statements are true. The difficult road between them is not an embarrassing interval to hurry past. It is the part of the history we most need to read.