Iceland's Banking Collapse of 2008

Key Takeaways

• The size problem was stated plainly by Iceland's own investigators. The Special Investigation Commission wrote that a financial system three times a country's output can still be governed by it, and that at nine times the roles are reversed. The Bank for International Settlements puts the three banks at ten.
• The guarantee behind Icesave was a rounding error. The fund held about ISK 13 billion at the end of September 2008, plus ISK 6 billion of letters of guarantee written by the very banks about to fail, against a guaranteed amount the regulator put at ISK 722 billion on 2 October 2008.
• The banks had financed much of their own equity: roughly ISK 300 billion of what the Commission called weak equity against a capital base of about ISK 1,186 billion in mid-2008. During 2008 the banks were the buyer in an average of 45 percent of automatically matched trades in their own shares, the seller in fewer than 2 percent.
• The currency did not so much fall as stop. The European Central Bank's daily euro reference rate went from 156.13 on 3 October 2008 to 305.00 on 9 October, a loss of 48.8 percent of the króna's euro value in four trading days, then printed exactly 305.00 on twenty consecutive days before the daily series went dark on 3 December 2008 and did not resume for more than nine years.
• Losses were allocated, not absorbed. Deposit claims were satisfied in full because the Emergency Act gave them priority, general claims recovered about 29 percent, and shareholders and subordinated creditors recovered nothing. Loans made so borrowers could buy the banks' own shares recovered 4 to 6 percent.

What Happened in the First Ten Days of October 2008?

The sequence began the week before. On 25 September 2008 Glitnir's chairman told the Central Bank's governors that the bank would struggle to repay loans falling due in mid-October, and asked for a collateralised loan of EUR 600 million. Four days later the government announced instead that it would inject the same sum for 75 percent of the bank. The Commission's verdict is blunt: the request was not properly processed, the sum was close to a quarter of the Central Bank's foreign currency reserves, and the proposal was not credible.

It also did the opposite of what was intended. Announcing that the state was rescuing one bank told every creditor of the other two exactly how much the state had, and deposits began leaving Landsbanki and Kaupthing immediately. Rating-contingent clauses then accelerated EUR 425 million of Glitnir's loans, and margin calls on it reached about EUR 1,100 million.

Events from the Special Investigation Commission, the EFTA Court judgment and the Bank for International Settlements. The exchange rate is the European Central Bank daily reference rate, Icelandic krónur per euro.

Three banks that were, between them, the country's banking system were gone inside seventy-two hours. By March 2009, on figures in the EFTA Court's judgment, 93 percent of the Icelandic commercial banking sector had failed, and institutions representing 99 percent of the market eventually entered winding-up or restructuring.

How Did Three Banks Get Larger Than Their Country?

Iceland had about 315,000 residents at the start of 2008. What let three banks headquartered there lend across Europe was the single passport under the European Economic Area Agreement, in force from 1994: a bank licensed in one member state may open a branch in another without applying to the host regulator, because the rules are treated as equivalent. A branch is supervised by the home state and, critically for what followed, its deposits fall under the home state's guarantee scheme.

Privatisation and merger left three large banks by the end of 2003, and the balance sheets then did something the Icelandic supervisory apparatus was never scaled for. Total assets more than doubled in 2004 and doubled again in 2005, with lending by the parent companies growing at an average of nearly 50 percent a year until the collapse. In 2005 alone the three raised around EUR 14 billion in foreign debt securities markets, slightly more than Iceland's entire gross domestic product that year, priced about 20 basis points over the benchmark. When European markets tightened the American market opened, because a high agency rating paired with an unusually generous coupon lifts the average rating of a collateralised debt obligation cheaply. That made the Icelandic banks, in the Commission's phrase, the cheapest banks available for their rating.

There had already been a rehearsal. In early 2006 Iceland's current account deficit exceeded 16 percent of gross domestic product, the króna fell sharply and credit default swap spreads widened. The episode was named the Geyser crisis and it passed. The Commission's assessment is that the calm was the dangerous part, because the imbalances then grew faster than before, and because no later than 2006 the authorities would have had to act if the banks were to shrink without destroying the value of their assets. Instead the governing coalition's agreement of May 2007 committed to letting financial activity keep growing at home.

Why Could the Central Bank Not Act as Lender of Last Resort?

A central bank can always meet a run on liabilities denominated in the currency it issues, because it creates the money. The Icelandic banks' liabilities were overwhelmingly in euros, sterling and dollars, which the Central Bank of Iceland could not create. That converts an ordinary liquidity backstop into a reserves problem, and Iceland's reserves were sized for a small open economy rather than a banking system many times its output.

The dependence had been deepening for a year. Collateralised borrowing by the three banks was around EUR 2 billion in the autumn of 2007 and roughly EUR 9 billion by the collapse, nearly half of it from the European Central Bank. The Central Bank of Iceland's own collateralised lending rose from ISK 30 billion in the autumn of 2005 to ISK 500 billion by early October 2008, about ISK 300 billion of it lent to the three large banks against their own securities. The Commission holds that lending on that collateral while aware of the banks' weaknesses can barely be reconciled with the statutory requirement that collateral be eligible.

The obvious remedy was a foreign currency swap line, and Iceland asked. On 15 April 2008 the chairman of the Central Bank's board of governors wrote formally to the Bank of England. The reply of 23 April refused the swap and offered something else: help in finding an effective way to reduce the size of the Icelandic banking system, which the Governor described as the only practical solution. The Central Bank of Iceland did not take up that offer. It asked for the swap to be reconsidered instead, and received no reply. After the meeting of central bank governors in Basel on 4 May 2008 only the Danish, Norwegian and Swedish central banks would provide facilities at all, on condition that the government press the banks to shrink. Ministers signed that declaration on 15 May 2008. The Commission records that the promises were not kept, that the declaration was never made public, and that by October the Icelandic authorities had very few people left to call.

What Was Icesave, and Why Was It Beyond Reykjavik's Control?

Icesave was an online savings account offered by a branch of Landsbanki, launched in the United Kingdom in October 2006 and accepting deposits in Amsterdam from 29 May 2008. Kaupthing ran a comparable product under the Edge name. These were not foreign subsidiaries but branches operating under the single passport, so the deposits were guaranteed in the first instance by Iceland's Depositors' and Investors' Guarantee Fund up to the ISK 1,700,000 per depositor set by Icelandic law. Both host schemes topped up above that: the British scheme to GBP 50,000 for each retail depositor, and the Dutch scheme from EUR 40,000 to EUR 100,000 per depositor.

The shift in funding is easy to understand and hard to overstate. Deposits in the Icelandic banks stood at just over ISK 250 billion at the start of 2000. By October 2008 those held by parties other than financial institutions exceeded ISK 3,100 billion, with a little over ISK 1,700 billion in branches abroad, and by 2007 deposits owned by foreign parties exceeded half the total. It was not really a diversification of funding: the Commission found total deposits shrinking from the autumn of 2007 as wholesale money left all three banks, and the outflow from Landsbanki's British and Dutch branches in the final year exceeded the inflow of Icesave retail money.

What makes Icesave a governance story rather than a marketing one is that everyone understood the branch structure was the problem and nobody made it stop. Conversion into a subsidiary would have moved the guarantee obligation to the host state, which was the cheapest defensive measure available. The Commission records that Landsbanki itself first drew up such plans in February 2008, that officials agreed among themselves it was necessary to press for the transfer and raised it with the banks, and that in the Landsbanki case it simply never happened. The party that pushed hardest for it was the British regulator, not Reykjavik. There had already been a warning shot: the EFTA Court's judgment records a run on the British Icesave accounts between February and April 2008, and at a meeting on 1 April 2008 the Central Bank's governors told the Prime Minister and the Foreign Minister that GBP 193 million had gone out of Icesave in London over a single weekend, and that Landsbanki could withstand that rate for six days.

Why Was the Icelandic Deposit Guarantee Fund Never Going to Work?

The Depositors' and Investors' Guarantee Fund, known by its Icelandic initials as TIF, began operating on 1 January 2000, implementing the European deposit-guarantee directive of 1994. It was not a public agency but a private foundation, and four of the six seats on its board were filled by the financial industry it insured. Member contributions were capped at 0.15 percent of the previous year's average guaranteed deposits, and there was no explicit government backstop: the state had no legal obligation to fund it at all.

Those choices were survivable while the banking system was small. When the fund started its assets were about ISK 2.9 billion against deposits of just over ISK 250 billion, around 1.2 percent. At the end of September 2008 its assets were ISK 13 billion, plus ISK 6 billion of letters of guarantee issued by banks and savings banks. The Commission's observation about those letters is the detail worth keeping: when the banks collapsed, the guarantees they had written to the fund that insured them became almost worthless.

The denominator arrived late, and from a meeting. At a gathering of three ministries, the Central Bank and the regulator on 2 October 2008, four days before the Emergency Act, the regulator's Director General stated that calculations made that summer put the amount guaranteed by the fund at ISK 722 billion. Set ISK 13 billion of real assets against that and the coverage ratio is about 1.8 percent; the Bank for International Settlements measures the same fund on its own basis and reports 0.41 percent. Both figures are given with their source rather than reconciled, and either way the fund could not pay. The point outlives the specifics, and is covered in general terms under FDIC deposit insurance: a deposit guarantee is a promise made by a balance sheet, and it is worth whatever that balance sheet is worth.

Who Owned the Banks, and Who Were They Lending To?

Largely the same people. The Commission examined the largest exposures at all three banks and found the principal shareholders among the largest borrowers in every case. At Glitnir the biggest borrower group was Baugur Group and its affiliates, owing more than ISK 250 billion at the collapse, about 70 percent of Glitnir's equity base. Kaupthing's largest shareholder, Exista, was its second-largest debtor and owed well over ISK 200 billion. At Landsbanki the largest debtor was Bjorgolfur Thor Bjorgolfsson with his affiliated companies and the third largest was his father Bjorgolfur Gudmundsson, and their obligations together came to well over ISK 200 billion, more than the whole of the group's equity. Baugur and its affiliates owed the three banks EUR 5.5 billion at the peak, about 53 percent of their combined equity base: one borrower group against half the banking system's capital. The regulator saw the problem and could not make it stick. In March 2007 it told Landsbanki that Actavis Group, in which Bjorgolfur Thor held a 38.84 percent stake, had to be counted alongside his other exposures, which took the bank's exposure to him and related parties to at least ISK 51.3 billion, or 49.7 percent of its regulatory capital. The bank rejected that reading in a side letter, the regulator nonetheless authorised it to report the exposures separately that quarter, and they were still reported separately at the next reporting date. The case was dropped after Actavis was refinanced that September.

Ownership and lending are one subject here because much of that lending bought the banks' own shares. They lent to buyers of their stock, took the stock as collateral, and counted the resulting equity as capital. The Commission calls the result weak equity and sizes it at roughly ISK 300 billion by mid-2008 against a capital base of about ISK 1,186 billion. Against the core component alone, meaning shareholders' equity less intangibles, weak equity exceeded 50 percent, and cross-financing between the banks takes it to roughly 70 percent. The Commission's statement of why this matters is the cleanest in the report: if equity no longer provides a cushion protecting depositors and creditors, it is not equity in the economic sense.

The share price was itself being managed. In automatically matched trades during 2008 the banks were the buyer in an average of 45 percent of trades in their own shares and the seller in fewer than 2 percent, which the Commission concluded was an attempt to create demand rather than market-making. Every published capital ratio nonetheless sat above the statutory minimum, and the regulator's own stress test, published in August 2008, reported that the banks' solvency was not at risk. What the collateral was worth appears in the recovery on those share-purchase loans: between 4 and 6 percent. Reading the notes rather than the ratios is the skill involved, and the subject of fundamental analysis.

What Did the Emergency Act of 6 October 2008 Actually Do?

Act No. 125/2008 was adopted by Althingi late on 6 October 2008 and did three things. It empowered the Ministry of Finance to fund and capitalise new banks or restructure an existing one. It authorised the regulator to take over a failing bank, assuming the powers of the shareholders' meeting and suspending the board. And it gave deposits priority over general unsecured claims in bankruptcy, equally in domestic branches and branches abroad.

Most of it was written in about seventy-two hours. Apart from the provision that became Article 100a of the Financial Undertakings Act, the Act was drafted between 4 and 6 October 2008 under enormous time pressure, and the drafters were not assisted by specialists in the fields the bill depended on, insolvency law among them. The contingency planning that should have preceded it barely existed. Three separate groups had been given the task at one time or another, and the last two were convened inside a single week: one by the Central Bank on 30 September, and one by the Prime Minister on the morning of 4 October, which was not initially given the others' material. Its first task, the Commission records, was to print out the annual accounts of the large financial corporations from a member's office at the University of Reykjavik. When the banks collapsed there was no governmental contingency plan at all, which the Commission calls unacceptable.

The resolution model the Act enabled was a domestic and foreign split rather than a good bank and bad bank split. Each bank's domestic loans, deposits and Icelandic infrastructure went to a new institution, while the foreign operations, wholesale funding, bonds and derivative positions stayed in the old estate as claims against it. A good bank and bad bank split would have stranded most domestic household and corporate credit in the bad bank, making later debt restructuring close to impossible, and there were no buyers for anything in any case. The deposit priority was reviewed by the EFTA Surveillance Authority in December 2010, which concluded it did not amount to unlawful discrimination under the EEA Agreement.

Why Did Britain Use Counter-Terrorism Law Against an Icelandic Bank?

The Landsbanki Freezing Order 2008 was made on 8 October 2008 and came into force the same morning at 10:10, under sections 4 and 14 of, and Schedule 3 to, the Anti-terrorism, Crime and Security Act 2001. The statutory test the Treasury had to satisfy was not about terrorism at all: the Order recites its belief that action to the detriment of the United Kingdom's economy had been or was likely to be taken by persons resident outside the United Kingdom or by a foreign government. That limb of the Act was drafted for economic threats and had never been used against a friendly state's bank.

What it froze mattered more than what it was called. The Order reached cash, deposits and securities held by Landsbanki, and initially also funds relating to Landsbanki held by the Central Bank of Iceland, the Icelandic regulator and the Government of Iceland. Assets of a sovereign state's central bank were caught by an instrument whose title carried the word terrorism, which is why the measure became a lasting political fact in Iceland rather than a merely legal one. It was revoked by a further order made on 10 June 2009, taking effect on 15 June.

The same morning produced a second British action with larger immediate consequences. Kaupthing's London subsidiary, Kaupthing Singer and Friedlander, had been told on 3 October to receive GBP 1,600 million from its parent by 6 October. On the morning of 8 October the regulator demanded a further GBP 300 million and Kaupthing did not pay. The UK Treasury transferred the subsidiary's Edge deposits to ING Direct at 10:00, the subsidiary was placed under cessation of payments later that day, and the resulting cross-default put the Icelandic parent into moratorium. The regulator took control of Kaupthing on the eve of 9 October. The sequencing is worth noting, because retellings reverse it: Landsbanki had already been taken over on 7 October, the day before the freezing order, which did not cause that failure but did set the terms of the four-year dispute that followed.

What Did the Króna Do, and Why Did the ECB Stop Quoting It?

The European Central Bank publishes daily euro reference rates computed from a market it can observe, which makes the series unusually honest here: when the market disappears, the series shows it.

The strongest reading for the króna in the Bank's 2007 and 2008 record was 81.68 per euro on 24 July 2007. The rate stood at 91.99 on 2 January 2008, at 143.38 on 29 September as the Glitnir stake was announced, and at 156.13 on Friday 3 October. Then came the week: 195.00 on Monday 6 October, 197.00 on the Tuesday, 265.00 on the Wednesday and 305.00 on Thursday 9 October. From 3 October to 9 October the króna lost 48.8 percent of its euro value in four trading days. Measured from the July 2007 high, it lost 73.2 percent.

The strange part comes next. The Bank published exactly 305.00 on 9 October and again on every publication day up to and including 5 November: twenty consecutive readings without a single decimal of movement. A reference rate that does not move for a month is not a stable currency. It is the absence of a currency market, held in place by an unusable interbank market and an official rate with nothing trading against it. The last daily rate the Bank published before that gap is dated 3 December 2008 at 290.00. It did not publish another one for more than nine years: the series resumes on 1 February 2018, by which point Iceland's capital controls had been lifted and a real market again existed to quote.

For an investor the transferable point is not the size of the depreciation but its character: a price series can go dark for years and still come back. Any asset whose value depends on converting one currency into another carries an implicit assumption that a market exists in which to do the converting, and that assumption is rarely priced until the week it fails. The general problem is covered under international ETFs, currency risk and hedging.

Who Warned About Iceland, and Why Did the Warnings Change Nothing?

The warnings were early, public and specific. Fitch revised Iceland's outlook from stable to negative in February 2006, and Danske Bank published its Geyser Crisis research the following month, arguing that a financial crisis was a substantial risk and noting a current account deficit heading toward 20 percent of output. The Commission's own view of early 2006 is that all the prerequisites for a crisis were already in place, with net foreign currency liabilities close to the country's entire annual output.

Then nothing happened, which is the part worth studying. The market calmed by mid-2006, the banks returned to funding markets, and the imbalances resumed growing faster than before. A warning followed by two profitable years is indistinguishable, in the moment, from a warning that was wrong, and the institutions alarmed in 2006 had spent their credibility by 2007.

Inside the government the information was better than the response. The Commission documents a meeting on 7 February 2008 at which the Central Bank's chairman gave the Prime Minister and two colleagues a bleak account of the banks' prospects, and at least five such meetings between February and May 2008. The minister actually responsible for banking was not summoned to any of them and, with one exception, was not told they had happened. The Commission is equally hard on how the warnings were delivered: the Central Bank did not bring documents, summarise information or attach proposals for action, ministers did not ask, and no minutes were kept. The result was a government that had heard alarming things without ever receiving a proposal it could accept or refuse.

The lesson is not that the risk was invisible. It is that identifying a fragility gives you no clock. Iceland's vulnerabilities were legible from 2006 and the system kept working for two and a half more years, long enough to end the career of anyone who acted early and short enough to ruin anyone who did not act at all. The same asymmetry runs through Japan's asset price bubble.

What Did the IMF Programme and the Capital Controls Do?

The International Monetary Fund approved a two-year stand-by arrangement on 19 November 2008 amounting to SDR 1.4 billion, around USD 2.1 billion at the time, disbursed in seven tranches over three years, each with a five-year term and interest-only payments in the first two years. Bilateral lending came alongside: USD 2.5 billion from the Nordic countries, USD 200 million from Poland and USD 50 million from the Faroe Islands. All carried a spread of 2.75 percent and all were repaid early.

Board approval did not come easily, and the reason is instructive. Fund rules require a programme to be fully financed before approval. The Dutch and British authorities held up board approval until Iceland agreed to guarantee the Icesave deposits, and the other Nordic countries withheld their bilateral contributions pending a settlement. Because that money was part of the financing, withholding it made the programme unfinanceable and therefore unapprovable. The deposit-guarantee dispute was not a quarrel running alongside the rescue. It was the gate the rescue had to pass through.

The programme's most consequential condition was one the Icelandic authorities had not proposed themselves. Capital controls were introduced on 28 November 2008, and the Fund's willingness to endorse them was a departure from decades of institutional preference. They were capital controls rather than exchange controls: transfers to pay for goods and services were exempt, so importers could still buy foreign currency. What was blocked was the exit of financial capital, specifically a stock of króna-denominated claims held by non-residents that peaked at roughly 40 percent of gross domestic product in 2009. Iceland notified them under Article 43 of the EEA Agreement, and the EFTA Court held in December 2011 that one such control was compatible with it.

Removing them took eight years and three months, because that stock had to be shrunk first. The Central Bank ran twenty-three currency auctions between June 2011 and February 2015, which cleared about half the offshore stock, and a bill passed in March 2012 revoked the failed estates' exemption from the controls, converting them from a threat to the balance of payments into a party with a reason to negotiate. In June 2015 the government announced a stability levy of 39 percent unless composition agreements meeting the Central Bank's conditions were reached by year end. They were reached that November, with creditors accepting haircuts in exchange for exemptions, and the non-residents' share of those contributions came to about 17 percent of output. Rules no. 200/2017 lifted the controls on 14 March 2017.

Who Lost Money, and Who Was Made Whole?

The gap between the banks' books and reality was measured almost immediately. An appraisal in November 2008 valued their assets at 40 percent of booked value at the time they failed, a difference the Commission puts at over ISK 7,000 billion. That is the number the dispute was really about, because every argument over who should be paid was an argument over how to divide a hole of that size.

The Emergency Act decided the answer in advance by reordering the queue. Deposit claims were satisfied in full because deposits had been given priority. General claims, by far the largest class, recovered about 29 percent on average, and holders of equity and subordinated debt recovered nothing at all. The aggregate recovery came to slightly over 58 percent, a figure that looks respectable only until it is decomposed into a fully paid priority class and a general class that lost roughly seven-tenths.

Domestic depositors were separately and completely protected. The Prime Minister announced a blanket Treasury guarantee for all deposits in domestic banks and local branches, regardless of size or holder, on 6 October 2008; it was removed in 2016. Depositors in the foreign branches were not covered, which is the distinction the Icesave litigation turned on. The British government paid its retail Icesave depositors in full and the Dutch government paid holders to EUR 100,000, and both then claimed against the Landsbanki estate, which paid its final priority instalment in January 2016.

Icelandic households lost in a way none of these categories captures. A large share of household and corporate borrowing was either indexed to inflation or linked to foreign currency, so the collapse of the króna and an inflation rate that reached 18.6 percent in January 2009 raised the balances people owed while their incomes fell. The Supreme Court ruled in 2010 that currency-linking of such loans breached the law on interest and indexation. Household and corporate debt, which reached 350 percent of gross domestic product in 2008, had more than halved by the end of 2017. People also left: Iceland recorded net immigration of 5,132 in 2007 and net emigration of 4,835 in 2009, from a population of roughly 319,000 at the peak.

Why Did Iceland Win the Icesave Case?

Before the case there were two referendums, and the official results are more striking than the summaries suggest. On the first bill, Statistics Iceland records 144,231 votes cast for a turnout of 62.7 percent, of which 2,599 were yes, 134,392 no and 7,240 blank or spoiled: 98.1 percent against on valid votes, with blank and void papers outnumbering supporters of the bill by nearly three to one. On the second bill turnout rose to 75.3 percent and the result narrowed to 59.8 percent against and 40.2 percent in favour. The second agreement was better and it still lost by twenty points.

The EFTA Surveillance Authority then took Iceland to the EFTA Court. The application was lodged on 15 December 2011 and judgment was delivered on 28 January 2013 in Case E-16/11, with the European Commission intervening in support of the Authority.

The Authority argued three things: that Directive 94/19/EC imposed an obligation of result, so Iceland itself had to ensure the minimum compensation was paid; that Iceland had breached the Directive read with the non-discrimination provision of the EEA Agreement; and that it had breached non-discrimination on its own. The Court rejected all three. It held that the Directive required states to adopt national rules providing a coverage level of at least EUR 20,000 and to establish a scheme accordingly, but did not require the state to step in and pay when the scheme itself could not, at least in a systemic crisis of this magnitude, and that the Authority had produced insufficient evidence that the fund was an emanation of the state. The discrimination plea failed partly on the Authority's own drafting: having declined to challenge the transfer of domestic deposits into the new bank, it was left arguing that Article 4 of the EEA Agreement independently generated the payment obligation the Directive did not contain. The Court held it did not, and ordered the Authority to pay Iceland's costs.

What Iceland won was narrower than the popular account. The Court did not hold that a country may walk away from foreign depositors, only that a directive obliging states to set up a funded scheme does not convert into a sovereign guarantee when the scheme proves too small.

Why Did Output Recover Faster Than Iceland's Reputation Did?

Because the currency was allowed to do the work. Iceland had no peg to defend and no monetary union to leave, so the exchange rate absorbed the shock in weeks rather than being defended for months and then abandoned. On Statistics Iceland's volume indices, exports rose about 11.7 percent between 2007 and 2009 while imports fell about 36.8 percent. An economy that had been running a current account deficit above 16 percent of output in 2006 rebalanced externally by making everything foreign expensive.

The bill was paid domestically. Gross fixed capital formation fell about 60.8 percent between 2007 and 2010, and private consumption about 18.3 percent from its 2007 peak to its 2010 trough. Inflation went from an annual 5.8 percent in January 2008 to a peak of 18.6 percent in January 2009 before falling to 2.5 percent by December 2010, and unemployment rose from 2.5 percent in 2007 to a peak of 8.3 percent in 2010.

The recovery clocks then ran at very different speeds, which is why any single sentence about how long Iceland took is wrong. Real gross domestic product fell about 10.4 percent from its 2008 peak to its 2010 trough and first exceeded the 2008 level in 2015. Unemployment did not return to a comparable figure until 2016 and private consumption not until 2017, while inflation normalised inside about two years. The capital controls outlasted all of them.

It is worth being precise about what Iceland avoided rather than what it achieved. It did not recover quickly because it let its banks fail. It recovered at the speed it did because a small open economy with its own floating currency and a large tradable sector can rebalance externally in a way a currency union member cannot, and because the losses were left with foreign creditors who had no vote in Reykjavik. Compare the Asian financial crisis, where devaluation bankrupted domestic borrowers whose debts were in dollars while their revenues were not.

What Happened to the People Who Ran the Banks?

More than in most countries, and less than the legend suggests. The regulator began investigating securities trading and lending practices immediately after October 2008, and an Office of the Special Prosecutor was created the same year. The last crisis cases closed in 2013, by which point the regulator had completed 205 cases and referred 103 to prosecutors. Charges followed in 23 cases, nineteen reached final judgment, and those produced fourteen convictions and five acquittals. More than forty people were convicted and twenty-seven received prison sentences, the longest of six years, the maximum statutory sentence for most economic crimes in Iceland.

The most common finding was market manipulation, and the scheme the courts described is the one the Commission had reconstructed from trading data: all three banks bought their own shares and resold them to selected clients financed at 100 percent with no additional collateral, a practice the rulings found common from at least early 2007 until the collapse. Senior managers and some owners were convicted of manipulation and breach of fiduciary duty.

The public sector was assessed separately, and not by a court. Acting under its statute, the Commission concluded that the Prime Minister Geir H. Haarde, the Finance Minister Árni M. Mathiesen and the Business Affairs Minister Björgvin G. Sigurðsson had shown negligence in failing to respond to the danger developing in the banks, and that the regulator's Director General Jónas Fr. Jónsson and the three Central Bank governors, Davíð Oddsson, Eiríkur Guðnason and Ingimundur Friðriksson, had shown negligence in supervising the rules on financial activity. Negligence in that statutory sense is a finding about the discharge of public duty, not a criminal verdict.

The rules changed afterwards in ways that map onto what went wrong. Loans secured by a bank's own shares are prohibited, connected lending is limited and monitored, and a credit register was introduced. Capital requirements were raised and liquidity rules now limit maturity and foreign currency mismatches. In January 2020 the regulator was merged into the Central Bank of Iceland, ending the split structure that had let each institution assume the other was watching.

Common Myths About the Iceland Banking Collapse

"Iceland let its banks fail while everyone else bailed theirs out." Iceland did not choose a philosophy, it ran out of options. A full rescue was arithmetically impossible against a banking system roughly ten times national output, and attempting it would have risked sovereign default. Nor did Iceland let everything fail: domestic deposits were guaranteed in full from 6 October 2008, and the domestic operations of all three banks were carved out and recapitalised with public funds. What it did not protect was foreign wholesale creditors, because it could not have.

"Iceland jailed its bankers." More than forty people were convicted and twenty-seven went to prison, which is genuinely unusual by international comparison. The longest sentence was six years, and that was the statutory ceiling for most economic crimes rather than a judicial statement about severity. Twenty-three sets of charges out of 205 regulatory cases is a reminder that most of what investigators found never reached court.

"Iceland refused to pay Icesave and got away with it." Two referendums rejected bills that would have put a state guarantee behind a negotiated settlement, and the EFTA Court later held that no such guarantee was owed. Neither fact meant the depositors went unpaid: British and Dutch depositors were compensated by their own governments, which then took priority claims against the Landsbanki estate under the very Emergency Act Iceland had passed. What Iceland refused was a sovereign guarantee on top of an estate claim, not the claim itself.

"The króna collapse was the disaster." The currency move was both the transmission mechanism and the repair mechanism, and separating the two is the analytical task. It devastated anyone with foreign-currency debt and domestic income, a great many Icelandic households among them, and simultaneously delivered the external adjustment that let output stabilise.

"Nobody could have seen it coming." Fitch moved to a negative outlook in February 2006 and Danske Bank published a paper the following month explicitly warning of a financial crisis. In April 2008 the Central Bank told ministers that GBP 193 million had left one bank's British deposit book over a single weekend and that the bank could survive six days of it. The risk was not hidden; the response was absent. The regulator published a stress test in August 2008, weeks before the collapse, finding the banks' solvency not at risk, which is a statement about the test rather than the banks.

What a Reader Can Actually Carry Forward

Iceland is easy to file away as a curiosity: a country of 315,000 people, three banks and a currency almost nobody holds. Filing it away discards the part most likely to matter to anyone who keeps money in a bank.

What generalises

• A deposit guarantee is a promise made by a balance sheet. Iceland's fund held ISK 13 billion of real assets against a guaranteed amount of ISK 722 billion, part of it in guarantees written by the banks it insured, and the scheme was compliant with European law throughout. Compliance and capacity are different properties, and only one pays out. See FDIC deposit insurance.
• Where a deposit sits legally decides who rescues it. Icesave money sat in a branch, so the obligation stayed in Reykjavik with a fund that could not meet it. In a subsidiary it would have been the host country's problem, and nothing on the statement told a saver which they held.
• The currency your liabilities are denominated in determines who can rescue you. A central bank is a lender of last resort only in its own money. Everything else is a reserves question, and reserves are finite in a way domestic currency is not. See international ETFs, currency risk and hedging.
• Reported capital can be financed by the thing it is meant to protect against. More than half the core equity of the Icelandic banking system was funded by that system itself, and every published ratio was still above the minimum. To know whether capital is real, ask what happens when the share price falls.
• Stress the funding, not only the assets. A test asking what happens if these loans lose 15 percent would have shown Icelandic banks surviving. One asking what happens if wholesale funding does not roll and the currency halves would have shown them failing. Which question gets asked is the whole game, and stress testing and scenario analysis is where that choice lives.

What does not generalise

• The arithmetic of a banking system ten times national output. That ratio removed every option a larger country would have had, and the external rebalancing depended on a floating currency and a large tradable sector, which a currency union member does not have.
• The willingness to leave general creditors with 29 percent. That required an emergency statute passed overnight, courts that upheld it and a creditor body that was almost entirely foreign.

The question worth asking now

Not whether a bank might fail, which almost nobody can answer usefully, but a narrower one. For any balance above your jurisdiction's insured limit, do you know which legal entity holds it, in which country that entity is supervised, and which scheme would pay if it stopped? For most people the balances sit below the line and none of it matters. For anyone holding a large operating balance or a foreign-branch deposit, the answer is specific, and finding it out is administrative rather than predictive.

Related Reading

• The 2008 financial crisis, the backdrop: Lehman Brothers filed on 15 September 2008 and Glitnir was asking the Central Bank for rescue money ten days later.
• Silicon Valley Bank and the 2023 regional banking stress, the sharpest contrast on policy: a government that could and did guarantee every uninsured deposit.
• The Asian financial crisis, for what happens when a currency mismatch meets a devaluation.
• Japan's asset price bubble, for a banking system that recognised its losses slowly rather than in a week.
• All Swoopr market history case studies.

References

Every figure on this page was verified against the following sources, each retrieved on 26 August 2026:

• Special Investigation Commission of the Althingi: Chapter 2, Summary of the Report's Main Conclusions: the nine times output formulation, lending growth and 2005 issuance, weak equity and cross-financing, own-share trading, both collateralised lending figures, the owner exposures, the swap-line correspondence, the October 2008 sequence, the November 2008 appraisal and the negligence findings.
• Special Investigation Commission of the Althingi: Chapter 17, The Depositors' and Investors' Guarantee Fund and Deposit Guarantees in General: deposit growth, the foreign-branch share and every guarantee fund figure.
• Special Investigation Commission of the Althingi: Chapter 21, Causes of the Collapse of the Icelandic Banks: the large-exposure correspondence with Landsbanki over Actavis, including the ISK 51.3 billion figure and the 49.7 percent of regulatory capital it represented.
• EFTA Court: Judgment of 28 January 2013 in Case E-16/11, EFTA Surveillance Authority v Iceland: the Icesave dates and coverage limits, the October 2008 legislative dates, the IMF terms, the capital controls and their Article 43 notification, the 93 and 99 percent figures, the EUR 20,000 minimum and the dismissal with costs.
• Bank for International Settlements Financial Stability Institute: The Banking Crisis in Iceland, FSI Crisis Management Series No 1: the ten times output and over 80 percent figures, the asset doubling of 2004 and 2005, the guarantee fund's cap, missing backstop and 0.41 percent coverage, all recovery rates, the domestic deposit guarantee and its removal, the IMF and bilateral terms, private debt of 350 percent of output, the stability contributions, the sentencing totals and the August 2008 stress test.
• UK Statutory Instruments: The Landsbanki Freezing Order 2008, SI 2008/2668: the date, the 10:10 commencement, the enabling powers, the Treasury's recited belief and the funds caught.
• UK Statutory Instruments: The Landsbanki Freezing (Revocation) Order 2009, SI 2009/1392: the date the freezing order was revoked and the date it ceased to have effect.
• Ministry of Finance and Economic Affairs of Iceland: Progress of the Plan for Removal of Capital Controls, 25 October 2017: Rules no. 200/2017 and the date the controls were lifted, and the offshore króna stock peaking at about 40 percent of output in 2009.
• European Central Bank: Icelandic Krona Euro Reference Exchange Rate, Series EXR.D.ISK.EUR.SP00.A: every exchange rate, the twenty identical publications, the multi-year gap after 3 December 2008 and the resumption of publication on 1 February 2018.
• Statistics Iceland: Indices of GDP and its Components 1945-2024, Table THJ01104: the output, consumption, investment and trade volume indices.
• Statistics Iceland: Activity Rate, Unemployment, Labour Force and Working Hours, Table VIN01002: every unemployment rate.
• Statistics Iceland: Consumer Price Index and Changes, Base 1988 Equals 100, Table VIS01000: every inflation rate.
• Statistics Iceland: Results for Referenda 1908-2011, Table KOS04101: every referendum figure. That blank and void ballots outnumbered yes votes in the first is arithmetic on that table.
• Statistics Iceland: Population Key Figures 1703-2026, Table MAN00000: population and net migration.

Rules that can change, and when this page was checked. Deposit-guarantee coverage levels move: the minimum during this episode was EUR 20,000, Directive 2009/14 raised it to EUR 50,000, and the European Union level when the Bank for International Settlements wrote its study was EUR 100,000. That study also recorded that as of March 2020 Iceland had not implemented the 2014 recast of the directive, a status that may since have changed. Iceland's blanket Treasury guarantee for domestic deposits was removed in 2016, and its capital account rules have been amended repeatedly since 2017. Last checked on 26 August 2026.

Figures deliberately not stated. No policy rate for the Central Bank of Iceland, no level or decline for any Icelandic equity index, no count of Icesave depositors, no króna total for the stability contributions and no gross domestic product figure, because no verified source supplied them in a form that could be quoted safely. The widely repeated 93 percent for the first Icesave referendum is not used either: it is the no vote as a share of all ballots cast, while the official table reports 98.1 percent on valid votes.

Method note: percentages described as computed were derived by Swoopr Investment from the named series. Exchange rate moves express the change in the value of the króna against the euro, which is why they differ from the change in the quoted ISK per EUR figure. Output and trade comparisons use annual volume indices, so they measure real quantities rather than values. The Commission reports the banking system at nine times output and the Bank for International Settlements at ten; both are stated with their source rather than reconciled.

This is educational content about a historical episode. It is not investment advice, it is not a forecast, and nothing here should be read as a claim about the condition of any bank, deposit scheme or currency today.

Frequently Asked Questions